1 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
5 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
6 and similarly for not instead of neg.
8 2017-11-08 Andi Kleen <ak@linux.intel.com>
10 * config/i386/i386.opt: Add -mforce-indirect-call.
11 * config/i386/predicates.md: Check for flag_force_indirect_call.
12 * doc/invoke.texi: Document -mforce-indirect-call
14 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
16 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
18 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
19 config/riscv/riscv.c (predict.h): New include.
20 (riscv_slow_unaligned_access_p): No longer static.
21 (riscv_block_move_straight): Add require.
22 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
24 2017-11-08 Jakub Jelinek <jakub@redhat.com>
27 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
28 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
29 nonimmediate_operand predicate for operand 1 instead of
32 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
36 * config/aarch64/constraints.md (Uml): New constraint.
37 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
40 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
42 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
43 of two vec_duplicates into a vec_concat.
45 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
47 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
48 Simplify vec_merge of vec_duplicate and vec_concat.
49 * config/aarch64/constraints.md (Utq): New constraint.
50 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
53 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
55 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
56 Simplify vec_merge of vec_duplicate and const_vector.
57 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
59 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
60 mode iterator. Update predicate on operand 1 to
61 handle non-const_vec constants. Delete constraints.
62 (*aarch64_combinez_be<mode>): Likewise for operand 2.
64 2017-11-08 Jakub Jelinek <jakub@redhat.com>
66 PR tree-optimization/78821
67 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
69 (store_operand_info::store_operand_info): Initialize it to false.
70 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
71 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
72 store in the group, and if chain_info is non-NULL, to ignore altogether
74 (compatible_load_p): Fail if bit_not_p does not match.
75 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
76 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
77 (pass_store_merging::process_store): Adjust
78 terminate_all_aliasing_chains calls to pass NULL in all current spots,
79 call terminate_all_aliasing_chains newly when adding a store into
80 a chain with non-NULL chain_info.
82 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
84 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
86 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
88 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
90 (aarch64_layout_frame): Initialise emit_frame_chain.
91 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
92 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
94 2017-11-08 Martin Liska <mliska@suse.cz>
96 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
97 of gimple_call_internal_p.
99 2017-11-07 Tom de Vries <tom@codesourcery.com>
101 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
103 2017-11-07 Tom de Vries <tom@codesourcery.com>
105 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
108 2017-11-08 Martin Liska <mliska@suse.cz>
111 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
113 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
115 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
117 2017-11-07 Jakub Jelinek <jakub@redhat.com>
120 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
121 (*cmp<mode>_ccz_1): New insn with $k alternative.
124 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
125 RTX_COMM_COMPARE as commutative as well.
126 (ix86_binary_operator_ok): Formatting fix.
127 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
128 *<code><mode>3<mask_name><round_saeonly_name>,
129 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
130 *<s>mul<mode>3_highpart<mask_name>,
131 *vec_widen_umult_even_v16si<mask_name>,
132 *vec_widen_umult_even_v8si<mask_name>,
133 *vec_widen_umult_even_v4si<mask_name>,
134 *vec_widen_smult_even_v16si<mask_name>,
135 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
136 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
137 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
138 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
139 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
140 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
141 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
142 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
143 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
144 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
145 ix86_binary_operator_ok. Formatting fixes.
146 (*<plusminus_insn><mode>3<mask_name><round_name>,
147 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
150 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
152 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
153 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
154 eq and ne if TARGET_ISEL.
155 (cmp): New code_iterator.
156 (UNS, UNSU_, UNSIK): New code_attrs.
157 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
158 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
160 ("eq<mode>3"): ... this.
161 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
163 ("ne<mode>3"): ... this.
165 2017-11-07 Julia Koval <julia.koval@intel.com>
168 * common/config/i386/i386-common.c
169 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
170 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
171 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
172 * config/i386/i386.opt: Ditto.
173 * config/i386/i386.c (ix86_target_string): Ditto.
174 (ix86_option_override_internal): Ditto.
175 (ix86_init_mpx_builtins): Move MPX to args2.
176 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
177 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
178 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
179 __builtin_ia32_vgf2p8affineinvqb_v32qi,
180 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
181 __builtin_ia32_vgf2p8affineinvqb_v16qi,
182 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
184 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
187 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
188 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
189 (zero-extendsidi peephole2): Remove peephole.
191 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
194 * stor-layout.c: Include attribs.h.
195 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
196 explicit lookup of "aligned" attribute.
198 2017-11-07 Andrew Waterman <andrew@sifive.com>
200 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
201 (riscv_expand_block_move): Likewise.
202 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
204 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
205 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
206 * config/riscv/riscv.c (riscv_block_move_straight): New function.
207 (riscv_adjust_block_mem): Likewise.
208 (riscv_block_move_loop): Likewise.
209 (riscv_expand_block_move): Likewise.
210 gcc/config/riscv/riscv.md (movmemsi): New pattern.
212 2017-11-07 Michael Clark <michaeljclark@mac.com>
214 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
215 (MUSL_DYNAMIC_LINKER): Likewise.
217 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
219 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
220 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
223 2017-11-07 Richard Biener <rguenther@suse.de>
225 * match.pd: Fix build.
227 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
228 Jackson Woodruff <jackson.woodruff@arm.com>
230 PR tree-optimization/71026
231 * match.pd: Canonicalize negate in division.
233 2017-11-07 Sudakshina Das <sudi.das@arm.com>
236 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
238 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
240 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
241 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
243 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
245 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
246 non-scalar integral types.
247 * match.pd (negate_expr_p): Handle MINUS_EXPR.
248 (-(A-B), -(~A)): New transformations.
250 2017-11-07 Tom de Vries <tom@codesourcery.com>
252 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
253 semicolon after "do {} while (0)".
254 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
255 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
256 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
257 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
258 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
260 2017-11-07 Tom de Vries <tom@codesourcery.com>
262 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
263 after "do {} while (0)".
264 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
265 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
266 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
267 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
268 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
270 2017-11-07 Tom de Vries <tom@codesourcery.com>
273 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
275 (arm_rtx_costs_internal): Add missing semicolon after
276 HANDLE_NARROW_SHIFT_ARITH call.
278 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
280 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
281 disable isel if it was not set explicitly.
283 2017-11-06 James Bowman <james.bowman@ftdichip.com>
285 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
287 (add_type_attribute) likewise.
289 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
291 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
292 of struct ix86_frame.
293 (ix86_initial_elimination_offset): Likewise.
294 (ix86_expand_split_stack_prologue): Likewise.
296 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
298 * tree-vrp.h (enum value_range_type): Update stale comment.
300 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
301 Alan Hayward <alan.hayward@arm.com>
302 David Sherwood <david.sherwood@arm.com>
304 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
305 (aarch64_expand_vec_perm_const): Take the number of units too.
306 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
307 (aarch64_expand_vec_perm_const): Likewise.
308 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
309 (vec_perm<mode>): Update accordingly.
311 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
312 Alan Hayward <alan.hayward@arm.com>
313 David Sherwood <david.sherwood@arm.com>
315 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
316 Take the number of units too.
317 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
318 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
319 but check for a vector mode before rather than after the call.
320 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
321 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
322 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
323 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
324 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
325 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
326 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
327 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
328 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
329 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
330 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
331 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
332 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
333 (aarch64_sqdmull2_n<mode>): Update accordingly.
335 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
336 Alan Hayward <alan.hayward@arm.com>
337 David Sherwood <david.sherwood@arm.com>
339 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
340 the number of units too.
341 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
342 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
343 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
344 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
345 (vec_store_lanesxi<mode>): Update accordingly.
347 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
348 Alan Hayward <alan.hayward@arm.com>
349 David Sherwood <david.sherwood@arm.com>
351 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
352 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
353 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
354 of units rather than the mode.
355 * config/aarch64/iterators.md (nunits): New mode attribute.
356 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
357 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
358 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
359 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
360 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
361 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
362 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
363 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
364 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
365 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
366 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
367 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
368 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
369 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
370 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
371 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
372 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
373 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
374 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
375 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
376 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
377 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
378 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
379 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
380 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
381 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
382 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
383 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
384 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
385 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
386 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
387 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
388 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
389 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
390 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
391 (aarch64_simd_vec_setv2di): Likewise.
393 2017-11-06 Carl Love <cel@us.ibm.com>
395 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
397 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
398 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
400 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
402 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
403 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
404 Add power 8 macro expansions.
405 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
406 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
407 power 8 instructions. (VSX_XXBR): Add iterator.
409 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
411 * config/arm/arm.md (predicable_short_it): Change default to "no",
412 improve documentation, remove uses that are identical to the default.
413 (enabled_for_depr_it): Rename to enabled_for_short_it.
414 * gcc/config/arm/arm-fixed.md (predicable_short_it): Remove default uses.
415 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
416 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
417 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
418 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
420 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
423 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
424 float128 helper macros, which are no longer used after deleting
425 the old 'q' built-in functions, and moving the round to odd
426 built-in functions to being special built-in functions.
427 (BU_FLOAT128_2): Likewise.
428 (BU_FLOAT128_1_HW): Likewise.
429 (BU_FLOAT128_2_HW): Likewise.
430 (BU_FLOAT128_3_HW): Likewise.
431 (FABSQ): Delete old 'q' built-in functions.
432 (COPYSIGNQ): Likewise.
433 (SQRTF128_ODD): Move round to odd built-in functions to be
434 special built-in functions, so that we can handle
435 -mabi=ieeelongdouble.
436 (TRUNCF128_ODD): Likewise.
437 (ADDF128_ODD): Likewise.
438 (SUBF128_ODD): Likewise.
439 (MULF128_ODD): Likewise.
440 (DIVF128_ODD): Likewise.
441 (FMAF128_ODD): Likewise.
442 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
443 built-in names to 'f128'.
444 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
445 old 'q' built-in functions, as the machine independent code for
446 'f128' built-in functions handles this.
447 (rs6000_expand_builtin): Add expansion for float128 round to odd
448 functions, keying off on -mabi=ieeelongdouble of whether to use
449 the KFmode or TFmode variant.
450 (rs6000_init_builtins): Initialize the _Float128 round to odd
452 * doc/extend.texi (PowerPC Built-in Functions): Document the old
453 _Float128 'q' built-in functions are now mapped into the new
454 'f128' built-in functions.
456 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
458 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
459 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
461 2017-11-06 David Malcolm <dmalcolm@redhat.com>
464 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
465 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
466 * toplev.c: Include "ipa-fnsummary.h".
467 (toplev::finalize): Call ipa_fnsummary_c_finalize.
469 2017-11-06 Jakub Jelinek <jakub@redhat.com>
471 PR tree-optimization/82838
472 * gimple-ssa-store-merging.c
473 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
474 on a separate gimple_seq which is then appended to seq.
476 2017-11-06 Jeff Law <law@redhat.com>
479 * config/i386/i386.c (PROBE_INTERVAL): Remove.
480 (get_probe_interval): New functions.
481 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
482 (ix86_adjust_stack_and_probe): Likewise.
483 (output_adjust_stack_and_probe): Likewise.
484 (ix86_emit_probe_stack_range): Likewise.
485 (ix86_expand_prologue): Likewise.
487 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
489 PR tree-optimization/82816
490 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
491 if the modes of the two types are the same.
492 (convert_plusminus_to_widen): Likewise.
494 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
496 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
498 (usadv16qi): New define_expand.
499 (usadv8hi): New define_expand.
501 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
504 * ipa-inline-transform.c (update_noncloned_frequencies): Always
506 (inline_transform): Likewise.
507 * predict.c (counts_to_freqs): Remove useless conditional.
508 * profile-count.h (profile_count::apply_scale): Move sanity check.
509 * tree-inline.c (copy_bb): Always scale.
510 (copy_cfg_body): Likewise.
512 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
515 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
517 (*cmp_ite0): Add enabled_for_depr_it attribute.
518 (*cmp_ite1): Likewise.
520 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
522 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
525 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
527 * tree-vrp.c (vrp_int_const_binop): Return true on success and
528 return the value by pointer.
529 (extract_range_from_multiplicative_op_1): Update accordingly.
530 Return as soon as an operation fails.
532 2017-11-05 Tom de Vries <tom@codesourcery.com>
535 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
536 (DEF_SANITIZER_BUILTIN): ... here.
537 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
538 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
540 2017-11-05 Tom de Vries <tom@codesourcery.com>
543 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
545 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
546 ASM_OUTPUT_BEFORE_CASE_LABEL call.
547 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
549 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
550 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
551 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
553 2017-11-05 Tom de Vries <tom@codesourcery.com>
556 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
559 2017-11-04 Michael Clark <michaeljclark@mac.com>
561 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
562 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
564 (*addsi3_extended): Likewise.
565 (*addsi3_extended2): Likewise.
566 (<optab>si3): Likewise.
567 (<optab>di3): Likewise.
568 (<optab><mode>3): Likewise.
569 (<*optabe>si3_internal): Likewise.
570 (zero_extendqi<SUPERQI:mode>2): Likewise.
571 (*add<mode>hi3): Likewise.
572 (*xor<mode>hi3): Likewise.
573 (<optab>di3): Likewise.
574 (*<optab>si3_extend): Likewise.
575 (*sge<u>_<X:mode><GPR:mode>): Likewise.
576 (*slt<u>_<X:mode><GPR:mode>): Likewise.
577 (*sle<u>_<X:mode><GPR:mode>): Likewise.
579 2017-11-04 Andrew Waterman <andrew@sifive.com>
581 * config/riscv/riscv.c (riscv_option_override): Conditionally set
582 TARGET_STRICT_ALIGN based upon -mtune argument.
584 2017-11-04 Andrew Waterman <andrew@sifive.com>
586 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
588 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
590 config/i386/i386.c (choose_basereg): Use optional scratch
591 register and add assertion.
592 (x86_emit_outlined_ms2sysv_save): Use scratch register when
593 needed, and don't allocate stack.
594 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
595 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
596 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
598 2017-11-03 Jeff Law <law@redhat.com>
600 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
601 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
602 to probe at the start of a noreturn function.
604 2017-11-03 Jakub Jelinek <jakub@redhat.com>
606 PR tree-optimization/78821
607 * gimple-ssa-store-merging.c: Update the file comment.
608 (MAX_STORE_ALIAS_CHECKS): Define.
609 (struct store_operand_info): New type.
610 (store_operand_info::store_operand_info): New constructor.
611 (struct store_immediate_info): Add rhs_code and ops data members.
612 (store_immediate_info::store_immediate_info): Add rhscode, op0r
613 and op1r arguments to the ctor, initialize corresponding data members.
614 (struct merged_store_group): Add load_align_base and load_align
616 (merged_store_group::merged_store_group): Initialize them.
617 (merged_store_group::do_merge): Update them.
618 (merged_store_group::apply_stores): Pick the constant for
619 encode_tree_to_bitpos from one of the two operands, or skip
620 encode_tree_to_bitpos if neither operand is a constant.
621 (class pass_store_merging): Add process_store method decl. Remove
622 bool argument from terminate_all_aliasing_chains method decl.
623 (pass_store_merging::terminate_all_aliasing_chains): Remove
624 var_offset_p argument and corresponding handling.
625 (stmts_may_clobber_ref_p): New function.
626 (compatible_load_p): New function.
627 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
628 if there is overlap and rhs_code is not INTEGER_CST. For
629 non-overlapping stores terminate group if rhs is not mergeable.
630 (get_alias_type_for_stmts): Change first argument from
631 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
632 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
633 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
635 (get_location_for_stmts): Change first argument from
636 auto_vec<gimple *> & to vec<gimple *> &.
637 (struct split_store): Remove orig_stmts data member, add orig_stores.
638 (split_store::split_store): Create orig_stores rather than orig_stmts.
639 (find_constituent_stmts): Renamed to ...
640 (find_constituent_stores): ... this. Change second argument from
641 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
642 to info structures rather than the statements.
643 (split_group): Rename ALLOW_UNALIGNED argument to
644 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
645 it. Adjust find_constituent_stores caller.
646 (imm_store_chain_info::output_merged_store): Handle rhs_code other
647 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
648 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
649 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
650 (mem_valid_for_store_merging): New function.
651 (handled_load): New function.
652 (pass_store_merging::process_store): New method.
653 (pass_store_merging::execute): Use process_store method. Adjust
654 terminate_all_aliasing_chains caller.
656 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
658 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
659 Return true for more constants, symbols and label references.
660 (aarch64_valid_floating_const): Remove unused function.
662 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
665 * config/aarch64/aarch64.c (aarch64_layout_frame):
666 Undo forcing of LR at bottom of frame.
668 2017-11-03 Jeff Law <law@redhat.com>
671 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
672 for int_registers_saved.
674 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
675 extracted from tree-ssa-dom.c.
676 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
677 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
678 (record_equivalences_from_incoming_edge): Add additional argument
679 to single_pred_edge_ignoring_loop_edges call.
680 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
681 (uncprop_dom_walker::before_dom_children): Add additional argument
682 to single_pred_edge_ignoring_loop_edges call.
683 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
684 single_pred_edge_ignoring_loop_edges rather than open coding.
685 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
687 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
689 * match.pd (-(-A)): Rewrite.
691 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
693 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
694 (rs6000_emit_int_cmove): New declaration.
695 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
696 (rs6000_emit_sISEL): Delete.
697 (rs6000_emit_int_cmove): Make non-static.
698 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
699 instead of rs6000_emit_sISEL.
701 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
703 * asan.c (create_cond_insert_point): Maintain profile.
704 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
706 * basic-block.h (struct basic_block_def): Remove frequency.
707 (EDGE_FREQUENCY): Use to_frequency
708 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
710 (find_traces): Update to use to_frequency.
711 (find_traces_1_round): Likewise; use only IPA counts.
712 (bb_to_key): Likewise.
713 (connect_traces): Use IPA counts only.
714 (copy_bb_p): Update to use to_frequency.
715 (fix_up_crossing_landing_pad): Likewise.
716 (sanitize_hot_paths): Likewise.
717 * bt-load.c (basic_block_freq): Likewise.
718 * cfg.c (init_flow): Set count_max to uninitialized.
719 (check_bb_profile): Remove frequencies; check counts.
720 (dump_bb_info): Do not dump frequencies.
721 (update_bb_profile_for_threading): Update counts only.
722 (scale_bbs_frequencies_int): Likewise.
723 (MAX_SAFE_MULTIPLIER): Remove.
724 (scale_bbs_frequencies_gcov_type): Update counts only.
725 (scale_bbs_frequencies_profile_count): Update counts only.
726 (scale_bbs_frequencies): Update counts only.
727 * cfg.h (struct control_flow_graph): Add count-max.
728 (update_bb_profile_for_threading): Update prototype.
729 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
730 (find_many_sub_basic_blocks): Likewise.
731 * cfgcleanup.c (try_forward_edges): Likewise.
732 (try_crossjump_to_edge): Likewise.
733 * cfgexpand.c (expand_gimple_cond): Likewise.
734 (expand_gimple_tailcall): Likewise.
735 (construct_init_block): Likewise.
736 (construct_exit_block): Likewise.
737 * cfghooks.c (verify_flow_info): Check consistency of counts.
738 (dump_bb_for_graph): Do not dump frequencies.
739 (split_block_1): Do not update frequencies.
740 (split_edge): Do not update frequencies.
741 (make_forwarder_block): Do not update frequencies.
742 (duplicate_block): Do not update frequencies.
743 (account_profile_record): Do not update frequencies.
744 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
745 for global heuristics.
746 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
747 (expected_loop_iterations_unbounded): Use counts only.
748 * cfgloopmanip.c (scale_loop_profile): Simplify.
749 (create_empty_loop_on_edge): Simplify
751 (duplicate_loop_to_header_edge): Simplify
752 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
753 (update_br_prob_note): Take care of removing note when profile
755 (relink_block_chain): Do not dump frequency.
756 (rtl_account_profile_record): Use to_frequency.
757 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
758 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
759 (cgraph_update_edges_for_call_stmt_node): Likewise.
760 (cgraph_edge::verify_count_and_frequency): Update.
761 (cgraph_node::verify_node): Temporarily disable frequency verification.
762 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
764 (cgraph_edge::rebuild_edges): Convert to ipa counts.
765 * cgraphunit.c (init_lowered_empty_function): Do not initialize
767 (cgraph_node::expand_thunk): Update profile.
768 * except.c (dw2_build_landing_pads): Do not update frequency.
769 * final.c (compute_alignments): Use to_frequency.
770 (dump_basic_block_info): Do not dump frequency.
771 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
772 (dump_gimple_bb_header): Do not dump frequency.
773 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
775 * gimple-streamer-in.c (input_bb): Do not stream frequency.
776 * gimple-streamer-out.c (output_bb): Do not stream frequency.
777 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
778 (init_before_recovery): Do not update frequency.
779 (sched_create_recovery_edges): Do not update frequency.
780 * hsa-gen.c (convert_switch_statements): Do not update frequency.
781 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
782 (ipa_cp_c_finalize): Set max_count to uninitialized.
783 * ipa-fnsummary.c (get_minimal_bb): Use counts.
784 (param_change_prob): Use counts.
785 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
787 * ipa-split.c (consider_split): Use to_frequency.
788 (split_function): Use to_frequency.
789 * ira-build.c (loop_compare_func): Likewise.
790 (mark_loops_for_removal): Likewise.
791 (mark_all_loops_for_removal): Likewise.
792 * loop-doloop.c (doloop_modify): Do not update frequency.
793 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
795 * lto-streamer-in.c (input_function): Update count_max.
796 * omp-expand.c (expand_omp_taskreg): Update count_max.
797 * omp-simd-clone.c (simd_clone_adjust): Update profile.
798 * predict.c (maybe_hot_frequency_p): Use to_frequency.
799 (maybe_hot_count_p): Use ipa counts only.
800 (maybe_hot_bb_p): Simplify.
801 (maybe_hot_edge_p): Simplify.
802 (probably_never_executed): Do not take frequency argument.
803 (probably_never_executed_bb_p): Do not pass frequency.
804 (probably_never_executed_edge_p): Likewise.
805 (combine_predictions_for_bb): Check that profile is nonzero.
806 (propagate_freq): Do not set frequency.
807 (drop_profile): Simplify.
808 (counts_to_freqs): Simplify.
809 (expensive_function_p): Use to_frequency.
810 (propagate_unlikely_bbs_forward): Simplify.
811 (determine_unlikely_bbs): Simplify.
812 (estimate_bb_frequencies): Add hack to silence graphite issues.
813 (compute_function_frequency): Use ipa counts.
814 (pass_profile::execute): Update.
815 (rebuild_frequencies): Use counts only.
816 (force_edge_cold): Use counts only.
817 * profile-count.c (profile_count::dump): Dump new count types.
818 (profile_count::differs_from_p): Check compatiblity.
819 (profile_count::to_frequency): New function.
820 (profile_count::to_cgraph_frequency): New function.
821 * profile-count.h (struct function): Declare.
822 (enum profile_quality): Add profile_guessed_local and
823 profile_guessed_global0.
824 (class profile_proability): Decrease number of bits to 29;
825 update from_reg_br_prob_note and to_reg_br_prob_note.
826 (class profile_count: Update comment; decrease number of bits
827 to 61. Check compatibility.
828 (profile_count::compatible_p): New private member function.
829 (profile_count::ipa_p): New member function.
830 (profile_count::operator<): Handle global zero correctly.
831 (profile_count::operator>): Handle global zero correctly.
832 (profile_count::operator<=): Handle global zero correctly.
833 (profile_count::operator>=): Handle global zero correctly.
834 (profile_count::nonzero_p): New member function.
835 (profile_count::force_nonzero): New member function.
836 (profile_count::max): New member function.
837 (profile_count::apply_scale): Handle IPA scalling.
838 (profile_count::guessed_local): New member function.
839 (profile_count::global0): New member function.
840 (profile_count::ipa): New member function.
841 (profile_count::to_frequency): Declare.
842 (profile_count::to_cgraph_frequency): Declare.
843 * profile.c (OVERLAP_BASE): Delete.
844 (compute_frequency_overlap): Delete.
845 (compute_branch_probabilities): Do not use compute_frequency_overlap.
846 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
847 * sched-ebb.c (rank): Use counts only.
848 * shrink-wrap.c (handle_simple_exit): Use counts only.
849 (try_shrink_wrapping): Use counts only.
850 (place_prologue_for_one_component): Use counts only.
851 * tracer.c (find_best_predecessor): Use to_frequency.
852 (find_trace): Use to_frequency.
853 (tail_duplicate): Use to_frequency.
854 * trans-mem.c (expand_transaction): Do not update frequency.
855 * tree-call-cdce.c: Do not update frequency.
856 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
857 (gimple_merge_blocks): Likewise.
858 (gimple_split_edge): Likewise.
859 (gimple_duplicate_sese_region): Likewise.
860 (gimple_duplicate_sese_tail): Likewise.
861 (move_sese_region_to_fn): Likewise.
862 (gimple_account_profile_record): Likewise.
863 (insert_cond_bb): Likewise.
864 * tree-complex.c (expand_complex_div_wide): Likewise.
865 * tree-eh.c (lower_resx): Update profile.
866 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
868 (initialize_cfun): Do not initialize frequencies
869 (freqs_to_counts): Delete.
870 (copy_cfg_body): Ignore count parameter.
872 (expand_call_inline): Update count_max.
873 (optimize_inline_calls): Update count_max.
874 (tree_function_versioning): Update count_max.
875 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
876 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
878 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
879 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
880 (try_peel_loop): Likewise.
881 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
883 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
884 (tree_transform_and_unroll_loop): Do not use frequencies
885 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
886 Use reliable prediction only.
887 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
888 * tree-ssa-sink.c (select_best_block): Use to_frequency.
889 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
891 * tree-ssa-threadupdate.c (create_block_for_threading): Do
893 (any_remaining_duplicated_blocks): Likewise.
894 (update_profile): Likewise.
895 (estimated_freqs_path): Delete.
896 (freqs_to_counts_path): Delete.
897 (clear_counts_path): Delete.
898 (ssa_fix_duplicate_block_edges): Likewise.
899 (duplicate_thread_path): Likewise.
900 * tree-switch-conversion.c (gen_inbound_check): Use counts.
901 * tree-tailcall.c (decrease_profile): Do not update frequency.
902 (eliminate_tail_call): Likewise.
903 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
904 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
905 (optimize_mask_stores): Likewise.
906 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
907 * ubsan.c (ubsan_expand_null_ifn): Update profile.
908 (ubsan_expand_ptr_ifn): Update profile.
909 * value-prof.c (gimple_ic): Simplify.
910 * value-prof.h (gimple_ic): Update prototype.
911 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
912 * ipa-inline.c (compute_uninlined_call_time): Be sure that
914 (want_inline_self_recursive_call_p): Likewise.
915 (resolve_noninline_speculation): Only cummulate defined counts.
916 (inline_small_functions): Use nonzero_p.
917 (ipa_inline): Do not access freed node.
919 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
921 * config/aarch64/aarch64.c (aarch64_override_options_internal):
922 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
924 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
926 * config/riscv/riscv.c (riscv_legitimize_move): Handle
927 non-legitimate address.
929 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
931 * config/rs6000/rs6000.md (*lt0_disi): Delete.
932 (*lt0_<mode>di, *lt0_<mode>si): New.
934 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
936 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
939 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
940 Jim Wilson <jim.wilson@linaro.org>
942 * config/aarch64/aarch64-cores.def (saphira): New CPU.
943 * config/aarch64/aarch64-tune.md: Regenerated.
944 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
945 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
947 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
949 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
950 (arc_expand_prologue): Restore blink for millicode.
951 * config/arc/linux.h (LINK_EH_SPEC): Defined.
953 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
956 * config/i386/i386.c (ix86_vector_duplicate_value): Use
957 gen_vec_duplicate after forcing the scalar into a register.
959 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
961 * combine (try_combine): Print the insns input to try_combine to the
964 2017-11-02 Steve Ellcey <sellcey@cavium.com>
967 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
968 Remove second argument from aarch64_process_target_attr call.
969 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
971 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
973 (aarch64_handle_attr_arch): Remove second argument.
974 (aarch64_handle_attr_cpu): Ditto.
975 (aarch64_handle_attr_tune): Ditto.
976 (aarch64_handle_attr_isa_flags): Ditto.
977 (aarch64_process_one_target_attr): Ditto.
978 (aarch64_process_target_attr): Ditto.
979 (aarch64_option_valid_attribute_p): Remove second argument.
980 on aarch64_process_target_attr call.
982 2017-11-02 David Malcolm <dmalcolm@redhat.com>
984 * diagnostic.c: Include "selftest-diagnostic.h".
985 (selftest::assert_location_text): New function.
986 (selftest::test_diagnostic_get_location_text): New function.
987 (selftest::diagnostic_c_tests): Call it.
989 2017-11-02 David Malcolm <dmalcolm@redhat.com>
991 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
992 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
993 (class selftest::test_diagnostic_context): Move to...
994 * selftest-diagnostic.c: New file.
995 * selftest-diagnostic.h: New file.
997 2017-11-02 James Bowman <james.bowman@ftdichip.com>
999 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
1000 offset range for FT32B.
1001 * config/ft32/ft32.h: option "mcompress" enables relaxation.
1002 * config/ft32/ft32.md: Add TARGET_NOPM.
1003 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
1004 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
1006 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
1008 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
1010 2017-11-02 Jeff Law <law@redhat.com>
1012 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
1013 virtual keyword on FINAL OVERRIDE members.
1015 * tree-ssa-propagate.h (ssa_propagation_engine): Group
1016 virtuals together. Add virtual destructor.
1017 (substitute_and_fold_engine): Similarly.
1019 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
1021 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
1023 2017-11-02 Richard Biener <rguenther@suse.de>
1025 PR tree-optimization/82795
1026 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
1028 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1030 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
1032 (gcc_GAS_CHECK_FEATURE): Remove.
1033 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
1034 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
1035 * configure: Regenerate.
1037 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
1039 * config/arc/arc.c (hwloop_optimize): Account for empty
1042 2017-11-02 Richard Biener <rguenther@suse.de>
1045 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
1046 Truncate ARRAY_REF index and element size.
1048 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1050 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
1052 2017-11-01 Jeff Law <law@redhat.com>
1054 * tree-ssa-ccp.c (ccp_folder): New class derived from
1055 substitute_and_fold_engine.
1056 (ccp_folder::get_value): New member function.
1057 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
1058 (ccp_fold_stmt): Remove prototype.
1059 (ccp_finalize): Call substitute_and_fold from the ccp_class.
1060 * tree-ssa-copy.c (copy_folder): New class derived from
1061 substitute_and_fold_engine.
1062 (copy_folder::get_value): Renamed from get_value.
1063 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
1064 * tree-vrp.c (vrp_folder): New class derived from
1065 substitute_and_fold_engine.
1066 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
1067 (vrp_folder::get_value): New member function.
1068 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
1069 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
1070 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
1071 provide a class interface to folder/substitute routines.
1072 (ssa_prop_fold_stmt_fn): Remove typedef.
1073 (ssa_prop_get_value_fn): Likewise.
1074 (subsitute_and_fold): Remove prototype.
1075 (replace_uses_in): Likewise.
1076 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
1077 Renamed from replace_uses_in. Call the virtual member function
1078 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
1079 (substitute_and_fold_dom_walker): Remove initialization of
1080 data member entries for calbacks. Add substitute_and_fold_engine
1081 member and initialize it.
1082 (substitute_and_fold_dom_walker::before_dom_children0: Use the
1083 member functions for get_value, replace_phi_args_in c
1084 replace_uses_in, and fold_stmt calls.
1085 (substitute_and_fold_engine::substitute_and_fold): Renamed from
1086 substitute_and_fold. Remove assert. Update ctor call.
1088 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
1089 (ssa_prop_visit_phi_fn): Likewise.
1090 (class ssa_propagation_engine): New class to provide an interface
1092 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
1094 (ssa_prop_visit_phi): Likewise.
1095 (ssa_propagation_engine::simulate_stmt): Moved into class.
1096 Call visit_phi/visit_stmt from the class rather than via
1097 file scoped static variables.
1098 (ssa_propagation_engine::simulate_block): Moved into class.
1099 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
1100 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
1101 set file scoped statics for the visit_stmt/visit_phi callbacks.
1102 * tree-complex.c (complex_propagate): New class derived from
1103 ssa_propagation_engine.
1104 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
1105 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
1106 (tree_lower_complex): Call ssa_propagate via the complex_propagate
1108 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
1109 ssa_propagation_engine.
1110 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
1111 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
1112 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
1113 * tree-ssa-copy.c (copy_prop): New class derived from
1114 ssa_propagation_engine.
1115 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
1116 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
1117 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
1118 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
1119 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
1120 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
1121 (execute_vrp): Call ssa_propagate from the vrp_prop class.
1123 2017-11-01 Jakub Jelinek <jakub@redhat.com>
1125 PR rtl-optimization/82778
1126 PR rtl-optimization/82597
1127 * compare-elim.c (struct comparison): Add in_a_setter field.
1128 (find_comparison_dom_walker::before_dom_children): Remove killed
1129 bitmap and df_simulate_find_defs call, instead walk the defs.
1130 Compute last_setter and initialize in_a_setter. Merge definitions
1131 with first initialization for a few variables.
1132 (try_validate_parallel): Use insn_invalid_p instead of
1133 recog_memoized. Return insn rather than just the pattern.
1134 (try_merge_compare): Fix up comment. Don't uselessly test if
1135 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
1137 (execute_compare_elim_after_reload): Remove df_chain_add_problem
1140 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1141 Alan Hayward <alan.hayward@arm.com>
1142 David Sherwood <david.sherwood@arm.com>
1144 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
1145 aarch64_hard_regno_nregs to get the number of registers
1148 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1149 Alan Hayward <alan.hayward@arm.com>
1150 David Sherwood <david.sherwood@arm.com>
1152 * config/aarch64/constraints.md (Upl): Rename to...
1154 * config/aarch64/aarch64.md
1155 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
1158 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1159 Alan Hayward <alan.hayward@arm.com>
1160 David Sherwood <david.sherwood@arm.com>
1162 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
1163 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
1166 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1167 Alan Hayward <alan.hayward@arm.com>
1168 David Sherwood <david.sherwood@arm.com>
1170 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
1171 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
1172 (aarch64_evpc_dup): Generate rtl direcly, rather than using
1174 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
1175 of a single element.
1176 * config/aarch64/iterators.md: Add a comment above the permute
1177 unspecs to say that they are generated directly by
1178 aarch64_expand_vec_perm_const.
1179 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
1181 2017-11-01 Nathan Sidwell <nathan@acm.org>
1183 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
1185 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1187 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
1188 and medany code models, and describe what they do.
1190 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1192 Revert accidental duplicate:
1194 * combine.c (can_change_dest_mode): Reject changes in
1195 REGMODE_NATURAL_SIZE.
1197 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
1199 PR rtl-optimization/64682
1200 PR rtl-optimization/69567
1201 PR rtl-optimization/69737
1202 PR rtl-optimization/82683
1203 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
1204 register mentioned in the note, drop the note, unless it came from I3,
1205 in which case it should go to I3 again.
1207 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1209 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
1210 and return false if not.
1211 (clear_bytes_written_by, live_bytes_read): Update accordingly.
1213 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1215 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
1216 range is known to be empty.
1218 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1219 Alan Hayward <alan.hayward@arm.com>
1220 David Sherwood <david.sherwood@arm.com>
1222 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
1223 and CONST_VECTOR_NUNITS instead of computing the number of units from
1224 the byte sizes of the vector and element.
1225 (simplify_binary_operation_1): Likewise.
1226 (simplify_const_binary_operation): Likewise.
1227 (simplify_ternary_operation): Likewise.
1229 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1230 Alan Hayward <alan.hayward@arm.com>
1231 David Sherwood <david.sherwood@arm.com>
1233 * var-tracking.c (INT_MEM_OFFSET): Replace with...
1234 (int_mem_offset): ...this new function.
1235 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
1236 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
1237 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
1240 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1241 Alan Hayward <alan.hayward@arm.com>
1242 David Sherwood <david.sherwood@arm.com>
1244 * lower-subreg.c (interesting_mode_p): New function.
1245 (compute_costs, find_decomposable_subregs, decompose_register)
1246 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
1247 (resolve_clobber, dump_choices): Use it.
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 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
1255 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1256 Alan Hayward <alan.hayward@arm.com>
1257 David Sherwood <david.sherwood@arm.com>
1259 * alias.c (find_base_value, find_base_term): Only process integer
1260 truncations. Check the precision rather than the size.
1262 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1263 Alan Hayward <alan.hayward@arm.com>
1264 David Sherwood <david.sherwood@arm.com>
1266 * machmode.h (is_narrower_int_mode): New function
1267 * optabs.c (expand_float, expand_fix): Use it.
1268 * dwarf2out.c (rotate_loc_descriptor): Likewise.
1270 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1271 Alan Hayward <alan.hayward@arm.com>
1272 David Sherwood <david.sherwood@arm.com>
1274 * rtl.h (narrower_subreg_mode): New function.
1275 * ira-color.c (update_costs_from_allocno): Use it.
1277 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1278 Alan Hayward <alan.hayward@arm.com>
1279 David Sherwood <david.sherwood@arm.com>
1281 * optabs-query.h (convert_optab_p): New function, split out from...
1282 (convert_optab_handler): ...here.
1283 (widening_optab_handler): Delete.
1284 (find_widening_optab_handler): Remove permit_non_widening parameter.
1285 (find_widening_optab_handler_and_mode): Likewise. Provide an
1286 override that operates on mode class wrappers.
1287 * optabs-query.c (widening_optab_handler): Delete.
1288 (find_widening_optab_handler_and_mode): Remove permit_non_widening
1289 parameter. Assert that the two modes are the same class and that
1290 the "from" mode is narrower than the "to" mode. Use
1291 convert_optab_handler instead of widening_optab_handler.
1292 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
1293 instead of widening_optab_handler.
1294 * expr.c (expand_expr_real_2): Update calls to
1295 find_widening_optab_handler.
1296 * optabs.c (expand_widen_pattern_expr): Likewise.
1297 (expand_binop_directly): Take the insn_code as a parameter.
1298 (expand_binop): Only call find_widening_optab_handler for
1299 conversion optabs; use optab_handler otherwise. Update calls
1300 to find_widening_optab_handler and expand_binop_directly.
1301 Use convert_optab_handler instead of widening_optab_handler.
1302 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
1303 find_widening_optab_handler and use scalar_mode rather than
1305 (convert_plusminus_to_widen): Likewise.
1307 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1308 Alan Hayward <alan.hayward@arm.com>
1309 David Sherwood <david.sherwood@arm.com>
1311 * machmode.h (fixed_size_mode): New class.
1312 * rtl.h (get_pool_mode): Return fixed_size_mode.
1313 * gengtype.c (main): Add fixed_size_mode.
1314 * target.def (get_raw_result_mode): Return a fixed_size_mode.
1315 (get_raw_arg_mode): Likewise.
1316 * doc/tm.texi: Regenerate.
1317 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
1318 * targhooks.c (default_get_reg_raw_mode): Likewise.
1319 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
1320 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
1321 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
1322 (msp430_get_raw_result_mode): Likewise.
1323 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
1324 * dbxout.c (dbxout_parms): Require fixed-size modes.
1325 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
1326 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
1327 * omp-low.c (lower_oacc_reductions): Likewise.
1328 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
1329 (simplify_subreg): Update accordingly.
1330 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
1331 (force_const_mem): Update accordingly. Return NULL_RTX for modes
1332 that aren't fixed-size.
1333 (get_pool_mode): Return a fixed_size_mode.
1334 (output_constant_pool_2): Take a fixed_size_mode.
1336 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1337 Alan Hayward <alan.hayward@arm.com>
1338 David Sherwood <david.sherwood@arm.com>
1340 * doc/rtl.texi (vec_series): Document.
1341 (const): Say that the operand can be a vec_series.
1342 * rtl.def (VEC_SERIES): New rtx code.
1343 * rtl.h (const_vec_series_p_1): Declare.
1344 (const_vec_series_p): New function.
1345 * emit-rtl.h (gen_const_vec_series): Declare.
1346 (gen_vec_series): Likewise.
1347 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
1348 (gen_vec_series): Likewise.
1349 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
1350 * simplify-rtx.c (simplify_unary_operation): Handle negations
1352 (simplify_binary_operation_series): New function.
1353 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
1354 (test_vector_ops_series): New function.
1355 (test_vector_ops): Call it.
1356 * config/powerpcspe/altivec.md (altivec_lvsl): Use
1357 gen_const_vec_series.
1358 (altivec_lvsr): Likewise.
1359 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
1361 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1362 Alan Hayward <alan.hayward@arm.com>
1363 David Sherwood <david.sherwood@arm.com>
1365 * doc/rtl.texi (const): Update description of address constants.
1366 Say that vector constants are allowed too.
1367 * common.md (E, F): Use CONSTANT_P instead of checking for
1369 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
1370 checking for CONST_VECTOR.
1371 * expmed.c (make_tree): Use build_vector_from_val for a CONST
1373 * expr.c (expand_expr_real_2): Check for vector modes instead
1374 of checking for CONST_VECTOR.
1375 * rtl.h (const_vec_p): New function.
1376 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
1377 (unwrap_const_vec_duplicate): Handle them here too.
1379 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1380 David Malcolm <dmalcolm@redhat.com>
1381 Alan Hayward <alan.hayward@arm.com>
1382 David Sherwood <david.sherwood@arm.com>
1384 * rtl.h (vec_duplicate_p): New function.
1385 * selftest-rtl.c (assert_rtx_eq_at): New function.
1386 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
1387 (assert_rtx_eq_at): Declare.
1388 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
1389 * selftest-run-tests.c (selftest::run_tests): Call it.
1390 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
1391 (simplify_unary_operation_1): Recursively handle vector duplicates.
1392 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
1394 (simplify_subreg): Handle subregs of vector duplicates.
1395 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
1396 (selftest::simplify_rtx_c_tests): New functions.
1398 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1399 Alan Hayward <alan.hayward@arm.com>
1400 David Sherwood <david.sherwood@arm.com>
1402 * emit-rtl.h (gen_const_vec_duplicate): Declare.
1403 (gen_vec_duplicate): Likewise.
1404 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
1406 (gen_const_vector): ...here.
1407 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
1408 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
1409 whose elements are all equal.
1410 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
1411 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
1412 (simplify_relational_operation): Likewise.
1413 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
1415 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
1416 (aarch64_expand_vector_init): Likewise.
1417 * config/arm/arm.c (neon_vdup_constant): Likewise.
1418 (neon_expand_vector_init): Likewise.
1419 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
1420 (arm_block_set_unaligned_vect): Likewise.
1421 (arm_block_set_aligned_vect): Likewise.
1422 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
1423 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
1424 (expand_vec_perm_even_odd_pack): Likewise.
1425 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
1426 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
1427 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
1428 gen_const_vec_duplicate.
1429 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
1430 * config/mips/mips.c (mips_gen_const_int_vector): Use
1431 gen_const_vec_duplicate.
1432 (mips_expand_vector_init): Use CONST0_RTX.
1433 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
1434 (define_split): Use gen_const_vec_duplicate.
1435 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
1436 (define_split): Use gen_const_vec_duplicate.
1437 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
1438 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
1439 * config/spu/spu.c (spu_const): Likewise.
1441 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1442 Alan Hayward <alan.hayward@arm.com>
1443 David Sherwood <david.sherwood@arm.com>
1445 * combine.c (can_change_dest_mode): Reject changes in
1446 REGMODE_NATURAL_SIZE.
1448 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
1450 * configure.ac (--enable-libssp): New.
1451 (gcc_cv_libc_provides_ssp): Check for explicit setting before
1452 trying to determine target-specific default. Adjust indentation.
1453 * configure: Regenerated.
1454 * doc/install.texi (Configuration): Expand --disable-libssp
1457 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
1459 config/i386/i386.c (ix86_expand_epilogue): Correct stack
1462 2017-10-31 Martin Jambor <mjambor@suse.cz>
1465 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
1467 2017-10-31 David Malcolm <dmalcolm@redhat.com>
1469 * auto-profile.c (autofdo_source_profile::read): Use
1470 UNKNOWN_LOCATION rather than 0.
1471 * diagnostic-core.h (warning_at_rich_loc): Rename to...
1472 (warning_at): ...this overload.
1473 (warning_at_rich_loc_n): Rename to...
1474 (warning_n): ...this overload.
1475 (error_at_rich_loc): Rename to...
1476 (error_at): ...this overload.
1477 (pedwarn_at_rich_loc): Rename to...
1478 (pedwarn): ...this overload.
1479 (permerror_at_rich_loc): Rename to...
1480 (permerror): ...this overload.
1481 (inform_at_rich_loc): Rename to...
1482 (inform): ...this overload.
1483 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
1484 (diagnostic_n_impl_richloc): Rename to...
1485 (diagnostic_n_impl): ...this rich_location *-based decl.
1486 (inform_at_rich_loc): Rename to...
1487 (inform): ...this, and add an assertion.
1488 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
1489 (warning_at_rich_loc): Rename to...
1490 (warning_at): ...this, and add an assertion.
1491 (warning_at_rich_loc_n): Rename to...
1492 (warning_n): ...this, and add an assertion.
1493 (warning_n): Update location_t-based implementation for removal of
1494 location_t-based diagnostic_n_impl.
1495 (pedwarn_at_rich_loc): Rename to...
1496 (pedwarn): ...this, and add an assertion.
1497 (permerror_at_rich_loc): Rename to...
1498 (permerror): ...this, and add an assertion.
1499 (error_n): Update for removal of location_t-based diagnostic_n_impl.
1500 (error_at_rich_loc): Rename to...
1501 (error_at): ...this, and add an assertion.
1502 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
1503 (driver::do_spec_on_infiles): Likewise.
1504 * substring-locations.c (format_warning_va): Update for renaming
1505 of inform_at_rich_loc.
1507 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
1509 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
1510 _Float<N>X built-in functions so that the variant without the
1511 "__builtin_" prefix is only enabled for the GNU C and Objective C
1512 languages when they are in non-strict ANSI/ISO mode.
1513 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
1514 * target.def (floatn_builtin_p): Add a target hook to control
1515 whether _Float<N> and _Float<N>X built-in functions without the
1516 "__builtin_" prefix are enabled, and return true for C and
1517 Objective C in the default hook. Include langhooks.h in
1519 * targhooks.h (default_floatn_builtin_p): Likewise.
1520 * targhooks.c (default_floatn_builtin_p): Likewise.
1521 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
1522 floatn_builtin_p target hook.
1523 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
1525 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
1526 Eric Botcazou <ebotcazou@adacore.com>
1528 PR rtl-optimization/81803
1529 * lra-constraints.c (curr_insn_transform): Also reload the whole
1530 register for a strict subreg no wider than a word if this is for
1531 a WORD_REGISTER_OPERATIONS target.
1533 2017-10-31 Jason Merrill <jason@redhat.com>
1535 * gdbinit.in: Skip over inlines from timevar.h.
1537 2017-10-31 Martin Liska <mliska@suse.cz>
1539 * doc/gcov.texi: Document new option.
1540 * gcov.c (print_usage): Likewise print it.
1541 (process_args): Support the argument.
1542 (format_count): New function.
1543 (format_gcov): Use the function.
1545 2017-10-31 Martin Liska <mliska@suse.cz>
1547 * gcov.c (struct name_map): do not use typedef.
1548 Define operator== and operator<.
1549 (name_search): Remove.
1550 (name_sort): Remove.
1551 (main): Do not allocate names.
1552 (process_file): Add vertical space.
1553 (generate_results): Use std::find.
1554 (release_structures): Do not release memory.
1555 (find_source): Use std::find.
1557 2017-10-31 Martin Liska <mliska@suse.cz>
1559 * gcov.c (struct line_info): Remove it's typedef.
1560 (line_info::line_info): Add proper ctor.
1561 (line_info::has_block): Do not use a typedef.
1562 (struct source_info): Do not use typedef.
1563 (circuit): Likewise.
1564 (get_cycles_count): Likewise.
1565 (output_intermediate_file): Iterate via vector iterator.
1566 (add_line_counts): Use std::vector methods.
1567 (accumulate_line_counts): Likewise.
1568 (output_lines): Likewise.
1570 2017-10-31 Martin Liska <mliska@suse.cz>
1572 * gcov.c (struct source_info): Remove typedef.
1573 (source_info::source_info): Add proper ctor.
1574 (accumulate_line_counts): Use struct, not it's typedef.
1575 (output_gcov_file): Likewise.
1576 (output_lines): Likewise.
1577 (main): Do not allocate an array.
1578 (output_intermediate_file): Use size of vector container.
1579 (process_file): Resize the vector.
1580 (generate_results): Do not preallocate, use newly added vector
1582 (release_structures): Do not release sources.
1583 (find_source): Use vector methods.
1584 (add_line_counts): Do not use typedef.
1586 2017-10-31 Martin Liska <mliska@suse.cz>
1588 * doc/gcov.texi: Document that.
1589 * gcov.c (add_line_counts): Mark lines with a non-executed
1591 (output_line_beginning): Handle such lines.
1592 (output_lines): Pass new argument.
1593 (output_intermediate_file): Print it in intermediate format.
1595 2017-10-31 Martin Liska <mliska@suse.cz>
1597 * color-macros.h: New file.
1598 * diagnostic-color.c: Factor out color related to macros to
1600 * doc/gcov.texi: Document -k option.
1601 * gcov.c (INCLUDE_STRING): Include string.h.
1602 (print_usage): Add -k option.
1603 (process_args): Parse it.
1604 (pad_count_string): New function.
1605 (output_line_beginning): Likewise.
1606 (DEFAULT_LINE_START): New macro.
1607 (output_lines): Support color output.
1609 2017-10-31 Martin Liska <mliska@suse.cz>
1611 PR gcov-profile/82633
1612 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
1613 their interaction with GCOV infrastructure.
1614 * configure.ac: Add -fkeep-{inline,static}-functions to
1616 * configure: Regenerate.
1618 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
1621 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
1623 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
1626 * config/rs6000/rs6000.md (allocate_stack): Force update interval
1627 into a register if it does not fit into an immediate offset field.
1629 2017-10-31 Olivier Hainque <hainque@adacore.com>
1631 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
1633 2017-10-31 Julia Koval <julia.koval@intel.com>
1635 * config.gcc: Add gfniintrin.h.
1636 * config/i386/gfniintrin.h: New.
1637 * config/i386/i386-builtin-types.def
1638 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1639 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1640 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1641 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1642 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1643 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
1644 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
1645 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
1646 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
1647 V64QI_FTYPE_V64QI_V64QI_INT): New types.
1648 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1649 * config/i386/immintrin.h: Include gfniintrin.h.
1650 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
1652 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
1654 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
1656 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1658 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
1659 (arm_ashldi3_1bit): Remove pattern.
1660 (ashrdi3): Remove shift by 1 expansion.
1661 (arm_ashrdi3_1bit): Remove pattern.
1662 (lshrdi3): Remove shift by 1 expansion.
1663 (arm_lshrdi3_1bit): Remove pattern.
1664 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
1665 cost of ashldi3 by 1.
1666 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
1667 (<shift>di3_neon): Likewise.
1669 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
1671 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
1672 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
1673 and (*aarch64_simd_mov<VQ:mode>).
1674 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
1675 pattern alternative.
1676 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
1677 attributes to match pattern alternative.
1679 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
1681 * config.gcc (powerpc*-*-*): Add emmintrin.h.
1682 * config/rs6000/emmintrin.h: New file.
1683 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
1685 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1687 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
1688 * (movdi_vfp_cortexa8): Remove pattern.
1690 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1692 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
1694 (Specific, alpha*-dec-osf5.1): Remove.
1695 (Specific, mips-sgi-irix5): Remove.
1696 (Specific, mips-sgi-irix6): Remove.
1698 2017-10-30 Jakub Jelinek <jakub@redhat.com>
1701 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
1702 arguments to clear_bit_region_be.
1704 2017-10-30 Jim Wilson <wilson@tuliptree.org>
1706 * gimplify.c: Include memmodel.h.
1708 2017-10-30 Martin Jambor <mjambor@suse.cz>
1710 * omp-grid.c (grid_attempt_target_gridification): Also insert a
1711 condition whether loop should be executed at all.
1713 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
1715 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1716 gimple folding of vec_madd() intrinsics.
1717 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
1718 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
1719 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
1721 2017-10-30 Richard Biener <rguenther@suse.de>
1723 PR tree-optimization/82762
1725 2017-10-23 Richard Biener <rguenther@suse.de>
1727 PR tree-optimization/82129
1729 2017-08-01 Richard Biener <rguenther@suse.de>
1731 PR tree-optimization/81181
1732 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
1733 (compute_antic): ... end of iteration here.
1735 2017-10-30 Joseph Myers <joseph@codesourcery.com>
1737 * doc/invoke.texi (C Dialect Options): Document -std=c17,
1738 -std=iso9899:2017 and -std=gnu17.
1739 * doc/standards.texi (C Language): Document C17 support.
1740 * doc/cpp.texi (Overview): Mention -std=c17.
1741 (Standard Predefined Macros): Document C11 and C17 values of
1742 __STDC_VERSION__. Do not refer to C99 support as incomplete.
1743 * doc/extend.texi (Inline): Do not list individual options for
1744 standards newer than C99.
1745 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
1747 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
1750 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
1752 * asan.c (asan_finish_file): Align asan globals array by shadow
1755 2017-10-30 Jakub Jelinek <jakub@redhat.com>
1758 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
1759 (struct store_immediate_info): Add bitregion_start and bitregion_end
1761 (store_immediate_info::store_immediate_info): Add brs and bre
1762 arguments and initialize bitregion_{start,end} from those.
1763 (struct merged_store_group): Add bitregion_start, bitregion_end,
1764 align_base and mask fields. Drop unnecessary struct keyword from
1765 struct store_immediate_info. Add do_merge method.
1766 (clear_bit_region_be): Use memset instead of loop storing zeros.
1767 (merged_store_group::do_merge): New method.
1768 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
1769 stores as long as the surrounding bitregions have no gaps.
1770 (merged_store_group::merge_overlapping): Use do_merge.
1771 (merged_store_group::apply_stores): Test that bitregion_{start,end}
1772 is byte aligned, rather than requiring that start and width are
1773 byte aligned. Drop unnecessary struct keyword from
1774 struct store_immediate_info. Allocate and populate also mask array.
1775 Make start of the arrays relative to bitregion_start rather than
1776 start and size them according to bitregion_{end,start} difference.
1777 (struct imm_store_chain_info): Drop unnecessary struct keyword from
1778 struct store_immediate_info.
1779 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
1780 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
1781 struct keyword from struct store_immediate_info.
1782 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
1783 between stores as long as the surrounding bitregions have no gaps.
1785 (struct split_store): Add orig non-static data member.
1786 (split_store::split_store): Initialize orig to false.
1787 (find_constituent_stmts): Return store_immediate_info *, non-NULL
1788 if there is exactly a single original stmt. Change stmts argument
1789 to pointer from reference, if NULL, don't push anything to it. Add
1790 first argument, use it to optimize skipping over orig stmts that
1791 are known to be before bitpos already. Simplify.
1792 (split_group): Return unsigned int count how many stores are or
1793 would be needed rather than a bool. Add allow_unaligned argument.
1794 Change split_stores argument from reference to pointer, if NULL,
1795 only do a dry run computing how many stores would be produced.
1796 Rewritten algorithm to use both alignment and misalign if
1797 !allow_unaligned and handle bitfield stores with gaps.
1798 (imm_store_chain_info::output_merged_store): Set start_byte_pos
1799 from bitregion_start instead of start. Compute allow_unaligned
1800 here, if true, do 2 split_group dry runs to compute which one
1801 produces fewer stores and prefer aligned if equal. Punt if
1802 new count is bigger or equal than original before emitting any
1803 statements, rather than during that. Remove no longer needed
1804 new_ssa_names tracking. Replace num_stmts with
1805 split_stores.length (). Use 32-bit stack allocated entries
1806 in split_stores auto_vec. Try to reuse original store lhs/rhs1
1807 if possible. Handle bitfields with gaps.
1808 (pass_store_merging::execute): Ignore bitsize == 0 stores.
1809 Compute bitregion_{start,end} for the stores and construct
1810 store_immediate_info with that. Formatting fixes.
1812 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
1815 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
1816 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
1818 2017-10-29 Jim Wilson <wilson@tuliptree.org>
1820 * gimplify.c: Include tm_p.h.
1822 * common.opt (gcoff): Re-add as ignored option.
1823 (gcoff1, gcoff2, gcoff3): Likewise.
1825 * Makefile.in (OBJS): Delete sdbout.o.
1826 (GTFILES): Delete $(srcdir)/sdbout.c.
1827 * debug.h: Delete sdb_debug_hooks.
1828 * final.c: Delete sdbout.h include.
1829 (final_scan_insn): Delete SDB_DEBUG check.
1830 (rest_of_clean_state): Likewise.
1831 * output.h: Delete sdb_begin_function_line.
1834 * toplev.c: Delete sdbout.h include.
1835 (process_options): Delete SDB_DEBUG check.
1836 * tree-core.h (tree_type_common): Delete pointer field of
1838 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
1839 TYPE_SYMTAB_POINTER.
1840 * tree.h (TYPE_SYMTAB_POINTER): Delete.
1841 (TYPE_SYMTAB_IS_POINTER): Delete.
1842 (TYPE_SYMTAB_IS_DIE): Renumber.
1843 * xcoffout.c: Refer to former sdbout.c file.
1844 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
1846 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
1847 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
1848 (Debugging Options): Delete -gcoff.
1849 (-gstabs): Delete SDB reference.
1851 (-gcoff@var{level}): Delete.
1852 * doc/passes.texi (Debugging information output): Delete SDB and
1853 sdbout.c references.
1854 * doc/tm.texi: Regenerate.
1855 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
1856 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
1858 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
1859 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
1860 (SDB_DEBUGGING_INFO): Delete.
1861 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
1862 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
1863 * target.def (output_source_filename): Delete COFF reference.
1865 * common.opt (gcoff): Delete.
1866 (gxcoff+): Update Negative chain.
1867 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
1869 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
1870 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
1871 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
1873 (expand_function_start): Change sdb reference to past tense.
1874 (expand_function_end): Change sdb reference to past tense.
1875 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
1876 * opts.c (debug_type_names): Delete coff entry.
1877 (common_handle_option): Delete OPT_gcoff case.
1878 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
1880 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
1881 * config/cris/cris.h: Delete SDB reference in comment.
1882 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
1883 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
1884 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
1885 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
1887 (ix86_expand_prologue): Likewise.
1888 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
1889 * config/ia64/ia64.h: Likewise.
1890 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
1891 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
1893 * config/mmix/mmix.h: Likewise.
1894 * config/nds32/nds32.c: Likewise.
1895 * config/stormy/storym16.h: Likewise.
1896 * config/visium/visium.h: Likewise.
1897 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
1899 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
1901 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
1902 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
1903 FRAME_POINTER_REGNUM point at high end of local var area.
1905 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
1907 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
1908 Move comment around. Do not reset best_edge for a copiable
1909 destination if the copy would cause a partition change.
1910 (better_edge_p): Remove redundant check.
1912 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
1914 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
1916 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
1918 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
1919 for math functions that have _Float<N> and _Float<N>X variants.
1920 (mathfn_built_in_2): Add support for math functions that have
1921 _Float<N> and _Float<N>X variants.
1922 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
1923 (expand_builtin_mathfn_ternary): Add support for fma with
1924 _Float<N> and _Float<N>X variants.
1925 (expand_builtin): Likewise.
1926 (fold_builtin_3): Likewise.
1927 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
1928 create math function _Float<N> and _Float<N>X variants as external
1930 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
1931 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
1932 the __builtin_ prefix and if not strict ansi, without the prefix.
1933 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
1934 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
1935 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
1936 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
1937 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
1938 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
1939 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
1940 function signatures for fma _Float<N> and _Float<N>X variants.
1941 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
1942 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
1943 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
1944 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
1945 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
1946 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
1947 * gencfn-macros.c (print_case_cfn): Add support for math functions
1948 that have _Float<N> and _Float<N>X variants.
1949 (print_define_operator_list): Likewise.
1950 (fltfn_suffixes): Likewise.
1952 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
1953 for math functions that have _Float<N> and _Float<N>X variants.
1954 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
1955 and _Float<N>X variants.
1956 (COPYSIGN): Likewise.
1959 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
1960 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
1962 (integer_valued_read_call_p): Likewise.
1963 * fold-const-call.c (fold_const_call_ss): Likewise.
1964 (fold_const_call_sss): Add support for copysign, fmin, and fmax
1965 _Float<N> and _Float<N>X variants.
1966 (fold_const_call_ssss): Add support for fma _Float<N> and
1967 _Float<N>X variants.
1968 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
1969 support for copysign and fma _Float<N> and _Float<N>X variants.
1970 (backprop::process_builtin_call_use): Likewise.
1971 * tree-call-cdce.c (can_test_argument_range); Add support for
1972 sqrt _Float<N> and _Float<N>X variants.
1973 (edom_only_function): Likewise.
1974 (get_no_error_domain): Likewise.
1975 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
1976 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
1977 copysign _Float<N> and _Float<N>X variants.
1978 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
1979 handled by machine independent code.
1980 (FMAF128): Likewise.
1981 * doc/cpp.texi (Common Predefined Macros): Document defining
1982 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
1983 fma _Float<N> and _Float<N>X variants.
1985 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
1988 * config/i386/i386-modes.def (CCFPU): Remove definition.
1989 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
1990 (ix86_cc_modes_compatible): Ditto.
1991 (ix86_expand_carry_flag_compare): Ditto.
1992 (ix86_expand_int_movcc): Ditto.
1993 (ix86_expand_int_addcc): Ditto.
1994 (ix86_reverse_condition): Ditto.
1995 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
1996 Return true/false for unordered/ordered fp comparisons.
1997 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
1998 (ix86_prepare_fp_compare_args): Update for rename.
1999 (ix86_expand_fp_compare): Update for rename. Generate unordered
2000 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
2001 (ix86_expand_sse_compare_and_jump): Ditto.
2002 * config/i386/predicates.md (fcmov_comparison_operator):
2003 Remove CCFPU mode handling.
2004 (ix86_comparison_operator): Ditto.
2005 (ix86_carry_flag_operator): Ditto.
2006 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
2007 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
2008 (*cmpu<mode>_cc_i387): Ditto.
2009 (FPCMP): Remove mode iterator.
2010 (unord): Remove mode attribute.
2011 (unord_subst): New define_subst transformation
2012 (unord): New define_subst attribute.
2014 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
2015 (*cmpi<unord>xf_i387): Ditto.
2016 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
2017 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
2018 using unord_subst transformation.
2019 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
2020 (round_saeonly): Also handle CCFP mode.
2021 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
2022 Remove UNSPEC_SAHF unspec handling.
2024 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
2026 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
2028 2017-10-27 Jeff Law <law@redhat.com>
2030 * gimple-ssa-sprintf.c: Include domwalk.h.
2031 (class sprintf_dom_walker): New class, derived from dom_walker.
2032 (sprintf_dom_walker::before_dom_children): New function.
2033 (struct call_info): Moved into sprintf_dom_walker class
2034 (compute_formath_length, handle_gimple_call): Likewise.
2035 (sprintf_length::execute): Call the dominator walker rather
2036 than walking the statements.
2038 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
2039 gimple statement locations.
2040 (check_array_bounds): Corresponding changes. Get the statement's
2041 location directly from wi->stmt.
2043 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
2046 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
2048 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
2050 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
2051 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
2053 2017-10-27 Jakub Jelinek <jakub@redhat.com>
2056 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
2057 * config/i386/i386.c (maybe_get_pool_constant): Removed.
2058 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
2059 maybe_get_pool_constant.
2060 * config/i386/predicates.md (zero_extended_scalar_load_operand):
2063 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2065 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
2066 2.26 caveat. Update gas and gld versions.
2067 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
2070 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2072 * cgraph.h (set_malloc_flag): Declare.
2073 * cgraph.c (set_malloc_flag_1): New function.
2074 (set_malloc_flag): Likewise.
2075 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
2076 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
2078 (read_ipa_call_summary): Add support for reading is_return_callee.
2079 (write_ipa_call_summary): Stream is_return_callee.
2080 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
2081 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
2082 ipa-prop.h, ipa-fnsummary.h.
2083 (pure_const_names): Change to static.
2084 (malloc_state_e): Define.
2085 (malloc_state_names): Define.
2086 (funct_state_d): Add field malloc_state.
2087 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
2088 (check_retval_uses): New function.
2089 (malloc_candidate_p): Likewise.
2090 (analyze_function): Add support for malloc attribute.
2091 (pure_const_write_summary): Stream malloc_state.
2092 (pure_const_read_summary): Add support for reading malloc_state.
2093 (dump_malloc_lattice): New function.
2094 (propagate_malloc): New function.
2095 (warn_function_malloc): New function.
2096 (ipa_pure_const::execute): Call propagate_malloc and
2097 ipa_free_fn_summary.
2098 (pass_local_pure_const::execute): Add support for malloc attribute.
2099 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
2100 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
2102 2017-10-27 Martin Liska <mliska@suse.cz>
2104 PR gcov-profile/82457
2105 * doc/invoke.texi: Document that one needs a non-strict ISO mode
2106 for fork-like functions to be properly instrumented.
2108 2017-10-27 Richard Biener <rguenther@suse.de>
2111 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
2112 info when we redirected EH.
2114 2017-10-26 Michael Collison <michael.collison@arm.com>
2116 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
2118 (<optab>_trunchf<GPI:mode>2: New pattern.
2119 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
2120 * config/aarch64/iterators.md (wv): New mode attribute.
2121 (vf, VF): New mode attributes.
2122 (vgp, VGP): New mode attributes.
2123 (s): Update attribute with SImode and DImode prefixes.
2125 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2127 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
2128 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
2129 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
2130 (nios2_option_overide): Initialize it. Don't allow R0-relative
2131 addressing with PIC.
2132 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
2133 (nios2_symbolic_constant_p): Likewise.
2134 (nios2_legitimate_address_p): Likewise.
2135 (nios2_r0rel_section_name_p): New.
2136 (nios2_symbol_ref_in_r0rel_data_p): New.
2137 (nios2_emit_move_sequence): Handle r0rel_constant_p.
2138 (r0rel_constant_p): New.
2139 (nios2_print_operand_address): Handle r0rel_constant_p.
2140 (nios2_cdx_narrow_form_p): Likewise.
2141 * config/nios2/nios2.opt (mr0rel-sec=): New option.
2142 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
2143 (Nios II Options): Document -mr0rel-sec.
2145 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2147 * config/nios2/nios2.c: Include xregex.h.
2148 (nios2_gprel_sec_regex): New.
2149 (nios2_option_overide): Initialize it. Don't allow GP-relative
2150 addressing with PIC.
2151 (nios2_small_section_name_p): Check for regex match.
2152 * config/nios2/nios2.opt (mgprel-sec=): New option.
2153 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
2154 (Nios II Options): Document -mgprel-sec.
2156 2017-10-26 Jim Wilson <wilson@tuliptree.org>
2158 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
2160 2017-10-26 Tom de Vries <tom@codesourcery.com>
2162 PR tree-optimization/82707
2163 * gimple.c (gimple_copy): Fix unsharing of
2164 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
2166 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
2168 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
2169 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
2170 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
2171 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
2172 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
2173 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
2174 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
2175 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
2176 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
2177 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
2178 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
2179 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
2180 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
2181 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
2182 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
2183 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
2184 _mm512_mask_cmpunord_ps_mask): New intrinsics.
2186 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2188 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
2190 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
2191 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
2192 warning to rs6000.c. Remove the Undocumented flag, since it has
2194 (-mabi=ibmlongdouble): Likewise.
2195 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
2196 already set, set the default format for long double.
2197 (rs6000_debug_reg_global): Print whether long double is IBM or
2199 (rs6000_option_override_internal): Rework setting long double
2200 format. Only warn if the user is changing the long double default
2201 and they did not use -Wno-psabi.
2202 * doc/invoke.texi (PowerPC options): Update the documentation for
2203 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
2205 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2206 Alan Hayward <alan.hayward@arm.com>
2207 David Sherwood <david.sherwood@arm.com>
2209 * rtl.h (wider_subreg_mode): New function.
2210 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
2211 rather than an unsigned int *.
2212 * ira-color.c (regno_max_ref_width): Replace with...
2213 (regno_max_ref_mode): ...this new variable.
2214 (coalesced_pseudo_reg_slot_compare): Update accordingly.
2215 Use wider_subreg_mode.
2216 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
2217 rather than an unsigned int *.
2218 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
2219 (process_alt_operands): Likewise.
2220 (invariant_p): Likewise.
2221 * lra-spills.c (assign_mem_slot): Likewise.
2222 (add_pseudo_to_slot): Likewise.
2223 * lra.c (collect_non_operand_hard_regs): Likewise.
2224 (add_regs_to_insn_regno_info): Likewise.
2225 * reload1.c (regno_max_ref_width): Replace with...
2226 (regno_max_ref_mode): ...this new variable.
2227 (reload): Update accordingly. Update call to
2228 ira_sort_regnos_for_alter_reg.
2229 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
2230 (init_eliminable_invariants): Update to use regno_max_ref_mode.
2231 (scan_paradoxical_subregs): Likewise.
2233 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2235 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
2236 (aarch64_frame): Add emit_frame_chain boolean.
2237 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2238 Move eh_return case to aarch64_layout_frame.
2239 (aarch64_layout_frame): Initialize emit_frame_chain.
2240 (aarch64_expand_prologue): Use emit_frame_chain.
2242 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2244 * config/aarch64/aarch64.c (aarch64_layout_frame):
2245 Ensure LR is always stored at the bottom of the callee-saves.
2246 Remove rarely used frame layout which saves callee-saves at top of
2247 frame, so the store of LR can be used as a valid probe in all cases.
2249 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2251 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
2252 Improve unaligned TImode/TFmode base/offset split.
2254 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2255 Alan Hayward <alan.hayward@arm.com>
2256 David Sherwood <david.sherwood@arm.com>
2258 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
2259 * combine.c (find_single_use_1): Likewise.
2260 (expand_field_assignment): Likewise.
2261 (move_deaths): Likewise.
2262 * lra-constraints.c (simplify_operand_subreg): Likewise.
2263 (curr_insn_transform): Likewise.
2264 * lra.c (collect_non_operand_hard_regs): Likewise.
2265 (add_regs_to_insn_regno_info): Likewise.
2266 * rtlanal.c (reg_referenced_p): Likewise.
2267 (covers_regno_no_parallel_p): Likewise.
2269 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2271 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
2272 Don't print any bits outside the precision of the value.
2273 * wide-int.cc (test_printing): Add some new tests.
2275 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2277 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
2278 supports -xbrace_comment option.
2279 * configure: Regenerate.
2280 * config.in: Regenerate.
2281 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
2282 (ASM_CPU_SPEC): Use it.
2284 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2286 * target.def (static_rtx_alignment): New hook.
2287 * targhooks.h (default_static_rtx_alignment): Declare.
2288 * targhooks.c (default_static_rtx_alignment): New function.
2289 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
2290 * doc/tm.texi: Regenerate.
2291 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
2292 instead of targetm.constant_alignment. Remove call to
2294 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2295 (cris_preferred_mininum_alignment): New function, split out from...
2296 (cris_constant_alignment): ...here.
2297 (cris_static_rtx_alignment): New function.
2298 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
2300 (ix86_constant_alignment): ...here.
2301 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2302 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2303 (mmix_static_rtx_alignment): New function.
2304 * config/spu/spu.c (spu_static_rtx_alignment): New function.
2305 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2307 2017-10-26 Tamar Christina <tamar.christina@arm.com>
2310 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
2311 Add flag_trapping_math and flag_fp_int_builtin_inexact.
2313 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
2315 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
2316 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
2317 mark as a sign-extending load.
2318 (local_pic_load_u): Define.
2320 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
2323 * fold-const.c (operand_equal_for_comparison_p): Also return true
2324 if ARG0 is a simple variant of ARG1 with narrower precision.
2325 (fold_ternary_loc): Always pass unstripped operands to the predicate.
2327 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
2329 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
2331 * i386.h (processor_costs): Add gather_static, gather_per_elt,
2332 scatter_static, scatter_per_elt.
2333 * x86-tune-costs.h: Add new cost entries.
2335 2017-10-25 Richard Biener <rguenther@suse.de>
2337 * tree-ssa-sccvn.h (vn_eliminate): Declare.
2338 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
2339 class pass_fre): Move to ...
2340 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
2341 class pass_fre): ... here and adjust for statistics.
2343 2017-10-25 Jakub Jelinek <jakub@redhat.com>
2346 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
2347 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
2348 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
2349 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
2352 2017-10-25 Richard Biener <rguenther@suse.de>
2354 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
2355 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
2356 eliminate_push_avail, eliminate_insert): Move inside...
2357 (class eliminate_dom_walker): ... this class in preparation
2359 (fini_eliminate): Remove by merging with ...
2360 (eliminate): ... this function. Adjust for class changes.
2361 (pass_pre::execute): Remove fini_eliminate call.
2362 (pass_fre::execute): Likewise.
2364 2017-10-24 Jakub Jelinek <jakub@redhat.com>
2367 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
2368 (VPERMI2, VPERMI2I): New mode iterators.
2369 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
2370 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
2372 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
2373 mode iterator. Remove 3 old define_insn patterns.
2374 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
2375 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
2376 VPERMI2 mode iterator, remove the other two expanders.
2377 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
2378 to use VPERMI2 mode iterator, add another alternative for vpermi2*
2379 instructions, remove the other two patterns.
2380 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
2381 mode iterator, remove the other two patterns.
2382 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
2383 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
2384 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
2385 and adjust argument order accordingly.
2386 (ix86_expand_vec_perm): Adjust caller.
2387 (expand_vec_perm_1): Likewise.
2388 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
2389 (expand_vec_perm_vpermt2_vpshub2): ... this.
2390 (ix86_expand_vec_perm_const_1): Adjust caller.
2391 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
2394 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
2395 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
2396 (vec_shl_<mode>): Remove unused expander.
2397 (avx512bw_<shift_insn><mode>3): New define_insn.
2398 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
2399 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
2401 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
2404 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
2407 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
2409 PR rtl-optimization/82396
2410 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
2411 (autopref_multipass_init): Simplify initialization.
2412 (autopref_rank_data): Simplify sort order.
2413 * gcc/sched-int.h (autopref_multipass_data_): Remove
2414 multi_mem_insn_p, min_offset and max_offset.
2416 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
2419 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2420 Check special value of flag_omit_frame_pointer.
2421 (aarch64_can_eliminate): Likewise.
2422 (aarch64_override_options_after_change_1): Simplify handling of
2423 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
2425 2017-10-24 Richard Biener <rguenther@suse.de>
2427 PR tree-optimization/82697
2428 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
2429 zero for conditional load and unconditional store.
2431 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
2433 * doc/install.texi: Document bootstrap-cet.
2435 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
2438 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
2439 ENDBR instruction at function entrance if function is only
2442 2017-10-24 Jakub Jelinek <jakub@redhat.com>
2445 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
2446 patterns to better describe from which operation the CF is computed.
2447 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
2448 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
2449 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
2450 is 0, use _0 suffixed expanders instead of emitting a comparison
2453 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
2455 * config/i386/i386.md(*movsf_internal, *movdf_internal):
2456 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
2458 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
2461 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
2462 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
2463 * loop-iv.c (iv_get_reaching_def): Likewise.
2464 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
2465 variable is promoted and the partition contains undefined values.
2467 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2469 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
2471 (nios2_address_cost): Define.
2472 (nios2_legitimize_address): Recognize (exp + constant) directly.
2473 (TARGET_ADDRESS_COST): Define.
2475 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2477 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
2478 (nios2_symbolic_memory_operand_p): Declare.
2479 (nios2_split_large_constant): Declare.
2480 (nios2_split_symbolic_memory_operand): Declare.
2481 * config/nios2/nios2.c: Adjust includes.
2482 (nios2_symbolic_constant_allowed): New.
2483 (nios2_symbolic_constant_p): New.
2484 (nios2_plus_symbolic_constant_p): New.
2485 (nios2_valid_addr_expr_p): Recognize addresses involving
2487 (nios2_legitimate_address_p): Likewise, also LO_SUM.
2488 (nios2_symbolic_memory_operand_p): New.
2489 (nios2_large_constant_p): New.
2490 (nios2_split_large_constant): New.
2491 (nios2_split_plus_large_constant): New.
2492 (nios2_split_symbolic_memory_operand): New.
2493 (nios2_legitimize_address): Code refactoring. Handle addresses
2494 involving symbolic constants.
2495 (nios2_emit_move_sequence): Likewise.
2496 (nios2_print_operand): Improve error output.
2497 (nios2_print_operand_address): Handle LO_SUM.
2498 (nios2_cdx_narrow_form_p): Likewise.
2499 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
2500 operands involving symbolic constants.
2501 (movhi_internal, movsi_internal): Likewise.
2502 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
2503 (extendhisi2, extendqi<mode>2): Likewise.
2505 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2507 * tree-pass.h (PROP_rtl_split_insns): Define.
2508 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
2510 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2512 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
2514 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2517 * target.def (const_not_ok_for_debug_p): Default to
2518 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
2519 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
2520 * targhooks.c (default_const_not_ok_for_debug_p): New function.
2521 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
2522 which targetm.const_not_ok_for_debug_p returned true.
2523 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
2525 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
2527 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
2528 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
2529 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
2531 (ix86_const_not_ok_for_debug_p): New function.
2532 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
2533 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
2535 2017-10-23 David Malcolm <dmalcolm@redhat.com>
2538 * system.h: Conditionally include "unique-ptr.h" if
2539 INCLUDE_UNIQUE_PTR is defined.
2540 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
2541 of defining INCLUDE_UNIQUE_PTR before including "system.h".
2543 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
2545 * config/rl78/rl78.md: New define_expand "subdi3".
2547 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
2550 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
2551 DF_REF_INSN if DF_REF_INSN_INFO is false.
2553 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
2555 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
2556 xmm_move instead of sse_move.
2557 (sse_store_index): New function.
2558 (ix86_register_move_cost): Be more sensible about mismatch stall;
2559 model AVX moves correctly; make difference between sse->integer and
2561 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
2562 moves; make difference between SSE and AVX.
2563 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
2564 and zmm_move. Increase size of sse load and store tables;
2565 add unaligned load and store tables; add ssemmx_to_integer.
2566 * x86-tune-costs.h: Update all entries according to real
2567 move latencies from Agner Fog's manual and chip documentation.
2569 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2572 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
2573 * config/i386/constraints.md (Wf): New constraint.
2574 * config/i386/i386.md (UNSPEC_SBB): New unspec.
2575 (cmp<dwi>_doubleword): Removed.
2576 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
2577 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
2578 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
2579 expand with cmp<dwi>_doubleword. For LTU and GEU use
2580 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
2582 * common.opt (gcolumn-info): Enable by default.
2583 * doc/invoke.texi (gcolumn-info): Document new default.
2585 2017-10-23 Richard Biener <rguenther@suse.de>
2587 PR tree-optimization/82672
2588 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
2589 Fold the stmt if we propagated into it.
2591 2017-10-23 Richard Biener <rguenther@suse.de>
2593 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
2594 (bitmap_remove_expr_from_set): ... this. All callers call this
2595 for non-constant values.
2596 (bitmap_set_subtract): Rename to...
2597 (bitmap_set_subtract_expressions): ... this. Adjust and
2599 (bitmap_set_contains_value): Remove superfluous check.
2600 (bitmap_set_replace_value): Inline into single caller ...
2601 (bitmap_value_replace_in_set): ... here and simplify.
2602 (dependent_clean): Merge into ...
2603 (clean): ... this using an overload. Adjust.
2604 (prune_clobbered_mems): Adjust.
2605 (compute_antic_aux): Likewise.
2606 (compute_partial_antic_aux): Likewise.
2608 2017-10-23 Richard Biener <rguenther@suse.de>
2610 PR tree-optimization/82129
2612 2017-08-01 Richard Biener <rguenther@suse.de>
2614 PR tree-optimization/81181
2615 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
2616 (compute_antic): ... end of iteration here.
2618 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2620 * target.def (starting_frame_offset): New hook.
2621 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
2622 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
2623 * doc/tm.texi.in: Regenerate.
2624 * hooks.h (hook_hwi_void_0): Declare.
2625 * hooks.c (hook_hwi_void_0): New function.
2626 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
2627 STARTING_FRAME_OFFSET.
2628 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
2629 * reload1.c (reload): Likewise.
2630 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
2631 instead of STARTING_FRAME_OFFSET.
2632 * function.c (try_fit_stack_local): Likewise.
2633 (assign_stack_local_1): Likewise
2634 (instantiate_virtual_regs): Likewise.
2635 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
2636 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
2637 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
2638 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
2639 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
2640 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
2641 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
2642 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
2643 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
2644 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
2645 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
2646 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
2647 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
2648 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
2649 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
2650 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
2651 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
2652 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
2653 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
2654 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
2655 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
2656 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
2657 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
2658 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
2659 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
2660 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
2661 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
2662 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
2663 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
2664 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
2665 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
2666 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
2667 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
2668 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
2669 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
2670 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
2671 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
2672 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
2673 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
2674 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
2675 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
2676 * config/avr/avr.c (avr_starting_frame_offset): Make static and
2677 return a HOST_WIDE_INT.
2678 (avr_builtin_setjmp_frame_value): Use it instead of
2679 STARTING_FRAME_OFFSET.
2680 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2681 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
2682 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
2684 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2685 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
2686 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
2687 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2688 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
2689 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
2690 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2691 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
2692 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
2693 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2694 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
2695 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
2697 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2698 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
2699 * config/mips/mips.c (mips_compute_frame_info): Refer to
2700 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
2701 (mips_starting_frame_offset): New function.
2702 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2703 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
2704 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
2705 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
2706 and return a HOST_WIDE_INT.
2707 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2708 (mmix_initial_elimination_offset): Refer to
2709 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
2710 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
2711 * config/pa/pa.c (pa_starting_frame_offset): New function.
2712 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
2713 (pa_expand_prologue): Likewise.
2714 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2715 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
2716 !FRAME_GROWS_DOWNWARD handling to...
2717 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2718 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
2719 !FRAME_GROWS_DOWNWARD handling to...
2720 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2721 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
2722 !FRAME_GROWS_DOWNWARD handling to...
2723 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2724 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
2726 (rs6000_starting_frame_offset): New function.
2727 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
2728 !FRAME_GROWS_DOWNWARD handling to...
2729 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2730 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
2731 !FRAME_GROWS_DOWNWARD handling to...
2732 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2733 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
2734 !FRAME_GROWS_DOWNWARD handling to...
2735 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2736 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
2737 (rs6000_starting_frame_offset): New function.
2738 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
2739 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
2740 * config/vax/vax.c (vax_starting_frame_offset): New function.
2741 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
2742 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2743 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
2744 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
2745 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2746 * system.h (STARTING_FRAME_OFFSET): Poison.
2748 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2750 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
2751 SCALAR_TYPE_MODE instead of TYPE_MODE.
2753 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2754 Alan Hayward <alan.hayward@arm.com>
2755 David Sherwood <david.sherwood@arm.com>
2757 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
2759 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2760 Alan Hayward <alan.hayward@arm.com>
2761 David Sherwood <david.sherwood@arm.com>
2763 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
2765 2017-10-23 Richard Biener <rguenther@suse.de>
2767 PR tree-optimization/82129
2768 * tree-ssa-pre.c (bitmap_set_and): Remove.
2769 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
2770 canonicalizing expressions in the set to those with lowest
2771 ID rather than taking that from the first edge.
2773 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2775 * combine.c (rtx_equal_for_field_assignment_p): Use
2776 byte_lowpart_offset.
2778 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2779 Alan Hayward <alan.hayward@arm.com>
2780 David Sherwood <david.sherwood@arm.com>
2782 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
2783 to a SUBREG_PROMOTED_VAR.
2785 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2786 Alan Hayward <alan.hayward@arm.com>
2787 David Sherwood <david.sherwood@arm.com>
2789 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
2790 (expand_debug_source_expr): Likewise.
2791 * combine.c (combine_simplify_rtx): Likewise.
2792 * cse.c (fold_rtx): Likewise.
2793 * optabs.c (expand_float): Likewise.
2794 * simplify-rtx.c (simplify_unary_operation_1): 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 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
2802 (record_promoted_value): Likewise.
2803 * expr.c (expand_expr_real_2): Likewise.
2804 * ree.c (update_reg_equal_equiv_notes): Likewise.
2805 (combine_set_extension): Likewise.
2806 * rtlanal.c (low_bitmask_len): Likewise.
2807 * simplify-rtx.c (neg_const_int): Likewise.
2808 (simplify_binary_operation_1): Likewise.
2810 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2811 Alan Hayward <alan.hayward@arm.com>
2812 David Sherwood <david.sherwood@arm.com>
2814 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
2815 * regcprop.c (maybe_mode_change): Likewise.
2816 * reload1.c (alter_reg): Likewise.
2818 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2820 * inchash.h (inchash::hash::add_wide_int): New function.
2821 * lto-streamer-out.c (hash_tree): Use it.
2823 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2825 * inchash.h (inchash::hash::add_wide_int): Rename to...
2826 (inchash::hash::add_hwi): ...this.
2827 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
2828 (polymorphic_call_target_hasher::hash): Likewise.
2829 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
2830 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
2831 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
2832 * lto-streamer-out.c (hash_tree): Likewise.
2833 * optc-save-gen.awk: Likewise.
2834 * tree.c (add_expr): Likewise.
2836 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
2839 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
2840 for ordered inequality comparisons even with TARGET_IEEE_FP.
2842 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
2845 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
2846 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
2847 Expand with cmp<dwi>_doubleword.
2849 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2851 * extend.texi: Add x86 specific to 'nocf_check' attribute.
2852 List CET intrinsics.
2853 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
2854 specific to -fcf-protection option.
2856 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2858 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
2859 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
2860 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
2861 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
2862 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
2863 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
2864 (extra_objs): Add cet.o for Linux/x86 targets.
2865 (tmake_file): Add i386/t-cet for Linux/x86 targets.
2866 * config/i386/cet.c: New file.
2867 * config/i386/cetintrin.h: Likewise.
2868 * config/i386/t-cet: Likewise.
2869 * config/i386/cpuid.h (bit_SHSTK): New.
2870 (bit_IBT): Likewise.
2871 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
2872 pass IBT and SHSTK bits.
2873 * config/i386/i386-builtin-types.def
2874 (VOID_FTYPE_UNSIGNED_PVOID): New.
2875 (VOID_FTYPE_UINT64_PVOID): Likewise.
2876 * config/i386/i386-builtin.def: Add CET intrinsics.
2877 * config/i386/i386-c.c (ix86_target_macros_internal): Add
2878 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
2879 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
2880 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
2882 * config/i386/i386.c (rest_of_insert_endbranch): New.
2883 (pass_data_insert_endbranch): Likewise.
2884 (pass_insert_endbranch): Likewise.
2885 (make_pass_insert_endbranch): Likewise.
2886 (ix86_notrack_prefixed_insn_p): Likewise.
2887 (ix86_target_string): Add -mibt, -mshstk flags.
2888 (ix86_option_override_internal): Add flag_cf_protection
2890 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
2891 (ix86_print_operand): Add 'notrack' prefix output.
2892 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
2893 (ix86_expand_builtin): Expand CET intrinsics.
2894 (x86_output_mi_thunk): Add 'endbranch' instruction.
2895 * config/i386/i386.h (TARGET_IBT): New.
2896 (TARGET_IBT_P): Likewise.
2897 (TARGET_SHSTK): Likewise.
2898 (TARGET_SHSTK_P): Likewise.
2899 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
2900 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
2901 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
2902 (builtin_setjmp_setup): New pattern.
2903 (builtin_longjmp): Likewise.
2904 (rdssp<mode>): Likewise.
2905 (incssp<mode>): Likewise.
2906 (saveprevssp): Likewise.
2907 (rstorssp): Likewise.
2908 (wrss<mode>): Likewise.
2909 (wruss<mode>): Likewise.
2910 (setssbsy): Likewise.
2911 (clrssbsy): Likewise.
2912 (nop_endbr): Likewise.
2913 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
2915 * config/i386/immintrin.h: Include <cetintrin.h>.
2916 * config/i386/linux-common.h
2917 (file_end_indicate_exec_stack_and_cet): New prototype.
2918 (TARGET_ASM_FILE_END): New.
2920 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
2922 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
2923 latencies instead of having separate table; make difference between
2924 integer and float costs.
2925 * i386.h (processor_costs): Remove scalar_stmt_cost,
2926 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
2927 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
2929 * x86-tune-costs.h: Remove entries which has been removed in
2930 procesor_costs from all tables; make cond_taken_branch_cost
2931 and cond_not_taken_branch_cost COST_N_INSNS based.
2933 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
2935 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
2937 2017-10-20 Jakub Jelinek <jakub@redhat.com>
2939 * config/i386/i386.md (isa): Remove fma_avx512f.
2940 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
2941 <avx512>_fmadd_<mode>_mask3<round_name>,
2942 <avx512>_fmsub_<mode>_mask<round_name>,
2943 <avx512>_fmsub_<mode>_mask3<round_name>,
2944 <avx512>_fnmadd_<mode>_mask<round_name>,
2945 <avx512>_fnmadd_<mode>_mask3<round_name>,
2946 <avx512>_fnmsub_<mode>_mask<round_name>,
2947 <avx512>_fnmsub_<mode>_mask3<round_name>,
2948 <avx512>_fmaddsub_<mode>_mask<round_name>,
2949 <avx512>_fmaddsub_<mode>_mask3<round_name>,
2950 <avx512>_fmsubadd_<mode>_mask<round_name>,
2951 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
2952 (*vec_widen_umult_even_v16si<mask_name>,
2953 *vec_widen_smult_even_v16si<mask_name>): Likewise.
2954 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
2956 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2958 * extend.texi: Add 'nocf_check' documentation.
2959 * gimple.texi: Add second parameter to
2960 gimple_build_call_from_tree.
2961 * invoke.texi: Add -fcf-protection documentation.
2962 * rtl.texi: Add REG_CALL_NOTRACK documenation.
2964 2017-10-20 Richard Biener <rguenther@suse.de>
2966 PR tree-optimization/82473
2967 * tree-vect-loop.c (vectorizable_reduction): Properly get at
2968 the largest input type.
2970 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2972 * c-attribs.c (handle_nocf_check_attribute): New function.
2973 (c_common_attribute_table): Add 'nocf_check' handling.
2974 * gimple-parser.c: Add second argument NULL to
2975 gimple_build_call_from_tree.
2976 * attrib.c (comp_type_attributes): Check nocf_check attribute.
2977 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
2979 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
2980 * common.opt: Add fcf-protection flag.
2981 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
2982 * flag-types.h: Add enum cf_protection_level.
2983 * gimple.c (gimple_build_call_from_tree): Add second parameter.
2984 Add 'nocf_check' attribute propagation to gimple call.
2985 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
2986 (gimple_build_call_from_tree): Update prototype.
2987 (gimple_call_nocf_check_p): New function.
2988 (gimple_call_set_nocf_check): Likewise.
2989 * gimplify.c: Add second argument to gimple_build_call_from_tree.
2990 * ipa-icf.c: Add nocf_check attribute in statement hash.
2991 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
2992 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
2993 * toplev.c (process_options): Add flag_cf_protection handling.
2995 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
2997 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
2999 2017-10-20 Richard Biener <rguenther@suse.de>
3001 PR tree-optimization/82603
3002 * tree-if-conv.c (predicate_mem_writes): Make sure to only
3003 remove false predicated stores.
3005 2017-10-20 Richard Biener <rguenther@suse.de>
3007 * graphite-isl-ast-to-gimple.c
3008 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
3009 Remove return value and simplify, dump copied stmt after lhs
3011 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
3012 Reduce dump verbosity.
3013 (gsi_insert_earliest): Likewise.
3014 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
3015 * graphite.c (print_global_statistics): Adjust dumping.
3016 (print_graphite_scop_statistics): Likewise.
3017 (print_graphite_statistics): Do not dump loops here.
3018 (graphite_transform_loops): But here.
3020 2017-10-20 Nicolas Roche <roche@adacore.com>
3022 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
3023 * configure: Regenerate.
3025 2017-10-20 Jakub Jelinek <jakub@redhat.com>
3028 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
3029 functions when optimizing replace GIMPLE_RETURN stmts with
3030 calls to __builtin_unreachable ().
3033 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
3034 for -fsanitize=thread link of executables.
3035 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
3036 link of executables.
3039 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
3041 (<shift_insn><mode>3<mask_name>): Change the last of the 3
3042 define_insns for logical vector shifts to use VI248_AVX512BW
3043 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
3044 condition, useless isa and prefix attributes. Change the first
3045 2 of these define_insns to ...
3046 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
3047 define_insn for avx512vl.
3048 (<shift_insn><mode>3): ... and this, new define_insn without
3049 masking for non-avx512vl.
3052 * config/i386/sse.md (*andnot<mode>3,
3053 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
3054 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
3055 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
3056 not applied use empty suffix even for TARGET_AVX512VL.
3057 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
3058 is applied, supply evex,evex or evex,evex,evex instead of just
3061 2017-10-20 Julia Koval <julia.koval@intel.com>
3063 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
3064 (OPTION_MASK_ISA_GFNI_UNSET): New.
3065 (ix86_handle_option): Handle OPT_mgfni.
3066 * config/i386/cpuid.h (bit_GFNI): New.
3067 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
3068 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
3069 * config/i386/i386.c (ix86_target_string): Add -mgfni.
3070 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
3071 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
3072 * config/i386/i386.opt: Add mgfni.
3074 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
3076 * config/msp430/msp430.c (msp430_option_override): Disable
3077 -fdelete-null-pointer-checks.
3078 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
3080 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3082 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
3083 of x87 and SSE instructions.
3085 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3087 * asan.c (create_cond_insert_point): Do not update edge count.
3088 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
3089 (afdo_propagate_circuit): Likewise.
3090 (afdo_calculate_branch_prob): Likewise.
3091 (afdo_annotate_cfg): Likewise.
3092 * basic-block.h (struct edge_def): Remove count.
3093 (edge_def::count): New accessor.
3094 * bb-reorder.c (rotate_loop): Update.
3095 (find_traces_1_round): Update.
3096 (connect_traces): Update.
3097 (sanitize_hot_paths): Update.
3098 * cfg.c (unchecked_make_edge): Update.
3099 (make_single_succ_edge): Update.
3100 (check_bb_profile): Update.
3101 (dump_edge_info): Update.
3102 (update_bb_profile_for_threading): Update.
3103 (scale_bbs_frequencies_int): Update.
3104 (scale_bbs_frequencies_gcov_type): Update.
3105 (scale_bbs_frequencies_profile_count): Update.
3106 (scale_bbs_frequencies): Update.
3107 * cfganal.c (connect_infinite_loops_to_exit): Update.
3108 * cfgbuild.c (compute_outgoing_frequencies): Update.
3109 (find_many_sub_basic_blocks): Update.
3110 * cfgcleanup.c (try_forward_edges): Update.
3111 (try_crossjump_to_edge): Update
3112 * cfgexpand.c (expand_gimple_cond): Update
3113 (expand_gimple_tailcall): Update
3114 (construct_exit_block): Update
3115 * cfghooks.c (verify_flow_info): Update
3116 (redirect_edge_succ_nodup): Update
3117 (split_edge): Update
3118 (make_forwarder_block): Update
3119 (duplicate_block): Update
3120 (account_profile_record): Update
3121 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
3122 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
3123 * cfgloopmanip.c (scale_loop_profile): Update.
3125 (lv_adjust_loop_entry_edge): Update.
3126 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
3127 (force_nonfallthru_and_redirect): Update.
3128 (purge_dead_edges): Update.
3129 (rtl_flow_call_edges_add): Update.
3130 * cgraphunit.c (init_lowered_empty_function): Update.
3131 (cgraph_node::expand_thunk): Update.
3132 * gimple-pretty-print.c (dump_probability): Update.
3133 (dump_edge_probability): Update.
3134 * gimple-ssa-isolate-paths.c (isolate_path): Update.
3135 * haifa-sched.c (sched_create_recovery_edges): Update.
3136 * hsa-gen.c (convert_switch_statements): Update.
3137 * ifcvt.c (dead_or_predicable): Update.
3138 * ipa-inline-transform.c (inline_transform): Update.
3139 * ipa-split.c (split_function): Update.
3140 * ipa-utils.c (ipa_merge_profiles): Update.
3141 * loop-doloop.c (add_test): Update.
3142 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
3143 * lto-streamer-in.c (input_cfg): Update.
3144 (input_function): Update.
3145 * lto-streamer-out.c (output_cfg): Update.
3146 * modulo-sched.c (sms_schedule): Update.
3147 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
3148 * predict.c (maybe_hot_edge_p): Update.
3149 (unlikely_executed_edge_p): Update.
3150 (probably_never_executed_edge_p): Update.
3151 (dump_prediction): Update.
3152 (drop_profile): Update.
3153 (propagate_unlikely_bbs_forward): Update.
3154 (determine_unlikely_bbs): Update.
3155 (force_edge_cold): Update.
3156 * profile.c (compute_branch_probabilities): Update.
3157 * reg-stack.c (better_edge): Update.
3158 * shrink-wrap.c (handle_simple_exit): Update.
3159 * tracer.c (better_p): Update.
3160 * trans-mem.c (expand_transaction): Update.
3161 (split_bb_make_tm_edge): Update.
3162 * tree-call-cdce.c: Update.
3163 * tree-cfg.c (gimple_find_sub_bbs): Update.
3164 (gimple_split_edge): Update.
3165 (gimple_duplicate_sese_region): Update.
3166 (gimple_duplicate_sese_tail): Update.
3167 (gimple_flow_call_edges_add): Update.
3168 (insert_cond_bb): Update.
3169 (execute_fixup_cfg): Update.
3170 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
3171 * tree-complex.c (expand_complex_div_wide): Update.
3172 * tree-eh.c (lower_resx): Update.
3173 (unsplit_eh): Update.
3174 (cleanup_empty_eh_move_lp): Update.
3175 * tree-inline.c (copy_edges_for_bb): Update.
3176 (freqs_to_counts): Update.
3177 (copy_cfg_body): Update.
3178 * tree-ssa-dce.c (remove_dead_stmt): Update.
3179 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
3180 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
3181 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
3182 (unloop_loops): Update.
3183 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3184 * tree-ssa-loop-split.c (connect_loops): Update.
3185 (split_loop): Update.
3186 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
3187 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
3188 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
3189 * tree-ssa-reassoc.c (branch_fixup): Update.
3190 * tree-ssa-tail-merge.c (replace_block_by): Update.
3191 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
3192 (compute_path_counts): Update.
3193 (update_profile): Update.
3194 (recompute_probabilities): Update.
3195 (update_joiner_offpath_counts): Update.
3196 (estimated_freqs_path): Update.
3197 (freqs_to_counts_path): Update.
3198 (clear_counts_path): Update.
3199 (ssa_fix_duplicate_block_edges): Update.
3200 (duplicate_thread_path): Update.
3201 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
3202 (case_bit_test_cmp): Update.
3203 (collect_switch_conv_info): Update.
3204 (gen_inbound_check): Update.
3205 (do_jump_if_equal): Update.
3206 (emit_cmp_and_jump_insns): Update.
3207 * tree-tailcall.c (decrease_profile): Update.
3208 (eliminate_tail_call): Update.
3209 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
3210 (vect_do_peeling): Update.
3211 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3212 * ubsan.c (ubsan_expand_null_ifn): Update.
3213 (ubsan_expand_ptr_ifn): Update.
3214 * value-prof.c (gimple_divmod_fixed_value): Update.
3215 (gimple_mod_pow2): Update.
3216 (gimple_mod_subtract): Update.
3217 (gimple_ic): Update.
3218 (gimple_stringop_fixed_value): Update.
3220 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3223 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
3225 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
3227 PR rtl-optimization/82395
3228 * ira-color.c (allocno_priority_compare_func): Fix comparison step
3229 based on non_spilled_static_chain_regno_p.
3231 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3233 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
3234 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
3235 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
3237 2017-10-19 Martin Sebor <msebor@redhat.com>
3239 PR tree-optimization/82596
3240 * tree.c (array_at_struct_end_p): Handle STRING_CST.
3242 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3244 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
3245 (get_mem_refs_of_builtin_call): Likewise.
3246 * builtins.c (expand_builtin_apply): Adjust call to
3247 allocate_dynamic_stack_space.
3248 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
3249 the third argument to allocate_dynamic_stack_space, otherwise -1.
3250 (expand_builtin): Deal with all alloca variants.
3251 (is_inexpensive_builtin): Likewise.
3252 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
3253 * calls.c (special_function_p): Deal with all alloca variants.
3254 (initialize_argument_information): Adjust call to
3255 allocate_dynamic_stack_space.
3256 (expand_call): Likewise.
3257 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
3258 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
3259 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
3260 use it for the stack usage computation.
3261 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
3262 * function.c (gimplify_parameters): Call build_alloca_call_expr.
3263 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
3264 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
3265 (in_loop_p): Remove first argument and useless check.
3266 (pass_walloca::execute): Remove useless test and adjust call to above.
3267 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
3268 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
3269 (gimplify_call_expr): Deal with all alloca variants.
3270 * hsa-gen.c (gen_hsa_alloca): Likewise.
3271 (gen_hsa_insns_for_call): Likewise.
3272 * ipa-pure-const.c (special_builtin_state): Likewise.
3273 * tree-chkp.c (chkp_build_returned_bound): Likewise.
3274 * tree-object-size.c (alloc_object_size): Likewise.
3275 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
3276 (call_may_clobber_ref_p_1): Likewise.
3277 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
3278 (ccp_fold_stmt): Likewise.
3279 (optimize_stack_restore): Likewise.
3280 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
3281 (mark_all_reaching_defs_necessary_1): Likewise.
3282 (propagate_necessity): Likewise.
3283 (eliminate_unnecessary_stmts): Likewise.
3284 * tree.c (build_common_builtin_nodes): Build
3285 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
3286 (build_alloca_call_expr): New function.
3287 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
3288 (CASE_BUILT_IN_ALLOCA): Likewise.
3289 (build_alloca_call_expr): Declare.
3290 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
3292 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3295 * dwarf2out.c (new_die_raw): New static inline function.
3296 (new_die): Use it to create the DIE.
3297 (add_AT_external_die_ref): Likewise.
3298 (clone_die): Likewise.
3299 (clone_as_declaration): Likewise.
3300 (dwarf2out_vms_debug_main_pointer): Likewise.
3301 (base_type_die): Likewise. Remove early return for corner cases.
3302 Do not call add_pubtype on the DIE here.
3303 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
3304 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
3305 typedefs for base types with DW_AT_endianity. Make sure a DIE with
3306 native order exists for base types, attach the DIE manually and call
3307 add_pubtype on it. Do not equate a reverse order DIE to the type.
3309 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
3311 * config/arm/arm.c (align_ok_ldrd_strd): New function.
3312 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
3314 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
3316 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3318 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
3319 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
3320 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
3321 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
3322 * opts.c (sanitizer_opts): Add builtin.
3323 * ubsan.c (instrument_builtin): New function.
3324 (pass_ubsan::execute): Call it.
3325 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
3326 * doc/invoke.texi: Document -fsanitize=builtin.
3328 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
3329 builtins, store max (log2 (align), 0) into uchar field instead of
3330 align into uptr field.
3331 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
3332 store uchar 0 field instead of uptr 0 field.
3333 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
3334 instead of passing one address of struct with 2 locations pass
3335 two addresses of structs with 1 location each.
3336 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
3337 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
3338 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
3339 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
3340 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
3341 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
3342 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
3343 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
3345 2017-10-19 Martin Liska <mliska@suse.cz>
3348 * file-find.c (remove_prefix): Remove.
3349 * file-find.h (remove_prefix): Likewise.
3350 * gcc-ar.c: Remove smartness of lookup.
3352 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
3354 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
3355 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
3356 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
3358 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3361 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
3362 (setcc + and to xor + setcc): New peephole2.
3364 2017-10-19 Tom de Vries <tom@codesourcery.com>
3366 * doc/sourcebuild.texi (Test Directives, Variants of
3367 dg-require-support): Add dg-require-stack-size.
3369 2017-10-19 Martin Liska <mliska@suse.cz>
3372 * gimplify.c (gimplify_decl_expr): Do not instrument variables
3373 that have a large alignment.
3374 (gimplify_target_expr): Likewise.
3376 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
3378 PR rtl-optimization/82602
3379 * ira.c (rtx_moveable_p): Return false for volatile asm.
3381 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
3384 * config/i386/i386-modes.def (CCGZ): New CC mode.
3385 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
3386 * config/i386/predicates.md (ix86_comparison_operator):
3388 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
3389 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
3390 with double-word subtraction.
3391 (put_condition_code): Handle CCGZmode.
3393 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
3395 * wide-int.cc (debug (const wide_int &)): New.
3396 (debug (const wide_int *)): New.
3397 (debug (const widest_int &)): New.
3398 (debug (const widest_int *)): New.
3400 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
3403 * lra-constraints.c (curr_insn_transform): Use non-input operand
3404 instead of output one for matched reload.
3406 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3408 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
3409 (tree-ssa-loop-ivopts.h): New header file.
3410 (struct builtin_info): New fields.
3411 (classify_builtin_1): Compute and record base and offset parts for
3412 memset builtin partition by calling strip_offset.
3413 (offset_cmp, fuse_memset_builtins): New functions.
3414 (finalize_partitions): Fuse adjacent memset partitions by calling
3416 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
3417 Expose the interface.
3418 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
3420 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3422 PR tree-optimization/82574
3423 * tree-loop-distribution.c (find_single_drs): New parameter. Check
3424 that data reference must be executed exactly once per iteration
3425 against the outermost loop in nest.
3426 (classify_partition): Update call to above function.
3428 2017-10-18 Richard Biener <rguenther@suse.de>
3430 PR tree-optimization/82591
3431 * graphite.c (graphite_transform_loops): Move code gen message
3433 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3434 Here. Handle scop_to_isl_ast failing.
3435 (scop_to_isl_ast): Limit the number of ISL operations.
3437 2017-10-18 Richard Biener <rguenther@suse.de>
3439 * graphite-isl-ast-to-gimple.c
3440 (translate_isl_ast_to_gimple::set_rename): Simplify.
3441 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
3442 (graphite_copy_stmts_from_block): ... here.
3443 (copy_bb_and_scalar_dependences): Simplify.
3444 (add_parameters_to_ivs_params): Canonicalize.
3445 (generate_entry_out_of_ssa_copies): Simplify.
3446 * graphite-sese-to-poly.c (extract_affine_name): Simplify
3447 by passing in ISL dimension.
3448 (parameter_index_in_region_1): Rename to ...
3449 (parameter_index_in_region): ... this.
3450 (extract_affine): Adjust assert, pass down parameter index.
3451 (add_param_constraints): Use range-info when available.
3452 (build_scop_context): Adjust.
3453 * sese.c (new_sese_info): Adjust.
3454 (free_sese_info): Likewise.
3455 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
3456 Remove unused typedefs.
3457 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
3459 2017-10-18 Martin Liska <mliska@suse.cz>
3461 * combine.c (simplify_compare_const): Add gcc_fallthrough.
3463 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
3465 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
3466 (s390_sched_init): Do not reset s390_sched_state if we entered the
3467 current basic block via a fallthru edge and all others are unlikely.
3469 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
3471 * config/s390/s390.c (NUM_SIDES): New variable.
3472 (LONGRUNNING_THRESHOLD): New variable.
3473 (LATENCY_FACTOR): New variable.
3474 (s390_sched_score): Decrease score for long-running instructions on
3476 (s390_sched_variable_issue): Perform bookkeeping for long-running
3479 2017-10-18 Richard Biener <rguenther@suse.de>
3481 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
3482 Simplify with removal of the parameter rename map.
3483 (set_rename): Likewise.
3484 (should_copy_to_new_region): Likewise.
3485 (graphite_copy_stmts_from_block): Likewise.
3486 (copy_bb_and_scalar_dependences): Remove initialization of
3487 unused copied_bb_map.
3489 (copy_internal_parameters): Likewise.
3490 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
3491 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
3492 Use INTEGRAL_TYPE_P.
3493 (parameter_index_in_region_1): Rename to ...
3494 (assign_parameter_index_in_region): ... this. Assert we have
3495 a parameter we handle.
3496 (scan_tree_for_params): Adjust.
3497 * sese.h (parameter_rename_map_t): Remove.
3498 (struct sese_info_t): Remove unused parameter_rename_map and
3499 copied_bb_map members.
3500 * sese.c (new_sese_info): Adjust.
3501 (free_sese_info): Likewise.
3503 2017-10-18 Martin Liska <mliska@suse.cz>
3506 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
3507 on an abnormal edge.
3509 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
3511 * doc/invoke.texi (ffunction-sections and fdata-sections):
3514 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3516 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
3517 the use statement can throw internally.
3519 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3521 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
3522 any RTX present on the RHS of a SET.
3523 * compare-elim.c (try_eliminate_compare): Restore comment.
3525 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3527 * langhooks.h (struct lang_hooks): Document that tree_size langhook
3528 may be also called on tcc_type nodes.
3529 * langhooks.c (lhd_tree_size): Likewise.
3531 2017-10-17 David Malcolm <dmalcolm@redhat.com>
3533 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
3534 format_warning_at_substring.
3535 (maybe_warn): Convert source_range * param to a location_t. Pass
3536 UNKNOWN_LOCATION rather than NULL to fmtwarn.
3537 (format_directive): Remove code to extract source_ranges and
3538 source_range * in favor of just a location_t.
3539 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
3541 * substring-locations.c (format_warning_va): Convert
3542 source_range * param to a location_t.
3543 (format_warning_at_substring): Likewise.
3544 * substring-locations.h (format_warning_va): Likewise.
3545 (format_warning_at_substring): Likewise.
3547 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
3549 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
3551 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
3552 and scatter/gather ops.
3554 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
3555 vec_gather_load and vec_scatter_store.
3556 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
3557 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
3558 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
3559 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
3560 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
3561 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
3563 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
3565 * reg-stack.c (compare_for_stack_reg): Add bool argument.
3566 Detect FTST instruction and handle its register pops. Only pop
3567 second operand if can_pop_second_op is true.
3568 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
3569 set can_pop_second_op to false in the compare_for_stack_reg call.
3571 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
3572 output_fp_compare for stack register operands.
3573 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
3574 instructions here. Do not emit stack register pops here. Assert
3575 that FCOMPP pops next to top stack register. Rewrite function.
3577 2017-10-17 Nathan Sidwell <nathan@acm.org>
3580 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
3581 use DECL_ASSEMBLER_NAME_RAW.
3584 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
3587 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
3588 Wilco Dijkstra <wilco.dijkstra@arm.com>
3590 * builtins.c (expand_builtin_update_setjmp_buf): Add a
3591 converstion to Pmode from the buf_addr.
3593 2017-10-17 Richard Biener <rguenther@suse.de>
3595 * graphite-dependences.c (scop_get_reads_and_writes): Change
3596 output parameters to references.
3598 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
3600 PR 71026/tree-optimization
3601 * fold-const.c (distribute_real_division): Removed.
3602 (fold_binary_loc): Remove calls to distribute_real_divison.
3604 2017-10-17 Richard Biener <rguenther@suse.de>
3606 * graphite-scop-detection.c
3607 (scop_detection::stmt_has_simple_data_refs_p): Always use
3608 the full nest as region.
3609 (try_generate_gimple_bb): Likewise.
3610 * sese.c (scalar_evolution_in_region): Simplify now that
3611 SCEV can handle instantiation in regions.
3612 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
3613 in the non-loop part of a function if requested.
3615 2017-10-17 Richard Biener <rguenther@suse.de>
3617 PR tree-optimization/82563
3618 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
3620 (graphite_regenerate_ast_isl): Call it.
3621 * graphite-scop-detection.c (build_scops): Remove entry edge split.
3623 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3625 PR tree-optimization/82549
3626 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
3627 Formatting fixes. Instead of calling make_bit_field_ref with negative
3630 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
3632 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
3633 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
3634 _mm_maskz_reduce_ss): New.
3635 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
3636 __builtin_ia32_reducess_mask): Ditto..
3637 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
3638 * config/i386/sse.md (reduces<mode>): Renamed to ...
3639 (reduces<mode><mask_scalar_name>): ... this.
3640 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
3642 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
3643 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
3645 2017-10-16 David Malcolm <dmalcolm@redhat.com>
3647 * Makefile.in (OBJS): Add unique-ptr-tests.o.
3648 * selftest-run-tests.c (selftest::run_tests): Call
3649 selftest::unique_ptr_tests_cc_tests.
3650 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
3651 * unique-ptr-tests.cc: New file.
3653 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
3656 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
3658 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
3659 (make_hard_regno_born, make_hard_regno_dead): Update
3660 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
3662 2017-10-16 Jeff Law <law@redhat.com>
3664 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
3666 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
3668 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
3670 2017-10-16 Olivier Hainque <hainque@adacore.com>
3672 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
3673 with_cpu if we were configured for an e500v2 target cpu name.
3675 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
3677 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
3678 * doc/invoke.texi: Document +nodsp as a valid extension for
3681 2017-10-16 Martin Liska <mliska@suse.cz>
3683 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
3684 (test_set_range): Likewise.
3685 (test_range_functions): Rename to ...
3686 (test_bit_in_range): ... this.
3687 (sbitmap_c_tests): Add new test.
3689 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3691 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
3693 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
3694 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
3696 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3698 * config/aarch64/aarch64-builtins.c
3699 (aarch64_types_quadopu_lane_qualifiers): New.
3700 (TYPES_QUADOPU_LANE): New.
3701 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
3702 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
3703 (aarch64_<sur>dot_laneq<vsi2qi>): New.
3704 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
3705 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
3706 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
3707 (Vdottype, DOTPROD): New.
3708 (sur): Add SDOT and UDOT.
3710 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3712 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
3713 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
3714 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
3716 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
3717 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
3718 Enable TARGET_DOTPROD.
3719 (cortex-a75.cortex-a55): Likewise.
3720 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
3722 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3724 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
3725 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
3727 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
3729 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
3730 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
3731 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
3732 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
3733 * config/arm/types.md (neon_dot, neon_dot_q): New.
3734 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
3736 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3738 * config/arm/arm.h (TARGET_DOTPROD): New.
3739 * config/arm/arm.c (arm_arch_dotprod): New.
3740 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
3741 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
3742 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
3743 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
3744 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
3745 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
3746 * doc/invoke.texi (armv8.2-a): Document dotprod
3748 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
3750 * i386.c (ix86_vec_cost): New function.
3751 (ix86_rtx_costs): Handle vector operations better.
3752 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
3753 * x86-tune-costs.h: Add new costs to all tables.
3755 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
3757 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
3759 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
3760 divsd, sqrtss and sqrtsd
3761 * x86-tune-costs.h: Add new entries to all costs.
3762 (znver1_cost): Fix to match real instruction latencies.
3764 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3765 Michael Collison <michael.collison@arm.com>
3767 * compare-elim.c: Include emit-rtl.h.
3768 (can_merge_compare_into_arith): New function.
3769 (try_validate_parallel): Likewise.
3770 (try_merge_compare): Likewise.
3771 (try_eliminate_compare): Call the above when no previous clobber
3773 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
3776 2017-10-14 Jakub Jelinek <jakub@redhat.com>
3780 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
3781 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
3785 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
3786 to be any operand_equal_p operands. For & (B - 1) require
3787 B to be power of 2. Recognize
3788 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
3790 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3793 * optabs.c (expand_memory_blockage): Fix call of
3794 targetm.have_memory_blockage.
3796 2017-10-14 Jakub Jelinek <jakub@redhat.com>
3799 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
3800 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
3801 objects to extra_objs instead of overwriting it.
3803 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3805 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
3806 Use any_fp_register_operand as operand[3] predicate. Simplify
3807 equality test for operands[2] and operands[4] memory location.
3808 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
3809 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
3810 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
3811 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
3812 any_fp_register_operand as operand[1] predicate. Simplify
3813 equality test for operands[0] and operands[3] memory location.
3814 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
3815 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
3816 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
3818 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3820 * target-insns.def: Add memory_blockage.
3821 * optabs.c (expand_memory_blockage): New function.
3822 (expand_asm_memory_barrier): Rename ...
3823 (expand_asm_memory_blockage): ... to this.
3824 (expand_mem_thread_fence): Call expand_memory_blockage
3825 instead of expand_asm_memory_barrier.
3826 (expand_mem_singnal_fence): Ditto.
3827 (expand_atomic_load): Ditto.
3828 (expand_atomic_store): Ditto.
3829 * doc/md.texi (Standard Pattern Names For Generation):
3830 Document memory_blockage instruction pattern.
3832 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
3834 * config/rl78/rl78.c (rl78_emit_libcall): New function.
3835 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
3836 * config/rl78/rl78.md: New define_expand "adddi3".
3838 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
3840 * cfghooks.c (verify_flow_info): Disable check that all probabilities
3843 2017-10-13 Jeff Law <law@redhat.com>
3845 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
3847 2017-10-13 Jakub Jelinek <jakub@redhat.com>
3850 * internal-fn.c (expand_mul_overflow): If both operands have
3851 the same highpart of -1 or 0 and the topmost bit of lowpart
3852 is different, overflow is if res <= 0 rather than res < 0.
3854 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
3856 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
3857 TARGET_P9_VECTOR code for unaligned_load case.
3859 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
3861 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
3863 2017-10-13 Nathan Sidwell <nathan@acm.org>
3865 * tree-core.h (tree_contains_struct): Make bool.
3866 * tree.c (tree_contains_struct): Likewise.
3867 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
3868 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
3869 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
3870 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
3871 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
3873 2017-10-13 Richard Biener <rguenther@suse.de>
3875 * graphite-isl-ast-to-gimple.c
3876 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
3877 parameters and dominance check.
3878 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
3879 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
3880 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
3881 Do not update SSA form here or do intermediate IL verification.
3882 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
3883 (graphite_initialize): Remove check on the number of loops in
3884 the function and inline into graphite_transform_loops.
3885 (graphite_finalize): Inline into graphite_transform_loops.
3886 (graphite_transform_loops): Perform SSA update and IL verification
3888 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
3890 2017-10-13 Richard Biener <rguenther@suse.de>
3892 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
3893 graphite_expression_type_precision): Avoid global constructor
3895 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
3896 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
3897 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
3898 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
3899 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
3900 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
3902 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
3905 * config/i386/i386.h (ix86_red_zone_size): New.
3906 * config/i386/i386.md (push peephole2s): Replace
3907 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
3909 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3910 Alan Hayward <alan.hayward@arm.com>
3911 David Sherwood <david.sherwood@arm.com>
3913 * combine.c (can_change_dest_mode): Reject changes in
3914 REGMODE_NATURAL_SIZE.
3916 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3917 Alan Hayward <alan.hayward@arm.com>
3918 David Sherwood <david.sherwood@arm.com>
3920 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
3921 (expand_debug_source_expr): Likewise.
3922 * combine.c (combine_simplify_rtx): Likewise.
3923 * cse.c (fold_rtx): Likewise.
3924 * fwprop.c (canonicalize_address): Likewise.
3925 * targhooks.c (default_shift_truncation_mask): Likewise.
3927 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3928 Alan Hayward <alan.hayward@arm.com>
3929 David Sherwood <david.sherwood@arm.com>
3931 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
3932 (widened_mode): Likewise.
3933 (expand_unop): Likewise.
3934 * ree.c (transform_ifelse): Likewise.
3935 (merge_def_and_ext): Likewise.
3936 (combine_reaching_defs): Likewise.
3937 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3939 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3940 Alan Hayward <alan.hayward@arm.com>
3941 David Sherwood <david.sherwood@arm.com>
3943 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
3944 * combine.c (gen_lowpart_for_combine): Likewise.
3945 * dwarf2out.c (rtl_for_decl_location): Likewise.
3946 * final.c (alter_subreg): Likewise.
3947 * rtlhooks.c (gen_lowpart_general): Likewise.
3948 (gen_lowpart_if_possible): Likewise.
3950 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3951 Alan Hayward <alan.hayward@arm.com>
3952 David Sherwood <david.sherwood@arm.com>
3954 * calls.c (expand_call): Use subreg_lowpart_offset.
3955 * cse.c (cse_insn): Likewise.
3956 * regcprop.c (copy_value): Likewise.
3957 (copyprop_hardreg_forward_1): Likewise.
3959 2017-10-13 Jakub Jelinek <jakub@redhat.com>
3962 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
3963 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
3964 =Q constraints to +Q and into insn condition add check
3965 that operands[0] and operands[1] are equal.
3966 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
3967 =Q constraints to +Q and into insn condition add check
3968 that operands[0] is equal to either operands[1] or operands[2].
3971 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
3972 instead of handling MINUS_EXPR twice (once for each argument),
3973 canonicalize operand order and handle just once, use rtype where
3974 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
3977 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
3978 any values of __C while still being pattern recognizable as a simple
3981 2017-10-13 Richard Biener <rguenther@suse.de>
3983 PR tree-optimization/82451
3985 2017-10-02 Richard Biener <rguenther@suse.de>
3987 PR tree-optimization/82355
3988 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
3989 a mapping for the enclosing loop but avoid generating one for
3991 (copy_bb_and_scalar_dependences): Remove premature codegen
3992 error on PHIs in blocks duplicated into multiple places.
3993 * graphite-scop-detection.c
3994 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
3995 in the region use it as loop and nest to analyze the DR in.
3996 (try_generate_gimple_bb): Likewise.
3997 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
3998 (add_loop_constraints): For blocks in a loop not in the region
3999 create a dimension with a single iteration.
4000 * sese.h (gbb_loop_at_index): Remove assert.
4002 * cfgloop.c (loop_preheader_edge): For the loop tree root
4003 return the single successor of the entry block.
4004 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4005 Reset the SCEV hashtable and niters.
4006 * graphite-scop-detection.c
4007 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
4008 assert that we only have POLYNOMIAL_CHREC that vary in loops
4009 contained in the region.
4010 (scop_detection::graphite_can_represent_expr): Adjust.
4011 (scop_detection::stmt_has_simple_data_refs_p): For loops
4012 not in the region set loop to NULL. The nest is now the
4013 entry edge to the region.
4014 (try_generate_gimple_bb): Likewise.
4015 * sese.c (scalar_evolution_in_region): Adjust for
4016 instantiate_scev change.
4017 * tree-data-ref.h (graphite_find_data_references_in_stmt):
4018 Make nest parameter the edge into the region.
4019 (create_data_ref): Likewise.
4020 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
4021 entry edge into a region and adjust instantiate_scev calls.
4022 (create_data_ref): Likewise.
4023 (graphite_find_data_references_in_stmt): Likewise.
4024 (find_data_references_in_stmt): Pass the loop preheader edge
4025 from the nest argument.
4026 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
4027 parameter the edge into the region.
4028 (instantiate_parameters): Use the loop preheader edge as entry.
4029 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
4031 (get_instantiated_value_entry): Make instantiate_below parameter
4032 the edge into the region.
4033 (instantiate_scev_name): Likewise. Adjust dominance checks,
4034 when we cannot use loop-based instantiation instantiate by
4035 walking use-def chains.
4036 (instantiate_scev_poly): Adjust.
4037 (instantiate_scev_binary): Likewise.
4038 (instantiate_scev_convert): Likewise.
4039 (instantiate_scev_not): Likewise.
4040 (instantiate_array_ref): Remove.
4041 (instantiate_scev_3): Likewise.
4042 (instantiate_scev_2): Likewise.
4043 (instantiate_scev_1): Likewise.
4044 (instantiate_scev_r): Do not blindly handle N-operand trees.
4045 Do not instantiate array-refs. Handle all constants and invariants.
4046 (instantiate_scev): Make instantiate_below parameter
4047 the edge into the region.
4048 (resolve_mixers): Use the loop preheader edge for the region
4049 parameter to instantiate_scev_r.
4050 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
4052 2017-10-13 Richard Biener <rguenther@suse.de>
4054 PR tree-optimization/82525
4055 * graphite-isl-ast-to-gimple.c
4056 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
4058 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
4059 Fail code generation when we cannot represent the isl integer.
4060 (binary_op_to_tree): Elide modulo operations that are no-ops
4061 in the type we code generate. Remove now superfluous code
4064 2017-10-13 Richard Biener <rguenther@suse.de>
4066 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
4067 (scop_detection::harmful_loop_in_region): Remove premature
4068 IV type restriction.
4069 (scop_detection::graphite_can_represent_scev): We can handle
4070 pointer IVs just fine.
4072 2017-10-13 Alan Modra <amodra@gmail.com>
4074 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
4075 "Clobbers and Scratch Registers". Add paragraph on
4076 alternative to clobbers for scratch registers and OpenBLAS
4079 2017-10-13 Alan Modra <amodra@gmail.com>
4081 * doc/extend.texi (Clobbers): Correct vax example. Delete old
4082 example of a memory input for a string of known length. Move
4083 commentary out of table. Add a number of new examples
4084 covering array memory inputs.
4086 2017-10-12 Martin Liska <mliska@suse.cz>
4088 PR tree-optimization/82493
4089 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
4090 (test_range_functions): New function.
4091 (sbitmap_c_tests): Likewise.
4092 * selftest-run-tests.c (selftest::run_tests): Run new tests.
4093 * selftest.h (sbitmap_c_tests): New function.
4095 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
4097 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
4099 * config/rs6000/amo.h: Fix spacing issue.
4101 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4104 * config/i386/i386.md (*ashl<mode>3_mask_1,
4105 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
4106 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
4109 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4111 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
4112 (profile_probability): Set max_probability
4113 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
4115 * profile-count.c (profile_probability::differs_from_p): Do not
4116 rely on max_probaiblity == 10000
4118 2017-10-12 Jeff Law <law@redhat.com>
4120 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
4123 2017-10-12 Martin Sebor <msebor@redhat.com>
4127 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
4128 (handle_alias_pairs): Call it.
4129 * common.opt (-Wattribute-alias): New option.
4130 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
4131 * doc/invoke.texi (-Wattribute-alias): Document.
4133 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
4136 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4138 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4140 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4141 (make_hard_regno_born, make_hard_regno_dead): Update
4142 bb_killed_pseudos and bb_gen_pseudos.
4144 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4146 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
4148 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
4150 * config/alpha/alpha.c (alpha_split_conditional_move):
4151 Use std::swap instead of manually swapping.
4152 (alpha_stdarg_optimize_hook): Ditto.
4153 (alpha_canonicalize_comparison): Ditto.
4155 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4157 * tree-loop-distribution.c (struct builtin_info): New struct.
4158 (struct partition): Refactor fields into struct builtin_info.
4159 (partition_free): Free struct builtin_info.
4160 (build_size_arg_loc, build_addr_arg_loc): Delete.
4161 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
4162 information from struct builtin_info.
4163 (find_single_drs): New function refactored from classify_partition.
4164 Also moved builtin validity checks to this function.
4165 (compute_access_range, alloc_builtin): New functions.
4166 (classify_builtin_st, classify_builtin_ldst): New functions.
4167 (classify_partition): Refactor code into functions find_single_drs,
4168 classify_builtin_st and classify_builtin_ldst.
4169 (distribute_loop): Don't do runtime alias check when distributing
4171 (find_seed_stmts_for_distribution): New function.
4172 (pass_loop_distribution::execute): Refactor code finding seed
4173 stmts into above function. Support distribution for the innermost
4174 two-level loop nest. Adjust dump information.
4176 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4178 * tree-loop-distribution.c: Adjust the general comment.
4179 (NUM_PARTITION_THRESHOLD): New macro.
4180 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
4181 (classify_partition): Skip builtin pattern of loop nest's inner loop.
4182 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
4183 in call to build_partition_graph.
4184 (finalize_partitions): New parameter. Make loop distribution more
4185 conservative by fusing more partitions.
4186 (distribute_loop): Don't do runtime alias check in case of loop nest
4188 (find_seed_stmts_for_distribution): New function.
4189 (prepare_perfect_loop_nest): New function.
4190 (pass_loop_distribution::execute): Refactor code finding seed stmts
4191 and loop nest into above functions. Support loop nest distribution.
4192 Adjust dump information accordingly.
4194 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4196 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
4197 and set PTYPE_SEQUENTIAL for merged partition.
4199 2017-10-12 Richard Biener <rguenther@suse.de>
4201 PR tree-optimization/69728
4203 2017-09-19 Richard Biener <rguenther@suse.de>
4205 PR tree-optimization/69728
4206 * graphite-sese-to-poly.c (schedule_error): New global.
4207 (add_loop_schedule): Handle empty domain by failing the
4209 (build_original_schedule): Handle schedule_error.
4211 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
4212 domain by returning an unchanged schedule.
4214 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4216 * genrecog.c (validate_pattern): For VEC_SELECT verify that
4217 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
4219 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
4221 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
4224 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4227 * expr.c (store_field): Don't optimize away bitsize == 0 store
4228 from CALL_EXPR with addressable return type.
4230 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4232 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
4233 * config/rs6000/rs6000.md (sel): Delete mode attribute.
4234 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
4235 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
4236 TARGET_ISEL instead of TARGET_ISEL<sel>.
4238 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
4240 * config/rs6000/rs6000.c
4241 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
4243 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4245 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
4247 (reg_or_cint_operand): Fix comment.
4248 (reg_or_zero_operand): New predicate.
4249 * config/rs6000/rs6000-protos.h (output_isel): Delete.
4250 * config/rs6000/rs6000.c (output_isel): Delete.
4251 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
4252 instead of reg_or_cint_operand. Output instruction directly (not via
4254 (isel_unsigned_<mode>): Ditto.
4255 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
4256 gpc_reg_operand. Add an instruction alternative for this. Output
4257 instruction directly.
4258 (*isel_reversed_unsigned_<mode>): Ditto.
4260 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4262 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
4263 (TARGET_CANONICALIZE_COMPARISON): Define.
4265 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
4268 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
4269 Check whether the dest is REG before adding REG_EQUIV note.
4271 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4274 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4276 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4277 (make_hard_regno_born, make_hard_regno_dead): Update
4278 bb_killed_pseudos and bb_gen_pseudos.
4280 2017-10-11 Nathan Sidwell <nathan@acm.org>
4282 * incpath.h (enum incpath_kind): Name enum, prefix values.
4283 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
4284 * incpath.c (heads, tails): Use INC_MAX.
4285 (add_env_var_paths, add_standard_paths): Use incpath_kind.
4286 (merge_include_chains, split_quote_chain,
4287 register_include_chains): Update incpath_kind names.
4288 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
4289 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
4291 (add_framework_path, darwin_register_objc_includes): Likewise.
4292 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
4294 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4296 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
4297 Do not use float_operator operator predicate.
4298 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
4299 * config/i386/predicates.md (float_operator): Remove predicate.
4301 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4303 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
4304 (*jccxf_i387): Ditto.
4305 (*jcc<mode>_i387): Ditto.
4306 (*jccu<mode>_i387): Ditto.
4307 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
4308 (*jcc_*_i387 splitters): Remove.
4309 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
4310 * config/i386/i386.c (ix86_split_fp_branch): Remove.
4311 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
4314 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4316 * profile-count.h (slow_safe_scale_64bit): New function.
4317 (safe_scale_64bit): New inline.
4318 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
4319 * profile-count.c: Include wide-int.h
4320 (slow_safe_scale_64bit): New.
4322 2017-10-11 Nathan Sidwell <nathan@acm.org>
4324 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
4325 HAS_DECL_ASSEMBLER_NAME_P.
4326 * gimple-expr.c (gimple_decl_printable_name: Check
4327 HAS_DECL_ASSEMBLER_NAME_P too.
4328 * ipa-utils.h (type_in_anonymous_namespace_p): Check
4329 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
4330 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
4331 * passes.c (rest_of_decl_compilation): Check
4332 HAS_DECL_ASSEMBLER_NAME_P too.
4333 * recog.c (verify_changes): Likewise.
4334 * tree-pretty-print.c (dump_decl_name): Likewise.
4335 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
4337 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
4338 (SET_DECL_ASSEMBLER_NAME): Use it.
4339 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
4340 (COPY_DECL_ASSEMBLER_NAME): Likewise.
4341 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
4343 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4345 * config.gcc (i386, x86_64): Add extra objects.
4346 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
4347 (ix86_min_insn_size): Declare.
4348 (ix86_issue_rate): Declare.
4349 (ix86_adjust_cost): Declare.
4350 (ia32_multipass_dfa_lookahead): Declare.
4351 (ix86_macro_fusion_p): Declare.
4352 (ix86_macro_fusion_pair_p): Declare.
4353 (ix86_bd_has_dispatch): Declare.
4354 (ix86_bd_do_dispatch): Declare.
4355 (ix86_core2i7_init_hooks): Declare.
4356 (ix86_atom_sched_reorder): Declare.
4357 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
4358 (COSTS_N_BYTES): Move to x86-tune-costs.h.
4359 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
4360 (rip_relative_addr_p): Rename to ...
4361 (ix86_rip_relative_addr_p): ... this one; export.
4362 (memory_address_length): Update.
4363 (ix86_issue_rate): Move to x86-tune-sched.c.
4364 (ix86_flags_dependent): Move to x86-tune-sched.c.
4365 (ix86_agi_dependent): Move to x86-tune-sched.c.
4366 (exact_dependency_1): Move to x86-tune-sched.c.
4367 (exact_store_load_dependency): Move to x86-tune-sched.c.
4368 (ix86_adjust_cost): Move to x86-tune-sched.c.
4369 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
4370 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
4371 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
4372 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
4373 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
4374 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
4375 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
4376 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
4377 (min_insn_size): Rename to ...
4378 (ix86_min_insn_size): ... this one; export.
4379 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
4380 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
4381 (core2i7_first_cycle_multipass_backtrack): Move to
4382 x86-tune-sched-core.c.
4383 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
4384 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
4385 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
4386 (ix86_avoid_jump_mispredicts): Update.
4387 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
4388 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
4389 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
4390 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
4391 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
4392 (MAX_INSN): Move to ix86-tune-sched-bd.c.
4393 (MAX_IMM): Move to ix86-tune-sched-bd.c.
4394 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
4395 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
4396 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
4397 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
4398 (MAX_STORE): Move to ix86-tune-sched-bd.c.
4399 (BIG): Move to ix86-tune-sched-bd.c.
4400 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
4401 (enum insn_path): Move to ix86-tune-sched-bd.c.
4402 (get_mem_group): Move to ix86-tune-sched-bd.c.
4403 (is_cmp): Move to ix86-tune-sched-bd.c.
4404 (dispatch_violation): Move to ix86-tune-sched-bd.c.
4405 (is_branch): Move to ix86-tune-sched-bd.c.
4406 (is_prefetch): Move to ix86-tune-sched-bd.c.
4407 (init_window): Move to ix86-tune-sched-bd.c.
4408 (allocate_window): Move to ix86-tune-sched-bd.c.
4409 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
4410 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
4411 (process_end_window): Move to ix86-tune-sched-bd.c.
4412 (allocate_next_window): Move to ix86-tune-sched-bd.c.
4413 (find_constant): Move to ix86-tune-sched-bd.c.
4414 (get_num_immediates): Move to ix86-tune-sched-bd.c.
4415 (has_immediate): Move to ix86-tune-sched-bd.c.
4416 (get_insn_path): Move to ix86-tune-sched-bd.c.
4417 (get_insn_group): Move to ix86-tune-sched-bd.c.
4418 (count_num_restricted): Move to ix86-tune-sched-bd.c.
4419 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
4420 (add_insn_window): Move to ix86-tune-sched-bd.c.
4421 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
4422 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
4423 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
4424 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
4425 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
4426 (do_dispatch): Move to ix86-tune-sched-bd.c.
4427 (has_dispatch): Move to ix86-tune-sched-bd.c.
4428 * config/i386/t-i386: Add new object files.
4429 * config/i386/x86-tune-costs.h: New file.
4430 * config/i386/x86-tune-sched-atom.c: New file.
4431 * config/i386/x86-tune-sched-bd.c: New file.
4432 * config/i386/x86-tune-sched-core.c: New file.
4433 * config/i386/x86-tune-sched.c: New file.
4435 2017-10-11 Liu Hao <lh_mouse@126.com>
4437 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
4438 the generic version below instead.
4439 (should_colorize): Recognize Windows consoles as terminals
4441 * pretty-print.c [__MINGW32__] (write_all): New function.
4442 [__MINGW32__] (find_esc_head): Likewise.
4443 [__MINGW32__] (find_esc_terminator): Likewise.
4444 [__MINGW32__] (eat_esc_sequence): Likewise.
4445 [__MINGW32__] (mingw_ansi_fputs): New function that handles
4447 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
4450 2017-10-11 Richard Biener <rguenther@suse.de>
4452 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
4453 Properly call analyze_scalar_evolution with the loop of the stmt.
4455 2017-10-11 Richard Biener <rguenther@suse.de>
4457 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
4458 * tree-core.h (tree_base): Add chrec_var union member.
4459 * tree.h (CHREC_VAR): Remove.
4460 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
4461 * tree-chrec.h (build_polynomial_chrec): Adjust.
4462 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
4463 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
4465 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
4467 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
4468 * match.pd: ... here.
4469 ((T) X == (T) Y): Relax condition.
4471 2017-10-11 Bin Cheng <bin.cheng@arm.com>
4473 PR tree-optimization/82472
4474 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
4476 (break_alias_scc_partitions): Update postorder number.
4478 2017-10-11 Martin Liska <mliska@suse.cz>
4481 * opts.c (parse_no_sanitize_attribute): Do not use error_value
4483 * opts.h (parse_no_sanitize_attribute): Remove last argument.
4485 2017-10-11 Martin Liska <mliska@suse.cz>
4487 * print-rtl.c (print_insn): Move declaration of idbuf
4488 to same scope as name.
4490 2017-10-11 Martin Liska <mliska@suse.cz>
4495 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
4498 2017-10-11 Martin Liska <mliska@suse.cz>
4501 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
4504 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
4506 * config.gcc (default_gnu_indirect_function): Default to yes for
4507 arm*-*-linux* with glibc.
4509 2017-10-11 Richard Biener <rguenther@suse.de>
4511 * tree-scalar-evolution.c (get_scalar_evolution): Handle
4512 default-defs and types we do not want to analyze.
4513 (interpret_loop_phi): Replace unreachable code with an assert.
4514 (compute_scalar_evolution_in_loop): Remove and inline ...
4515 (analyze_scalar_evolution_1): ... here, replacing condition with
4516 what makes the intent clearer. Remove handling of cases
4517 get_scalar_evolution now handles.
4519 2017-10-10 Jim Wilson <wilson@tuliptree.org>
4521 PR rtl-optimization/81434
4522 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
4523 comment for main loop. In sched_group_found if, also add checks for
4524 pass and min_cost_group.
4526 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
4528 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
4529 (rs6000_insn_cost): New function.
4530 * config/rs6000/rs6000.md (cost): New attribute.
4532 2017-10-10 Jakub Jelinek <jakub@redhat.com>
4533 H.J. Lu <hongjiu.lu@intel.com>
4537 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
4538 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
4539 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
4540 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
4541 to OPTION_MASK_ISA_AVX512VL - builtins that have both
4542 OPTION_MASK_ISA_MMX and some other bit set require both
4543 mmx and the ISAs without the mmx bit.
4544 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
4545 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
4546 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
4547 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
4548 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
4549 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
4550 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
4551 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
4552 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
4553 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
4554 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
4555 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
4556 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
4557 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
4558 Add OPTION_MASK_ISA_MMX.
4560 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
4562 * config.gcc (armv7*-*-freebsd*): New target.
4563 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
4565 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
4567 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
4568 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
4571 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
4573 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
4574 a template parameter.
4575 (WIDE_INT_REF_FOR): Update accordingly.
4576 * tree.h (wi::int_traits <const_tree>): Delete.
4577 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
4578 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
4579 (wi::tree_to_wide_ref): New typedef.
4580 (wi::to_wide): New function.
4581 * calls.c (get_size_range): Use wi::to_wide when operating on
4583 * cgraph.c (cgraph_node::create_thunk): Likewise.
4584 * config/i386/i386.c (ix86_data_alignment): Likewise.
4585 (ix86_local_alignment): Likewise.
4586 * dbxout.c (stabstr_O): Likewise.
4587 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
4588 * expr.c (const_vector_from_tree): Likewise.
4589 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
4590 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
4591 (fold_negate_expr_1, int_const_binop_1, const_binop)
4592 (fold_convert_const_int_from_real, optimize_bit_field_compare)
4593 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
4594 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
4595 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
4596 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
4597 (fold_not_const, round_up_loc): Likewise.
4598 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
4599 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
4600 (alloca_call_type): Likewise.
4601 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
4602 * godump.c (go_output_typedef): Likewise.
4603 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
4604 * internal-fn.c (get_min_precision): Likewise.
4605 * ipa-cp.c (ipcp_store_vr_results): Likewise.
4606 * ipa-polymorphic-call.c
4607 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
4608 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
4609 (ipa_modify_call_arguments): Likewise.
4610 * match.pd: Likewise.
4611 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
4612 * print-tree.c (print_node_brief, print_node): Likewise.
4613 * stmt.c (expand_case): Likewise.
4614 * stor-layout.c (layout_type): Likewise.
4615 * tree-affine.c (tree_to_aff_combination): Likewise.
4616 * tree-cfg.c (group_case_labels_stmt): Likewise.
4617 * tree-data-ref.c (dr_analyze_indices): Likewise.
4618 (prune_runtime_alias_test_list): Likewise.
4619 * tree-dump.c (dequeue_and_dump): Likewise.
4620 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
4621 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
4622 * tree-pretty-print.c (dump_generic_node): Likewise.
4623 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
4624 (simple_iv_with_niters): Likewise.
4625 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
4626 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
4627 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
4628 * tree-ssa-loop-niter.c (split_to_var_and_offset)
4629 (refine_value_range_using_guard, number_of_iterations_ne_max)
4630 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
4631 (get_cst_init_from_scev, record_nonwrapping_iv)
4632 (scev_var_range_cant_overflow): Likewise.
4633 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
4634 * tree-ssa-pre.c (compute_avail): Likewise.
4635 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
4636 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
4637 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
4638 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
4639 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
4640 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
4641 (dump_case_nodes, try_switch_expansion): Likewise.
4642 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
4643 (vect_do_peeling): Likewise.
4644 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
4645 * tree-vect-stmts.c (vectorizable_load): Likewise.
4646 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
4647 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
4648 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
4649 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
4650 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
4651 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
4652 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
4653 (evrp_dom_walker::before_dom_children): Likewise.
4654 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
4655 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
4656 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
4657 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
4658 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
4659 (get_range_pos_neg): Likewise.
4660 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
4661 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
4662 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
4663 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
4664 * config/avr/avr.c (avr_fold_builtin): Likewise.
4665 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
4666 * config/msp430/msp430.c (msp430_attr): Likewise.
4667 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
4668 * config/powerpcspe/powerpcspe-c.c
4669 (altivec_resolve_overloaded_builtin): Likewise.
4670 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
4671 (rs6000_expand_ternop_builtin): Likewise.
4672 * config/rs6000/rs6000-c.c
4673 (altivec_resolve_overloaded_builtin): Likewise.
4674 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
4675 (rs6000_expand_ternop_builtin): Likewise.
4676 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
4678 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4680 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
4681 when copying loop nest with only one inner loop.
4683 2017-10-10 Richard Biener <rguenther@suse.de>
4685 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
4686 blocks if SCEV is active.
4687 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
4689 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
4690 (scev_initialize): Assert we are not yet initialized.
4692 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4694 * tree-loop-distribution.c (generate_loops_for_partition): Remove
4695 inner loop's exit stmt by making it always exit the loop, otherwise
4696 we would generate an infinite empty loop.
4698 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4700 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
4701 renaming variables in new preheader if it's deleted.
4703 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4705 * tree-loop-distribution.c (struct partition): Remove unused field
4706 loops of the structure.
4707 (partition_alloc, partition_free): Ditto.
4708 (build_rdg_partition_for_vertex): Ditto.
4710 2017-10-09 Jeff Law <law@redhat.com>
4712 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
4713 return type to match prototype and documentation.
4715 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4717 * config/rs6000/rs6000.c (processor_costs): Move to ...
4718 * config/rs6000/rs6000.h: ... here.
4719 (rs6000_cost): Declare.
4721 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
4723 * except.c (setjmp_fn): New global variable.
4724 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
4725 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
4726 if DONT_USE_BUILTIN_SETJMP is defined.
4728 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4730 * target.def (insn_cost): New hook.
4731 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
4732 * doc/tm.texi: Regenerate.
4733 * rtlanal.c (insn_cost): Use the new hook.
4735 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4737 * combine.c (combine_validate_cost): Compute the new insn_cost,
4738 not just pattern_cost.
4739 (try_combine): Adjust comment.
4741 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4743 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
4745 * combine.c (uid_insn_cost): Adjust comment.
4746 (combine_validate_cost): Adjust comment. Use pattern_cost instead
4748 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
4749 * dse.c (find_shift_sequence): Ditto.
4750 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
4751 (bb_valid_for_noce_process_p): Use pattern_cost.
4752 * rtl.h (insn_rtx_cost): Delete.
4753 (pattern_cost): New prototype.
4754 (insn_cost): New prototype.
4755 * rtlanal.c (insn_rtx_cost): Rename to...
4756 (pattern_cost): ... this.
4759 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
4761 * config/i386/i386.md (*jcc_2): Remove insn pattern.
4762 (*jcc<mode>_0_r_i387): Ditto.
4763 (*jccxf_r_i387): Ditto.
4764 (*jcc<mode>_r_i387): Ditto.
4765 (*jccu<mode>_r_i387): Ditto.
4766 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
4767 (*jcc): Rename from *jcc_1.
4769 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4771 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
4772 deferred rescans after the lvx/stvx recombination pre-pass.
4774 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
4776 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
4777 memory operation instruction support.
4778 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
4779 (rs6000-ibm-aix[789]*): Likewise.
4780 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
4781 Document new functions.
4783 2017-10-09 Richard Biener <rguenther@suse.de>
4785 PR tree-optimization/82397
4786 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
4787 equality only for semantically equal trees.
4789 2017-10-09 Richard Biener <rguenther@suse.de>
4791 PR tree-optimization/82449
4792 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
4793 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
4794 allow constant addresses.
4795 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
4798 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4800 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
4803 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4806 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
4809 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4812 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
4814 2017-10-09 Jakub Jelinek <jakub@redhat.com>
4817 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
4818 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
4820 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
4822 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
4823 (WI_BINARY_PREDICATE_RESULT): Likewise.
4824 (wi::binary_traits::operator_result): New type.
4825 (wi::binary_traits::predicate_result): Likewise.
4826 (generic_wide_int::operator~, unary generic_wide_int::operator-)
4827 (generic_wide_int::operator==, generic_wide_int::operator!=)
4828 (generic_wide_int::operator&, generic_wide_int::and_not)
4829 (generic_wide_int::operator|, generic_wide_int::or_not)
4830 (generic_wide_int::operator^, generic_wide_int::operator+
4831 (binary generic_wide_int::operator-, generic_wide_int::operator*):
4833 (operator~, unary operator-, operator==, operator!=, operator&)
4834 (operator|, operator^, operator+, binary operator-, operator*): New
4836 * expr.c (get_inner_reference): Use wi::bit_and_not.
4837 * fold-const.c (fold_binary_loc): Likewise.
4838 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
4839 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
4840 (bit_value_binop): Likewise.
4841 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
4842 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
4843 (extract_range_from_binary_expr_1): Likewise.
4844 (masked_increment): Likewise.
4845 (simplify_bit_ops_using_ranges): Likewise.
4847 2017-10-09 Martin Jambor <mjambor@suse.cz>
4850 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
4851 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
4852 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
4854 (hsa_fixup_mov_insn_type): New function.
4855 (hsa_op_with_type::get_in_type): Use it.
4856 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
4857 immediates in an assert.
4858 (hsa_op_with_type::extend_int_to_32bit): New method.
4859 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
4860 types. Convert to dest type if necessary.
4861 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
4862 (reg_for_gimple_ssa): Pass false as min32int to
4863 hsa_type_for_scalar_tree_type.
4864 (gen_hsa_addr): Fixup type when creating addresable temporary.
4865 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
4866 (gen_hsa_unary_operation): Extend operands and convert to dest type if
4867 necessary. Call hsa_fixup_mov_insn_type.
4868 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
4869 extend operands and convert to dest type if necessary.
4870 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
4871 to dest type if necessary.
4872 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
4873 if conversion nt necessary and size matches.
4874 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
4875 to dest type if necessary.
4876 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
4877 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
4879 (gen_hsa_clrsb): Likewise.
4880 (gen_hsa_ffs): Likewise.
4881 (gen_hsa_divmod): Extend operands and convert to dest type if
4883 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
4885 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
4887 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
4888 Remove empty default arguments. Use a brace block as output
4890 (conditional return): Ditto.
4892 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
4893 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
4895 (group_ending_nop): Ditto.
4896 (doloop_end): Ditto.
4897 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
4898 (splitters for those): Ditto.
4900 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
4902 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
4903 a conditional jump (and the compare for it) so that pc_rtx is the
4905 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
4906 for the deleted and renamed ctr<mode>_internal[234] patterns.
4907 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
4908 Delete second conditional return pattern.
4909 (ctr<mode>_internal2): Delete this second bdnz pattern.
4910 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
4911 (ctr<mode>_internal4): Delete this second bdz pattern.
4913 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
4915 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
4916 (always_initialized_rtx_for_ssa_name_p): New predicate.
4917 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
4918 (finish_out_of_ssa): Free new field of SA.
4919 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
4920 * tree-ssa-coalesce.c: Include tree-ssa.h.
4921 (get_parm_default_def_partitions): Remove extern keyword.
4922 (get_undefined_value_partitions): New function.
4923 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
4924 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
4926 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
4928 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
4930 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
4932 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
4934 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
4935 for targets that preffer 128bit.
4937 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
4939 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
4941 2017-10-08 Olivier Hainque <hainque@adacore.com>
4943 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
4944 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
4946 (thumb_set_return_address): Likewise.
4948 2017-10-08 Olivier Hainque <hainque@adacore.com>
4950 * common/config/arm/arm-common.c (arm_except_unwind_info):
4951 Handle DWARF2_UNWIND_INFO.
4953 2017-10-07 Michael Collison <michael.collison@arm.com>
4955 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
4958 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
4960 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
4961 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
4962 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
4963 defined, force the creation of a new block for a dispatch label.
4965 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
4967 * invoke.texi (Wsuggest-attribute=cold): Document.
4968 * common.opt (Wsuggest-attribute=cold): New
4969 * ipa-pure-const.c (warn_function_cold): New function.
4970 * predict.c (compute_function_frequency): Use it.
4971 * predict.h (warn_function_cold): Declare.
4973 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
4975 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
4978 2017-10-06 Martin Liska <mliska@suse.cz>
4980 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
4981 keyword for member functions.
4982 (struct sanopt_tree_couple): New struct.
4983 (struct sanopt_tree_couple_hash): New function.
4984 (struct sanopt_ctx): Add new hash_map.
4985 (has_dominating_ubsan_ptr_check): New function.
4986 (record_ubsan_ptr_check_stmt): Likewise.
4987 (maybe_optimize_ubsan_ptr_ifn): Likewise.
4988 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
4989 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
4991 2017-10-06 Sudakshina Das <sudi.das@arm.com>
4994 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
4995 aarch64_simd_valid_immediate on CONST_VECTORs.
4996 (aarch64_reg_or_bic_imm): Likewise.
4998 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
5000 PR rtl-optimization/82396
5001 * haifa-sched.c (ready_sort_real): Disable qsort checking.
5003 2017-10-06 Sebastian Pop <sebpop@gmail.com>
5005 * graphite-dependences.c (scop_get_reads): Move code to...
5006 (scop_get_must_writes): Move code to...
5007 (scop_get_may_writes): Move code to...
5008 (scop_get_reads_and_writes): ... here.
5009 (scop_get_dependences): Call scop_get_reads_and_writes.
5011 2017-10-06 Jakub Jelinek <jakub@redhat.com>
5013 PR tree-optimization/82434
5014 * fold-const.h (can_native_encode_type_p,
5015 can_native_encode_string_p): Remove.
5016 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
5017 don't encode anything, just return what would be otherwise returned.
5018 (native_encode_fixed, native_encode_complex, native_encode_vector):
5020 (native_encode_string): Likewise. Inline by hand
5021 can_native_encode_string_p.
5022 (can_native_encode_type_p): Remove.
5023 (can_native_encode_string_p): Remove.
5024 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
5025 STRING_CSTs using can_native_encode_string_p, test all
5026 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
5027 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
5028 argument from native_encode_expr.
5029 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
5030 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
5033 2017-10-06 Richard Biener <rguenther@suse.de>
5035 PR tree-optimization/82397
5036 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
5037 operand_equal_p but rely on data_ref_compare_tree for detecting
5039 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
5040 to match up with dr_group_sort_cmp.
5042 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5045 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
5047 * config/s390/s390-builtin-types.def: Regenerate.
5049 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5052 * config/s390/s390-builtin-types.def: Regenerate.
5053 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
5054 Change flag from B_VXE to B_VX.
5055 (s390_vec_min_dbl): Remove B_VXE flag.
5057 2017-10-06 Richard Biener <rguenther@suse.de>
5059 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
5060 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
5061 translate_isl_ast_to_gimple::is_valid_rename,
5062 translate_isl_ast_to_gimple::get_rename,
5063 translate_isl_ast_to_gimple::get_def_bb_for_const,
5064 translate_isl_ast_to_gimple::get_new_name,
5065 translate_isl_ast_to_gimple::collect_all_ssa_names,
5066 translate_isl_ast_to_gimple::copy_loop_phi_args,
5067 translate_isl_ast_to_gimple::collect_all_ssa_names,
5068 translate_isl_ast_to_gimple::copy_loop_phi_args,
5069 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
5070 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
5071 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
5072 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
5073 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
5074 translate_isl_ast_to_gimple::copy_cond_phi_args,
5075 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
5076 translate_isl_ast_to_gimple::edge_for_new_close_phis,
5077 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
5078 translate_isl_ast_to_gimple::rename_uses,
5079 translate_isl_ast_to_gimple::rename_all_uses): Remove.
5080 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
5081 (set_rename_for_each_def): Likewise.
5082 (graphite_copy_stmts_from_block): Handle debug stmt resetting
5083 here. Handle rewriting SCEV analyzable uses here.
5084 (copy_bb_and_scalar_dependences): Generate code for PHI
5086 (graphite_regenerate_ast_isl): Adjust.
5087 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
5088 (add_write, add_read): New functions.
5089 (build_cross_bb_scalars_def): Use it and simplify.
5090 (build_cross_bb_scalars_use): Likewise.
5091 (graphite_find_cross_bb_scalar_vars): Inline into...
5092 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
5093 simulating out-of-SSA. Compute liveout and add dependencies.
5094 (build_scops): Force an empty entry block.
5095 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
5097 (sese_build_liveouts): Declare.
5098 (sese_trivially_empty_bb_p): Likewise.
5099 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
5100 compute liveout and debug_liveout.
5101 (sese_bad_liveouts_use): Remove.
5102 (sese_reset_debug_liveouts_bb): Likewise.
5103 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
5104 (sese_build_liveouts): Build liveout and debug_liveout and store
5106 (new_sese_info): Adjust.
5107 (free_sese_info): Likewise.
5108 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
5109 do not build liveout here.
5110 (move_sese_in_condition): Adjust region entry.
5111 (scev_analyzable_p): Match up with chrec_apply requirements.
5112 (sese_trivially_empty_bb_p): New.
5113 * tree-into-ssa.c (get_reaching_def): Properly support generating
5114 default-defs for incremental rewrite of anonymous names.
5116 2017-10-06 Richard Biener <rguenther@suse.de>
5118 * graphite-sese-to-poly.c (extract_affine): For casts increasing
5119 precision do not perform modulo reduction.
5121 2017-10-06 Richard Biener <rguenther@suse.de>
5123 PR tree-optimization/82436
5124 * tree-vect-slp.c (vect_supported_load_permutation_p): More
5125 conservatively choose the vectorization factor when checking
5126 whether we can perform the required load permutation.
5127 (vect_transform_slp_perm_load): Assert when we may not fail.
5129 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
5131 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
5132 message for incompatible -msdata=* and -mcall-* options.
5134 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5136 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
5137 rate for post-reload scheduling.
5139 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5141 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
5143 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5145 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
5146 to improve monte carlo in scimark.
5148 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5150 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
5151 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
5152 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
5153 pentium4_cost, nocona_cost): Set reassociation width to 1.
5154 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
5155 width to 2 for fp operations and 1 otherwise.
5156 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
5158 (atom_cost): Set reassociation width to 2.
5159 (slm_cost, generic_cost): Set fp reassociation width
5160 to 2 and 1 otherwise.
5161 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
5162 (core_cost): Set fp reassociation width to 4 and vector to 2.
5163 (ix86_reassociation_width): Rewrite using cost table; special case
5164 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
5165 and TARGET_AVX128_OPTIMAL.
5166 * config/i386/i386.h (processor_costs): Add
5167 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
5168 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
5169 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
5170 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
5171 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
5172 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
5174 2017-10-05 Nathan Sidwell <nathan@acm.org>
5176 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
5178 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5180 * config/arm/arm.c (arm_test_fpu_data): New.
5181 (arm_run_selftests): Call arm_test_fpu_data.
5183 2017-10-04 Nathan Sidwell <nathan@acm.org>
5185 * toplev.c (toplev::main): Remove excess parens on pretty_printer
5187 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
5189 2017-10-04 Sudakshina Das <sudi.das@arm.com>
5191 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
5192 check type for aarch64_simd_valid_immediate.
5193 (aarch64_output_simd_mov_immediate): Update prototype.
5194 (aarch64_simd_valid_immediate): Update prototype.
5195 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
5196 support for ORR-immediate.
5197 (and<mode>3): modified pattern to add support for BIC-immediate.
5198 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
5199 now checks for valid immediate for BIC and ORR based on new enum
5201 (aarch64_output_simd_mov_immediate): Function now used to output
5202 BIC/ORR imm as well based on new enum argument.
5203 * config/aarch64/constraints.md (Do): New vector immediate constraint.
5205 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
5206 (aarch64_reg_or_bic_imm): Likewise.
5208 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5210 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
5211 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
5213 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5217 PR rtl-optimization/82396
5218 * haifa-sched.c (autopref_multipass_init): Simplify
5220 (autopref_rank_data): Simplify sort order.
5221 * sched-int.h (autopref_multipass_data_): Remove
5222 multi_mem_insn_p, min_offset and max_offset.
5224 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5226 * doc/sourcebuild.texi: Document vect_peeling_profitable.
5228 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5230 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
5233 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5235 * doc/sourcebuild.texi: Document vect_long_mult.
5237 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
5239 PR tree-optimization/82413
5240 * fold-const.c (build_range_check): Use widest_int when comparing
5241 the maximum ETYPE value with HIGH.
5243 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5245 PR rtl-optimization/82396
5246 * haifa-sched.c (autopref_multipass_init): Simplify
5248 (autopref_rank_data): Simplify sort order.
5249 * sched-int.h (autopref_multipass_data_): Remove
5250 multi_mem_insn_p, min_offset and max_offset.
5252 2017-10-04 Jakub Jelinek <jakub@redhat.com>
5254 PR tree-optimization/82381
5255 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
5256 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
5259 PR tree-optimization/82374
5260 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
5261 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
5262 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
5263 current_function_decl to the new decl.
5265 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
5267 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
5268 helper macro for IEEE float128 hardware built-in functions.
5269 (SQRTF128_ODD): Add built-in functions with the round-to-odd
5271 (TRUNCF128_ODD): Likewise.
5272 (ADDF128_ODD): Likewise.
5273 (SUBF128_ODD): Likewise.
5274 (MULF128_ODD): Likewise.
5275 (DIVF128_ODD): Likewise.
5276 (FMAF128_ODD): Likewise.
5277 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
5278 UNSPEC_TRUNC_ROUND_TO_ODD.
5279 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
5280 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
5281 floating point round to odd instructions.
5282 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
5283 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
5284 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
5285 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
5286 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
5287 (trunc<mode>sf2_hw): Change the truncate with round to odd
5288 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
5289 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
5290 to odd hardware instructions.
5291 (sub<mode>3_odd): Likewise.
5292 (mul<mode>3_odd): Likewise.
5293 (div<mode>3_odd): Likewise.
5294 (sqrt<mode>2_odd): Likewise.
5295 (fma<mode>4_odd): Likewise.
5296 (fms<mode>4_odd): Likewise.
5297 (nfma<mode>4_odd): Likewise.
5298 (nfms<mode>4_odd): Likewise.
5299 (trunc<mode>df2_odd): Change the truncate with round to odd
5300 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
5302 * doc/extend.texi (PowerPC built-in functions): Update documentation
5303 for existing IEEE float128-bit built-in functions. Add built-in
5304 functions that generate the IEEE 128-bit floating point round to
5307 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
5309 PR rtl-optimization/77729
5310 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
5311 to (X&(C1&~C2))|C2 transformations.
5313 2017-10-03 Martin Jambor <mjambor@suse.cz>
5315 PR tree-optimization/82363
5316 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
5317 mismatch, mark lacc written regardless of racc.
5319 2017-10-03 Jakub Jelinek <jakub@redhat.com>
5321 PR tree-optimization/82381
5322 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
5323 stmt_to_insert nor wheather SSA_NAMEs are default defs.
5324 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
5325 fallthrough into reassoc_stmt_dominates_stmt_p.
5328 * combine.c (combine_instructions): Don't combine in unreachable
5331 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
5334 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
5335 function to not use the have_cpu variable. Do not set cpu_index,
5336 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
5338 (rs6000_valid_attribute_p): Remove duplicate initializations of
5339 old_optimize and func_optimize.
5340 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
5341 (rs6000_activate_target_options): Make global.
5342 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
5345 2017-10-02 Jakub Jelinek <jakub@redhat.com>
5347 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
5348 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
5349 Set *poffset to 0 and *psize and *pmax_size to -1 if
5350 *poffset + *psize overflows in HOST_WIDE_INT.
5352 PR tree-optimization/82387
5353 PR tree-optimization/82388
5354 PR tree-optimization/82389
5355 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
5356 instead of live_bytes non-NULL.
5358 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
5361 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
5364 2017-10-02 Richard Biener <rguenther@suse.de>
5366 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
5367 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
5368 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
5370 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
5372 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
5373 requested precision matches the type's.
5374 * calls.c (alloc_max_size): Calculate the new candidate size as
5375 a widest_int and use wi::to_widest when comparing it with the
5376 current candidate size.
5377 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
5378 zero rather than integer_zero_node.
5379 * match.pd: Check for a no-op conversion before using wi::add
5380 rather than after. Use tree_to_uhwi when summing small shift
5381 counts into an unsigned int.
5383 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
5384 Alan Hayward <alan.hayward@arm.com>
5385 David Sherwood <david.sherwood@arm.com>
5388 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
5389 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
5390 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
5391 POINTER_AND_FP_REGS.
5393 2017-10-02 Richard Biener <rguenther@suse.de>
5395 PR tree-optimization/82355
5396 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5397 a mapping for the enclosing loop but avoid generating one for
5399 (copy_bb_and_scalar_dependences): Remove premature codegen
5400 error on PHIs in blocks duplicated into multiple places.
5401 * graphite-scop-detection.c
5402 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5403 in the region use it as loop and nest to analyze the DR in.
5404 (try_generate_gimple_bb): Likewise.
5405 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5406 (add_loop_constraints): For blocks in a loop not in the region
5407 create a dimension with a single iteration.
5408 * sese.h (gbb_loop_at_index): Remove assert.
5410 2017-10-01 Kevin Buettner <kevinb@redhat.com>
5412 * omp-expand.c (adjust_context_scope): New function.
5413 (expand_parallel_call): Call adjust_context_scope.
5415 2017-10-01 Jeff Law <law@redhat.com>
5417 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
5418 dom_opt_dom_walker class with direct access to private members.
5419 Add comments. Call test_for_singularity.
5420 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
5421 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
5422 m_dummy_cond anymore.
5423 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
5425 (pass_dominator:execute): Build the dummy_cond here and pass it
5426 to the dom_opt_dom_walker ctor.
5427 (test_for_singularity): New function.
5429 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
5430 Maya Rashish <coypu@sdf.org>
5432 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
5433 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
5434 (alpha*-*-netbsd*) Use nbsd_tm_file.
5435 (arm*-*-netbsdelf*) Likewise.
5436 (i[34567]86-*-netbsdelf*) Likewise.
5437 (x86_64-*-netbsd*) Likewise.
5438 (mips*-*-netbsd*) Likewise.
5439 (powerpc-*-netbsd*) Likewise.
5440 (sh*-*-netbsd*) Likewise.
5441 (sparc-*-netbsdelf*) Likewise.
5442 (sparc64-*-netbsd*) Likewise.
5443 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
5445 (vax-*-netbsdelf*) Likewise.
5446 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
5447 (UINT_FAST8_TYPE) Likewise.
5448 (INT_FAST16_TYPE) Check CHAR_FAST16.
5449 (UINT_FAST16_TYPE) Likewise.
5451 2017-09-30 Jakub Jelinek <jakub@redhat.com>
5454 * config/i386/i386.md
5455 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
5456 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
5457 *divmodsi4_zext_2): New define_insn_and_split.
5458 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
5459 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
5460 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
5461 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
5462 New define_insn_and_split.
5463 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
5464 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
5465 operands[1] having DImode when mode is SImode.
5467 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
5468 always SImode for DIV and MOD in REG_EQUAL notes.
5470 2017-09-29 Yury Gribov <tetra2005@gmail.com>
5473 * match.pd: Fix handling of NaNs in pattern.
5475 2017-09-29 Jeff Law <law@redhat.com>
5477 * sbitmap.c (bitmap_bit_in_range_p): New function.
5478 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
5479 * tree-ssa-dse.c (live_bytes_read): New function.
5480 (dse_classify_store): Ignore reads of dead bytes.
5482 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
5483 typos and whitespace errors.
5484 * config/i386/predicates.md (address_no_seg_operand): Likewise.
5485 * config/s390/s390.c (s390_emit_prologue): Likewise.
5487 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
5490 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
5491 with a symbol for LRA.
5493 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
5495 PR rtl-optimization/82338
5496 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
5498 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
5500 * genmodes.c (calc_wider_mode): Suppress qsort macro.
5501 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
5502 (qsort_chk): Declare.
5503 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
5504 (qsort_chk): New function.
5506 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5508 PR tree-optimization/82337
5509 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
5510 phi definition if the PHI result appears in an abnormal PHI.
5511 (find_basis_for_base_expr): Don't record a basis if the LHS of the
5512 basis appears in an abnormal PHI.
5514 2017-09-29 Richard Biener <rguenther@suse.de>
5516 * graphite-isl-ast-to-gimple.c
5517 (translate_isl_ast_to_gimple::set_codegen_error): New function.
5518 (binary_op_to_tree): Use it.
5519 (get_rename_from_scev): Likewise.
5520 (copy_loop_phi_nodes): Likewise.
5521 (copy_bb_and_scalar_dependences): Likewise.
5522 (translate_pending_phi_nodes): Likewise.
5524 2017-09-29 Jakub Jelinek <jakub@redhat.com>
5527 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
5528 for movabsq $(i32 << shift), r64.
5530 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
5532 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
5533 index when encoding %esp as %rsp to avoid 0x67 prefix.
5535 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
5537 * config/i386/i386.md (*movsf_internal, *movdf_internal):
5538 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
5540 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
5542 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
5543 Extensions with more than 16 double VFP registers.
5544 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
5545 to_clear_mask and all code related to it. Replace the remaining
5546 entry by a sbitmap and adapt code accordingly.
5548 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
5550 * brig-builtins.def: Change pure attributes to const.
5552 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5554 * config.gcc (default_gnu_indirect_function): Default to yes for
5555 sparc*-*-linux* with glibc.
5557 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5559 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
5560 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
5561 when creating .init_array and .fini_array sections with priority
5564 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
5567 * config/aarch64/aarch64.c
5568 (aarch64_builtin_support_vector_misalignment): Always return false
5569 when misalignment is unknown.
5571 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
5573 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
5574 this function to return false if the definition used by the swap
5575 instruction is artificial, or if the memory address from which the
5576 constant value is loaded is not represented by a base address held
5577 in a register or if the base address register is a frame or stack
5578 pointer. Additionally, return false if the base address of the
5579 loaded constant is a SYMBOL_REF but is not considered to be a
5581 (replace_swapped_load_constant): New function.
5582 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
5583 loaded constant vector with a load of a swapped constant vector.
5585 2017-09-27 Carl Love <cel@us.ibm.com>
5587 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
5588 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
5589 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
5592 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
5594 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
5595 first, always call autopref_rank_data otherwise.
5597 2017-09-27 Richard Biener <rguenther@suse.de>
5599 * graphite-scop-detection.c (find_scop_parameters): Move
5600 loop bound handling ...
5601 (gather_bbs::before_dom_children): ... here, avoiding the need
5602 to build scop_info->loop_nest.
5603 (record_loop_in_sese): Remove.
5604 * sese.h (sese_info_t::loop_nest): Remove.
5605 * sese.c (new_sese_info): Do not allocate loop_nest.
5606 (free_sese_info): Do not free loop_nest.
5608 2017-09-27 Jakub Jelinek <jakub@redhat.com>
5611 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
5612 lhs from calls if the lhs has addressable type.
5614 2017-09-27 Richard Biener <rguenther@suse.de>
5616 * graphite.h (scop::max_alias_set): New member.
5617 * graphite-scop-detection.c: Remove references to non-existing
5618 --param in comments.
5619 (build_alias_sets): Record the maximum alias set used for drs.
5620 (build_scops): Support zero as unlimited for
5621 --param graphite-max-arrays-per-scop.
5622 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
5624 (build_poly_sr_1): ... here. Compute alias set based on the
5625 maximum alias set used for drs rather than
5626 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
5628 2017-09-27 Richard Biener <rguenther@suse.de>
5630 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
5631 --param loop-block-tile-size=0 to disable tiling.
5633 2017-09-27 Richard Biener <rguenther@suse.de>
5635 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
5636 (graphite-max-nb-scop-params): Document special value zero.
5637 * domwalk.h (dom_walker::STOP): New symbolical constant.
5638 (dom_walker::dom_walker): Add optional parameter for bb to
5640 (dom_walker::~dom_walker): Declare.
5641 (dom_walker::before_dom_children): Document STOP return value.
5642 (dom_walker::m_user_bb_to_rpo): New member.
5643 (dom_walker::m_bb_to_rpo): Likewise.
5644 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
5645 mapping here if not provided by the user.
5646 (dom_walker::~dom_walker): Free bb to RPO mapping if not
5647 provided by the user.
5648 (dom_walker::STOP): Define.
5649 (dom_walker::walk): Do not compute bb to RPO mapping here.
5650 Support STOP return value from before_dom_children to stop
5652 * graphite-optimize-isl.c (optimize_isl): If the schedule
5653 is the same still generate code if -fgraphite-identity
5654 or -floop-parallelize-all are given.
5655 * graphite-scop-detection.c: Include cfganal.h.
5656 (gather_bbs::gather_bbs): Get and pass through bb to RPO
5658 (gather_bbs::before_dom_children): Return STOP for BBs
5660 (build_scops): Compute bb to RPO mapping and pass it to
5661 the domwalk. Treat --param graphite-max-nb-scop-params=0
5662 as not limiting the number of params.
5663 * graphite.c (graphite_initialize): Remove limit on the
5664 number of basic-blocks in a function.
5665 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
5666 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
5667 default value of 10.
5669 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5671 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
5672 Adjust code to eliminate needing to do the shift right 32-bits
5673 operation after XSCVDPSPN.
5675 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5677 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
5678 ((X / Y) != 0 -> X >= Y): Likewise.
5680 2017-09-26 Carl Love <cel@us.ibm.com>
5682 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
5683 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
5684 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
5685 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
5686 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
5687 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
5688 definitions and overloading.
5689 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
5690 statement for P9V_BUILTIN_XST_LEN_R.
5691 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
5692 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
5693 define_expand and define_insn for the instructions and builtins.
5694 * doc/extend.texi: Update the built-in documentation file for the new
5696 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
5697 define_insn for the instructions
5699 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
5702 * gcc/config/netbsd-protos.h: New file.
5703 * gcc/config/netbsd.c: New file.
5704 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
5705 * gcc/config/t-netbsd: New file.
5706 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
5707 (tmake_file) Add t-netbsd.
5708 (extra_objs) Add netbsd.o.
5710 2017-09-26 Janus Weil <janus@gcc.gnu.org>
5714 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
5716 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5718 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
5719 sign extension from a vector register to a GPR by doing a 32-bit
5720 direct move and then an EXTSW.
5721 (extendsi<mode>2 splitter): Likewise.
5722 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
5723 right or vector extract after doing XSCVDPSPN. Use
5724 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
5726 (movdi_from_sf_zero_ext): Likewise.
5727 (reload_gpr_from_vsxsf): Likewise.
5728 (p8_mfvsrd_4_disf): Delete, no longer used.
5729 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
5730 then needing to move the SFmode to a GPR to use the XSCVDPSP
5731 instruction instead of FRSP and XSCVDPSPN.
5732 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
5733 it is adjacent to the other XSCVSPDP insns.
5734 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
5735 SFmode to be in traditional Altivec registers.
5736 (vsx_xscvdpspn): Eliminate useless alternative constraint.
5737 (vsx_xscvspdpn): Likewise.
5738 (vsx_xscvspdpn_scalar): Likewise.
5740 2017-09-26 Martin Jambor <mjambor@suse.cz>
5742 * tree-sra.c (compare_access_positions): Put integral types first,
5743 stabilize sorting of integral types, remove conditions putting
5744 non-full-precision integers last.
5745 (sort_and_splice_var_accesses): Disable scalarization if a
5746 non-integert would be represented by a non-full-precision integer.
5748 2017-09-26 Joseph Myers <joseph@codesourcery.com>
5750 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
5751 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
5752 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
5753 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
5754 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
5755 conditionals inside the function instead of around it. Call
5756 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
5757 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
5759 2017-09-26 Richard Biener <rguenther@suse.de>
5761 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
5763 (scop_detection::build_scop_breadth): ... this. Removed.
5764 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
5765 (scop_detection::harmful_stmt_in_bb): Likewise.
5766 (scop_detection::graphite_can_represent_stmt): Likewise.
5767 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
5768 (scop_detection::can_represent_loop): Remove recursion, fold in ...
5769 (scop_detection::can_represent_loop_1): ... this. Removed.
5770 (scop_detection::harmful_loop_in_region): Simplify after inlining
5771 the above and remove more quadraticness.
5772 (build_scops): Adjust.
5773 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
5776 2017-09-26 Jakub Jelinek <jakub@redhat.com>
5779 * config/i386/i386.c (ix86_print_operand_address_as): Only test
5780 REGNO (base) == SP_REG if base is a REG.
5783 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
5784 if it is different SSA_NAME.
5785 (optimize_range_tests_cmp_bitwise): New function.
5786 (optimize_range_tests): Call it.
5788 2017-09-26 Richard Biener <rguenther@suse.de>
5790 PR tree-optimization/82321
5791 * graphite.c (canonicalize_loop_closed_ssa): Properly check
5792 for the def being inside the loop.
5794 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5796 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
5798 * config/s390/s390-builtins.def: Fix constraint on op4.
5800 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5802 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
5803 independent expanders.
5804 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
5805 ("vec_ordered", "vec_unordered"): New expanders.
5807 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5809 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
5812 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5814 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
5815 vec_unpacks_lo_v16qi.
5816 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
5818 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5820 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
5821 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
5822 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
5824 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5826 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
5828 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
5830 ("*vec_slb<mode>"): New insn pattern.
5831 ("vec_shr_<mode>"): New expander.
5832 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
5833 and force the shift count operand to V16QImode.
5834 ("vec_srb<mode>"): Set shift count mode to V16QI.
5836 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5838 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
5839 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
5840 ("vec_widen_smult_hi_<mode>"): New expander definitions.
5842 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
5845 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
5847 2017-09-26 Richard Biener <rguenther@suse.de>
5849 PR tree-optimization/82320
5850 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
5853 2017-09-25 Jeff Law <law@redhat.com>
5855 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
5856 prototype for new argument.
5857 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
5858 mostly extracted from rs6000_emit_allocate_stack.
5859 (rs6000_emit_probe_stack_range_stack_clash): New function.
5860 (rs6000_emit_allocate_stack): Call
5861 rs6000_emit_probe_stack_range_stack_clash as needed.
5862 (rs6000_emit_probe_stack_range): Add additional argument
5863 to call to gen_probe_stack_range{si,di}.
5864 (output_probe_stack_range): New.
5865 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
5866 (output_probe_stack_range_stack_clash): New.
5867 (rs6000_emit_prologue): Emit notes into dump file as requested.
5868 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
5869 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
5870 Add additional operand and pass it to output_probe_stack_range.
5872 2017-09-25 Bin Cheng <bin.cheng@arm.com>
5874 PR tree-optimization/82163
5875 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
5876 (checking_verify_loop_closed_ssa): New parameter.
5877 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
5878 (check_loop_closed_ssa_stmt): Delete.
5879 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
5880 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
5881 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
5884 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
5886 * brig-builtins.def: Treat HSAIL barrier builtins as
5887 setjmp/longjump style functions.
5889 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
5891 * target.def (constant_alignment): New hook.
5892 * defaults.h (CONSTANT_ALIGNMENT): Delete.
5893 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
5894 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
5895 * doc/tm.texi: Regenerate.
5896 * targhooks.h (default_constant_alignment): Declare.
5897 (constant_alignment_word_strings): Likewise.
5898 * targhooks.c (default_constant_alignment): New function.
5899 (constant_alignment_word_strings): Likewise.
5900 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
5901 instead of CONSTANT_ALIGNMENT.
5902 * varasm.c (align_variable, get_variable_align, build_constant_desc)
5903 (force_const_mem): Likewise.
5904 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
5905 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
5906 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
5907 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5908 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
5910 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
5911 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5912 constant_alignment_word_strings.
5913 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
5914 (CONSTANT_ALIGNMENT): Likewise.
5915 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5916 (arm_constant_alignment): New function.
5917 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
5918 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5919 constant_alignment_word_strings.
5920 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
5921 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5922 constant_alignment_word_strings.
5923 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
5924 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5925 (cris_constant_alignment): New function.
5926 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
5927 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5928 (epiphany_constant_alignment): New function.
5929 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
5930 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5931 constant_alignment_word_strings.
5932 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
5933 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5934 constant_alignment_word_strings.
5935 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
5936 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5937 constant_alignment_word_strings.
5938 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
5939 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
5940 * config/i386/i386.c (ix86_constant_alignment): Make static.
5941 Use the same interface as the target hook.
5942 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5943 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
5944 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5945 constant_alignment_word_strings.
5946 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
5947 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
5948 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5949 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
5950 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5951 constant_alignment_word_strings.
5952 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
5953 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5954 constant_alignment_word_strings.
5955 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
5956 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5957 constant_alignment_word_strings.
5958 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
5959 * config/microblaze/microblaze.c (microblaze_constant_alignment):
5961 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5962 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
5963 * config/mips/mips.c (mips_constant_alignment): New function.
5964 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5965 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
5966 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
5967 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5968 (mmix_constant_alignment): Make static. Use the same interface
5970 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
5971 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5972 constant_alignment_word_strings.
5973 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
5974 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5975 constant_alignment_word_strings.
5976 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
5977 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5978 constant_alignment_word_strings.
5979 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
5980 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5981 (rs6000_constant_alignment): New function.
5982 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
5983 * config/riscv/riscv.c (riscv_constant_alignment): New function.
5984 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5985 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
5986 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5987 (rs6000_constant_alignment): New function.
5988 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
5989 * config/s390/s390.c (s390_constant_alignment): New function.
5990 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5991 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
5992 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5993 constant_alignment_word_strings.
5994 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
5995 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5996 (sparc_constant_alignment): New function.
5997 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
5998 * config/spu/spu.c (spu_constant_alignment): New function.
5999 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6000 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
6001 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6002 constant_alignment_word_strings.
6003 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
6004 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6005 constant_alignment_word_strings.
6006 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
6007 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6008 constant_alignment_word_strings.
6009 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
6010 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6011 (visium_constant_alignment): New function.
6012 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
6013 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6014 (xtensa_constant_alignment): New function.
6015 * system.h (CONSTANT_ALIGNMENT): Poison.
6017 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
6019 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6020 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
6021 (rs6000_builtin_valid_without_lhs): New helper function.
6022 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6023 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
6025 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
6027 * target.h (vec_perm_indices): Use unsigned short rather than
6029 (auto_vec_perm_indices): Likewise.
6030 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
6031 Use unsigned int rather than unsigned char.
6032 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
6034 2017-09-25 Richard Biener <rguenther@suse.de>
6036 * cfgloop.h (sort_sibling_loops): Declare.
6037 * cfgloop.c (sort_sibling_loops_cmp): New helper.
6038 (sort_sibling_loops): New function sorting the sibling loop list
6040 * graphite.c (graphite_transform_loops): Sort sibling loops.
6042 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
6044 * target.def (vec_perm_const_ok): Change sel parameter to
6046 * optabs-query.c (can_vec_perm_p): Update accordingly.
6047 * doc/tm.texi: Regenerate.
6048 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
6049 auto_vec_perm_indices and remove separate nelt field.
6050 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
6051 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
6052 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
6053 (aarch64_expand_vec_perm_const): Update accordingly.
6054 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
6055 to vec_perm_indices.
6056 * config/arm/arm.c (expand_vec_perm_d): Change perm to
6057 auto_vec_perm_indices and remove separate nelt field.
6058 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
6059 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
6060 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
6062 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
6063 to vec_perm_indices.
6064 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
6065 sel to vec_perm_indices.
6066 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
6067 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
6068 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
6070 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
6073 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
6076 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
6077 on the FUNCTION_DECL function context if it has a DIE that is a
6080 2017-09-25 Richard Biener <rguenther@suse.de>
6082 PR tree-optimization/82285
6083 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
6086 2017-09-25 Tom de Vries <tom@codesourcery.com>
6090 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
6093 2017-09-25 Richard Biener <rguenther@suse.de>
6095 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
6096 ISL errors other than isl_error_quota happen. Dump if the
6097 schedule is the same.
6098 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
6099 errors instead of aborting inside ISL.
6101 2017-09-25 Iain Sandoe <iain@codesourcery.com>
6104 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
6105 of libgcc_eh for m64.
6106 * config/i386/darwin64.h: Likewise.
6108 2017-09-25 Richard Biener <rguenther@suse.de>
6111 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
6112 attribute for incomplete types nor twice for complete ones.
6114 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
6117 * config/i386/i386.c (ix86_print_operand_address_as): Encode
6118 %esp as %rsp to avoid 0x67 prefix if there is no index or base
6121 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
6124 * config/i386/i386.opt (mprefer-avx256): Use
6125 ix86_target_flags variable.
6126 * config/i386/i386.c (ix86_target_string): Move
6127 -mprefer-avx256 to flag2_opts.
6129 2017-09-22 Jakub Jelinek <jakub@redhat.com>
6132 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
6133 and x != -1 | y != -1 into (x & y) != -1.
6135 2017-09-22 Steve Ellcey <sellcey@cavium.com>
6137 * config.gcc: Add new case statement to set
6138 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
6139 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
6140 s390x-*-linux* case statements. Added aarch64 to the list of
6141 supported architectures.
6143 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6145 PR tree-optimization/82289
6146 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
6147 STMT_VINFO_RELEVANT_P.
6149 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6150 Alan Hayward <alan.hayward@arm.com>
6151 David Sherwood <david.sherwood@arm.com>
6153 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
6154 for VR_RANGE only; don't allow VR_ANTI_RANGE.
6155 (extract_range_from_binary_expr_1): Don't call
6156 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
6158 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6159 Alan Hayward <alan.hayward@arm.com>
6160 David Sherwood <david.sherwood@arm.com>
6162 * target.def (preferred_vector_alignment): New hook.
6163 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
6165 * doc/tm.texi: Regenerate.
6166 * targhooks.h (default_preferred_vector_alignment): Declare.
6167 * targhooks.c (default_preferred_vector_alignment): New function.
6168 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
6170 (DR_TARGET_ALIGNMENT): New macro.
6171 (aligned_access_p): Update commentary.
6172 (vect_known_alignment_in_bytes): New function.
6173 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
6175 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
6176 Calculate the misalignment based on the target alignment rather than
6178 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
6179 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
6180 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
6181 the target alignment, rather than masking the element misalignment
6182 with the number of elements in a vector. Also use the target
6183 alignment when calculating the maximum number of peels.
6184 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
6185 instead of TYPE_ALIGN_UNIT.
6186 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
6187 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
6188 (vect_create_addr_base_for_vector_ref): Update call accordingly.
6189 (vect_create_data_ref_ptr): Likewise.
6190 (vect_setup_realignment): Realign by ANDing with
6191 -DR_TARGET_MISALIGNMENT.
6192 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
6193 the number of peels based on DR_TARGET_ALIGNMENT.
6194 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
6195 with the guaranteed alignment boundary when deciding whether
6197 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
6198 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
6199 (ensure_base_align): Remove stmt_info parameter. Get the
6200 target base alignment from DR_TARGET_ALIGNMENT.
6201 (vectorizable_store): Update call accordingly. Interpret
6202 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
6204 (vectorizable_load): Likewise.
6206 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6207 Alan Hayward <alan.hayward@arm.com>
6208 David Sherwood <david.sherwood@arm.com>
6210 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
6211 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
6212 (vect_enhance_data_refs_alignment): Likewise.
6214 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6216 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
6217 error. Only quit immediately if parsing is complete.
6218 (BEGIN): Initialize fatal_err and parse_done.
6219 (begin fpu, end fpu): Check number of arguments.
6220 (begin arch, end arch): Likewise.
6221 (begin cpu, end cpu): Likewise.
6222 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
6223 (optalias): Likewise.
6225 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6227 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
6228 * config/arm/arm-isa.h: Delete. Move definitions to ...
6229 * arm-cpus.in: ... here. Use new feature and fgroup values.
6230 * config/arm/arm.c (arm_option_override): Use lower case for feature
6232 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
6233 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
6234 * config/arm/parsecpu.awk (END): Add new command 'isa'.
6236 (print_isa_bits_for): New function.
6237 (gen_isa): New function.
6238 (gen_comm_data): Use print_isa_bits_for.
6239 (define feature): New keyword.
6240 (define fgroup): New keyword.
6241 * config/arm/t-arm (TM_H): Remove.
6242 (GTM_H): Add arm-isa.h.
6243 (arm-isa.h): Add rule to generate file.
6244 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
6245 case for feature bit names.
6247 2017-09-22 Richard Biener <rguenther@suse.de>
6249 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
6251 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
6252 print of no dependency loops ...
6253 * graphite.c (graphite_transform_loops): ... here.
6254 (canonicalize_loop_closed_ssa_form): Work from inner to outer
6256 (same_close_phi_node, remove_duplicate_close_phi,
6257 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
6258 (canonicalize_loop_closed_ssa): ... here and simplify.
6259 * graphite-optimize-isl.c: Include tree-vectorizer.h.
6260 (optimize_isl): Use dump_printf_loc to tell when we stopped
6261 optimizing because of an ISL timeout.
6263 2017-09-22 Richard Biener <rguenther@suse.de>
6265 PR tree-optimization/82291
6266 * tree-if-conv.c (predicate_mem_writes): Make sure to
6267 remove writes in blocks predicated with false.
6269 2017-09-22 Richard Biener <rguenther@suse.de>
6271 * sese.c: Include cfganal.h.
6272 (if_region_set_false_region): Remove.
6273 (create_if_region_on_edge): Likewise.
6274 (move_sese_in_condition): Re-implement without destroying
6277 2017-09-22 Richard Biener <rguenther@suse.de>
6279 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
6280 Verify both BBs contain loop PHI nodes before dispatching to
6282 (graphite_regenerate_ast_isl): Do not recompute dominators,
6283 do not verify three times. Restructure for clarity.
6284 * graphite-scop-detection.c (same_close_phi_node,
6285 remove_duplicate_close_phi, make_close_phi_nodes_unique,
6286 defined_in_loop_p, canonicalize_loop_closed_ssa,
6287 canonicalize_loop_closed_ssa_form): Simplify, remove excess
6288 checking and SSA rewrite, move to ...
6289 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
6290 (graphite_initialize): Do not pass in ctx, do not reset the
6291 SCEV cache, compute only dominators.
6292 (graphite_transform_loops): Allocate ISL ctx after
6293 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
6294 Maintain post-dominators only around build_scops.
6295 * sese.c (if_region_set_false_region): Make static. Free
6296 and recompute dominators.
6297 (move_sese_in_condition): Assert we don't get called with
6298 post-dominators computed.
6299 * sese.h (if_region_set_false_region): Remove.
6301 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
6303 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
6304 mode attribute for TARGET_AVX512VL.
6306 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
6308 * config/i386/i386.opt (mprefer-avx256): New option.
6309 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
6311 (ix86_preferred_simd_mode): Return 256-bit AVX modes
6312 for TARGET_PREFER_AVX256.
6313 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
6315 2017-09-21 Jeff Law <law@redhat.com>
6317 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
6318 Fix dump output if the only stack space is for pushed registers.
6320 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6322 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
6325 2017-09-21 Martin Sebor <msebor@redhat.com>
6328 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
6329 code (in C++) or code that triggers warnings.
6331 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
6333 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
6335 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
6337 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
6338 * sched-rgn.c: Ditto.
6339 * sel-sched-ir.c: Ditto.
6341 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
6343 * toplev.h (set_random_seed): Adjust return type.
6344 * toplev.c (init_local_tick): Move eager initialization of random_seed
6345 to get_random_seed. Adjust comment.
6346 (init_random_seed): Inline to get_random_seed, delete.
6347 (get_random_seed): Initialize random_seed lazily.
6348 (set_random_seed): Do not return previous value.
6349 (print_switch_value): Do not call get_random_seed.
6351 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
6353 * cgraph.c (delete_function_version): New, broken out from...
6354 (cgraph_node::delete_function_version): ...here. Rename to
6355 cgraph_node::delete_function_version_by_decl. Update all uses.
6356 (cgraph_node::remove): Call delete_function_version.
6358 2017-09-21 Jakub Jelinek <jakub@redhat.com>
6361 * tree-inline.c (expand_call_inline): Emit clobber stmts for
6362 VAR_DECLs to which addressable non-volatile parameters are mapped
6363 and for id->retvar after the return value assignment. Clear
6364 id->retval and id->retbnd after inlining.
6366 2017-09-21 Richard Biener <rguenther@suse.de>
6368 PR tree-optimization/82276
6369 PR tree-optimization/82244
6370 * tree-vrp.c (build_assert_expr_for): Set
6371 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
6373 (remove_range_assertions): Revert earlier change.
6375 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
6378 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
6380 2017-09-21 Richard Biener <rguenther@suse.de>
6382 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
6383 Restore valid IL after code generation errors.
6384 * graphite.c (graphite_transform_loops): Diagnose code
6385 generation issues as MSG_MISSED_OPTIMIZATION and continue
6386 with processing SCOPs.
6388 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6389 Alan Hayward <alan.hayward@arm.com>
6390 David Sherwood <david.sherwood@arm.com>
6392 * calls.c (compute_argument_addresses): Use simplify_gen_binary
6393 rather than choosing between plus_constant and gen_rtx_<CODE>.
6394 * expr.c (emit_push_insn): Likewise.
6395 (expand_expr_real_2): Likewise.
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 * loop-unroll.c (split_iv): Call copy_rtx on the step.
6403 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6404 Alan Hayward <alan.hayward@arm.com>
6405 David Sherwood <david.sherwood@arm.com>
6407 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
6408 calling tree_to_uhwi.
6410 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6411 Alan Hayward <alan.hayward@arm.com>
6412 David Sherwood <david.sherwood@arm.com>
6414 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
6415 INTEGER_CST rather than a negative test for ADDR_EXPR.
6417 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6418 Alan Hayward <alan.hayward@arm.com>
6419 David Sherwood <david.sherwood@arm.com>
6421 * tree-vrp.c (extract_range_from_binary_expr_1): Check
6422 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
6424 2017-09-21 Richard Biener <rguenther@suse.de>
6426 PR tree-optimization/71351
6427 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
6428 graphite_create_new_loop_guard): Remove, fold remaining parts
6430 (translate_isl_ast_node_for): ... here and simplify.
6432 2017-09-21 Jakub Jelinek <jakub@redhat.com>
6435 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
6436 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
6437 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
6438 alternative always use QI mode, for -Os imov (=R,R) alternative
6439 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
6442 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6443 Jeff Law <law@redhat.com>
6445 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
6446 (allocate_stack_space): New function, partially extracted from
6448 (s390_emit_prologue): Track offset to most recent stack probe.
6449 Code to allocate space moved into allocate_stack_space.
6450 Dump actions when no stack is allocated.
6451 (s390_prologue_plus_offset): New function.
6452 (s390_emit_stack_probe): Likewise.
6454 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
6456 * common.opt (Wa, Wl, Wp, g, gz=): Add
6458 (gno-column-info): Remove.
6459 (gcolumn-info): Drop RejectNegative.
6461 (gno-record-gcc-switches): Remove.
6462 (grecord-gcc-switches): Drop RejectNegative.
6463 (gno-split-dwarf): Remove.
6464 (gsplit-dwarf): Drop RejectNegative.
6465 (gno-strict-dwarf): Remove.
6466 (gstrict-dwarf): Drop RejectNegative.
6467 * config/darwin.opt (gfull, gused): Add RejectNegative.
6468 * dwarf2out.c (gen_producer_string): Drop
6469 gno-record-gcc-switches handler.
6470 * optc-gen.awk: Add g to prefixes with negative forms.
6471 * opts-common.c (remapping_prefix_p): New.
6472 (find_opt): Check it.
6473 (generate_canonical_option): Test g prefix.
6474 (option_map): Add -gno- mapping.
6475 (add_misspelling_candidates): Check remapping_prefix_p.
6477 2017-09-20 Jeff Law <law@redhat.com>
6479 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
6480 thinko in stack clash protection support.
6482 * explow.c (compute_stack_clash_protection_loop_data): Use
6483 CONST_INT_P instead of explicit test. Verify object is a
6484 CONST_INT_P before looking at INTVAL.
6485 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
6486 instead of explicit test.
6488 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
6491 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
6492 address instead of to r1 and r11.
6494 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
6496 * config.gcc: Support "knm".
6497 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
6498 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
6500 * config/i386/i386.c (m_KNM): Define.
6501 (processor_target_table): Add "knm".
6503 (ix86_option_override_internal): Add "knm".
6504 (ix86_issue_rate): Add PROCESSOR_KNM.
6505 (ix86_adjust_cost): Ditto.
6506 (ia32_multipass_dfa_lookahead): Ditto.
6507 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
6508 (fold_builtin_cpu): Add M_INTEL_KNM.
6509 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
6510 (processor_type): Add PROCESSOR_KNM.
6511 * config/i386/x86-tune.def: Add m_KNM.
6512 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
6514 2017-09-20 Richard Biener <rguenther@suse.de>
6516 PR tree-optimization/80213
6517 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
6518 are allowed in empty BBs as well.
6519 (canonicalize_loop_closed_ssa): Also look for other complex
6521 (scop_detection::get_sese): Include the loop-closed PHI block
6523 (scop_detection::merge_sese): Remove code adding extra blocks.
6524 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
6525 (build_scops): Assert the final returned scop is invalid.
6527 2017-09-20 Richard Biener <rguenther@suse.de>
6529 PR tree-optimization/82264
6530 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
6532 (vn_phi_lookup): Likewise.
6533 (vn_phi_insert): Likewise.
6535 2017-09-20 Jakub Jelinek <jakub@redhat.com>
6537 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
6538 that fits into uhwi or shwi, add DW_AT_const_value regardless
6539 of early_dwarf without going through RTL, using add_AT_unsigned
6542 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
6543 (DEBUG_LTO_ABBREV_SECTION): Likewise.
6544 (DEBUG_LTO_MACINFO_SECTION): Likewise.
6545 (DEBUG_MACRO_SECTION): Likewise.
6546 (DEBUG_LTO_MACRO_SECTION): Likewise.
6547 (DEBUG_STR_DWO_SECTION): Likewise.
6548 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
6549 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
6550 (DEBUG_LTO_DWO_LINE_SECTION): Define.
6551 (DEBUG_LTO_LINE_STR_SECTION): Define.
6552 (init_sections_and_labels): Initialize debug_line_str_section
6553 variable. Initialize debug_loc_section for -gdwarf-5 to
6554 DEBUG_LOCLISTS_SECTION. Formatting fixes.
6556 2017-09-20 Richard Biener <rguenther@suse.de>
6558 * graphite-sese-to-poly.c (extract_affine): Properly handle
6559 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
6561 2017-09-20 Richard Biener <rguenther@suse.de>
6563 PR tree-optimization/81373
6564 * graphite-scop-detection.c (build_cross_bb_scalars_def):
6565 Force SESE live-out defs to be handled even if they are
6568 2017-09-19 Jeff Law <law@redhat.com>
6570 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
6571 nothing for stack adjustments with REG_STACK_CHECK.
6572 * sched-deps.c (parse_add_or_inc): Reject insns with
6573 REG_STACK_CHECK from dependency breaking.
6574 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
6575 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
6576 * reg-notes.def (STACK_CHECK): New note.
6578 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
6579 (ix86_expand_prologue): Dump stack clash info as needed.
6580 Call ix86_adjust_stack_and_probe_stack_clash as needed.
6582 * function.c (dump_stack_clash_frame_info): New function.
6583 * function.h (dump_stack_clash_frame_info): Prototype.
6584 (enum stack_clash_probes): New enum.
6586 * config/alpha/alpha.c (alpha_expand_prologue): Also check
6587 flag_stack_clash_protection.
6588 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
6589 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
6590 (arm_frame_pointer_required): Likewise.
6591 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
6592 (ia64_expand_prologue): Likewise.
6593 * config/mips/mips.c (mips_expand_prologue): Likewise.
6594 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
6595 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6596 (sparc_flat_expand_prologue): Likewise.
6597 * config/spu/spu.c (spu_expand_prologue): Likewise.
6599 * explow.c: Include "params.h".
6600 (anti_adjust_stack_and_probe_stack_clash): New function.
6601 (get_stack_check_protect): Likewise.
6602 (compute_stack_clash_protection_loop_data): Likewise.
6603 (emit_stack_clash_protection_loop_start): Likewise.
6604 (emit_stack_clash_protection_loop_end): Likewise.
6605 (allocate_dynamic_stack_space): Use get_stack_check_protect.
6606 Use anti_adjust_stack_and_probe_stack_clash.
6607 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
6608 (emit_stack_clash_protection_loop_start): Likewise.
6609 (emit_stack_clash_protection_loop_end): Likewise.
6610 * rtl.h (get_stack_check_protect): Prototype.
6611 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
6612 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
6613 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
6615 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
6617 * doc/tm.texi: Rebuilt.
6618 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
6619 get_stack_check_protect.
6620 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
6621 * config/arm/arm.c (arm_expand_prologue): Likewise.
6622 (arm_frame_pointer_required): Likewise.
6623 * config/i386/i386.c (ix86_expand_prologue): Likewise.
6624 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
6625 * config/mips/mips.c (mips_expand_prologue): Likewise.
6626 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
6627 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
6628 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6629 (sparc_flat_expand_prologue): Likewise.
6631 * common.opt (-fstack-clash-protection): New option.
6632 * flag-types.h (enum stack_check_type): Note difference between
6633 -fstack-check= and -fstack-clash-protection.
6634 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
6635 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
6636 * toplev.c (process_options): Issue warnings/errors for cases
6637 not handled with -fstack-clash-protection.
6638 * doc/invoke.texi (-fstack-clash-protection): Document new option.
6639 (-fstack-check): Note additional problem with -fstack-check=generic.
6640 Note that -fstack-check is primarily for Ada and refer users
6641 to -fstack-clash-protection for stack-clash-protection.
6642 Document new params for stack clash protection.
6644 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
6646 * config/i386/i386.c (ix86_split_long_move): Do not handle
6647 address used for LEA in a special way.
6649 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
6651 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
6653 2017-09-19 Martin Sebor <msebor@redhat.com>
6656 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
6657 of incompatible types.
6659 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
6661 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6662 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
6663 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6664 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
6666 2017-09-19 Richard Biener <rguenther@suse.de>
6668 PR tree-optimization/82244
6669 * tree-vrp.c (remove_range_assertions): Do not propagate
6670 a constant to abnormals but replace the assert with a copy.
6672 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
6674 PR rtl-optimization/57878
6675 PR rtl-optimization/68988
6676 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
6677 avoidance test involving non_reload_pseudos. Move frequency test
6678 below the general fragmentation avoidance test.
6680 2017-09-19 Richard Biener <rguenther@suse.de>
6682 PR tree-optimization/69728
6683 * graphite-sese-to-poly.c (schedule_error): New global.
6684 (add_loop_schedule): Handle empty domain by failing the
6686 (build_original_schedule): Handle schedule_error.
6688 2017-09-19 Richard Biener <rguenther@suse.de>
6690 * graphite-scop-detection.c (scop_detection::can_represent_loop):
6691 Do not iterate to sibling loops but only to siblings of inner
6694 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
6697 * config/m68k/m68k.md (moveq feeding equality comparison): Check
6698 that the registers are different.
6700 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
6702 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
6703 to processor_model and "amdfam17h" to arch_names_table.
6704 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
6706 2017-09-18 Jakub Jelinek <jakub@redhat.com>
6709 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
6711 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6712 Alan Hayward <alan.hayward@arm.com>
6713 David Sherwood <david.sherwood@arm.com>
6715 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
6717 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
6719 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
6720 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
6721 vect_schedule_slp_instance.
6722 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
6723 Update call to vect_slp_analyze_node_operations. Simplify return
6725 (vect_slp_analyze_bb_1): Update call accordingly.
6726 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
6727 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
6728 (vect_schedule_slp): Update call accordingly.
6730 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6731 Alan Hayward <alan.hayward@arm.com>
6732 David Sherwood <david.sherwood@arm.com>
6734 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
6735 with types that aren't in fact scalar.
6737 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6739 * tree-vect-slp.c (vect_record_max_nunits): New function,
6741 (vect_build_slp_tree_1): ...here.
6742 (vect_build_slp_tree_2): Call it for phis too.
6744 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6746 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
6747 to vect_get_vec_def_for_operand when getting the mask operand.
6749 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6750 Alan Hayward <alan.hayward@arm.com>
6751 David Sherwood <david.sherwood@arm.com>
6753 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
6756 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6757 Alan Hayward <alan.hayward@arm.com>
6758 David Sherwood <david.sherwood@arm.com>
6760 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
6761 calculation for vector booleans.
6763 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6764 Alan Hayward <alan.hayward@arm.com>
6765 David Sherwood <david.sherwood@arm.com>
6767 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
6769 (vect_transform_stmt): ...here.
6770 (vect_analyze_stmt): Use it instead of calling
6771 vectorizable_live_operation directly.
6773 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
6775 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
6776 non-SIMT targets in acc vector loops.
6778 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
6780 * configure.ac: Add arc and check if assembler supports gdwarf2.
6781 * configure: Regenerate.
6783 2017-09-18 Richard Biener <rguenther@suse.de>
6785 PR tree-optimization/82220
6786 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
6787 epilogue niters from the min_profitable_iters compute.
6789 2017-09-18 Jakub Jelinek <jakub@redhat.com>
6792 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
6793 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
6794 (ix86_init_pic_reg): Revert 2017-09-01 changes.
6796 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
6799 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
6800 switching to a new text section.
6802 2017-09-18 Richard Biener <rguenther@suse.de>
6804 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
6806 (build_alias_set): Reject aliases with no access function.
6808 2017-09-18 Richard Biener <rguenther@suse.de>
6810 PR tree-optimization/79622
6811 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
6813 (build_cross_bb_scalars_use): Likewise.
6815 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
6817 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
6819 2017-09-18 Alan Modra <amodra@gmail.com>
6822 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
6823 stack_pointer_rtx for count 0. Update comments. Break up
6824 large rtl expression.
6826 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
6828 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
6829 Increase to 20 bytes.
6830 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
6831 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
6832 or avx version of the stub.
6834 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
6837 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
6838 compute the minimum stack alignment. Also update preferred stack
6839 boundary for leaf functions.
6841 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
6843 PR tree-optimization/82228
6844 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
6847 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
6849 * common/config/nds32/nds32-common.c
6850 (nds32_option_optimization_table): Refine formatting.
6851 (nds32_option_optimization_table): Use -fsched-pressure and
6852 -fomit-frame-pointer for specific optimization level.
6854 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
6856 * config/nds32/nds32.c: Refine formatting and comments.
6857 * config/nds32/nds32.h: Likewise.
6858 * config/nds32/nds32.md: Likewise.
6859 * config/nds32/nds32-cost.c: Likewise.
6860 * config/nds32/nds32-isr.c: Likewise.
6861 * config/nds32/nds32-md-auxiliary.c: Likewise.
6862 * config/nds32/nds32-multiple.md: Likewise.
6863 * config/nds32/nds32-predicates.c: Likewise.
6865 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
6866 Jakub Jelinek <jakub@redhat.com>
6868 Add support for -std=c++2a.
6869 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
6871 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
6873 2017-09-15 Steve Ellcey <sellcey@cavium.com>
6876 * doc/extend.texi (Common Function Attributes): Add
6877 references to ARM, AArch64, and S/390 specific attributes.
6878 (Function Specific Option Pragmas): Add AArch64 and S/390
6879 to list of back ends that support the target pragma.
6881 2017-09-15 Nathan Sidwell <nathan@acm.org>
6883 * doc/standards.texi: Fix C++17 description. Update URLs for
6886 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
6888 * common.opt (Wcast-align=strict): New warning option.
6889 * doc/invoke.texi: Document -Wcast-align=strict.
6891 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
6893 * cgraph.h (cgraph_thunk_info): Add comments.
6894 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
6895 assert for VIRTUAL_* arguments stricter.
6897 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
6899 PR tree-optimization/71026
6900 * match.pd: Move RDIV patterns from fold-const.c
6901 * fold-const.c (distribute_real_division): Removed.
6902 (fold_binary_loc): Remove calls to distribute_real_divison.
6904 2017-09-15 Jakub Jelinek <jakub@redhat.com>
6906 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
6907 c++1z and gnu++1z as deprecated. Change other references to
6908 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
6909 Change -Wc++1z-compat to -Wc++17-compat.
6910 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
6911 * dwarf2out.c (highest_c_language): Handle C++17.
6912 (gen_compile_unit_die): Likewise.
6914 2017-09-15 Jakub Jelinek <jakub@redhat.com>
6916 PR rtl-optimization/82192
6917 * combine.c (make_extraction): Don't look through non-paradoxical
6918 SUBREGs or TRUNCATE if pos + len is or might be bigger than
6921 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
6922 Alan Hayward <alan.hayward@arm.com>
6923 David Sherwood <david.sherwood@arm.com>
6925 * target.def (function_arg_offset): New hook.
6926 * targhooks.h (default_function_arg_offset): Declare.
6927 * targhooks.c (default_function_arg_offset): New function.
6928 * function.c (locate_and_pad_parm): Use
6929 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
6930 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
6931 (TARGET_FUNCTION_ARG_OFFSET): ...this.
6932 * doc/tm.texi: Regenerate.
6933 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
6934 * config/spu/spu.c (spu_function_arg_offset): New function.
6935 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
6936 * system.h (FUNCTION_ARG_OFFSET): Poison.
6938 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
6939 Alan Hayard <alan.hayward@arm.com>
6940 David Sherwood <david.sherwood@arm.com>
6942 * target.def (truly_noop_truncation): New hook.
6943 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
6944 than TRULY_NOOP_TRUNCATION.
6945 * hooks.h (hook_bool_uint_uint_true): Declare.
6946 * hooks.c (hook_bool_uint_uint_true): New function.
6947 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
6948 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
6949 * doc/tm.texi: Regenerate.
6950 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
6951 rather than TRULY_NOOP_TRUNCATION in comments.
6952 (simplify_comparison): Likewise.
6953 (record_truncated_value): Likewise.
6954 * expmed.c (extract_bit_field_1): Likewise.
6955 (extract_split_bit_field): Likewise.
6956 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
6957 instead of TRULY_NOOP_TRUNCATION.
6958 * function.c (assign_parm_setup_block): Likewise.
6959 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
6960 * rtlhooks.c: Include target.h.
6961 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
6962 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
6963 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
6964 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
6965 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
6966 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
6967 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
6968 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
6969 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
6970 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
6971 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
6972 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
6973 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
6974 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
6975 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
6976 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
6977 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
6978 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
6979 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
6980 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
6981 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
6982 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
6983 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
6984 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
6985 * config/mips/mips.c (mips_truly_noop_truncation): New function.
6986 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
6987 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
6988 rather than TRULY_NOOP_TRUNCATION in comments.
6989 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
6990 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
6991 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
6992 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
6993 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
6994 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
6995 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
6996 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
6997 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
6998 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
6999 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
7000 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7001 rather than TRULY_NOOP_TRUNCATION in comments.
7002 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
7003 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
7004 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
7005 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
7006 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
7007 TRULY_NOOP_TRUNCATION condition.
7008 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
7009 (TRULY_NOOP_TRUNCATION): Delete.
7010 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
7011 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
7012 * config/spu/spu.c (spu_truly_noop_truncation): New function.
7013 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7014 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
7015 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
7016 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
7017 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7018 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7019 rather than TRULY_NOOP_TRUNCATION in comments.
7020 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
7021 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
7022 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
7023 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
7024 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
7025 * system.h (TRULY_NOOP_TRUNCATION): Poison.
7027 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
7030 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
7031 (*cmp_ior): Likewise.
7032 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
7033 (*ior_scc_scc_cmp): Likewise.
7034 (*and_scc_scc): Likewise.
7035 (*and_scc_scc_cmp): Likewise.
7037 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7038 Alan Hayard <alan.hayward@arm.com>
7039 David Sherwood <david.sherwood@arm.com>
7041 * target.def (can_change_mode_class): New hook.
7042 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
7043 (hard_regno_nregs): Likewise.
7044 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
7045 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
7046 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
7047 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
7048 (LOAD_EXTEND_OP): Update accordingly.
7049 * doc/tm.texi: Regenerate.
7050 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
7051 CANNOT_CHANGE_MODE_CLASS.
7052 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
7053 (REG_CAN_CHANGE_MODE_P): ...this new macro.
7054 * combine.c (simplify_set): Update accordingly.
7055 * emit-rtl.c (validate_subreg): Likewise.
7056 * recog.c (general_operand): Likewise.
7057 * regcprop.c (mode_change_ok): Likewise.
7058 * reload1.c (choose_reload_regs): Likewise.
7059 (inherit_piecemeal_p): Likewise.
7060 * rtlanal.c (simplify_subreg_regno): Likewise.
7061 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
7062 instead of CANNOT_CHANGE_MODE_CLASS.
7063 (reload_cse_simplify_operands): Likewise.
7064 * reload.c (push_reload): Use targetm.can_change_mode_class
7065 instead of CANNOT_CHANGE_MODE_CLASS.
7066 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
7067 REG_CANNOT_CHANGE_MODE_P.
7068 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7069 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
7070 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7071 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7072 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7073 (arm_can_change_mode_class): New function.
7074 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7075 than CANNOT_CHANGE_MODE_CLASS in comments.
7076 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7077 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
7078 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
7079 (ix86_can_change_mode_class): ...this new function, inverting the
7080 sense of the return value.
7081 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7082 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7083 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7084 (ia64_can_change_mode_class): New function.
7085 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7086 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
7087 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
7088 (m32c_can_change_mode_class): ...this new function, inverting the
7089 sense of the return value.
7090 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7091 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7092 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
7093 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
7094 (mips_can_change_mode_class): ...this new function, inverting the
7095 sense of the return value.
7096 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7097 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7098 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7099 (msp430_can_change_mode_class): New function.
7100 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7101 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
7102 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7103 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7104 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7105 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
7106 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7107 (pa_cannot_change_mode_class): Replace with...
7108 (pa_can_change_mode_class): ...this new function, inverting the
7109 sense of the return value.
7110 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7111 than CANNOT_CHANGE_MODE_CLASS in comments.
7112 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7113 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
7114 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7115 (pdp11_cannot_change_mode_class): Replace with...
7116 (pdp11_can_change_mode_class): ...this new function, inverting the
7117 sense of the return value.
7118 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7119 * config/powerpcspe/powerpcspe-protos.h
7120 (rs6000_cannot_change_mode_class_ptr): Delete.
7121 * config/powerpcspe/powerpcspe.c
7122 (rs6000_cannot_change_mode_class_ptr): Delete.
7123 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7124 (rs6000_option_override_internal): Assign to
7125 targetm.can_change_mode_class instead of
7126 rs6000_cannot_change_mode_class_ptr.
7127 (rs6000_cannot_change_mode_class): Replace with...
7128 (rs6000_can_change_mode_class): ...this new function, inverting the
7129 sense of the return value.
7130 (rs6000_debug_cannot_change_mode_class): Replace with...
7131 (rs6000_debug_can_change_mode_class): ...this new function.
7132 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7133 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
7134 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7135 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7136 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
7138 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
7139 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7140 (rs6000_option_override_internal): Assign to
7141 targetm.can_change_mode_class instead of
7142 rs6000_cannot_change_mode_class_ptr.
7143 (rs6000_cannot_change_mode_class): Replace with...
7144 (rs6000_can_change_mode_class): ...this new function, inverting the
7145 sense of the return value.
7146 (rs6000_debug_cannot_change_mode_class): Replace with...
7147 (rs6000_debug_can_change_mode_class): ...this new function.
7148 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7149 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
7150 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
7151 (s390_can_change_mode_class): ...this new function, inverting the
7152 sense of the return value.
7153 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7154 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7155 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
7156 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7157 (sh_cannot_change_mode_class): Replace with...
7158 (sh_can_change_mode_class): ...this new function, inverting the
7159 sense of the return value.
7160 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7161 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7162 (sparc_can_change_mode_class): New function.
7163 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7164 * config/spu/spu.c (spu_can_change_mode_class): New function.
7165 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7166 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7167 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7168 (visium_can_change_mode_class): New function.
7169 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
7171 2017-09-15 Richard Biener <rguenther@suse.de>
7173 PR tree-optimization/82217
7174 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
7175 but not undefined case.
7177 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7180 * postreload.c (reload_cse_simplify_operands): Skip
7181 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
7183 2017-09-15 Richard Biener <rguenther@suse.de>
7185 PR tree-optimization/68823
7186 * graphite-scop-detection.c (build_alias_set): If we have a
7187 possible dependence check whether we can handle them by just
7188 looking at the DRs DR_ACCESS_FNs.
7189 (build_scops): If build_alias_set fails, fail the SCOP.
7191 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
7193 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
7194 to support float128 built-in functions that require the ISA 3.0
7196 (BU_FLOAT128_3_HW): Likewise.
7197 (SQRTF128): Add support for the IEEE 128-bit square root and fma
7199 (FMAF128): Likewise.
7201 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
7202 support for built-in functions that need the ISA 3.0 IEEE 128-bit
7203 floating point instructions.
7204 (rs6000_invalid_builtin): Likewise.
7205 (rs6000_builtin_mask_names): Likewise.
7206 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
7207 (RS6000_BTM_FLOAT128_HW): Likewise.
7208 (RS6000_BTM_COMMON): Likewise.
7209 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
7211 * doc/extend.texi (RS/6000 built-in functions): Document the
7212 IEEE 128-bit floating point square root and fused multiply-add
7215 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
7217 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
7218 reg (r2) isn't in the set of registers defined in the prologue.
7220 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7221 Alan Hayward <alan.hayward@arm.com>
7222 David Sherwood <david.sherwood@arm.com>
7224 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
7225 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
7226 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
7227 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
7228 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
7230 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
7231 max_vectorization_factor.
7232 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
7234 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7235 Alan Hayward <alan.hayward@arm.com>
7236 David Sherwood <david.sherwood@arm.com>
7238 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
7239 (vect_worthwhile_without_simd_p): Declare.
7240 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
7241 (vectorizable_reduction): Use it.
7242 * tree-vect-stmts.c (vectorizable_shift): Likewise.
7243 (vectorizable_operation): Likewise.
7245 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7246 Alan Hayward <alan.hayward@arm.com>
7247 David Sherwood <david.sherwood@arm.com>
7249 * tree-vectorizer.h (vect_get_num_copies): New function.
7250 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
7251 * tree-vect-loop.c (vectorizable_reduction): Likewise.
7252 (vectorizable_induction): Likewise.
7253 (vectorizable_live_operation): Likewise.
7254 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7255 (vectorizable_bswap): Likewise.
7256 (vectorizable_call): Likewise.
7257 (vectorizable_conversion): Likewise.
7258 (vectorizable_assignment): Likewise.
7259 (vectorizable_shift): Likewise.
7260 (vectorizable_operation): Likewise.
7261 (vectorizable_store): Likewise.
7262 (vectorizable_load): Likewise.
7263 (vectorizable_condition): Likewise.
7264 (vectorizable_comparison): Likewise.
7265 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
7267 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7268 Alan Hayward <alan.hayward@arm.com>
7269 David Sherwood <david.sherwood@arm.com>
7271 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
7272 of vect_init_vector.
7274 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7275 Alan Hayward <alan.hayward@arm.com>
7276 David Sherwood <david.sherwood@arm.com>
7278 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
7279 an inline wrapper that provides a location.
7280 (gimple_build_vector): Likewise.
7281 * gimple-fold.c (gimple_build_vector_from_val): New function.
7282 (gimple_build_vector): Likewise.
7283 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
7284 functions to build the initial value. Always return a gimple value.
7285 (get_initial_defs_for_reduction): Likewise. Only compute
7287 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
7288 vect_init_vector on the results from get_initial_def(s)_for_reduction.
7289 (vectorizable_induction): Use gimple_build_vector rather than
7292 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7293 Alan Hayward <alan.hayward@arm.com>
7294 David Sherwood <david.sherwood@arm.com>
7296 * target.h (vec_perm_indices): New typedef.
7297 (auto_vec_perm_indices): Likewise.
7298 * optabs-query.h: Include target.h
7299 (can_vec_perm_p): Take a vec_perm_indices *.
7300 * optabs-query.c (can_vec_perm_p): Likewise.
7301 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
7302 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7303 * tree-vect-generic.c (lower_vec_perm): Likewise.
7304 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
7305 (vect_grouped_load_supported): Likewise.
7306 (vect_shift_permute_load_chain): Likewise.
7307 (vect_permute_store_chain): Use auto_vec_perm_indices.
7308 (vect_permute_load_chain): Likewise.
7309 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
7310 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
7311 Update uses of can_vec_perm_p.
7312 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
7313 mode with a number of elements. Take a vec_perm_indices *.
7314 (vect_create_epilog_for_reduction): Update accordingly.
7315 Use auto_vec_perm_indices.
7316 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
7317 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
7318 (vect_transform_slp_perm_load): Likewise.
7319 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
7320 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
7321 (vect_gen_perm_mask_checked): Likewise.
7322 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
7323 (vect_gen_perm_mask_checked): Likewise.
7324 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
7325 (vectorizable_store): Likewise.
7326 (vectorizable_load): Likewise.
7327 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
7328 (vectorizable_bswap): Likewise.
7330 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7331 Alan Hayward <alan.hayward@arm.com>
7332 David Sherwood <david.sherwood@arm.com>
7334 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
7335 * tree.c (build_vector): Likewise.
7336 (build_vector_from_ctor): Update accordingly.
7337 (build_vector_from_val): Likewise.
7338 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
7339 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7340 * tree-vect-generic.c (add_rshift): Likewise.
7341 (expand_vector_divmod): Likewise.
7342 (optimize_vector_constructor): Likewise.
7343 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
7344 (vect_transform_slp_perm_load): Likewise.
7345 (vect_schedule_slp_instance): Likewise.
7346 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
7347 (vectorizable_call): Likewise.
7348 (vect_gen_perm_mask_any): Likewise. Add elements in order.
7349 * expmed.c (make_tree): Likewise.
7350 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
7351 a vector passed to build_vector.
7352 (fold_convert_const): Likewise.
7353 (exact_inverse): Likewise.
7354 (fold_ternary_loc): Likewise.
7355 (fold_relational_const): Likewise.
7356 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
7357 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
7358 (const_unop): Likewise. Store the reduction accumulator in a
7359 variable rather than an array.
7360 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
7361 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
7362 the new vector, rather than constructing it after the input arrays.
7363 (native_interpret_vector): Use auto_vec<tree> when building
7364 a vector passed to build_vector. Add elements in order.
7365 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
7366 auto_vec<tree> when building a vector passed to build_vector.
7367 (vect_create_epilog_for_reduction): Likewise.
7368 (vectorizable_induction): Likewise.
7369 (get_initial_def_for_reduction): Likewise. Fix indentation of
7371 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
7373 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
7374 passed to build_vector.
7376 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7377 Alan Hayward <alan.hayward@arm.com>
7378 David Sherwood <david.sherwood@arm.com>
7380 * tree-core.h (tree_base::u): Add an "nelts" field.
7381 (tree_vector): Use VECTOR_CST_NELTS as the length.
7382 * tree.c (tree_size): Likewise.
7383 (make_vector): Initialize VECTOR_CST_NELTS.
7384 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
7385 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
7386 TYPE_VECTOR_SUBPARTS.
7387 * expr.c (const_vector_mask_from_tree): Consistently use "units"
7388 as the number of units, setting it from VECTOR_CST_NELTS.
7389 (const_vector_from_tree): Likewise.
7390 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
7391 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
7392 (fold_negate_expr_1): Likewise.
7393 (fold_convert_const): Likewise.
7394 (const_binop): Likewise. Differentiate the number of output and
7396 (const_unop): Likewise.
7397 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
7398 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
7399 in cases that did the opposite.
7401 2017-09-14 Richard Biener <rguenther@suse.de>
7403 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
7406 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
7408 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
7410 2017-09-14 Jakub Jelinek <jakub@redhat.com>
7413 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
7414 if and where to split a bb, except for splitting before debug insn
7415 sequences followed by non-label real insn. Delete debug insns
7416 in between basic blocks.
7418 * combine.c (make_compound_operation_int): Formatting fixes.
7420 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
7421 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7422 * config/netbsd.h (LINK_EH_SPEC): Likewise.
7423 * config/sol2.h (LINK_EH_SPEC): Likewise.
7424 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7425 * config/s390/linux.h (LINK_SPEC): Likewise.
7426 * config/freebsd.h (LINK_EH_SPEC): Likewise.
7427 * config/openbsd.h (LINK_EH_SPEC): Likewise.
7428 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7429 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
7430 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
7431 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7432 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
7433 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
7435 2017-09-13 Jakub Jelinek <jakub@redhat.com>
7437 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
7439 (ENDFILE_LINUX_SPEC): Likewise.
7440 (LINK_EH_SPEC): Likewise.
7441 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
7442 (LINK_OS_LINUX_SPEC32): Likewise.
7443 (LINK_OS_LINUX_SPEC64): Likewise.
7444 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
7445 (LINK_OS_LINUX_SPEC): Likewise.
7447 2017-09-13 Martin Liska <mliska@suse.cz>
7450 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
7451 CASE_HIGH is NULL_TREE.
7453 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
7454 Alan Hayward <alan.hayward@arm.com>
7455 David Sherwood <david.sherwood@arm.com>
7457 * target.def (secondary_memory_needed): New hook.
7458 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
7459 instead of SECONDARY_MEMORY_NEEDED.
7460 (secondary_memory_needed_mode): Likewise.
7461 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
7462 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
7463 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
7464 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
7465 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
7466 * doc/tm.texi: Regenerate.
7467 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
7468 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
7469 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7470 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
7471 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
7472 * config/i386/i386.c (inline_secondary_memory_needed): Put the
7473 mode argument first and change the reg_class arguments to reg_class_t.
7474 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
7475 Make static. Update the call to inline_secondary_memory_needed.
7476 (ix86_register_move_cost): Update the call to
7477 inline_secondary_memory_needed.
7478 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7479 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
7481 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
7482 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7484 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
7485 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
7486 * config/mips/mips.c (mips_secondary_memory_needed): Make static
7487 and match hook interface. Add comment from mips.h.
7488 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7489 * config/mmix/mmix.md (truncdfsf2): Refer to
7490 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7492 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
7493 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
7494 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7495 (pa_secondary_memory_needed): New function.
7496 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
7497 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
7498 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7499 (pdp11_secondary_memory_needed): Make static and match hook interface.
7500 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
7501 * config/powerpcspe/powerpcspe-protos.h
7502 (rs6000_secondary_memory_needed_ptr): Delete.
7503 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
7505 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7506 (rs6000_option_override_internal): Assign to
7507 targetm.secondary_memory_needed rather than
7508 rs6000_secondary_memory_needed_ptr.
7509 (rs6000_secondary_memory_needed): Match hook interface.
7510 (rs6000_debug_secondary_memory_needed): Likewise.
7511 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
7512 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
7513 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
7514 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7515 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
7516 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
7518 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
7519 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7520 (rs6000_option_override_internal): Assign to
7521 targetm.secondary_memory_needed rather than
7522 rs6000_secondary_memory_needed_ptr.
7523 (rs6000_secondary_memory_needed): Match hook interface.
7524 (rs6000_debug_secondary_memory_needed): Likewise.
7525 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
7526 * config/s390/s390.c (s390_secondary_memory_needed): New function.
7527 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7528 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
7529 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7530 (sparc_secondary_memory_needed): New function.
7531 * lra-constraints.c (check_and_process_move): Refer to
7532 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7534 (curr_insn_transform): Likewise.
7535 (process_alt_operands): Use targetm.secondary_memory_needed
7536 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7537 (check_secondary_memory_needed_p): Likewise.
7538 (choose_split_class): Likewise.
7539 * reload.c: Unconditionally include code that was previously
7540 conditional on SECONDARY_MEMORY_NEEDED.
7541 (push_secondary_reload): Use targetm.secondary_memory_needed
7542 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7543 (push_reload): Likewise.
7544 * reload1.c: Unconditionally include code that was previously
7545 conditional on SECONDARY_MEMORY_NEEDED.
7546 (choose_reload_regs): Use targetm.secondary_memory_needed
7547 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7548 (gen_reload): Likewise.
7549 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
7551 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
7552 Alan Hayward <alan.hayward@arm.com>
7553 David Sherwood <david.sherwood@arm.com>
7555 * target.def (secondary_memory_needed_mode): New hook:
7556 * targhooks.c (default_secondary_memory_needed_mode): Declare.
7557 * targhooks.h (default_secondary_memory_needed_mode): New function.
7558 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
7559 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
7560 * doc/tm.texi: Regenerate.
7561 * lra-constraints.c (check_and_process_move): Use
7562 targetm.secondary_memory_needed_mode instead of
7563 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
7564 (curr_insn_transform): Likewise.
7565 * reload.c (get_secondary_mem): Likewise.
7566 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7567 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
7569 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7570 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7571 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
7572 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7573 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
7575 * config/powerpcspe/powerpcspe-protos.h
7576 (rs6000_secondary_memory_needed_mode): Delete.
7577 * config/powerpcspe/powerpcspe.c
7578 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7579 (rs6000_secondary_memory_needed_mode): Make static.
7580 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7581 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
7583 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7585 (rs6000_secondary_memory_needed_mode): Make static.
7586 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7587 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
7588 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7589 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7590 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7592 (sparc_secondary_memory_needed_mode): New function.
7593 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
7595 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
7597 * config/aarch64/constraints.md (Umq): New constraint.
7598 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
7600 (mov<mode>): Update condition.
7602 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7604 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
7605 when bitposition is the same.
7607 2017-09-13 Richard Biener <rguenther@suse.de>
7609 * dwarf2out.c (output_die_symbol): Remove.
7610 (output_die): Do not output a DIEs symbol.
7612 2017-09-13 Richard Biener <rguenther@suse.de>
7615 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
7616 default-def to avoid breaking iterator update with the weird
7617 interaction with cgraph_update_edges_for_call_stmt_node.
7619 2017-09-13 Richard Biener <rguenther@suse.de>
7621 * tree-cfg.c (verify_gimple_assign_binary): Add verification
7622 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
7623 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
7624 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
7626 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
7628 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
7629 Disable pc relative literal load irrespective of
7630 TARGET_FIX_ERR_A53_84341 for default.
7632 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
7634 * config/sparc/sparc.c (output_return): Output the source location of
7635 the insn in the delay slot, if any.
7636 (output_sibcall): Likewise.
7638 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
7641 * common.opt (-static-pie): New alias.
7642 (shared): Negate static-pie.
7643 (-no-pie): Update help text.
7645 (static-pie): New option.
7646 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
7647 -static-pie support.
7648 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
7649 (LINK_EH_SPEC): Likewise.
7650 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7651 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7652 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7653 * gcc.c (LINK_COMMAND_SPEC): Likewise.
7654 (init_gcc_specs): Likewise.
7655 (init_spec): Likewise.
7656 (display_help): Update help message for -pie.
7657 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
7660 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
7662 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
7663 (movdi_aarch64): Likewise.
7664 (movti_aarch64): Likewise.
7666 2017-09-12 Simon Wright <simon@pushface.org>
7669 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
7670 calculation of the minor version, always output as 0.
7672 2017-09-12 Jakub Jelinek <jakub@redhat.com>
7675 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
7676 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
7677 on it early, rather than manual conversion late. For
7678 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
7679 instead of performing manual conversion.
7681 2017-09-12 Carl Love <cel@us.ibm.com>
7683 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
7684 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
7686 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7687 VMULOSW): Add definitions.
7688 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7689 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7690 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7691 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7692 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7694 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
7696 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
7698 (movti_aarch64): Likewise.
7699 (movdf_aarch64): Likewise.
7700 (movtf_aarch64): Likewise.
7701 (load_pairdi): Likewise.
7702 (store_pairdi): Likewise.
7703 (load_pairdf): Likewise.
7704 (store_pairdf): Likewise.
7705 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
7706 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
7707 (ldr_got_small_<mode>): Likewise.
7708 (ldr_got_small_28k_<mode>): Likewise.
7709 (ldr_got_tiny): Likewise.
7710 * config/aarch64/iterators.md (ldst_sz): New.
7711 (ldpstp_sz): Likewise.
7712 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
7714 (thunderx_load): Split load_8 to load_16.
7715 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
7717 (thunderx2t99_storepair_basic): Split store_8 to store_16.
7718 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
7719 (xgene1_store_pair): Split store_8 to store_16.
7720 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
7721 (falkor_st_0_st_sd): Split store_8 to store_16.
7723 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
7725 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
7726 and store1/2/3/4 to store_4/8/12/16.
7727 * config/aarch64/aarch64.md: Update for rename.
7728 * config/arm/arm.md: Likewise.: Likewise.
7729 * config/arm/arm.c: Likewise.
7730 * config/arm/thumb1.md: Likewise.
7731 * config/arm/thumb2.md: Likewise.
7732 * config/arm/vfp.md: Likewise.
7733 * config/arm/arm-generic.md: Likewise.
7734 * config/arm/arm1020e.md: Likewise.
7735 * config/arm/arm1026ejs.md: Likewise.
7736 * config/arm/arm1136jfs.md: Likewise.
7737 * config/arm/arm926ejs.md: Likewise.
7738 * config/arm/cortex-a15.md: Likewise.
7739 * config/arm/cortex-a17.md: Likewise.
7740 * config/arm/cortex-a5.md: Likewise.
7741 * config/arm/cortex-a53.md: Likewise.
7742 * config/arm/cortex-a57.md: Likewise.
7743 * config/arm/cortex-a7.md: Likewise.
7744 * config/arm/cortex-a8.md: Likewise.
7745 * config/arm/cortex-a9.md: Likewise.
7746 * config/arm/cortex-m4.md: Likewise.
7747 * config/arm/cortex-m7.md: Likewise.
7748 * config/arm/cortex-r4.md: Likewise.
7749 * config/arm/exynos-m1.md: Likewise.
7750 * config/arm/fa526.md: Likewise.
7751 * config/arm/fa606te.md: Likewise.
7752 * config/arm/fa626te.md: Likewise.
7753 * config/arm/fa726te.md: Likewise.
7754 * config/arm/fmp626.md: Likewise.
7755 * config/arm/iwmmxt.md: Likewise.
7756 * config/arm/ldmstm.md: Likewise.
7757 * config/arm/marvell-pj4.md: Likewise.
7758 * config/arm/xgene1.md: Likewise.
7759 * config/aarch64/thunderx.md: Likewise.
7760 * config/aarch64/thunderx2t99.md: Likewise.
7761 * config/aarch64/falkor.md: Likewise.
7763 2017-09-12 Martin Liska <mliska@suse.cz>
7765 * attribs.c (private_lookup_attribute): New function.
7766 * attribs.h (private_lookup_attribute): Declared here.
7767 (lookup_attribute): Called from this place.
7769 2017-09-12 Richard Biener <rguenther@suse.de>
7771 PR tree-optimization/82157
7772 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
7773 stmts with side-effects.
7775 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7776 Alan Hayward <alan.hayward@arm.com>
7777 David Sherwood <david.sherwood@arm.com>
7779 * target.def (hard_regno_nregs): New hook.
7780 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
7781 * targhooks.h (default_hard_regno_nregs): Declare.
7782 * targhooks.c (default_hard_regno_nregs): New function.
7783 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
7784 (TARGET_HARD_REGNO_NREGS): ...this hook.
7785 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
7786 (CLASS_MAX_NREGS): Likewise.
7787 * doc/tm.texi: Regenerate.
7788 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
7789 instead of HARD_REGNO_NREGS.
7790 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
7791 HARD_REGNO_NREGS in the comment.
7792 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
7793 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
7794 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
7795 Return an unsigned int.
7796 (TARGET_HARD_REGNO_NREGS): Redefine.
7797 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
7798 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
7799 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
7800 (arc_hard_regno_nregs): New function.
7801 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
7802 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
7803 (arm_hard_regno_nregs): New function.
7804 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
7805 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
7806 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
7807 (TARGET_HARD_REGNO_NREGS): Redefine.
7808 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
7809 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
7810 (HARD_REGNO_NREGS): Delete.
7811 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
7812 (cr16_hard_regno_nregs): New function.
7813 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
7814 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
7815 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
7816 (cris_hard_regno_nregs): New function.
7817 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
7818 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
7819 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
7820 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
7821 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
7822 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
7823 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
7824 (frv_hard_regno_nregs): Make static. Take and return an
7826 (frv_class_max_nregs): Remove outdated copy of documentation.
7827 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
7828 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
7829 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
7830 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
7831 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
7832 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
7833 (TARGET_HARD_REGNO_NREGS): Redefine.
7834 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
7835 (CLASS_MAX_NREGS): Update comment.
7836 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
7837 (ia64_hard_regno_nregs): New function.
7838 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
7839 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
7840 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
7841 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
7842 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
7844 (m32c_hard_regno_nregs): Likewise. Make static.
7845 (TARGET_HARD_REGNO_NREGS): Redefine.
7846 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
7847 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
7848 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
7849 (m68k_hard_regno_nregs): New function.
7850 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
7851 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
7852 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
7853 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
7854 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
7855 Take and return an unsigned int.
7856 (TARGET_HARD_REGNO_NREGS): Redefine.
7857 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
7858 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
7859 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
7860 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
7861 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
7862 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
7863 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
7864 (msp430_hard_regno_nregs): Make static. Take and return an
7866 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
7867 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
7868 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
7869 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
7870 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
7871 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
7872 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
7873 (TARGET_HARD_REGNO_NREGS): Redefine.
7874 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
7875 (PA_HARD_REGNO_NREGS): ...this.
7876 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
7877 (PA_HARD_REGNO_NREGS): ...this.
7878 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
7879 (pa_hard_regno_nregs): New function.
7880 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
7881 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
7882 (pdp11_hard_regno_nregs): New function.
7883 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
7884 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
7885 (rs6000_hard_regno_nregs_hook): New function.
7886 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
7887 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
7888 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
7889 Take and return an unsigned int. Move earlier in file.
7890 (TARGET_HARD_REGNO_NREGS): Redefine.
7891 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
7892 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
7893 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
7894 (rl78_hard_regno_nregs): Make static. Take and return an
7896 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
7897 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
7898 (rs6000_hard_regno_nregs_hook): New function.
7899 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
7900 * config/rx/rx.c (rx_hard_regno_nregs): New function.
7901 (TARGET_HARD_REGNO_NREGS): Redefine.
7902 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
7903 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
7904 instead of HARD_REGNO_NREGS.
7905 (s390_hard_regno_nregs): New function.
7906 (s390_hard_regno_mode_ok): Add comment from s390.h.
7907 (TARGET_HARD_REGNO_NREGS): Redefine.
7908 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
7909 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
7910 (sh_hard_regno_nregs): New function.
7911 (sh_pass_in_reg_p): Use it.
7912 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
7913 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
7914 (sparc_hard_regno_nregs): New function.
7915 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
7916 * config/spu/spu.c (spu_hard_regno_nregs): New function.
7917 (spu_function_arg_advance): Use it, supplying a valid register number.
7918 (TARGET_HARD_REGNO_NREGS): Redefine.
7919 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
7920 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
7921 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
7922 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
7923 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
7924 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
7925 (CLASS_MAX_NREGS): Remove copy of old documentation.
7926 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
7927 (visium_hard_regno_nregs): New function.
7928 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
7929 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
7930 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
7931 xtensa_hard_regno_nregs): New function.
7932 * system.h (HARD_REGNO_NREGS): Poison.
7934 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7936 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
7937 hard_regno_nregs instead of HARD_REGNO_NREGS.
7938 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
7939 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
7940 (c6x_expand_epilogue): Likewise.
7941 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
7942 (frv_read_iacc_argument): Likewise.
7943 * config/sh/sh.c: Include regs.h.
7944 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
7945 (regs_used): Likewise.
7946 (output_stack_adjust): Likewise.
7947 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
7948 * expmed.c: Include regs.h.
7949 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
7950 * ree.c: Include regs.h.
7951 (combine_reaching_defs): Use hard_regno_nregs instead of
7953 (add_removable_extension): Likewise.
7955 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7957 * regs.h (hard_regno_nregs): Turn into a function.
7958 (end_hard_regno): Update accordingly.
7959 * caller-save.c (setup_save_areas): Likewise.
7960 (save_call_clobbered_regs): Likewise.
7961 (replace_reg_with_saved_mem): Likewise.
7962 (insert_restore): Likewise.
7963 (insert_save): Likewise.
7964 * combine.c (can_change_dest_mode): Likewise.
7965 (move_deaths): Likewise.
7966 (distribute_notes): Likewise.
7967 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
7968 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
7969 (rs6000_split_multireg_move): Likewise.
7970 (rs6000_register_move_cost): Likewise.
7971 (rs6000_memory_move_cost): Likewise.
7972 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
7973 (rs6000_split_multireg_move): Likewise.
7974 (rs6000_register_move_cost): Likewise.
7975 (rs6000_memory_move_cost): Likewise.
7976 * cselib.c (cselib_reset_table): Likewise.
7977 (cselib_lookup_1): Likewise.
7978 * emit-rtl.c (set_mode_and_regno): Likewise.
7979 * function.c (aggregate_value_p): Likewise.
7980 * ira-color.c (setup_profitable_hard_regs): Likewise.
7981 (check_hard_reg_p): Likewise.
7982 (calculate_saved_nregs): Likewise.
7983 (assign_hard_reg): Likewise.
7984 (improve_allocation): Likewise.
7985 (calculate_spill_cost): Likewise.
7986 * ira-emit.c (modify_move_list): Likewise.
7987 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
7988 (ira_hard_reg_in_set_p): Likewise.
7989 * ira.c (setup_reg_mode_hard_regset): Likewise.
7990 (clarify_prohibited_class_mode_regs): Likewise.
7991 (check_allocation): Likewise.
7992 * lra-assigns.c (find_hard_regno_for_1): Likewise.
7993 (lra_setup_reg_renumber): Likewise.
7994 (setup_try_hard_regno_pseudos): Likewise.
7995 (spill_for): Likewise.
7996 (assign_hard_regno): Likewise.
7997 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
7998 * lra-constraints.c (in_class_p): Likewise.
7999 (lra_constraint_offset): Likewise.
8000 (simplify_operand_subreg): Likewise.
8001 (lra_constraints): Likewise.
8002 (split_reg): Likewise.
8003 (split_if_necessary): Likewise.
8004 (invariant_p): Likewise.
8005 (inherit_in_ebb): Likewise.
8006 * lra-lives.c (process_bb_lives): Likewise.
8007 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
8008 (get_hard_regs): Likewise.
8009 (do_remat): Likewise.
8010 * lra-spills.c (assign_spill_hard_regs): Likewise.
8011 * mode-switching.c (create_pre_exit): Likewise.
8012 * postreload.c (reload_combine_recognize_pattern): Likewise.
8013 * recog.c (peep2_find_free_register): Likewise.
8014 * regcprop.c (kill_value_regno): Likewise.
8015 (set_value_regno): Likewise.
8016 (copy_value): Likewise.
8017 (maybe_mode_change): Likewise.
8018 (find_oldest_value_reg): Likewise.
8019 (copyprop_hardreg_forward_1): Likewise.
8020 * regrename.c (check_new_reg_p): Likewise.
8021 (regrename_do_replace): Likewise.
8022 * reload.c (push_reload): Likewise.
8023 (combine_reloads): Likewise.
8024 (find_dummy_reload): Likewise.
8025 (operands_match_p): Likewise.
8026 (find_reloads): Likewise.
8027 (find_equiv_reg): Likewise.
8028 (reload_adjust_reg_for_mode): Likewise.
8029 * reload1.c (count_pseudo): Likewise.
8030 (count_spilled_pseudo): Likewise.
8031 (find_reg): Likewise.
8032 (clear_reload_reg_in_use): Likewise.
8033 (free_for_value_p): Likewise.
8034 (allocate_reload_reg): Likewise.
8035 (choose_reload_regs): Likewise.
8036 (reload_adjust_reg_for_temp): Likewise.
8037 (emit_reload_insns): Likewise.
8038 (delete_output_reload): Likewise.
8039 * rtlanal.c (subreg_get_info): Likewise.
8040 * sched-deps.c (sched_analyze_reg): Likewise.
8041 * sel-sched.c (init_regs_for_mode): Likewise.
8042 (mark_unavailable_hard_regs): Likewise.
8043 (choose_best_reg_1): Likewise.
8044 (verify_target_availability): Likewise.
8045 * valtrack.c (dead_debug_insert_temp): Likewise.
8046 * var-tracking.c (track_loc_p): Likewise.
8047 (emit_note_insn_var_location): Likewise.
8048 * varasm.c (make_decl_rtl): Likewise.
8049 * reginfo.c (choose_hard_reg_mode): Likewise.
8050 (init_reg_modes_target): Refer directly to
8051 this_target_regs->x_hard_regno_nregs.
8053 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8055 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
8056 instead of hard_regno_nregs.
8058 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8060 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
8061 end_hard_regno instead of hard_regno_nregs.
8062 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
8063 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
8064 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
8065 * ira-color.c (improve_allocation): Likewise.
8066 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8067 * lra-lives.c (mark_regno_live): Likewise.
8068 (mark_regno_dead): Likewise.
8069 * lra-remat.c (operand_to_remat): Likewise.
8070 * lra.c (collect_non_operand_hard_regs): Likewise.
8071 * postreload.c (reload_combine_note_store): Likewise.
8072 (move2add_valid_value_p): Likewise.
8073 * reload.c (regno_clobbered_p): Likewise.
8075 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8077 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
8079 * config/v850/v850.c (v850_reorg): Likewise.
8080 * reload.c (refers_to_regno_for_reload_p): Likewise.
8081 (find_equiv_reg): Likewise.
8082 * reload1.c (reload_reg_reaches_end_p): Likewise.
8084 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8086 * caller-save.c (add_used_regs): Use REG_NREGS instead of
8088 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
8089 * config/arm/arm.c (output_move_neon): Likewise.
8090 (arm_attr_length_move_neon): Likewise.
8091 (neon_split_vcombine): Likewise.
8092 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
8093 (c6x_mark_reg_written): Likewise.
8094 (c6x_dwarf_register_span): Likewise.
8095 * config/i386/i386.c (ix86_save_reg): Likewise.
8096 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
8097 (rws_access_reg): Likewise.
8098 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
8099 * mode-switching.c (create_pre_exit): Likewise.
8100 * ree.c (combine_reaching_defs): Likewise.
8101 (add_removable_extension): Likewise.
8102 * regcprop.c (find_oldest_value_reg): Likewise.
8103 (copyprop_hardreg_forward_1): Likewise.
8104 * reload.c (reload_inner_reg_of_subreg): Likewise.
8105 (push_reload): Likewise.
8106 (combine_reloads): Likewise.
8107 (find_dummy_reload): Likewise.
8108 (reload_adjust_reg_for_mode): Likewise.
8109 * reload1.c (find_reload_regs): Likewise.
8110 (forget_old_reloads_1): Likewise.
8111 (reload_reg_free_for_value_p): Likewise.
8112 (reload_adjust_reg_for_temp): Likewise.
8113 (emit_reload_insns): Likewise.
8114 (delete_output_reload): Likewise.
8115 * sel-sched.c (choose_best_reg_1): Likewise.
8116 (choose_best_pseudo_reg): Likewise.
8118 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8119 Alan Hayward <alan.hayward@arm.com>
8120 David Sherwood <david.sherwood@arm.com>
8122 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
8123 * target.def (slow_unaligned_access): New hook.
8124 * targhooks.h (default_slow_unaligned_access): Declare.
8125 * targhooks.c (default_slow_unaligned_access): New function.
8126 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
8127 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
8128 * doc/tm.texi: Regenerate.
8129 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
8130 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
8131 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
8133 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
8134 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
8136 (rs6000_slow_unaligned_access): New function.
8137 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8138 (expand_block_compare): Likewise.
8139 (expand_strn_compare): Likewise.
8140 (rs6000_rtx_costs): Likewise.
8141 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
8142 (riscv_slow_unaligned_access): Likewise.
8143 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
8144 (riscv_slow_unaligned_access_p): ...this and make static.
8145 (riscv_option_override): Update accordingly.
8146 (riscv_slow_unaligned_access): New function.
8147 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8148 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
8149 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8150 (rs6000_slow_unaligned_access): New function.
8151 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8152 (rs6000_rtx_costs): Likewise.
8153 * config/rs6000/rs6000-string.c (expand_block_compare)
8154 (expand_strn_compare): Use targetm.slow_unaligned_access instead
8155 of SLOW_UNALIGNED_ACCESS.
8156 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
8157 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
8158 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
8159 of SLOW_UNALIGNED_ACCESS.
8160 * expmed.c (simple_mem_bitfield_p): Likewise.
8161 * expr.c (alignment_for_piecewise_move): Likewise.
8162 (emit_group_load_1): Likewise.
8163 (emit_group_store): Likewise.
8164 (copy_blkmode_from_reg): Likewise.
8165 (emit_push_insn): Likewise.
8166 (expand_assignment): Likewise.
8167 (store_field): Likewise.
8168 (expand_expr_real_1): Likewise.
8169 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
8170 * lra-constraints.c (simplify_operand_subreg): Likewise.
8171 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
8172 * gimple-ssa-store-merging.c: Likewise in block comment at start
8174 * tree-ssa-strlen.c: Include target.h.
8175 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
8176 of SLOW_UNALIGNED_ACCESS.
8177 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
8179 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8181 PR rtl-optimization/82185
8182 * expmed.c (emit_store_flag_int): Only test tem if it has been
8185 2017-09-12 Richard Biener <rguenther@suse.de>
8188 * match.pd ((FTYPE) N CMP CST): Fix typo.
8190 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
8192 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
8194 (mips_near_type_p): Add 'short_call' attribute as a synonym
8196 * doc/extend.texi (short_call): Document new function attribute.
8198 2017-09-12 Jakub Jelinek <jakub@redhat.com>
8201 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
8202 assertion check that in the condition.
8203 (get_atomic_generic_size): Likewise. Before testing if parameter
8204 has pointer type, if it has array type, call for C++
8205 default_conversion to perform array-to-pointer conversion.
8207 2017-09-12 Richard Biener <rguenther@suse.de>
8209 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
8210 for operations we cannot scalarize.
8212 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
8214 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
8215 vectors heap vectors. Clean up comments.
8216 Make visited_bbs a reference.
8217 (profitable_jump_thread_path): Make GC
8218 vectors heap vectors. Clean up comments.
8220 (convert_and_register_jump_thread_path): Make GC vectors heap
8222 (check_subpath_and_update_thread_path): Same. Clean up comments.
8223 Make visited_bbs a reference.
8224 (handle_phi): Abstract common code to to
8225 register_jump_thread_path_if_profitable.
8226 Rename VAR_BB to DEF_BB.
8228 Make GC vectors heap vectors.
8229 Make visited_bbs a reference.
8230 (handle_assignment): Same.
8231 (register_jump_thread_path_if_profitable): New.
8232 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
8234 Make GC vectors heap vectors. Clean up comments.
8235 Make visited_bbs a reference.
8236 (find_jump_threads_backwards): Make visited_bbs live in the stack.
8237 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
8240 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
8243 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
8244 words of E_DImode object are reachable by xtensa_uimm8x4 access.
8246 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
8248 Revert r251800 and r251799.
8250 2017-09-11 Martin Jambor <mjambor@suse.cz>
8253 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
8254 arguments in advance.
8255 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
8256 use it to find predecessor edges.
8257 (naive_outof_ssa): Collect vector of predecessors.
8259 2017-09-08 Jason Merrill <jason@redhat.com>
8261 PR c++/70029 - ICE with ref-qualifier and -flto
8262 * langhooks.h (struct lang_hooks_for_types): Add
8263 copy_lang_qualifiers.
8264 * attribs.c (build_type_attribute_qual_variant): Use it.
8265 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
8267 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
8268 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
8270 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
8273 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
8274 (*mulsi3_sp64): New instruction.
8275 (mulsi3): New expander.
8277 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
8279 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
8281 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8283 * sancov.c: Include memmodel.h.
8285 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
8288 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
8291 2017-09-07 Carl Love <cel@us.ibm.com>
8293 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
8294 the sldi instruction.
8296 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
8298 * sancov.c: Include tm_p.h.
8300 2017-09-07 Jakub Jelinek <jakub@redhat.com>
8303 * output.h (switch_to_other_text_partition): New declaration.
8304 * varasm.c (switch_to_other_text_partition): New function.
8305 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
8306 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
8307 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
8308 to the other text partition before emitting LCL label and switch back
8309 after emitting the word after it.
8311 2017-09-07 Richard Biener <rguenther@suse.de>
8313 * passes.def (pass_split_crit_edges): Remove instance before PRE.
8314 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
8315 critical edges here, after loop init.
8316 (pass_data_pre): Remove PROP_no_crit_edges flags.
8317 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
8318 for valueization of call args to avoid leaking VN_TOP.
8319 (visit_use): Assert we do not visit default defs.
8320 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
8321 Use error_mark_node to more easily detect leaking VN_TOP.
8322 All default-defs are varying, not VN_TOP. Mark them visited.
8323 (run_scc_vn): Make code match comment.
8325 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8327 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
8328 OPTION_MASK_FLOAT128_KEYWORD.
8329 (POWERPC_MASKS): Likewise.
8330 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
8331 support for the -mfloat128-type option, and make -mfloat128
8332 default on PowerPC Linux systems. Define or undefine
8333 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
8334 Define __float128 to be __ieee128 if IEEE 128-bit support is
8335 enabled, or undefine it.
8336 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
8337 Delete defining __FLOAT128_TYPE__.
8338 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
8339 -mfloat128-type option and make -mfloat128 default on PowerPC
8341 (TARGET_FLOAT128_TYPE): Likewise.
8342 (-mfloat128-type): Likewise.
8343 * config/rs6000/rs6000.c (rs6000_option_override_internal):
8344 Delete the -mfloat128-type option and make -mfloat128 default on
8345 PowerPC Linux systems. Always use __ieee128 to be the keyword for
8346 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
8347 128-bit floating point is enabled. Change tests from using
8348 -mfloat128-type to -mfloat128.
8349 (rs6000_mangle_type): Use the correct mangling for the __float128
8350 type even if normal long double is restricted to 64-bits.
8351 (floatn_mode): Enable the _Float128 type by default on VSX Linux
8353 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
8354 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
8355 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
8357 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
8358 documentation for -mfloat128.
8360 2017-09-06 Olivier Hainque <hainque@adacore.com>
8362 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
8364 2017-09-06 Wish Wu <wishwu007@gmail.com>
8365 Jakub Jelinek <jakub@redhat.com>
8367 * asan.c (initialize_sanitizer_builtins): Add
8368 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
8369 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
8370 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
8371 BT_FN_VOID_UINT64_PTR variables.
8372 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
8373 (BT_FN_VOID_UINT16_UINT16): Likewise.
8374 (BT_FN_VOID_UINT32_UINT32): Likewise.
8375 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
8376 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
8377 (BT_FN_VOID_UINT64_PTR): Likewise.
8378 * common.opt (flag_sanitize_coverage): New variable.
8379 (fsanitize-coverage=trace-pc): Remove.
8380 (fsanitize-coverage=): Add.
8381 * flag-types.h (enum sanitize_coverage_code): New enum.
8382 * fold-const.c (fold_range_test): Disable non-short-circuit
8383 optimization if flag_sanitize_coverage.
8384 (fold_truth_andor): Likewise.
8385 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
8386 * opts.c (COVERAGE_SANITIZER_OPT): Define.
8387 (coverage_sanitizer_opts): New array.
8388 (get_closest_sanitizer_option): Add OPTS argument, handle also
8389 OPT_fsanitize_coverage_.
8390 (parse_sanitizer_options): Adjusted to also handle
8391 OPT_fsanitize_coverage_.
8392 (common_handle_option): Add OPT_fsanitize_coverage_.
8393 * sancov.c (instrument_comparison, instrument_switch): New function.
8394 (sancov_pass): Add trace-cmp support.
8395 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
8396 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
8397 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
8398 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
8399 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
8400 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
8401 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
8402 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
8403 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
8404 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
8406 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
8408 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
8409 error. Only quit immediately if parsing is complete.
8410 (BEGIN): Initialize fatal_err and parse_done.
8411 (begin fpu, end fpu): Check number of arguments.
8412 (begin arch, end arch): Likewise.
8413 (begin cpu, end cpu): Likewise.
8414 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
8415 (optalias): Likewise.
8417 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
8419 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
8420 * config/arm/arm-isa.h: Delete. Move definitions to ...
8421 * arm-cpus.in: ... here. Use new feature and fgroup values.
8422 * config/arm/arm.c (arm_option_override): Use lower case for feature
8424 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
8425 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
8426 * config/arm/parsecpu.awk (END): Add new command 'isa'.
8428 (print_isa_bits_for): New function.
8429 (gen_isa): New function.
8430 (gen_comm_data): Use print_isa_bits_for.
8431 (define feature): New keyword.
8432 (define fgroup): New keyword.
8433 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
8434 (arm-isa.h): Add rule to generate file.
8435 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
8436 case for feature bit names.
8438 2017-09-06 Richard Biener <rguenther@suse.de>
8440 * tree-ssa-pre.c (NECESSARY): Remove.
8441 (create_expression_by_pieces): Do not touch pass-local flags.
8442 (insert_into_preds_of_block): Likewise.
8443 (do_pre_regular_insertion): Likewise.
8444 (eliminate_insert): Likewise.
8445 (eliminate_dom_walker::before_dom_children): Likewise.
8446 (fini_eliminate): Do not look at inserted_exprs.
8447 (mark_operand_necessary): Remove.
8448 (remove_dead_inserted_code): Replace with simple work-list
8449 algorithm based on inserted_exprs and SSA uses.
8450 (pass_pre::execute): Re-order fini_eliminate and
8451 remove_dead_inserted_code.
8453 2017-09-06 Olivier Hainque <hainque@adacore.com>
8455 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
8456 for VxWorks 7. Adjust surrounding comments.
8458 2017-09-06 Richard Biener <rguenther@suse.de>
8460 * gimple-ssa-strength-reduction.c
8461 (find_candidates_dom_walker::before_dom_children): Also allow
8464 2017-09-06 Richard Biener <rguenther@suse.de>
8466 PR tree-optimization/82108
8467 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
8468 for gap in the non-permutation SLP case.
8470 2017-09-06 Martin Jambor <mjambor@suse.cz>
8472 PR tree-optimization/82078
8473 * tree-sra.c (sort_and_splice_var_accesses): Move call to
8474 add_access_to_work_queue...
8475 (build_accesses_from_assign): ...here.
8476 (propagate_all_subaccesses): Make sure racc is the group
8477 representative, if there is one.
8479 2017-09-06 Jakub Jelinek <jakub@redhat.com>
8482 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
8485 2017-09-06 Richard Biener <rguenther@suse.de>
8487 * gimple-ssa-strength-reduction.c
8488 (find_candidates_dom_walker::before_doom_children): Use a
8489 type and not a mode check.
8491 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
8494 * config/arm/predicates.md (arm_general_adddi_operand): Create new
8496 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
8498 2017-09-05 Jeff Law <law@redhat.com>
8500 PR tree-optimization/64910
8501 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
8502 cases where we have 3 or more operands.
8504 2017-09-05 Jakub Jelinek <jakub@redhat.com>
8507 * omp-low.c (lower_omp_for): Recompute tree invariant if
8508 gimple_omp_for_initial/final is ADDR_EXPR.
8511 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
8512 into gimple val before gimplification fo the COND_EXPR.
8514 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
8516 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
8517 REGION_COPY argument.
8518 (thread_through_all_blocks): Remove unused argument to
8519 duplicate_thread_path.
8521 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8522 Alan Hayward <alan.hayward@arm.com>
8523 David Sherwood <david.sherwood@arm.com>
8525 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
8526 Take a scalar_mode rather than a machine_mode.
8527 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8528 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
8529 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8530 (aarch64_gen_adjusted_ldpstp): Likewise.
8531 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
8533 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8534 Alan Hayward <alan.hayward@arm.com>
8535 David Sherwood <david.sherwood@arm.com>
8537 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
8538 Take a scalar_int_mode instead of a machine_mode.
8539 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8540 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8541 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8542 (aarch64_simd_attr_length_rglist): Delete.
8543 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
8544 a scalar_int_mode instead of a machine_mode.
8545 (aarch64_add_offset): Likewise.
8546 (aarch64_internal_mov_immediate): Likewise
8547 (aarch64_add_constant_internal): Likewise.
8548 (aarch64_add_constant): Likewise.
8549 (aarch64_movw_imm): Likewise.
8550 (aarch64_rtx_arith_op_extract_p): Likewise.
8551 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8552 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8553 Remove assert that the mode isn't a vector.
8554 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8555 (aarch64_expand_mov_immediate): Update calls after above changes.
8556 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
8557 (aarch64_and_bitmask_imm): Check for scalar integer modes.
8558 (aarch64_move_imm): Likewise.
8559 (aarch64_can_const_movi_rtx_p): Likewise.
8560 (aarch64_strip_extend): Likewise.
8561 (aarch64_extr_rtx_p): Likewise.
8562 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
8563 a CONST_INT when the mode parameter is VOIDmode.
8564 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
8566 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8568 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
8569 * stor-layout.c (bitwise_mode_for_mode): Likewise.
8570 (bitwise_type_for_mode): Update accordingly.
8572 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8574 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
8575 * stor-layout.c (mode_for_size_tree): Likewise.
8576 (mode_for_array): Update accordingly.
8577 (layout_decl): Likewise.
8578 (compute_record_mode): Likewise. Only set the mode once.
8580 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8582 * target.def (get_mask_mode): Change return type to opt_mode.
8584 * doc/tm.texi: Regenerate.
8585 * targhooks.h (default_get_mask_mode): Return an opt_mode.
8586 * targhooks.c (default_get_mask_mode): Likewise.
8587 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
8588 * optabs-query.c (can_vec_mask_load_store_p): Update use of
8589 targetm.get_mask_mode.
8590 * tree.c (build_truth_vector_type): Likewise.
8592 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8594 * machmode.h (mode_for_vector): Return an opt_mode.
8595 * stor-layout.c (mode_for_vector): Likewise.
8596 (mode_for_int_vector): Update accordingly.
8597 (layout_type): Likewise.
8598 * config/i386/i386.c (emit_memmov): Likewise.
8599 (ix86_expand_set_or_movmem): Likewise.
8600 (ix86_expand_vector_init): Likewise.
8601 (ix86_get_mask_mode): Likewise.
8602 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
8604 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
8605 * expmed.c (extract_bit_field_1): Likewise.
8606 * expr.c (expand_expr_real_2): Likewise.
8607 * optabs-query.c (can_vec_perm_p): Likewise.
8608 (can_vec_mask_load_store_p): Likewise.
8609 * optabs.c (expand_vec_perm): Likewise.
8610 * targhooks.c (default_get_mask_mode): Likewise.
8611 * tree-vect-stmts.c (vectorizable_store): Likewise.
8612 (vectorizable_load): Likewise.
8613 (get_vectype_for_scalar_type_and_size): Likewise.
8615 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8617 * machmode.h (mode_for_int_vector): New function.
8618 * stor-layout.c (mode_for_int_vector): Likewise.
8619 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
8620 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
8621 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
8622 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
8623 (s390_expand_vcond): Likewise.
8625 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8627 * machmode.h (opt_machine_mode): New type.
8628 (opt_mode<T>): Allow construction from anything that can be
8630 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
8631 (mode_for_size): Return an opt_machine_mode.
8632 * stor-layout.c (mode_for_size): Likewise.
8633 (mode_for_size_tree): Update call accordingly.
8634 (bitwise_mode_for_mode): Likewise.
8635 (make_fract_type): Likewise.
8636 (make_accum_type): Likewise.
8637 * caller-save.c (replace_reg_with_saved_mem): Update call
8639 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8640 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8641 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8642 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8643 * expmed.c (extract_bit_field_1): Likewise.
8644 * reload.c (get_secondary_mem): Likewise.
8645 * varasm.c (assemble_integer): Likewise.
8646 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
8649 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8651 * machmode.h (decimal_float_mode_for_size): New function.
8652 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
8653 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
8654 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
8655 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
8656 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
8657 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
8659 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8661 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
8662 (get_builtin_sync_mode): Likewise.
8663 (expand_ifn_atomic_compare_exchange): Likewise.
8664 (expand_builtin_atomic_clear): Likewise.
8665 (expand_builtin_atomic_test_and_set): Likewise.
8666 (fold_builtin_atomic_always_lock_free): Likewise.
8667 * calls.c (compute_argument_addresses): Likewise.
8668 (emit_library_call_value_1): Likewise.
8669 (store_one_arg): Likewise.
8670 * combine.c (combine_instructions): Likewise.
8671 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
8672 * config/arm/arm.c (arm_function_value): Likewise.
8673 (aapcs_allocate_return_reg): Likewise.
8674 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
8675 * config/i386/i386.c (construct_container): Likewise.
8676 (ix86_gimplify_va_arg): Likewise.
8677 (ix86_expand_sse_cmp): Likewise.
8678 (emit_memmov): Likewise.
8679 (emit_memset): Likewise.
8680 (expand_small_movmem_or_setmem): Likewise.
8681 (ix86_expand_pextr): Likewise.
8682 (ix86_expand_pinsr): Likewise.
8683 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
8684 * config/microblaze/microblaze.c (microblaze_block_move_straight):
8686 * config/mips/mips.c (mips_function_value_1) Likewise.
8687 (mips_block_move_straight): Likewise.
8688 (mips_expand_ins_as_unaligned_store): Likewise.
8689 * config/powerpcspe/powerpcspe.c
8690 (rs6000_darwin64_record_arg_advance_flush): Likewise.
8691 (rs6000_darwin64_record_arg_flush): Likewise.
8692 * config/rs6000/rs6000.c
8693 (rs6000_darwin64_record_arg_advance_flush): Likewise.
8694 (rs6000_darwin64_record_arg_flush): Likewise.
8695 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
8696 (sparc_function_value_1): Likewise.
8697 * config/spu/spu.c (adjust_operand): Likewise.
8698 (spu_emit_branch_or_set): Likewise.
8699 (arith_immediate_p): Likewise.
8700 * emit-rtl.c (gen_lowpart_common): Likewise.
8701 * expr.c (expand_expr_real_1): Likewise.
8702 * function.c (assign_parm_setup_block): Likewise.
8703 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
8704 * reload1.c (alter_reg): Likewise.
8705 * stor-layout.c (mode_for_vector): Likewise.
8706 (layout_type): Likewise.
8708 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8710 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
8711 (spu_convert_move): Likewise.
8712 * lower-subreg.c (resolve_simple_move): Likewise.
8714 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8717 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
8718 define_insn to a define_expand.
8719 (altivec_vsum2sws_direct): New define_insn.
8720 (altivec_vsumsws): Convert from a define_insn to a define_expand.
8722 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
8724 * config/arm/arm.c (arm_option_params_internal): Improve setting of
8727 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
8732 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
8733 realign stack if stack alignment needed is less than incoming
8736 2017-09-05 Marek Polacek <polacek@redhat.com>
8739 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
8742 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
8744 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
8746 2017-09-05 Richard Biener <rguenther@suse.de>
8748 PR tree-optimization/82084
8749 * fold-const.c (can_native_encode_string_p): Handle wide characters.
8751 2017-09-05 Richard Biener <rguenther@suse.de>
8753 PR tree-optimization/82102
8754 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
8756 2017-09-05 Martin Liska <mliska@suse.cz>
8758 PR tree-optimization/82032
8759 * tree-cfg.c (generate_range_test): New function.
8760 * tree-cfg.h (generate_range_test): Declared here.
8761 * tree-cfgcleanup.c (convert_single_case_switch): New function.
8762 (cleanup_control_expr_graph): Use it.
8763 * tree-switch-conversion.c (try_switch_expansion): Remove
8765 (emit_case_nodes): Use generate_range_test.
8767 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
8770 * config/i386/i386.md (*<btsc><mode>_mask): Add
8771 TARGET_USE_BT to insn constraint.
8772 (*btr<mode>_mask): Ditto.
8774 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
8776 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
8777 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
8779 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
8782 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
8783 TARGET_NEON and TARGET_IWMMXT.
8784 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
8785 TARGET_NEON and TARGET_IWMMXT.
8786 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
8788 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
8790 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
8791 (ix86_rewrite_tls_address): Ditto.
8792 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
8793 (ix86_rewrite_tls_address_1): Ditto.
8794 (ix86_rewrite_tls_address): Ditto.
8795 * config/i386/predicates.md (tls_address_pattern): New predicate.
8796 * config/i386/i386.md (TLS address splitter): New splitter.
8798 2017-09-04 Richard Biener <rguenther@suse.de>
8800 PR tree-optimization/82084
8801 * fold-const.h (can_native_encode_string_p): Declare.
8802 * fold-const.c (can_native_encode_string_p): Factor out from ...
8803 (native_encode_string): ... here.
8804 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
8805 vectorizing stores from constants we later cannot handle.
8807 2017-09-04 Marek Polacek <polacek@redhat.com>
8810 * doc/invoke.texi: Update -Wtautological-compare documentation.
8812 2017-09-04 Jeff Law <law@redhat.com>
8814 PR tree-optimization/64910
8815 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
8816 swap the first and last operand if the last is a constant.
8818 2017-09-04 Marek Polacek <polacek@redhat.com>
8821 * convert.c (do_narrow): When sanitizing signed integer overflows,
8822 bail out for signed types.
8823 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
8825 2017-09-04 Richard Biener <rguenther@suse.de>
8827 PR tree-optimization/82060
8828 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
8829 Move devirtualization after stmt folding and before EH/AB/noreturn
8830 cleanup to get the stmt refs canonicalized. Use a bool instead
8831 of gimple_modified_p since that doesn't work for NOPs. Schedule
8832 NOPs generated by folding for removal.
8834 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8835 Alan Hayward <alan.hayward@arm.com>
8836 David Sherwood <david.sherwood@arm.com>
8838 * coretypes.h (pad_direction): New enum.
8839 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
8840 (FUNCTION_ARG_PADDING): Likewise.
8841 * target.def (function_arg_padding): New hook.
8842 * targhooks.h (default_function_arg_padding): Declare.
8843 * targhooks.c (default_function_arg_padding): New function.
8844 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
8845 (TARGET_FUNCTION_ARG_PADDING): ...this.
8846 * doc/tm.texi: Regenerate.
8847 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
8848 instead of direction.
8849 (compute_argument_addresses): Likewise.
8850 (load_register_parameters): Likewise.
8851 (emit_library_call_value_1): Likewise.
8852 (store_one_arg): Use targetm.calls.function_arg_padding instead
8853 of FUNCTION_ARG_PADDING.
8854 (must_pass_in_stack_var_size_or_pad): Likewise.
8855 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
8856 (emit_group_store): Likewise.
8857 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
8858 instead of FUNCTION_ARG_PADDING.
8859 (emit_push_insn): Likewise, and propagate enum change throughout
8861 * function.h (direction): Delete.
8862 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
8864 * function.c (assign_parm_find_stack_rtl): Likewise.
8865 (assign_parm_setup_block_p): Likewise.
8866 (assign_parm_setup_block): Likewise.
8867 (gimplify_parameters): Likewise.
8868 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
8869 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
8871 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
8872 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8873 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
8874 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
8875 (aarch64_function_arg_padding): ...this new function.
8876 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
8877 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8878 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
8879 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8880 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
8881 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8882 (arm_pad_arg_upward): Replace with...
8883 (arm_function_arg_padding): ...this new function.
8884 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
8886 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
8887 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
8888 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8889 (ia64_hpux_function_arg_padding): Replace with...
8890 (ia64_function_arg_padding): ...this new function. Use pad_direction
8891 instead of direction. Check for TARGET_HPUX.
8892 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
8893 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8894 (iq2000_function_arg_padding): New function.
8895 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
8896 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
8897 (mips_function_arg_padding): ...this new function.
8898 (mips_pad_reg_upward): Update accordingly.
8899 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8900 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
8901 targetm.calls.function_arg_padding.
8902 (FUNCTION_ARG_PADDING): Delete.
8903 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8904 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
8905 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8906 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
8907 (nios2_block_reg_padding): Return pad_direction instead of direction.
8908 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
8909 instead of direction.
8910 (nios2_function_arg_padding): Likewise. Make static.
8911 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8912 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
8913 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
8914 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
8915 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8916 (pa_function_arg_padding): Make static. Return pad_direction instead
8918 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
8919 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8920 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
8921 instead of direction. Use targetm.calls.function_arg_padding.
8922 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
8923 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
8924 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
8925 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
8926 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
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/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
8933 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8934 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
8935 instead of direction. Use targetm.calls.function_arg_padding.
8936 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
8937 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
8938 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
8939 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
8940 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8941 (function_arg_padding): Rename to...
8942 (rs6000_function_arg_padding): ...this. Make static. Return
8943 pad_direction instead of direction.
8944 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
8945 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
8946 * config/s390/s390.c (s390_function_arg_padding): New function.
8947 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8948 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
8949 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
8950 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8951 (function_arg_padding): Rename to...
8952 (sparc_function_arg_padding): ...this. Make static. Return
8953 pad_direction instead of direction.
8954 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
8955 * config/spu/spu.c (spu_function_arg_padding): New function.
8956 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8957 * system.h (FUNCTION_ARG_PADDING): Poison.
8959 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8960 Alan Hayward <alan.hayward@arm.com>
8961 David Sherwood <david.sherwood@arm.com>
8963 * target.def (modes_tieable_p): New hook.
8964 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
8965 (TARGET_MODES_TIEABLE_P): ...this.
8966 * doc/tm.texi.in: Regenerate.
8967 * hooks.h (hook_bool_mode_mode_true): Declare.
8968 * hooks.c (hook_bool_mode_mode_true): New function.
8969 * combine.c (subst): Use targetm.modes_tieable_p instead of
8971 * dse.c (find_shift_sequence): Likewise.
8972 * expmed.c (extract_low_bits): Likewise.
8973 * lower-subreg.c: Include target.h.
8974 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
8976 * rtlanal.c (rtx_cost): Likewise.
8977 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
8978 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
8979 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
8980 (TARGET_MODES_TIEABLE_P): Redefine.
8981 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
8982 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
8983 (TARGET_MODES_TIEABLE_P): Redefine.
8984 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
8985 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
8986 (arc_modes_tieable_p): New function.
8987 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
8988 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
8989 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
8990 (arm_modes_tieable_p): Make static.
8991 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
8992 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
8993 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
8994 (TARGET_MODES_TIEABLE_P): Redefine.
8995 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
8996 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
8997 (TARGET_MODES_TIEABLE_P): Redefine.
8998 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
8999 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
9000 (cr16_modes_tieable_p): New function.
9001 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
9002 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
9003 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
9004 (TRULY_NOOP_TRUNCATION): Update comment.
9005 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
9006 (TRULY_NOOP_TRUNCATION): Update comment.
9007 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
9008 (frv_modes_tieable_p): New function.
9009 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
9010 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
9011 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
9012 (TARGET_MODES_TIEABLE_P): Redefine.
9013 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
9014 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
9015 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
9016 (TARGET_MODES_TIEABLE_P): Redefine.
9017 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
9018 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
9019 (ia64_modes_tieable_p): New function.
9020 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
9021 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
9022 (iq2000_modes_tieable_p): New function.
9023 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
9024 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
9025 (lm32_modes_tieable_p): New function.
9026 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
9027 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
9028 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
9029 (TARGET_MODES_TIEABLE_P): Redefine.
9030 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
9031 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
9032 (m32r_modes_tieable_p): New function.
9033 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
9034 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
9035 (m68k_modes_tieable_p): New function.
9036 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
9037 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
9038 (mcore_modes_tieable_p): New function.
9039 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
9040 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
9042 (TARGET_MODES_TIEABLE_P): Redefine.
9043 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
9044 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
9045 * config/mips/mips.c (mips_modes_tieable_p): Make static.
9046 (TARGET_MODES_TIEABLE_P): Redefine.
9047 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
9048 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
9049 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
9050 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
9051 (mn10300_modes_tieable_p): ...this and make static.
9052 (TARGET_MODES_TIEABLE_P): Redefine.
9053 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
9054 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
9055 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
9056 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
9057 (msp430_modes_tieable_p): Make static.
9058 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
9059 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
9060 (TARGET_MODES_TIEABLE_P): Redefine.
9061 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
9062 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
9063 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
9064 (TARGET_MODES_TIEABLE_P): Redefine.
9065 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
9066 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
9067 * config/pa/pa.c (pa_modes_tieable_p): Make static.
9068 (TARGET_MODES_TIEABLE_P): Redefine.
9069 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
9070 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
9071 (pdp11_modes_tieable_p): New function.
9072 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
9073 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
9074 (rs6000_modes_tieable_p): New function.
9075 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9076 * config/powerpcspe/powerpcspe.md: Update comment.
9077 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
9078 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
9079 (TARGET_MODES_TIEABLE_P): Redefine.
9080 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
9081 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
9082 (rl78_modes_tieable_p): New function.
9083 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
9084 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
9085 (rs6000_modes_tieable_p): New function.
9086 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9087 * config/rs6000/rs6000.md: Update comment.
9088 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
9089 * config/rx/rx.c (rx_modes_tieable_p): New function.
9090 (TARGET_MODES_TIEABLE_P): Redefine.
9091 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
9092 * config/s390/s390.c (s390_modes_tieable_p): New function.
9093 (TARGET_MODES_TIEABLE_P): Redefine.
9094 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
9095 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
9096 (sh_modes_tieable_p): New function.
9097 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
9098 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
9099 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
9100 (sparc_modes_tieable_p): Make static.
9101 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
9102 * config/spu/spu.c (spu_modes_tieable_p): New function.
9103 (TARGET_MODES_TIEABLE_P): Redefine.
9104 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
9105 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
9106 (TARGET_MODES_TIEABLE_P): Redefine.
9107 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
9108 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
9109 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
9110 * config/v850/v850.c (v850_modes_tieable_p): New function.
9111 (TARGET_MODES_TIEABLE_P): Redefine.
9112 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
9113 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
9114 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
9115 (visium_modes_tieable_p): New function.
9116 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
9117 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
9118 (xtensa_modes_tieable_p): New function.
9119 * system.h (MODES_TIEABLE_P): Poison.
9121 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9122 Alan Hayward <alan.hayward@arm.com>
9123 David Sherwood <david.sherwood@arm.com>
9125 * target.def (hard_regno_mode_ok): New hook.
9126 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
9127 (TARGET_HARD_REGNO_MODE_OK): ...this.
9128 * doc/tm.texi.in: Regenerate.
9129 * hooks.h (hook_bool_uint_mode_true): Declare.
9130 * hooks.c (hook_bool_uint_mode_true): New function.
9131 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
9133 * genpreds.c (write_insn_preds_c): Add an include of target.h.
9134 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
9135 instead of HARD_REGNO_MODE_OK.
9136 * caller-save.c: Include target.h.
9137 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
9139 * combine.c (can_combine_p): Likewise.
9140 (combinable_i3pat): Likewise.
9141 (can_change_dest_mode): Likewise.
9142 * expr.c (init_expr_target): Likewise.
9143 (convert_move): Likewise.
9144 (convert_modes): Likewise.
9145 * ira.c (setup_prohibited_class_mode_regs): Likewise.
9146 (setup_prohibited_mode_move_regs): Likewise.
9147 * ira.h (target_ira): Likewise.
9148 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9149 * lra-constraints.c (process_alt_operands): Likewise.
9150 (split_reg): Likewise.
9151 * recog.c (peep2_find_free_register): Likewise.
9152 * ree.c (combine_reaching_defs): Likewise.
9153 * regcprop.c (maybe_mode_change): Likewise.
9154 * reginfo.c (init_reg_sets_1): Likewise.
9155 (choose_hard_reg_mode): Likewise.
9156 (simplifiable_subregs): Likewise.
9157 * regrename.c (check_new_reg_p): Likewise.
9158 * reload.c (find_valid_class): Likewise.
9159 (find_valid_class_1): Likewise.
9160 (reload_inner_reg_of_subreg): Likewise.
9161 (push_reload): Likewise.
9162 (combine_reloads): Likewise.
9163 (find_dummy_reload): Likewise.
9164 (find_reloads): Likewise.
9165 * reload1.c (find_reg): Likewise.
9166 (set_reload_reg): Likewise.
9167 (allocate_reload_reg): Likewise.
9168 (choose_reload_regs): Likewise.
9169 (reload_adjust_reg_for_temp): Likewise.
9170 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
9171 (simplify_subreg_regno): Likewise.
9172 * sel-sched.c (init_regs_for_mode): Likewise.
9173 * varasm.c (make_decl_rtl): Likewise.
9174 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
9175 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
9177 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
9178 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
9179 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9180 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
9181 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
9182 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9183 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
9184 (arc_mode_class): Delete.
9185 (HARD_REGNO_MODE_OK): Delete.
9186 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9187 (arc_hard_regno_mode_ok): Rename old array to...
9188 (arc_hard_regno_mode_ok_modes): ...this.
9189 (arc_conditional_register_usage): Update accordingly.
9190 (arc_mode_class): Make static.
9191 (arc_hard_regno_mode_ok): New function.
9192 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
9193 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
9194 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9195 (arm_hard_regno_mode_ok): Make static.
9196 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
9198 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
9199 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
9200 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
9202 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9203 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
9204 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
9205 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
9206 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
9207 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9208 * config/bfin/predicates.md (valid_reg_operand): Use
9209 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
9210 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
9211 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
9212 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9213 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
9214 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
9215 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9216 (cr16_hard_regno_mode_ok): Make static and return a bool.
9217 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
9218 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9219 (cris_hard_regno_mode_ok): New function.
9220 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
9221 (epiphany_mode_class): Delete.
9222 (HARD_REGNO_MODE_OK): Delete.
9223 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
9224 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9225 (hard_regno_mode_ok): Rename to...
9226 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
9227 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
9228 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
9230 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
9231 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
9232 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9233 (frv_hard_regno_mode_ok): Make static and return a bool.
9234 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
9236 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
9237 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
9238 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
9239 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
9241 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9242 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
9243 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
9244 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
9246 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9247 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
9248 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9249 (ia64_hard_regno_mode_ok): New function.
9250 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
9251 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9252 (iq2000_hard_regno_mode_ok): New function.
9253 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
9254 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9255 (lm32_hard_regno_mode_ok): New function.
9256 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
9257 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
9258 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
9259 instead of HARD_REGNO_MODE_OK.
9260 (m32c_hard_regno_ok): Rename to...
9261 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
9262 (m32c_cannot_change_mode_class): Update accordingly.
9263 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9264 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
9265 (m32r_mode_class): Delete.
9266 (HARD_REGNO_MODE_OK): Delete.
9267 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9268 (m32r_hard_regno_mode_ok): Rename to...
9269 (m32r_hard_regno_modes): ...this.
9270 (m32r_mode_class): Make static.
9271 (m32r_hard_regno_mode_ok): New function.
9272 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
9273 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
9274 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9275 (m68k_hard_regno_mode_ok): Make static.
9276 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
9277 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9278 (mcore_hard_regno_mode_ok): New function.
9279 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
9280 (HARD_REGNO_MODE_OK): Delete.
9281 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
9283 (microblaze_hard_regno_mode_ok_p): ...this and make static.
9284 (microblaze_hard_regno_mode_ok): New function.
9285 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9286 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
9287 (mips_hard_regno_mode_ok): Delete.
9288 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
9289 (mips_hard_regno_mode_ok_p): ...this and make static.
9290 (mips_hard_regno_mode_ok_p): Rename to...
9291 (mips_hard_regno_mode_ok_uncached): ...this.
9292 (mips_hard_regno_mode_ok): New function.
9293 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
9294 of HARD_REGNO_MODE_OK.
9295 (mips_option_override): Update after above name changes.
9296 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9297 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
9298 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
9299 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
9300 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
9301 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9302 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
9303 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
9304 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
9305 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9306 (msp430_hard_regno_mode_ok): Make static and return a bool.
9307 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
9308 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
9309 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
9311 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9312 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
9313 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
9314 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
9315 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
9316 (PA_HARD_REGNO_MODE_OK): ...this
9317 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
9318 (PA_HARD_REGNO_MODE_OK): ...this.
9319 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9320 (pa_hard_regno_mode_ok): New function.
9321 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
9322 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9323 (pdp11_hard_regno_mode_ok): New function.
9324 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
9325 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
9327 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
9329 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9330 (rs6000_hard_regno_mode_ok): Rename to...
9331 (rs6000_hard_regno_mode_ok_uncached): ...this.
9332 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9333 (rs6000_hard_regno_mode_ok): New function.
9334 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
9335 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
9336 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
9337 (riscv_hard_regno_mode_ok): ...this and make static.
9338 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9339 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
9340 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
9341 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9342 (rl78_hard_regno_mode_ok): Make static and return bool.
9343 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
9344 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
9346 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
9347 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9348 (rs6000_hard_regno_mode_ok): Rename to...
9349 (rs6000_hard_regno_mode_ok_uncached): ...this.
9350 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9351 (rs6000_hard_regno_mode_ok): New function.
9352 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
9353 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
9354 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9355 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
9356 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
9357 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
9358 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9359 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
9360 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
9361 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9362 (sh_hard_regno_mode_ok): Make static.
9363 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
9364 instead of HARD_REGNO_MODE_OK.
9365 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
9366 (sparc_mode_class): Delete.
9367 (HARD_REGNO_MODE_OK): Delete.
9368 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9369 (hard_regno_mode_classes): Make static.
9370 (sparc_mode_class): Likewise.
9371 (sparc_hard_regno_mode_ok): New function.
9372 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
9373 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
9374 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
9376 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9377 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
9378 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
9379 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
9380 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
9381 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9382 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
9383 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
9384 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9385 (visium_hard_regno_mode_ok): New function.
9386 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
9387 instead of HARD_REGNO_MODE_OK.
9388 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
9389 (HARD_REGNO_MODE_OK): Delete.
9390 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
9391 (xtensa_hard_regno_mode_ok_p): ...this and make static.
9392 (xtensa_option_override): Update accordingly.
9393 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9394 (xtensa_hard_regno_mode_ok): New function.
9395 * system.h (HARD_REGNO_MODE_OK): Poison.
9397 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9398 Alan Hayward <alan.hayward@arm.com>
9399 David Sherwood <david.sherwood@arm.com>
9401 * target.def (hard_regno_call_part_clobbered): New hook.
9402 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
9403 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
9404 * doc/tm.texi: Regenerate.
9405 * hooks.h (hook_bool_uint_mode_false): Declare.
9406 * hooks.c (hook_bool_uint_mode_false): New function.
9407 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9408 * cselib.c (cselib_process_insn): Use
9409 targetm.hard_regno_call_part_clobbered instead of
9410 HARD_REGNO_CALL_PART_CLOBBERED.
9411 * ira-conflicts.c (ira_build_conflicts): Likewise.
9412 * ira-costs.c (ira_tune_allocno_costs): Likewise.
9413 * lra-constraints.c (need_for_call_save_p): Likewise.
9414 * lra-lives.c: Include target.h.
9415 (check_pseudos_live_through_calls): Use
9416 targetm.hard_regno_call_part_clobbered instead of
9417 HARD_REGNO_CALL_PART_CLOBBERED.
9418 * regcprop.c: Include target.h.
9419 (copyprop_hardreg_forward_1): Use
9420 targetm.hard_regno_call_part_clobbered instead of
9421 HARD_REGNO_CALL_PART_CLOBBERED.
9422 * reginfo.c (choose_hard_reg_mode): Likewise.
9423 * regrename.c (check_new_reg_p): Likewise.
9424 * reload.c (find_equiv_reg): Likewise.
9425 * reload1.c (emit_reload_insns): Likewise.
9426 * sched-deps.c (deps_analyze_insn): Likewise.
9427 * sel-sched.c (init_regs_for_mode): Likewise.
9428 (mark_unavailable_hard_regs): Likewise.
9429 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
9430 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9431 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
9433 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9434 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9435 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
9437 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
9439 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9440 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9441 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
9443 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9444 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9445 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
9447 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9448 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
9450 * config/powerpcspe/powerpcspe.c
9451 (rs6000_hard_regno_call_part_clobbered): New function.
9452 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9453 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9454 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
9456 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9457 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9458 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
9460 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9461 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9462 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
9464 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9465 Alan Hayward <alan.hayward@arm.com>
9466 David Sherwood <david.sherwood@arm.com>
9468 * rtl.h (subreg_memory_offset): Declare.
9469 * emit-rtl.c (subreg_memory_offset): New function.
9470 * expmed.c (store_bit_field_1): Use it.
9471 * expr.c (undefined_operand_subword_p): Likewise.
9472 * simplify-rtx.c (simplify_subreg): Likewise.
9474 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
9476 PR rtl-optimization/57448
9479 * optabs.c (expand_atomic_load): Place compiler memory barriers if
9480 using atomic_load pattern.
9481 (expand_atomic_store): Likewise.
9483 2017-09-04 Jakub Jelinek <jakub@redhat.com>
9486 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
9487 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
9488 handling. Use replace_call_with_value with NULL instead of
9489 gsi_replace, unlink_stmt_vdef and release_defs.
9491 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
9494 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
9496 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9499 * rtl.h (emit_library_call_value_1): Declare.
9500 (emit_library_call): Replace declaration with a series of overloads.
9501 Remove the parameter count argument.
9502 (emit_library_call_value): Likewise.
9503 * calls.c (emit_library_call_value_1): Make global. Replace varargs
9504 with an "rtx_mode_t *".
9505 (emit_library_call_value): Delete.
9506 (emit_library_call): Likewise.
9507 * asan.c (asan_emit_stack_protection): Update calls accordingly.
9508 (asan_emit_allocas_unpoison): Likewise.
9509 * builtins.c (expand_builtin_powi): Likewise.
9510 (expand_asan_emit_allocas_unpoison): Likewise.
9511 * cfgexpand.c (expand_main_function): Likewise.
9512 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
9513 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
9514 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
9515 * config/arm/arm.c (arm_trampoline_init): Likewise.
9516 (arm_call_tls_get_addr): Likewise.
9517 (arm_expand_divmod_libfunc): Likewise.
9518 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
9519 (smulsi3_highpart): Likewise.
9520 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
9521 (c6x_expand_compare): Likewise.
9522 (c6x_expand_movmem): Likewise.
9523 * config/frv/frv.c (frv_trampoline_init): Likewise.
9524 * config/i386/i386.c (ix86_trampoline_init): Likewise.
9525 (ix86_expand_divmod_libfunc): Likewise.
9526 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
9527 (ia64_expand_compare): Likewise.
9528 (ia64_profile_hook): Likewise.
9529 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
9530 (nonlocal_goto): Likewise.
9531 (restore_stack_nonlocal): Likewise.
9532 * config/m32r/m32r.c (block_move_call): Likewise.
9533 (m32r_trampoline_init): Likewise.
9534 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
9535 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
9536 (m68k_call_m68k_read_tp): Likewise.
9537 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
9538 (microblaze_expand_divide): Likewise.
9539 * config/mips/mips.h (mips_args): Likewise.
9540 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
9541 (MIPS_ICACHE_SYNC): Likewise.
9542 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
9543 (nios2_trampoline_init): Likewise.
9544 * config/pa/pa.c (hppa_tls_call): Likewise.
9545 (pa_trampoline_init): Likewise.
9546 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
9547 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
9548 (expand_strn_compare): Likewise.
9549 (rs6000_generate_compare): Likewise.
9550 (rs6000_expand_float128_convert): Likewise.
9551 (output_profile_hook): Likewise.
9552 (rs6000_trampoline_init): Likewise.
9553 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
9554 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
9555 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
9556 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
9557 (rs6000_generate_compare): Likewise.
9558 (rs6000_expand_float128_convert): Likewise.
9559 (output_profile_hook): Likewise.
9560 (rs6000_trampoline_init): Likewise.
9561 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
9562 * config/sh/sh.c (sh_trampoline_init): Likewise.
9563 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
9564 (sparc_emit_float_lib_cmp): Likewise.
9565 (sparc32_initialize_trampoline): Likewise.
9566 (sparc64_initialize_trampoline): Likewise.
9567 (sparc_profile_hook): Likewise.
9568 * config/spu/spu.c (ea_load_store): Likewise.
9569 * config/spu/spu.md (floatunssidf2): Likewise.
9570 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
9571 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
9572 * config/visium/visium.c (expand_block_move_4): Likewise.
9573 (expand_block_move_2): Likewise.
9574 (expand_block_move_1): Likewise.
9575 (expand_block_set_4): Likewise.
9576 (expand_block_set_2): Likewise.
9577 (expand_block_set_1): Likewise.
9578 (visium_trampoline_init): Likewise.
9579 (visium_profile_hook): Likewise.
9580 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
9581 (xtensa_setup_frame_addresses): Likewise.
9582 (xtensa_trampoline_init): Likewise.
9583 * except.c (sjlj_emit_function_enter): Likewise.
9584 (sjlj_emit_function_exit): Likewise.
9585 * explow.c (allocate_dynamic_stack_space): Likewise.
9586 (probe_stack_range): Likewise.
9587 * expr.c (convert_mode_scalar): Likewise.
9588 * optabs.c (expand_binop): Likewise.
9589 (expand_twoval_binop_libfunc): Likewise.
9590 (expand_unop): Likewise.
9591 (prepare_cmp_insn): Likewise.
9592 (prepare_float_lib_cmp): Likewise.
9593 (expand_float): Likewise.
9594 (expand_fix): Likewise.
9595 (expand_fixed_convert): Likewise.
9596 (maybe_emit_sync_lock_test_and_set): Likewise.
9597 (expand_atomic_compare_and_swap): Likewise.
9598 (expand_mem_thread_fence): Likewise.
9599 (expand_atomic_fetch_op): Likewise.
9601 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
9603 * doc/generic.texi (OpenACC): Adjust URL.
9604 * doc/invoke.texi (C Dialect Options): Ditto.
9606 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
9608 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
9609 predicate for operand 1. Add (m,<S>) constraint.
9610 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
9611 Prevent memory operand 1 with register operand 2.
9613 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
9615 PR rtl-optimization/82024
9616 * combine.c (try_combine): If the combination result is a PARALLEL,
9617 and we only need to retain the SET in there that would be placed
9618 at I2, check that we can place that at I3 instead, before doing so.
9620 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9623 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
9625 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
9626 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
9629 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
9630 Jeff Law <law@redhat.com>
9632 * varasm.c (bss_initializer_p): Do not put constants into .bss
9633 (categorize_decl_for_section): Handle bss_initializer_p returning
9634 false when DECL_INITIAL is NULL.
9636 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9639 * config/s390/s390.c (s390_can_inline_p): New function.
9641 2017-09-01 Jeff Law <law@redhat.com>
9643 PR tree-optimization/82052
9644 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
9645 Always initialize the returned slot after a hash table miss
9646 when INSERT is true.
9648 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
9650 * config/s390/s390.md (mem_signal_fence): Remove.
9651 * doc/md.texi (mem_signal_fence): Remove.
9652 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
9654 * target-insns.def (mem_signal_fence): Remove.
9656 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9659 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
9662 * asan.c (create_odr_indicator): Strip name encoding from assembler
9663 name before appending it after __odr_asan_.
9665 2017-09-01 Martin Liska <mliska@suse.cz>
9667 PR tree-optimization/82059
9668 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
9669 frequency only when an edge is redirected.
9671 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9673 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
9674 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
9675 (arc_conditional_register_usage): Remove ARC600 lp_count
9677 (arc_file_start): Emit Tag_ARC_CPU_variation.
9678 (arc_can_use_doloop_p): New conditions to use ZOLs.
9679 (hwloop_fail): New function.
9680 (hwloop_optimize): Likewise.
9681 (hwloop_pattern_reg): Likewise.
9682 (arc_doloop_hooks): New struct, to be used with reorg_loops.
9683 (arc_reorg_loops): New function, calls reorg_loops.
9684 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
9685 (arc600_corereg_hazard): Remove ZOL checking, case handled by
9687 (arc_loop_hazard): Remove function, functionality moved into
9689 (arc_hazard): Remove arc_loop_hazard call.
9690 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
9691 into hwloop_optimize.
9692 (arc_label_align): Remove ZOL handling.
9693 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
9694 * config/arc/arc.md (doloop_begin): Remove pattern.
9695 (doloop_begin_i): Likewise.
9696 (doloop_end_i): Likewise.
9697 (doloop_fallback): Likewise.
9698 (doloop_fallback_m): Likewise.
9699 (doloop_end): Reimplement expand.
9700 (arc_lp): New pattern for LP instruction.
9701 (loop_end): New pattern.
9702 (loop_fail): Likewise.
9703 (decrement_and_branch_until_zero): Likewise.
9704 * config/arc/arc.opt (mlpc-width): New option.
9705 * doc/invoke.texi (mlpc-width): Document option.
9707 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9709 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
9710 (arc_ccfsm_advance): Fix checking for delay slots.
9711 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
9713 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9715 * config/arc/arc.md (movqi_insn): Add stores to save constant long
9717 (movhi_insn): Update store instruction constraint which are saving
9718 6-bit short immediates.
9719 (movsi_insn): Consider also short scaled load operations.
9720 (zero_extendhisi2_i): Use Usd constraint instead of T.
9721 (extendhisi2_i): Add q constraint.
9722 (arc_clzsi2): Add type and length attributes.
9723 (arc_ctzsi2): Likewise.
9724 * config/arc/constraints.md (Usc): Update constraint, the
9725 assembler can parse two relocations for a single instruction.
9727 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9729 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
9730 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
9732 2017-08-31 Olivier Hainque <hainque@adacore.com>
9734 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
9735 match as powerpc-wrs-vxworks*.
9737 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
9739 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
9740 register constraint for by-element operand.
9741 (aarch64_mls_elt_merge<mode>): Likewise.
9743 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9745 * config/arc/arc.c (arc_can_follow_jump): Check for short
9748 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9750 * config.gcc: Use g.opt for arc.
9751 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
9752 functionality moved to ...
9753 (legitimate_scaled_address_p): New function, ...here.
9754 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
9755 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
9756 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
9758 (arc_override_options): Handle G option.
9759 (arc_output_pic_addr_const): Correct function definition.
9760 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
9761 (arc_decl_anon_ns_mem_p): Delete.
9762 (arc_in_small_data_p): Overhaul this function to take into
9763 consideration the value given via G option.
9764 (arc_rewrite_small_data_1): Renamed and corrected old
9765 arc_rewrite_small_data function.
9766 (arc_rewrite_small_data): New function.
9767 (small_data_pattern): Don't use pic_offset_table_rtx.
9768 * config/arc/arc.h (CC1_SPEC): Recognize G option.
9769 * config/arc/simdext.md (movmisalignv2hi): Use
9770 prepare_move_operands function.
9772 (movmisalign*): Likewise.
9773 * doc/invoke.texi (ARC options): Document -G option.
9775 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9777 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
9779 * config/arc/arc.c (arc_print_operand): Output scalled address for
9780 sdata whenever is possible.
9781 (arc_in_small_data_p): Allow sdata for 64bit datum when double
9782 load/stores are available.
9783 (compact_sda_memory_operand): Check for the alignment required by
9784 code density instructions.
9785 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
9787 * config/arc/constraints.md (Usd): Update constraint.
9788 (Us0): New constraint.
9789 (Usc): Update constraint.
9791 2017-08-31 Richard Biener <rguenther@suse.de>
9794 * dwarf2out.c (dwarf2out_early_global_decl): Process each
9797 2017-08-31 Tamar Christina <tamar.christina@arm.com>
9799 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
9800 Resize type_signature.
9802 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9803 Alan Hayward <alan.hayward@arm.com>
9804 David Sherwood <david.sherwood@arm.com>
9806 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
9807 subregs whose inner modes can be stored in GPRs.
9808 (aarch64_classify_index): Likewise.
9810 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9811 Alan Hayward <alan.hayward@arm.com>
9812 David Sherwood <david.sherwood@arm.com>
9814 * config/aarch64/iterators.md (V_cmp_result): Rename to...
9815 (V_INT_EQUIV): ...this.
9816 (v_cmp_result): Rename to...
9817 (v_int_equiv): ...this.
9818 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
9819 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
9820 (copysign<mode>3): Likewise.
9821 (aarch64_simd_bsl<mode>_internal): Likewise.
9822 (aarch64_simd_bsl<mode>): Likewise.
9823 (vec_cmp<mode><mode>): Likewise.
9824 (vcond<mode><mode>): Likewise.
9825 (vcond<v_cmp_mixed><mode>): Likewise.
9826 (vcondu<mode><v_cmp_mixed>): Likewise.
9827 (aarch64_cm<optab><mode>): Likewise.
9828 (aarch64_cmtst<mode>): Likewise.
9829 (aarch64_fac<optab><mode>): Likewise.
9830 (vec_perm_const<mode>): Likewise.
9831 (vcond_mask_<mode><v_cmp_result>): Rename to...
9832 (vcond_mask_<mode><v_int_equiv>): ...this.
9833 (vec_cmp<mode><v_cmp_result>): Rename to...
9834 (vec_cmp<mode><v_int_equiv>): ...this.
9836 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9837 Alan Hayward <alan.hayward@arm.com>
9838 David Sherwood <david.sherwood@arm.com>
9840 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
9842 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
9843 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
9844 (UNSPEC_LD4_DREG): New unspecs.
9845 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
9846 (aarch64_ld2<mode>_dreg_be): Replace with...
9847 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
9849 (aarch64_ld3<mode>_dreg_le)
9850 (aarch64_ld3<mode>_dreg_be): Replace with...
9851 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
9853 (aarch64_ld4<mode>_dreg_le)
9854 (aarch64_ld4<mode>_dreg_be): Replace with...
9855 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
9858 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9860 PR tree-optimization/81987
9861 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
9862 insert an initializer in a location not dominated by the stride
9865 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
9867 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
9868 on the entire header of the finally block in the fallthru case.
9870 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
9872 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
9874 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
9876 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
9877 rs6000_emit_move_from_cr and call renamed function.
9878 (rs6000_emit_prologue): Call renamed functions.
9879 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
9880 movesi_from_cr, remove volatile CRs.
9882 2017-08-30 Jon Beniston <jon@beniston.com>
9883 Richard Biener <rguenther@suse.de>
9885 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
9886 of VECTOR_MODE_P check.
9887 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
9888 element vector types.
9890 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9892 * df.h (df_read_modify_subreg_p): Remove in favor of...
9893 * rtl.h (read_modify_subreg_p): ...this new function. Take a
9894 const_rtx instead of an rtx.
9895 * cprop.c (local_cprop_find_used_regs): Update accordingly.
9896 * df-problems.c (df_word_lr_mark_ref): Likewise.
9897 * ira-lives.c (mark_pseudo_reg_live): Likewise.
9898 (mark_pseudo_reg_dead): Likewise.
9899 (mark_ref_dead): Likewise.
9900 * reginfo.c (init_subregs_of_mode): Likewise.
9901 * sched-deps.c (sched_analyze_1): Likewise.
9902 * df-scan.c (df_def_record_1): Likewise.
9903 (df_uses_record): Likewise.
9904 (df_read_modify_subreg_p): Remove in favor of...
9905 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
9906 const_rtx instead of an rtx.
9908 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9909 Alan Hayward <alan.hayward@arm.com>
9910 David Sherwood <david.sherwood@arm.com>
9912 * rtl.h (partial_subreg_p): New function.
9913 * caller-save.c (save_call_clobbered_regs): Use it.
9914 * calls.c (expand_call): Likewise.
9915 * combine.c (combinable_i3pat): Likewise.
9916 (simplify_set): Likewise.
9917 (make_extraction): Likewise.
9918 (make_compound_operation_int): Likewise.
9919 (gen_lowpart_or_truncate): Likewise.
9920 (force_to_mode): Likewise.
9921 (make_field_assignment): Likewise.
9922 (reg_truncated_to_mode): Likewise.
9923 (record_truncated_value): Likewise.
9924 (move_deaths): Likewise.
9925 * cse.c (record_jump_cond): Likewise.
9926 (cse_insn): Likewise.
9927 * cselib.c (cselib_lookup_1): Likewise.
9928 * expmed.c (extract_bit_field_using_extv): Likewise.
9929 * function.c (assign_parm_setup_reg): Likewise.
9930 * ifcvt.c (noce_convert_multiple_sets): Likewise.
9931 * ira-build.c (create_insn_allocnos): Likewise.
9932 * lra-coalesce.c (merge_pseudos): Likewise.
9933 * lra-constraints.c (match_reload): Likewise.
9934 (simplify_operand_subreg): Likewise.
9935 (curr_insn_transform): Likewise.
9936 * lra-lives.c (process_bb_lives): Likewise.
9937 * lra.c (new_insn_reg): Likewise.
9938 (lra_substitute_pseudo): Likewise.
9939 * regcprop.c (mode_change_ok): Likewise.
9940 (maybe_mode_change): Likewise.
9941 (copyprop_hardreg_forward_1): Likewise.
9942 * reload.c (push_reload): Likewise.
9943 (find_reloads): Likewise.
9944 (find_reloads_subreg_address): Likewise.
9945 * reload1.c (alter_reg): Likewise.
9946 (eliminate_regs_1): Likewise.
9947 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9949 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
9951 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
9952 back to if statements, including unpack.
9954 2017-08-30 Martin Liska <mliska@suse.cz>
9957 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
9959 (func_checker::compare_asm_inputs_outputs): ... this function.
9960 (func_checker::compare_gimple_asm): Use the function to compare
9961 also ASM constrains.
9962 * ipa-icf-gimple.h: Rename the function.
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 (complex_mode): New type.
9969 * gdbhooks.py (build_pretty_printer): Handle it.
9970 * machmode.h (complex_mode): New class.
9971 (complex_mode::includes_p): New function.
9972 (is_complex_int_mode): Likewise.
9973 (is_complex_float_mode): Likewise.
9974 * genmodes.c (get_mode_class): Handle complex mode classes.
9975 * function.c (expand_function_end): Use is_complex_int_mode.
9977 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9978 Alan Hayward <alan.hayward@arm.com>
9979 David Sherwood <david.sherwood@arm.com>
9981 * coretypes.h (scalar_mode_pod): New typedef.
9982 * gdbhooks.py (build_pretty_printer): Handle it.
9983 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
9984 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
9985 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
9986 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
9989 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9990 Alan Hayward <alan.hayward@arm.com>
9991 David Sherwood <david.sherwood@arm.com>
9993 * machmode.h (mode_for_vector): Take a scalar_mode instead
9995 * stor-layout.c (mode_for_vector): Likewise.
9996 * explow.c (promote_mode): Use as_a <scalar_mode>.
9997 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
9999 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10000 Alan Hayward <alan.hayward@arm.com>
10001 David Sherwood <david.sherwood@arm.com>
10003 * target.def (preferred_simd_mode): Take a scalar_mode
10004 instead of a machine_mode.
10005 * targhooks.h (default_preferred_simd_mode): Likewise.
10006 * targhooks.c (default_preferred_simd_mode): Likewise.
10007 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
10008 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
10009 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
10010 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
10011 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
10012 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
10013 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
10014 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
10016 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
10017 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
10018 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
10019 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
10020 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
10021 * doc/tm.texi: Regenerate.
10022 * optabs-query.c (can_vec_mask_load_store_p): Return false for
10025 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10026 Alan Hayward <alan.hayward@arm.com>
10027 David Sherwood <david.sherwood@arm.com>
10029 * target.def (scalar_mode_supported_p): Take a scalar_mode
10030 instead of a machine_mode.
10031 * targhooks.h (default_scalar_mode_supported_p): Likewise.
10032 * targhooks.c (default_scalar_mode_supported_p): Likewise.
10033 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
10034 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
10035 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
10036 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
10037 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
10038 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
10039 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
10040 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
10041 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
10042 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
10043 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
10044 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
10046 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
10047 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
10048 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
10049 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
10050 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
10052 * doc/tm.texi: Regenerate.
10054 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10055 Alan Hayward <alan.hayward@arm.com>
10056 David Sherwood <david.sherwood@arm.com>
10058 * coretypes.h (opt_scalar_mode): New typedef.
10059 * gdbhooks.py (build_pretty_printers): Handle it.
10060 * machmode.h (mode_iterator::get_2xwider): Add overload for
10062 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
10064 * expr.c (convert_mode_scalar): Likewise.
10065 * omp-low.c (omp_clause_aligned_alignment): Likewise.
10066 * optabs.c (expand_float): Likewise.
10067 (expand_fix): Likewise.
10068 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10070 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10071 Alan Hayward <alan.hayward@arm.com>
10072 David Sherwood <david.sherwood@arm.com>
10074 * optabs.c (expand_float): Explicitly check for scalars before
10075 using a branching expansion.
10076 (expand_fix): Likewise.
10078 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10079 Alan Hayward <alan.hayward@arm.com>
10080 David Sherwood <david.sherwood@arm.com>
10082 * expr.c (convert_mode): Split scalar handling out into...
10083 (convert_mode_scalar): ...this new function. Treat the modes
10086 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10087 Alan Hayward <alan.hayward@arm.com>
10088 David Sherwood <david.sherwood@arm.com>
10090 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
10092 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
10094 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10095 Alan Hayward <alan.hayward@arm.com>
10096 David Sherwood <david.sherwood@arm.com>
10098 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
10099 rather than a machine_mode.
10100 (fixed_from_string): Likewise.
10101 (fixed_convert): Likewise.
10102 (fixed_convert_from_int): Likewise.
10103 (fixed_convert_from_real): Likewise.
10104 (real_convert_from_fixed): Likewise.
10105 * fixed-value.c (fixed_from_double_int): Likewise.
10106 (fixed_from_string): Likewise.
10107 (fixed_convert): Likewise.
10108 (fixed_convert_from_int): Likewise.
10109 (fixed_convert_from_real): Likewise.
10110 (real_convert_from_fixed): Likewise.
10111 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
10113 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10114 Alan Hayward <alan.hayward@arm.com>
10115 David Sherwood <david.sherwood@arm.com>
10117 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
10118 of separate mode class checks. Do not allow vector modes here.
10119 (immed_wide_int_const): Use as_a <scalar_mode>.
10120 * explow.c (trunc_int_for_mode): Likewise.
10121 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
10122 (wi::shwi): Likewise.
10123 (wi::min_value): Likewise.
10124 (wi::max_value): Likewise.
10125 * dwarf2out.c (loc_descriptor): Likewise.
10126 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
10127 for CONST_WIDE_INT.
10129 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10130 Alan Hayward <alan.hayward@arm.com>
10131 David Sherwood <david.sherwood@arm.com>
10133 * tree.h (SCALAR_TYPE_MODE): New macro.
10134 * expr.c (expand_expr_addr_expr_1): Use it.
10135 (expand_expr_real_2): Likewise.
10136 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
10137 (fold_convert_const_fixed_from_int): Likewise.
10138 (fold_convert_const_fixed_from_real): Likewise.
10139 (native_encode_fixed): Likewise
10140 (native_encode_complex): Likewise
10141 (native_encode_vector): Likewise.
10142 (native_interpret_fixed): Likewise.
10143 (native_interpret_real): Likewise.
10144 (native_interpret_complex): Likewise.
10145 (native_interpret_vector): Likewise.
10146 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
10147 (simd_clone_adjust_argument_types): Likewise.
10148 (simd_clone_init_simd_arrays): Likewise.
10149 (simd_clone_adjust): Likewise.
10150 * stor-layout.c (layout_type): Likewise.
10151 * tree.c (build_minus_one_cst): Likewise.
10152 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
10153 * tree-inline.c (estimate_move_cost): Likewise.
10154 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
10155 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
10156 (vectorizable_reduction): Likewise.
10157 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
10158 (vect_recog_mixed_size_cond_pattern): Likewise.
10159 (check_bool_pattern): Likewise.
10160 (adjust_bool_pattern): Likewise.
10161 (search_type_for_mask_1): Likewise.
10162 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
10163 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10164 (vectorizable_load): Likewise.
10165 (vectorizable_store): Likewise.
10166 * ubsan.c (ubsan_encode_value): Likewise.
10167 * varasm.c (output_constant): Likewise.
10169 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10170 Alan Hayward <alan.hayward@arm.com>
10171 David Sherwood <david.sherwood@arm.com>
10173 * coretypes.h (scalar_mode): New class.
10174 * machmode.h (scalar_mode): Likewise.
10175 (scalar_mode::includes_p): New function.
10176 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
10177 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
10178 * genmodes.c (get_mode_class): Handle remaining scalar modes.
10179 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
10180 * expmed.c (store_bit_field_1): Likewise.
10181 (extract_bit_field_1): Likewise.
10182 * expr.c (write_complex_part): Likewise.
10183 (read_complex_part): Likewise.
10184 (emit_move_complex_push): Likewise.
10185 (expand_expr_real_2): Likewise.
10186 * function.c (assign_parm_setup_reg): Likewise.
10187 (assign_parms_unsplit_complex): Likewise.
10188 * optabs.c (expand_binop): Likewise.
10189 * rtlanal.c (subreg_get_info): Likewise.
10190 * simplify-rtx.c (simplify_immed_subreg): Likewise.
10191 * varasm.c (output_constant_pool_2): Likewise.
10193 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10194 Alan Hayward <alan.hayward@arm.com>
10195 David Sherwood <david.sherwood@arm.com>
10197 * expmed.c (extract_high_half): Use scalar_int_mode and remove
10199 (expmed_mult_highpart_optab): Likewise.
10200 (expmed_mult_highpart): Likewise.
10202 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10203 Alan Hayward <alan.hayward@arm.com>
10204 David Sherwood <david.sherwood@arm.com>
10206 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
10207 instead of a machine_mode.
10208 (builtin_memset_read_str): Likewise.
10209 * builtins.c (c_readstr): Likewise.
10210 (builtin_memcpy_read_str): Likewise.
10211 (builtin_strncpy_read_str): Likewise.
10212 (builtin_memset_read_str): Likewise.
10213 (builtin_memset_gen_str): Likewise.
10214 (expand_builtin_signbit): Use scalar_int_mode for local variables.
10215 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
10216 instead of a machine_mode.
10217 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
10219 (make_extraction): Likewise.
10220 (try_widen_shift_mode): Take and return scalar_int_modes instead
10222 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
10223 a scalar_int_mode instead of a machine_mode.
10224 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
10225 (avr_addr_space_pointer_mode): Likewise.
10226 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
10227 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
10228 (msp430_unwind_word_mode): Likewise.
10229 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
10230 (spu_addr_space_pointer_mode): Likewise.
10231 (spu_addr_space_address_mode): Likewise.
10232 (spu_libgcc_cmp_return_mode): Likewise.
10233 (spu_libgcc_shift_count_mode): Likewise.
10234 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
10235 (rl78_addr_space_pointer_mode): Likewise.
10236 (fl78_unwind_word_mode): Likewise.
10237 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
10239 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
10240 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
10241 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
10242 (mips_valid_pointer_mode): Likewise.
10243 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
10244 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
10245 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
10247 (ft32_addr_space_address_mode): Likewise.
10248 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
10249 scalar_int_mode instead of a machine_mode.
10250 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
10252 (m32c_addr_space_address_mode): Likewise.
10253 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
10254 (rs6000_eh_return_filter_mode): Likewise.
10255 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
10256 (rs6000_eh_return_filter_mode): Likewise.
10257 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
10258 (s390_libgcc_shift_count_mode): Likewise.
10259 (s390_unwind_word_mode): Likewise.
10260 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
10262 * target.def (mode_rep_extended): Likewise.
10263 (valid_pointer_mode): Likewise.
10264 (addr_space.valid_pointer_mode): Likewise.
10265 (eh_return_filter_mode): Return a scalar_int_mode rather than
10267 (libgcc_cmp_return_mode): Likewise.
10268 (libgcc_shift_count_mode): Likewise.
10269 (unwind_word_mode): Likewise.
10270 (addr_space.pointer_mode): Likewise.
10271 (addr_space.address_mode): Likewise.
10272 * doc/tm.texi: Regenerate.
10273 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
10275 (do_jump): Use scalar_int_mode for local variables.
10276 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
10277 rather than a machine_mode.
10278 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
10279 (scompare_loc_descriptor_wide): Likewise.
10280 (scompare_loc_descriptor_narrow): Likewise.
10281 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
10283 * except.c (sjlj_emit_dispatch_table): Likewise.
10284 (expand_builtin_eh_copy_values): Likewise.
10285 * explow.c (convert_memory_address_addr_space_1): Likewise.
10286 Take a scalar_int_mode rather than a machine_mode.
10287 (convert_memory_address_addr_space): Take a scalar_int_mode rather
10288 than a machine_mode.
10289 (memory_address_addr_space): Use scalar_int_mode for local variables.
10290 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
10291 rather than a machine_mode.
10292 * expmed.c (mask_rtx): Likewise.
10293 (init_expmed_one_conv): Likewise.
10294 (expand_mult_highpart_adjust): Likewise.
10295 (extract_high_half): Likewise.
10296 (expmed_mult_highpart_optab): Likewise.
10297 (expmed_mult_highpart): Likewise.
10298 (expand_smod_pow2): Likewise.
10299 (expand_sdiv_pow2): Likewise.
10300 (emit_store_flag_int): Likewise.
10301 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
10303 (extract_low_bits): Likewise.
10304 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
10306 * expr.c (pieces_addr::adjust): Likewise.
10307 (can_store_by_pieces): Likewise.
10308 (store_by_pieces): Likewise.
10309 (clear_by_pieces_1): Likewise.
10310 (expand_expr_addr_expr_1): Likewise.
10311 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
10312 (expand_expr_real_1): Likewise.
10313 (try_casesi): Likewise.
10314 * final.c (shorten_branches): Likewise.
10315 * fold-const.c (fold_convert_const_int_from_fixed): Change the
10316 type of "mode" to machine_mode.
10317 * internal-fn.c (expand_arith_overflow_result_store): Take a
10318 scalar_int_mode rather than a machine_mode.
10319 (expand_mul_overflow): Use scalar_int_mode for local variables.
10320 * loop-doloop.c (doloop_modify): Likewise.
10321 (doloop_optimize): Likewise.
10322 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
10323 than a machine_mode.
10324 (expand_doubleword_shift_condmove): Likewise.
10325 (expand_doubleword_shift): Likewise.
10326 (expand_doubleword_clz): Likewise.
10327 (expand_doubleword_popcount): Likewise.
10328 (expand_doubleword_parity): Likewise.
10329 (expand_absneg_bit): Use scalar_int_mode for local variables.
10330 (prepare_float_lib_cmp): Likewise.
10331 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
10332 rather than a machine_mode.
10333 (convert_memory_address_addr_space): Likewise.
10334 (get_mode_bounds): Likewise.
10335 (get_address_mode): Return a scalar_int_mode rather than a
10337 * rtlanal.c (get_address_mode): Likewise.
10338 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
10339 than a machine_mode.
10340 * targhooks.c (default_mode_rep_extended): Likewise.
10341 (default_valid_pointer_mode): Likewise.
10342 (default_addr_space_valid_pointer_mode): Likewise.
10343 (default_eh_return_filter_mode): Return a scalar_int_mode rather
10344 than a machine_mode.
10345 (default_libgcc_cmp_return_mode): Likewise.
10346 (default_libgcc_shift_count_mode): Likewise.
10347 (default_unwind_word_mode): Likewise.
10348 (default_addr_space_pointer_mode): Likewise.
10349 (default_addr_space_address_mode): Likewise.
10350 * targhooks.h (default_eh_return_filter_mode): Likewise.
10351 (default_libgcc_cmp_return_mode): Likewise.
10352 (default_libgcc_shift_count_mode): Likewise.
10353 (default_unwind_word_mode): Likewise.
10354 (default_addr_space_pointer_mode): Likewise.
10355 (default_addr_space_address_mode): Likewise.
10356 (default_mode_rep_extended): Take a scalar_int_mode rather than
10358 (default_valid_pointer_mode): Likewise.
10359 (default_addr_space_valid_pointer_mode): Likewise.
10360 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
10362 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
10363 rather than a machine_mode.
10364 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
10365 for local variables.
10366 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10367 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
10368 than a machine_mode.
10370 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10371 Alan Hayward <alan.hayward@arm.com>
10372 David Sherwood <david.sherwood@arm.com>
10374 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
10375 the mode argument to scalar_int_mode.
10376 (do_jump_by_parts_zero_rtx): Likewise.
10377 (do_jump_by_parts_equality_rtx): Likewise.
10378 (do_jump_by_parts_greater): Take a mode argument.
10379 (do_jump_by_parts_equality): Likewise.
10380 (do_jump_1): Update calls accordingly.
10382 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10383 Alan Hayward <alan.hayward@arm.com>
10384 David Sherwood <david.sherwood@arm.com>
10386 * is-a.h (safe_dyn_cast): New function.
10387 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
10388 (jump_table_for_label): Likewise.
10389 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
10390 instead of an rtx_insn *.
10391 (shorten_branches): Use dyn_cast instead of LABEL_P and
10392 JUMP_TABLE_DATA_P. Use jump_table_for_label and
10393 rtx_jump_table_data::get_data_mode.
10394 (final_scan_insn): Likewise.
10396 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10397 Alan Hayward <alan.hayward@arm.com>
10398 David Sherwood <david.sherwood@arm.com>
10400 * combine.c (try_combine): Use is_a <scalar_int_mode> when
10401 trying to combine a full-register integer set with a subreg
10404 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10405 Alan Hayward <alan.hayward@arm.com>
10406 David Sherwood <david.sherwood@arm.com>
10408 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
10409 that is always either address_mode or pointer_mode.
10411 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10412 Alan Hayward <alan.hayward@arm.com>
10413 David Sherwood <david.sherwood@arm.com>
10415 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
10416 when the two are known to be equal.
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 * simplify-rtx.c (simplify_const_unary_operation): Use
10423 is_a <scalar_int_mode> instead of checking for a nonzero
10424 precision. Forcibly convert op_mode to a scalar_int_mode
10425 in that case. More clearly differentiate the operand and
10426 result modes and use the former when deciding what the value
10427 of a count-bits operation should be. Use is_int_mode instead
10428 of checking for a MODE_INT. Remove redundant check for whether
10429 this mode has a zero precision.
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 * optabs.c (widen_leading): Change the type of the mode argument
10436 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
10437 (widen_bswap): Likewise.
10438 (expand_parity): Likewise.
10439 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
10440 (expand_ffs): Likewise.
10441 (epand_unop): Check for scalar integer modes before calling the
10444 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10445 Alan Hayward <alan.hayward@arm.com>
10446 David Sherwood <david.sherwood@arm.com>
10448 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
10450 (expand_expr_real_1): Update call accordingly.
10452 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10453 Alan Hayward <alan.hayward@arm.com>
10454 David Sherwood <david.sherwood@arm.com>
10456 * expmed.c (store_bit_field_using_insv): Add op0_mode and
10457 value_mode arguments. Use scalar_int_mode internally.
10458 (store_bit_field_1): Rename the new integer mode from imode
10459 to op0_mode and use it instead of GET_MODE (op0). Update calls
10460 to store_split_bit_field, store_bit_field_using_insv and
10461 store_fixed_bit_field.
10462 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
10463 Use scalar_int_mode internally. Use a bit count rather than a mode
10464 when calculating the largest bit size for get_best_mode.
10465 Update calls to store_split_bit_field and store_fixed_bit_field_1.
10466 (store_fixed_bit_field_1): Add mode and value_mode arguments.
10467 Remove assertion that OP0 has a scalar integer mode.
10468 (store_split_bit_field): Add op0_mode and value_mode arguments.
10469 Update calls to extract_fixed_bit_field.
10470 (extract_bit_field_using_extv): Add an op0_mode argument.
10471 Use scalar_int_mode internally.
10472 (extract_bit_field_1): Rename the new integer mode from imode to
10473 op0_mode and use it instead of GET_MODE (op0). Update calls to
10474 extract_split_bit_field, extract_bit_field_using_extv and
10475 extract_fixed_bit_field.
10476 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
10477 to extract_split_bit_field and extract_fixed_bit_field_1.
10478 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
10479 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
10480 on the target mode.
10481 (extract_split_bit_field): Add an op0_mode argument. Update call
10482 to extract_fixed_bit_field.
10484 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10485 Alan Hayward <alan.hayward@arm.com>
10486 David Sherwood <david.sherwood@arm.com>
10488 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
10489 * explow.c (hard_function_value): Likewise.
10490 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
10491 convert_to_mode call outside the loop.
10492 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
10493 for the mode iterator. Require the mode specified by max_pieces
10495 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
10497 (copy_blkmode_to_reg): Likewise.
10498 (set_storage_via_setmem): Likewise.
10499 * optabs.c (prepare_cmp_insn): Likewise.
10500 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
10501 * stor-layout.c (finish_bitfield_representative): Likewise.
10503 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10504 Alan Hayward <alan.hayward@arm.com>
10505 David Sherwood <david.sherwood@arm.com>
10507 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
10508 * expr.c (convert_move): Use them.
10509 (convert_modes): Likewise.
10510 (store_expr_with_bounds): Likewise.
10512 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10513 Alan Hayward <alan.hayward@arm.com>
10514 David Sherwood <david.sherwood@arm.com>
10516 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
10517 parameter for the mode of "x". Remove the "known_x", "known_mode"
10518 and "known_ret" arguments. Change the type of the mode argument
10519 to scalar_int_mode.
10520 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
10521 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
10522 (reg_num_sign_bit_copies_for_combine): Likewise.
10523 * rtlanal.c (nonzero_bits1): Likewise.
10524 (num_sign_bit_copies1): Likewise.
10525 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
10526 (reg_num_sign_bit_copies_general): Likewise.
10527 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
10528 (reg_nonzero_bits_general): Likewise.
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 (num_sign_bit_copies): Handle VOIDmode here rather
10535 than in subroutines. Return 1 for non-integer modes.
10536 (cached_num_sign_bit_copies): Change the type of the mode parameter
10537 to scalar_int_mode.
10538 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
10539 classes. Handle CONST_INT_P first and then check whether X also
10540 has a scalar integer mode. Check the same thing for inner registers
10541 of a SUBREG and for values that are being extended or truncated.
10543 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10544 Alan Hayward <alan.hayward@arm.com>
10545 David Sherwood <david.sherwood@arm.com>
10547 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
10548 in subroutines. Return the mode mask for non-integer modes.
10549 (cached_nonzero_bits): Change the type of the mode parameter
10550 to scalar_int_mode.
10551 (nonzero_bits1): Likewise. Remove early exit for other mode
10552 classes. Handle CONST_INT_P first and then check whether X
10553 also has a scalar integer mode.
10555 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10556 Alan Hayward <alan.hayward@arm.com>
10557 David Sherwood <david.sherwood@arm.com>
10559 * expr.c (widest_int_mode_for_size): Make the comment match the code.
10560 Return a scalar_int_mode and assert that the size is greater than
10562 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
10563 (op_by_pieces_d::op_by_pieces_d): Likewise.
10564 (op_by_pieces_d::run): Likewise.
10565 (can_store_by_pieces): Likewise.
10567 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10568 Alan Hayward <alan.hayward@arm.com>
10569 David Sherwood <david.sherwood@arm.com>
10571 * combine.c (extract_left_shift): Add a mode argument and update
10573 (make_compound_operation_int): Change the type of the mode parameter
10574 to scalar_int_mode and update the call to extract_left_shift.
10576 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10577 Alan Hayward <alan.hayward@arm.com>
10578 David Sherwood <david.sherwood@arm.com>
10580 * combine.c (simplify_and_const_int): Change the type of the mode
10581 parameter to scalar_int_mode.
10582 (simplify_and_const_int_1): Likewise. Update recursive call.
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 (simplify_compare_const): Check that the mode is a
10589 scalar_int_mode (rather than VOIDmode) before testing its
10591 (simplify_comparison): Move COMPARISON_P handling out of the
10592 loop and restrict the latter part of the loop to scalar_int_modes.
10593 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
10594 and when considering SUBREG_REGs. Use is_int_mode instead of
10595 checking GET_MODE_CLASS against MODE_INT.
10597 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10598 Alan Hayward <alan.hayward@arm.com>
10599 David Sherwood <david.sherwood@arm.com>
10601 * combine.c (try_widen_shift_mode): Move check for equal modes to...
10602 (simplify_shift_const_1): ...here. Use scalar_int_mode for
10603 shift_unit_mode and for modes involved in scalar shifts.
10605 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10606 Alan Hayward <alan.hayward@arm.com>
10607 David Sherwood <david.sherwood@arm.com>
10609 * combine.c (force_int_to_mode): New function, split out from...
10610 (force_to_mode): ...here. Keep xmode up-to-date and use it
10611 instead of GET_MODE (x).
10613 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10614 Alan Hayward <alan.hayward@arm.com>
10615 David Sherwood <david.sherwood@arm.com>
10617 * optabs-query.h (extraction_insn::struct_mode): Change type to
10618 opt_scalar_int_mode and update comment.
10619 (extraction_insn::field_mode): Change type to scalar_int_mode.
10620 (extraction_insn::pos_mode): Likewise.
10621 * combine.c (make_extraction): Update accordingly.
10622 * optabs-query.c (get_traditional_extraction_insn): Likewise.
10623 (get_optab_extraction_insn): Likewise.
10624 * recog.c (simplify_while_replacing): Likewise.
10625 * expmed.c (narrow_bit_field_mem): Change the type of the mode
10626 parameter to opt_scalar_int_mode.
10628 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10629 Alan Hayward <alan.hayward@arm.com>
10630 David Sherwood <david.sherwood@arm.com>
10632 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
10633 to a scalar_int_mode instead of a machine_mode.
10634 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
10635 (get_best_mode): Return a boolean and use a pointer argument to store
10636 the selected mode. Replace the limit mode parameter with a bit limit.
10637 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
10638 for the values returned by bit_field_mode_iterator::next_mode.
10639 (store_bit_field): Update call to get_best_mode.
10640 (store_fixed_bit_field): Likewise.
10641 (extract_fixed_bit_field): Likewise.
10642 * expr.c (optimize_bitfield_assignment_op): Likewise.
10643 * fold-const.c (optimize_bit_field_compare): Likewise.
10644 (fold_truth_andor_1): Likewise.
10645 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
10646 Update for new type of m_mode.
10647 (get_best_mode): As above.
10649 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10650 Alan Hayward <alan.hayward@arm.com>
10651 David Sherwood <david.sherwood@arm.com>
10653 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
10654 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
10655 (store_bit_field): Check is_a <scalar_int_mode> before calling
10656 strict_volatile_bitfield_p.
10657 (extract_bit_field): Likewise.
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 * target.def (cstore_mode): Return a scalar_int_mode.
10664 * doc/tm.texi: Regenerate.
10665 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
10666 * targhooks.h (default_cstore_mode): Likewise.
10667 * targhooks.c (default_cstore_mode): Likewise, using a forced
10669 * expmed.c (emit_cstore): Expect the target of the cstore to be
10672 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10673 Alan Hayward <alan.hayward@arm.com>
10674 David Sherwood <david.sherwood@arm.com>
10676 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
10678 (niter_desc): Likewise mode.
10679 (iv_analyze): Add a mode parameter.
10681 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
10682 and change its type to scalar_int_mode.
10683 * loop-iv.c: Update commentary at head of file.
10684 (iv_constant): Pass the mode paraeter before the rtx it describes
10685 and change its type to scalar_int_mode. Remove VOIDmode handling.
10686 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
10687 (iv_extend): Likewise.
10688 (shorten_into_mode): Likewise.
10689 (iv_add): Use scalar_int_mode.
10690 (iv_mult): Likewise.
10691 (iv_shift): Likewise.
10692 (canonicalize_iv_subregs): Likewise.
10693 (get_biv_step_1): Pass the outer_mode parameter before the rtx
10694 it describes and change its mode to scalar_int_mode. Also change
10695 the type of the returned inner_mode to scalar_int_mode.
10696 (get_biv_step): Likewise, turning outer_mode from a pointer
10697 into a direct parameter. Update call to get_biv_step_1.
10698 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
10699 iv_constant and get_biv_step.
10700 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
10701 and change its type to scalar_int_mode. Don't initialise iv->mode
10702 to VOIDmode and remove later checks for its still being VOIDmode.
10703 Update calls to iv_analyze_op and iv_analyze_expr. Check
10704 is_a <scalar_int_mode> when changing the mode under consideration.
10705 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
10706 Update call to iv_analyze_expr.
10707 (iv_analyze_op): Add a mode parameter. Reject subregs whose
10708 inner register is not also a scalar_int_mode. Update call to
10710 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
10711 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
10712 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
10713 separate mode class checks. Update calls to iv_analyze. Remove
10714 fix-up of VOIDmodes after iv_analyze_biv.
10715 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
10716 don't have a scalar_int_mode. Update call to biv_p.
10718 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10719 Alan Hayward <alan.hayward@arm.com>
10720 David Sherwood <david.sherwood@arm.com>
10722 * cfgexpand.c (convert_debug_memory_address): Use
10723 as_a <scalar_int_mode>.
10724 * combine.c (expand_compound_operation): Likewise.
10725 (make_extraction): Likewise.
10726 (change_zero_ext): Likewise.
10727 (simplify_comparison): Likewise.
10728 * cse.c (cse_insn): Likewise.
10729 * dwarf2out.c (minmax_loc_descriptor): Likewise.
10730 (mem_loc_descriptor): Likewise.
10731 (loc_descriptor): Likewise.
10732 * expmed.c (init_expmed_one_mode): Likewise.
10733 (synth_mult): Likewise.
10734 (emit_store_flag_1): Likewise.
10735 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
10736 of a comparison with size.
10737 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
10738 (reduce_to_bit_field_precision): Likewise.
10739 * function.c (expand_function_end): Likewise.
10740 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
10741 * loop-doloop.c (doloop_modify): Likewise.
10742 * optabs.c (expand_binop): Likewise.
10743 (expand_unop): Likewise.
10744 (expand_copysign_absneg): Likewise.
10745 (prepare_cmp_insn): Likewise.
10746 (maybe_legitimize_operand): Likewise.
10747 * recog.c (const_scalar_int_operand): Likewise.
10748 * rtlanal.c (get_address_mode): Likewise.
10749 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10750 (simplify_cond_clz_ctz): Likewise.
10751 * tree-nested.c (get_nl_goto_field): Likewise.
10752 * tree.c (build_vector_type_for_mode): Likewise.
10753 * var-tracking.c (use_narrower_mode): Likewise.
10755 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10756 Alan Hayward <alan.hayward@arm.com>
10757 David Sherwood <david.sherwood@arm.com>
10759 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
10760 * builtins.c (expand_builtin_signbit): Use it.
10761 * cfgexpand.c (expand_debug_expr): Likewise.
10762 * dojump.c (do_jump): Likewise.
10763 (do_compare_and_jump): Likewise.
10764 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
10765 * expmed.c (make_tree): Likewise.
10766 * expr.c (expand_expr_real_2): Likewise.
10767 (expand_expr_real_1): Likewise.
10768 (try_casesi): Likewise.
10769 * fold-const-call.c (fold_const_call_ss): Likewise.
10770 * fold-const.c (unextend): Likewise.
10771 (extract_muldiv_1): Likewise.
10772 (fold_single_bit_test): Likewise.
10773 (native_encode_int): Likewise.
10774 (native_encode_string): Likewise.
10775 (native_interpret_int): Likewise.
10776 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
10777 * internal-fn.c (expand_addsub_overflow): Likewise.
10778 (expand_neg_overflow): Likewise.
10779 (expand_mul_overflow): Likewise.
10780 (expand_arith_overflow): Likewise.
10781 * match.pd: Likewise.
10782 * stor-layout.c (layout_type): Likewise.
10783 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
10784 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
10785 * tree-ssanames.c (get_range_info): Likewise.
10786 * tree-switch-conversion.c (array_value_type) Likewise.
10787 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
10788 (vect_recog_divmod_pattern): Likewise.
10789 (vect_recog_mixed_size_cond_pattern): Likewise.
10790 * tree-vrp.c (extract_range_basic): Likewise.
10791 (simplify_float_conversion_using_ranges): Likewise.
10792 * tree.c (int_fits_type_p): Likewise.
10793 * ubsan.c (instrument_bool_enum_load): Likewise.
10794 * varasm.c (mergeable_string_section): Likewise.
10795 (narrowing_initializer_constant_valid_p): Likewise.
10796 (output_constant): Likewise.
10798 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10799 Alan Hayward <alan.hayward@arm.com>
10800 David Sherwood <david.sherwood@arm.com>
10802 * machmode.h (NARROWEST_INT_MODE): New macro.
10803 * expr.c (alignment_for_piecewise_move): Use it instead of
10804 GET_CLASS_NARROWEST_MODE (MODE_INT).
10805 (push_block): Likewise.
10806 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
10808 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10810 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10811 Alan Hayward <alan.hayward@arm.com>
10812 David Sherwood <david.sherwood@arm.com>
10814 * postreload.c (move2add_valid_value_p): Change the type of the
10815 mode parameter to scalar_int_mode.
10816 (move2add_use_add2_insn): Add a mode parameter and use it instead
10818 (move2add_use_add3_insn): Likewise.
10819 (reload_cse_move2add): Update accordingly.
10821 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10822 Alan Hayward <alan.hayward@arm.com>
10823 David Sherwood <david.sherwood@arm.com>
10825 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
10827 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
10828 * optabs.c (expand_unop): Likewise.
10830 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10831 Alan Hayward <alan.hayward@arm.com>
10832 David Sherwood <david.sherwood@arm.com>
10834 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
10835 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
10836 (popcount_loc_descriptor): Likewise.
10837 (bswap_loc_descriptor): Likewise.
10838 (rotate_loc_descriptor): Likewise.
10839 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
10840 calling the functions above.
10842 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10843 Alan Hayward <alan.hayward@arm.com>
10844 David Sherwood <david.sherwood@arm.com>
10846 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
10848 (try_combine): Likewise.
10849 (simplify_if_then_else): Likewise.
10850 * cse.c (cse_insn): Likewise.
10851 * dwarf2out.c (mem_loc_descriptor): Likewise.
10852 * emit-rtl.c (gen_lowpart_common): Likewise.
10853 * simplify-rtx.c (simplify_truncation): Likewise.
10854 (simplify_binary_operation_1): Likewise.
10855 (simplify_const_relational_operation): Likewise.
10856 (simplify_ternary_operation): Likewise.
10857 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
10859 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10860 Alan Hayward <alan.hayward@arm.com>
10861 David Sherwood <david.sherwood@arm.com>
10863 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
10864 * reload.c (push_reload): Likewise.
10865 (find_reloads): Likewise.
10867 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10868 Alan Hayward <alan.hayward@arm.com>
10869 David Sherwood <david.sherwood@arm.com>
10871 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
10872 (make_compound_operation_int): Likewise.
10873 (change_zero_ext): Likewise.
10874 * expr.c (convert_move): Likewise.
10875 (convert_modes): Likewise.
10876 * fwprop.c (forward_propagate_subreg): Likewise.
10877 * loop-iv.c (get_biv_step_1): Likewise.
10878 * optabs.c (widen_operand): Likewise.
10879 * postreload.c (move2add_valid_value_p): Likewise.
10880 * recog.c (simplify_while_replacing): Likewise.
10881 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10882 (simplify_binary_operation_1): Likewise. Remove redundant
10883 mode equality check.
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 * combine.c (combine_simplify_rtx): Add checks for
10890 is_a <scalar_int_mode>.
10891 (simplify_if_then_else): Likewise.
10892 (make_field_assignment): Likewise.
10893 (simplify_comparison): Likewise.
10894 * ifcvt.c (noce_try_bitop): Likewise.
10895 * loop-invariant.c (canonicalize_address_mult): Likewise.
10896 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10898 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10899 Alan Hayward <alan.hayward@arm.com>
10900 David Sherwood <david.sherwood@arm.com>
10902 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
10903 is_a <scalar_int_mode> instead of != BLKmode.
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 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
10910 instead of != VOIDmode.
10911 * combine.c (if_then_else_cond): Likewise.
10912 (change_zero_ext): Likewise.
10913 * dwarf2out.c (mem_loc_descriptor): Likewise.
10914 (loc_descriptor): Likewise.
10915 * rtlanal.c (canonicalize_condition): Likewise.
10916 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
10918 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10919 Alan Hayward <alan.hayward@arm.com>
10920 David Sherwood <david.sherwood@arm.com>
10922 * simplify-rtx.c (simplify_binary_operation_1): Use
10923 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
10925 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10926 Alan Hayward <alan.hayward@arm.com>
10927 David Sherwood <david.sherwood@arm.com>
10929 * wide-int.h (int_traits<unsigned char>) New class.
10930 (int_traits<unsigned short>) Likewise.
10931 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
10932 Use GET_MODE_UNIT_PRECISION and remove redundant test for
10934 * combine.c (set_nonzero_bits_and_sign_copies): Use
10935 is_a <scalar_int_mode>.
10936 (find_split_point): Likewise.
10937 (combine_simplify_rtx): Likewise.
10938 (simplify_logical): Likewise.
10939 (expand_compound_operation): Likewise.
10940 (expand_field_assignment): Likewise.
10941 (make_compound_operation): Likewise.
10942 (extended_count): Likewise.
10943 (change_zero_ext): Likewise.
10944 (simplify_comparison): Likewise.
10945 * dwarf2out.c (scompare_loc_descriptor): Likewise.
10946 (ucompare_loc_descriptor): Likewise.
10947 (minmax_loc_descriptor): Likewise.
10948 (mem_loc_descriptor): Likewise.
10949 (loc_descriptor): Likewise.
10950 * expmed.c (init_expmed_one_mode): Likewise.
10951 * lra-constraints.c (lra_constraint_offset): Likewise.
10952 * optabs.c (prepare_libcall_arg): Likewise.
10953 * postreload.c (move2add_note_store): Likewise.
10954 * reload.c (operands_match_p): Likewise.
10955 * rtl.h (load_extend_op): Likewise.
10956 * rtlhooks.c (gen_lowpart_general): Likewise.
10957 * simplify-rtx.c (simplify_truncation): Likewise.
10958 (simplify_unary_operation_1): Likewise.
10959 (simplify_binary_operation_1): Likewise.
10960 (simplify_const_binary_operation): Likewise.
10961 (simplify_const_relational_operation): Likewise.
10962 (simplify_subreg): Likewise.
10963 * stor-layout.c (bitwise_mode_for_mode): Likewise.
10964 * var-tracking.c (adjust_mems): Likewise.
10965 (prepare_call_arguments): Likewise.
10967 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10968 Alan Hayward <alan.hayward@arm.com>
10969 David Sherwood <david.sherwood@arm.com>
10971 * machmode.h (is_int_mode): New fuction.
10972 * combine.c (find_split_point): Use it.
10973 (combine_simplify_rtx): Likewise.
10974 (simplify_if_then_else): Likewise.
10975 (simplify_set): Likewise.
10976 (simplify_shift_const_1): Likewise.
10977 (simplify_comparison): Likewise.
10978 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
10979 * cse.c (notreg_cost): Likewise.
10980 (cse_insn): Likewise.
10981 * cselib.c (cselib_lookup_1): Likewise.
10982 * dojump.c (do_jump_1): Likewise.
10983 (do_compare_rtx_and_jump): Likewise.
10984 * dse.c (get_call_args): Likewise.
10985 * dwarf2out.c (rtl_for_decl_init): Likewise.
10986 (native_encode_initializer): Likewise.
10987 * expmed.c (emit_store_flag_1): Likewise.
10988 (emit_store_flag): Likewise.
10989 * expr.c (convert_modes): Likewise.
10990 (store_field): Likewise.
10991 (expand_expr_real_1): Likewise.
10992 * fold-const.c (fold_read_from_constant_string): Likewise.
10993 * gimple-ssa-sprintf.c (get_format_string): Likewise.
10994 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
10995 * optabs.c (expand_binop): Likewise.
10996 (expand_unop): Likewise.
10997 (expand_abs_nojump): Likewise.
10998 (expand_one_cmpl_abs_nojump): Likewise.
10999 * simplify-rtx.c (mode_signbit_p): Likewise.
11000 (val_signbit_p): Likewise.
11001 (val_signbit_known_set_p): Likewise.
11002 (val_signbit_known_clear_p): Likewise.
11003 (simplify_relational_operation_1): Likewise.
11004 * tree.c (vector_type_mode): Likewise.
11006 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11007 Alan Hayward <alan.hayward@arm.com>
11008 David Sherwood <david.sherwood@arm.com>
11010 * machmode.h (smallest_mode_for_size): Fix formatting.
11011 (smallest_int_mode_for_size): New function.
11012 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
11013 instead of smallest_mode_for_size.
11014 * combine.c (make_extraction): Likewise.
11015 * config/arc/arc.c (arc_expand_movmem): Likewise.
11016 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
11017 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
11018 * config/s390/s390.c (s390_expand_insv): Likewise.
11019 * config/sparc/sparc.c (assign_int_registers): Likewise.
11020 * config/spu/spu.c (spu_function_value): Likewise.
11021 (spu_function_arg): Likewise.
11022 * coverage.c (get_gcov_type): Likewise.
11023 (get_gcov_unsigned_t): Likewise.
11024 * dse.c (find_shift_sequence): Likewise.
11025 * expmed.c (store_bit_field_1): Likewise.
11026 * expr.c (convert_move): Likewise.
11027 (store_field): Likewise.
11028 * internal-fn.c (expand_arith_overflow): Likewise.
11029 * optabs-query.c (get_best_extraction_insn): Likewise.
11030 * optabs.c (expand_twoval_binop_libfunc): Likewise.
11031 * stor-layout.c (layout_type): Likewise.
11032 (initialize_sizetypes): Likewise.
11033 * targhooks.c (default_get_mask_mode): Likewise.
11034 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
11036 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11037 Alan Hayward <alan.hayward@arm.com>
11038 David Sherwood <david.sherwood@arm.com>
11040 * machmode.h (opt_mode::else_blk): New function.
11041 (int_mode_for_mode): Declare.
11042 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
11043 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
11045 * cfgexpand.c (expand_debug_expr): Likewise.
11046 * combine.c (gen_lowpart_or_truncate): Likewise.
11047 (gen_lowpart_for_combine): Likewise.
11048 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
11049 * config/avr/avr.c (avr_to_int_mode): Likewise.
11050 (avr_out_plus_1): Likewise.
11051 (avr_out_plus): Likewise.
11052 (avr_out_round): Likewise.
11053 * config/i386/i386.c (ix86_split_to_parts): Likewise.
11054 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
11055 (s390_expand_vcond): Likewise.
11056 * config/spu/spu.c (spu_split_immediate): Likewise.
11057 (spu_expand_mov): Likewise.
11058 * dse.c (get_stored_val): Likewise.
11059 * expmed.c (store_bit_field_1): Likewise.
11060 (convert_extracted_bit_field): Use int_mode_for_mode instead of
11062 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
11063 (extract_low_bits): Likewise.
11064 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
11065 handling rather than repeating the check.
11066 (emit_group_store): Likewise.
11067 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
11068 * optabs.c (expand_absneg_bit): Likewise.
11069 (expand_copysign_absneg): Likewise.
11070 (expand_copysign_bit): Likewise.
11071 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
11072 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
11073 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
11074 * var-tracking.c (prepare_call_arguments): Likewise.
11075 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
11076 int_mode_for_mode instead of mode_for_size.
11077 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
11079 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11080 Alan Hayward <alan.hayward@arm.com>
11081 David Sherwood <david.sherwood@arm.com>
11083 * machmode.h (int_mode_for_size): New function.
11084 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
11085 instead of mode_for_size.
11086 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
11088 * combine.c (expand_field_assignment): Use int_mode_for_size
11089 instead of mode_for_size.
11090 (make_extraction): Likewise.
11091 (simplify_shift_const_1): Likewise.
11092 (simplify_comparison): Likewise.
11093 * dojump.c (do_jump): Likewise.
11094 * dwarf2out.c (mem_loc_descriptor): Likewise.
11095 * emit-rtl.c (init_derived_machine_modes): Likewise.
11096 * expmed.c (flip_storage_order): Likewise.
11097 (convert_extracted_bit_field): Likewise.
11098 * expr.c (copy_blkmode_from_reg): Likewise.
11099 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
11100 * internal-fn.c (expand_mul_overflow): Likewise.
11101 * lower-subreg.c (simple_move): Likewise.
11102 * optabs-libfuncs.c (init_optabs): Likewise.
11103 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11104 * tree.c (vector_type_mode): Likewise.
11105 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
11106 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
11107 * tree-vect-generic.c (expand_vector_parallel): Likewise.
11108 * tree-vect-stmts.c (vectorizable_load): Likewise.
11109 (vectorizable_store): Likewise.
11111 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11112 Alan Hayward <alan.hayward@arm.com>
11113 David Sherwood <david.sherwood@arm.com>
11115 * coretypes.h (pod_mode): New type.
11116 (scalar_int_mode_pod): New typedef.
11117 * machmode.h (pod_mode): New class.
11118 (int_n_data_t::m): Change type to scalar_int_mode_pod.
11119 * genmodes.c (emit_mode_int_n): Update accordingly.
11120 * lower-subreg.h (target_lower_subreg): Change type to
11121 scalar_int_mode_pod.
11122 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
11123 scalar_int_mode_pod.
11125 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11126 Alan Hayward <alan.hayward@arm.com>
11127 David Sherwood <david.sherwood@arm.com>
11129 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
11130 machine_mode to scalar_int_mode.
11131 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
11132 (rs6000_option_override_internal): Remove cast to int.
11133 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
11134 machine_mode to scalar_int_mode.
11135 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
11136 (rs6000_option_override_internal): Remove cast to int.
11137 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
11138 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
11140 * config/s390/s390.c (s390_expand_builtin): Likewise.
11141 * coretypes.h (scalar_int_mode): New type.
11142 (opt_scalar_int_mode): New typedef.
11143 * machmode.h (scalar_int_mode): New class.
11144 (scalar_int_mode::includes_p): New function.
11145 (byte_mode): Change type to scalar_int_mode.
11146 (word_mode): Likewise.
11147 (ptr_mode): Likewise.
11148 * emit-rtl.c (byte_mode): Likewise.
11149 (word_mode): Likewise.
11150 (ptr_mode): Likewise.
11151 (init_derived_machine_modes): Update accordingly.
11152 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
11153 and MODE_PARTIAL_INT.
11154 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
11155 opt_scalar_int_mode.
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 (libgcc_floating_mode_supported_p): Take a
11163 * doc/tm.texi: Regenerate.
11164 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
11166 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
11167 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
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 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
11175 * doc/tm.texi: Regenerate.
11176 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
11177 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
11178 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
11179 * targhooks.h (default_floatn_mode): Likewise.
11180 * targhooks.c (default_floatn_mode): Likewise.
11181 * tree.c (build_common_tree_nodes): Update accordingly.
11183 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11184 Alan Hayward <alan.hayward@arm.com>
11185 David Sherwood <david.sherwood@arm.com>
11187 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
11188 (mode_iterator::iterate_p): Likewise.
11189 (mode_iterator::get_wider): Likewise.
11190 * expr.c (init_expr_target): Use opt_scalar_float_mode.
11192 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11193 Alan Hayward <alan.hayward@arm.com>
11194 David Sherwood <david.sherwood@arm.com>
11196 * coretypes.h (opt_scalar_float_mode): New typedef.
11197 * machmode.h (float_mode_for_size): New function.
11198 * emit-rtl.c (double_mode): Delete.
11199 (init_emit_once): Use float_mode_for_size.
11200 * stor-layout.c (layout_type): Likewise.
11201 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
11203 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11204 Alan Hayward <alan.hayward@arm.com>
11205 David Sherwood <david.sherwood@arm.com>
11207 * output.h (assemble_real): Take a scalar_float_mode.
11208 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
11209 * config/arm/arm.md (consttable_4): Likewise.
11210 (consttable_8): Likewise.
11211 (consttable_16): Likewise.
11212 * config/mips/mips.md (consttable_float): Likewise.
11213 * config/s390/s390.c (s390_output_pool_entry): Likewise.
11214 * varasm.c (assemble_real): Take a scalar_float_mode.
11215 (output_constant_pool_2): Update accordingly.
11216 (output_constant): Likewise.
11218 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11219 Alan Hayward <alan.hayward@arm.com>
11220 David Sherwood <david.sherwood@arm.com>
11222 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
11223 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
11224 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
11225 (native_encode_real): Likewise.
11226 (native_interpret_real): Likewise.
11227 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
11228 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11230 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11231 Alan Hayward <alan.hayward@arm.com>
11232 David Sherwood <david.sherwood@arm.com>
11234 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
11235 <scalar_float_mode>. Simplify.
11236 (gen_extend_conv_libfunc): Likewise.
11238 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11239 Alan Hayward <alan.hayward@arm.com>
11240 David Sherwood <david.sherwood@arm.com>
11242 * coretypes.h (scalar_float_mode): New type.
11243 * machmode.h (mode_traits::from_int): Use machine_mode if
11244 USE_ENUM_MODES is defined.
11245 (is_a): New function.
11247 (dyn_cast): Likewise.
11248 (scalar_float_mode): New class.
11249 (scalar_float_mode::includes_p): New function.
11250 (is_float_mode): Likewise.
11251 * gdbhooks.py (MachineModePrinter): New class.
11252 (build_pretty_printer): Use it for scalar_float_mode.
11253 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
11254 (format_helper::format_helper): Turn into a template.
11255 * genmodes.c (get_mode_class): New function.
11256 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
11257 or machine_mode if none.
11258 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
11259 as_a <scalar_float_mode>.
11260 * dwarf2out.c (mem_loc_descriptor): Likewise.
11261 (insert_float): Likewise.
11262 (add_const_value_attribute): Likewise.
11263 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11264 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
11265 (expand_unop): Update accordingly.
11266 (expand_abs_nojump): Likewise.
11267 (expand_copysign_absneg): Take a scalar_float_mode.
11268 (expand_copysign_bit): Likewise.
11269 (expand_copysign): Update accordingly.
11271 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11272 Alan Hayward <alan.hayward@arm.com>
11273 David Sherwood <david.sherwood@arm.com>
11275 * coretypes.h (opt_mode): New class.
11276 * machmode.h (opt_mode): Likewise.
11277 (opt_mode::else_void): New function.
11278 (opt_mode::require): Likewise.
11279 (opt_mode::exists): Likewise.
11280 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
11281 (GET_MODE_2XWIDER_MODE): Likewise.
11282 (mode_iterator::get_wider): Update accordingly.
11283 (mode_iterator::get_2xwider): Likewise.
11284 (mode_iterator::get_known_wider): Likewise, turning into a template.
11285 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
11286 forcing a wider mode to exist.
11287 * config/cr16/cr16.h (LONG_REG_P): Likewise.
11288 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11289 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
11290 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
11291 * lower-subreg.c (init_lower_subreg): Likewise.
11292 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
11293 on the final iteration.
11294 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
11295 a wider mode exists before asking for a move pattern.
11296 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
11297 forcing a wider mode to exist.
11298 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
11299 returning false if no such mode exists.
11300 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
11301 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
11302 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
11303 Avoid checking for a MODE_INT if we already know the mode is not a
11305 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
11306 forcing a wider mode to exist.
11307 (expmed_mult_highpart_optab): Likewise.
11308 (expmed_mult_highpart): Likewise.
11309 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
11311 * lto-streamer-in.c (lto_input_mode_table): Likewise.
11312 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
11313 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
11314 * internal-fn.c (expand_mul_overflow): Update use of
11315 GET_MODE_2XWIDER_MODE.
11316 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11317 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
11318 GET_MODE_WIDER_MODE.
11319 (convert_plusminus_to_widen): Likewise.
11320 * tree-switch-conversion.c (array_value_type): Likewise.
11321 * var-tracking.c (emit_note_insn_var_location): Likewise.
11322 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11323 Return false inside rather than outside the loop if no wider mode
11325 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
11326 and GET_MODE_2XWIDER_MODE
11327 (can_compare_p): Use else_void.
11328 * gdbhooks.py (OptMachineModePrinter): New class.
11329 (build_pretty_printer): Use it for opt_mode.
11331 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11332 Alan Hayward <alan.hayward@arm.com>
11333 David Sherwood <david.sherwood@arm.com>
11335 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
11336 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
11338 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11339 Alan Hayward <alan.hayward@arm.com>
11340 David Sherwood <david.sherwood@arm.com>
11342 * machmode.h (mode_traits): New structure.
11343 (get_narrowest_mode): New function.
11344 (mode_iterator::start): Likewise.
11345 (mode_iterator::iterate_p): Likewise.
11346 (mode_iterator::get_wider): Likewise.
11347 (mode_iterator::get_known_wider): Likewise.
11348 (mode_iterator::get_2xwider): Likewise.
11349 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
11350 (FOR_EACH_MODE): Likewise.
11351 (FOR_EACH_MODE_FROM): Likewise.
11352 (FOR_EACH_MODE_UNTIL): Likewise.
11353 (FOR_EACH_WIDER_MODE): Likewise.
11354 (FOR_EACH_2XWIDER_MODE): Likewise.
11355 * builtins.c (expand_builtin_strlen): Use new mode iterators.
11356 * combine.c (simplify_comparison): Likewise
11357 * config/i386/i386.c (type_natural_mode): Likewise.
11358 * cse.c (cse_insn): Likewise.
11359 * dse.c (find_shift_sequence): Likewise.
11360 * emit-rtl.c (init_derived_machine_modes): Likewise.
11361 (init_emit_once): Likewise.
11362 * explow.c (hard_function_value): Likewise.
11363 * expmed.c (extract_fixed_bit_field_1): Likewise.
11364 (extract_bit_field_1): Likewise.
11365 (expand_divmod): Likewise.
11366 (emit_store_flag_1): Likewise.
11367 * expr.c (init_expr_target): Likewise.
11368 (convert_move): Likewise.
11369 (alignment_for_piecewise_move): Likewise.
11370 (widest_int_mode_for_size): Likewise.
11371 (emit_block_move_via_movmem): Likewise.
11372 (copy_blkmode_to_reg): Likewise.
11373 (set_storage_via_setmem): Likewise.
11374 (compress_float_constant): Likewise.
11375 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11376 * optabs-query.c (get_best_extraction_insn): Likewise.
11377 * optabs.c (expand_binop): Likewise.
11378 (expand_twoval_unop): Likewise.
11379 (expand_twoval_binop): Likewise.
11380 (widen_leading): Likewise.
11381 (widen_bswap): Likewise.
11382 (expand_parity): Likewise.
11383 (expand_unop): Likewise.
11384 (prepare_cmp_insn): Likewise.
11385 (prepare_float_lib_cmp): Likewise.
11386 (expand_float): Likewise.
11387 (expand_fix): Likewise.
11388 (expand_sfix_optab): Likewise.
11389 * postreload.c (move2add_use_add2_insn): Likewise.
11390 * reg-stack.c (reg_to_stack): Likewise.
11391 * reginfo.c (choose_hard_reg_mode): Likewise.
11392 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11393 * stor-layout.c (mode_for_size): Likewise.
11394 (smallest_mode_for_size): Likewise.
11395 (mode_for_vector): Likewise.
11396 (finish_bitfield_representative): Likewise.
11397 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
11398 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
11399 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11400 * var-tracking.c (prepare_call_arguments): Likewise.
11402 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11403 Alan Hayward <alan.hayward@arm.com>
11404 David Sherwood <david.sherwood@arm.com>
11406 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
11407 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
11408 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
11409 * machmode.h (mode_size): Move earlier in file.
11410 (mode_precision): Likewise.
11411 (mode_inner): Likewise.
11412 (mode_nunits): Likewise.
11413 (mode_unit_size): Likewise.
11414 (unit_unit_precision): Likewise.
11415 (mode_wider): Likewise.
11416 (mode_2xwider): Likewise.
11417 (machine_mode): New class.
11418 (mode_to_bytes): New function.
11419 (mode_to_bits): Likewise.
11420 (mode_to_precision): Likewise.
11421 (mode_to_inner): Likewise.
11422 (mode_to_unit_size): Likewise.
11423 (mode_to_unit_precision): Likewise.
11424 (mode_to_nunits): Likewise.
11425 (GET_MODE_SIZE): Use mode_to_bytes.
11426 (GET_MODE_BITSIZE): Use mode_to_bits.
11427 (GET_MODE_PRECISION): Use mode_to_precision.
11428 (GET_MODE_INNER): Use mode_to_inner.
11429 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
11430 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
11431 (GET_MODE_NUNITS): Use mode_to_nunits.
11432 * system.h (ALWAYS_INLINE): New macro.
11433 * config/powerpcspe/powerpcspe-c.c
11434 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
11435 int for arg1_mode and arg2_mode.
11437 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11438 Alan Hayward <alan.hayward@arm.com>
11439 David Sherwood <david.sherwood@arm.com>
11441 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
11442 Prefix mode names with E_ in case statements.
11443 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11444 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
11445 (aarch64_split_simd_move): Likewise.
11446 (aarch64_gen_storewb_pair): Likewise.
11447 (aarch64_gen_loadwb_pair): Likewise.
11448 (aarch64_gen_store_pair): Likewise.
11449 (aarch64_gen_load_pair): Likewise.
11450 (aarch64_get_condition_code_1): Likewise.
11451 (aarch64_constant_pool_reload_icode): Likewise.
11452 (get_rsqrte_type): Likewise.
11453 (get_rsqrts_type): Likewise.
11454 (get_recpe_type): Likewise.
11455 (get_recps_type): Likewise.
11456 (aarch64_gimplify_va_arg_expr): Likewise.
11457 (aarch64_simd_container_mode): Likewise.
11458 (aarch64_emit_load_exclusive): Likewise.
11459 (aarch64_emit_store_exclusive): Likewise.
11460 (aarch64_expand_compare_and_swap): Likewise.
11461 (aarch64_gen_atomic_cas): Likewise.
11462 (aarch64_emit_bic): Likewise.
11463 (aarch64_emit_atomic_swap): Likewise.
11464 (aarch64_emit_atomic_load_op): Likewise.
11465 (aarch64_evpc_trn): Likewise.
11466 (aarch64_evpc_uzp): Likewise.
11467 (aarch64_evpc_zip): Likewise.
11468 (aarch64_evpc_ext): Likewise.
11469 (aarch64_evpc_rev): Likewise.
11470 (aarch64_evpc_dup): Likewise.
11471 (aarch64_gen_ccmp_first): Likewise.
11472 (aarch64_gen_ccmp_next): Likewise.
11473 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11474 (alpha_emit_xfloating_libcall): Likewise.
11475 (emit_insxl): Likewise.
11476 (alpha_arg_type): Likewise.
11477 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
11478 (arc_preferred_simd_mode): Likewise.
11479 (arc_secondary_reload): Likewise.
11480 (get_arc_condition_code): Likewise.
11481 (arc_print_operand): Likewise.
11482 (arc_legitimate_constant_p): Likewise.
11483 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11484 * config/arc/arc.md (casesi_load): Likewise.
11485 (casesi_compact_jump): Likewise.
11486 * config/arc/predicates.md (proper_comparison_operator): Likewise.
11487 (cc_use_register): Likewise.
11488 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11489 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
11490 (arm_init_iwmmxt_builtins): Likewise.
11491 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
11492 (neon_expand_vector_init): Likewise.
11493 (arm_attr_length_move_neon): Likewise.
11494 (maybe_get_arm_condition_code): Likewise.
11495 (arm_emit_vector_const): Likewise.
11496 (arm_preferred_simd_mode): Likewise.
11497 (arm_output_iwmmxt_tinsr): Likewise.
11498 (thumb1_output_casesi): Likewise.
11499 (thumb2_output_casesi): Likewise.
11500 (arm_emit_load_exclusive): Likewise.
11501 (arm_emit_store_exclusive): Likewise.
11502 (arm_expand_compare_and_swap): Likewise.
11503 (arm_evpc_neon_vuzp): Likewise.
11504 (arm_evpc_neon_vzip): Likewise.
11505 (arm_evpc_neon_vrev): Likewise.
11506 (arm_evpc_neon_vtrn): Likewise.
11507 (arm_evpc_neon_vext): Likewise.
11508 (arm_validize_comparison): Likewise.
11509 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
11510 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
11511 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
11512 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
11513 (c6x_preferred_simd_mode): Likewise.
11514 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
11515 (epiphany_rtx_costs): Likewise.
11516 * config/epiphany/predicates.md (proper_comparison_operator):
11518 * config/frv/frv.c (condexec_memory_operand): Likewise.
11519 (frv_emit_move): Likewise.
11520 (output_move_single): Likewise.
11521 (output_condmove_single): Likewise.
11522 (frv_hard_regno_mode_ok): Likewise.
11523 (frv_matching_accg_mode): Likewise.
11524 * config/h8300/h8300.c (split_adds_subs): Likewise.
11525 (h8300_rtx_costs): Likewise.
11526 (h8300_print_operand): Likewise.
11527 (compute_mov_length): Likewise.
11528 (output_logical_op): Likewise.
11529 (compute_logical_op_length): Likewise.
11530 (compute_logical_op_cc): Likewise.
11531 (h8300_shift_needs_scratch_p): Likewise.
11532 (output_a_shift): Likewise.
11533 (compute_a_shift_length): Likewise.
11534 (compute_a_shift_cc): Likewise.
11535 (expand_a_rotate): Likewise.
11536 (output_a_rotate): Likewise.
11537 * config/i386/i386.c (classify_argument): Likewise.
11538 (function_arg_advance_32): Likewise.
11539 (function_arg_32): Likewise.
11540 (function_arg_64): Likewise.
11541 (function_value_64): Likewise.
11542 (ix86_gimplify_va_arg): Likewise.
11543 (ix86_legitimate_constant_p): Likewise.
11544 (put_condition_code): Likewise.
11545 (split_double_mode): Likewise.
11546 (ix86_avx256_split_vector_move_misalign): Likewise.
11547 (ix86_expand_vector_logical_operator): Likewise.
11548 (ix86_split_idivmod): Likewise.
11549 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
11550 (ix86_build_const_vector): Likewise.
11551 (ix86_build_signbit_mask): Likewise.
11552 (ix86_match_ccmode): Likewise.
11553 (ix86_cc_modes_compatible): Likewise.
11554 (ix86_expand_branch): Likewise.
11555 (ix86_expand_sse_cmp): Likewise.
11556 (ix86_expand_sse_movcc): Likewise.
11557 (ix86_expand_int_sse_cmp): Likewise.
11558 (ix86_expand_vec_perm_vpermi2): Likewise.
11559 (ix86_expand_vec_perm): Likewise.
11560 (ix86_expand_sse_unpack): Likewise.
11561 (ix86_expand_int_addcc): Likewise.
11562 (ix86_split_to_parts): Likewise.
11563 (ix86_vectorize_builtin_gather): Likewise.
11564 (ix86_vectorize_builtin_scatter): Likewise.
11565 (avx_vpermilp_parallel): Likewise.
11566 (inline_memory_move_cost): Likewise.
11567 (ix86_tieable_integer_mode_p): Likewise.
11568 (x86_maybe_negate_const_int): Likewise.
11569 (ix86_expand_vector_init_duplicate): Likewise.
11570 (ix86_expand_vector_init_one_nonzero): Likewise.
11571 (ix86_expand_vector_init_one_var): Likewise.
11572 (ix86_expand_vector_init_concat): Likewise.
11573 (ix86_expand_vector_init_interleave): Likewise.
11574 (ix86_expand_vector_init_general): Likewise.
11575 (ix86_expand_vector_set): Likewise.
11576 (ix86_expand_vector_extract): Likewise.
11577 (emit_reduc_half): Likewise.
11578 (ix86_emit_i387_round): Likewise.
11579 (ix86_mangle_type): Likewise.
11580 (ix86_expand_round_sse4): Likewise.
11581 (expand_vec_perm_blend): Likewise.
11582 (canonicalize_vector_int_perm): Likewise.
11583 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
11584 (expand_vec_perm_1): Likewise.
11585 (expand_vec_perm_interleave3): Likewise.
11586 (expand_vec_perm_even_odd_pack): Likewise.
11587 (expand_vec_perm_even_odd_1): Likewise.
11588 (expand_vec_perm_broadcast_1): Likewise.
11589 (ix86_vectorize_vec_perm_const_ok): Likewise.
11590 (ix86_expand_vecop_qihi): Likewise.
11591 (ix86_expand_mul_widen_hilo): Likewise.
11592 (ix86_expand_sse2_abs): Likewise.
11593 (ix86_expand_pextr): Likewise.
11594 (ix86_expand_pinsr): Likewise.
11595 (ix86_preferred_simd_mode): Likewise.
11596 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
11597 * config/i386/sse.md (*andnot<mode>3): Likewise.
11598 (<mask_codefor><code><mode>3<mask_name>): Likewise.
11599 (*<code><mode>3): Likewise.
11600 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
11601 (ia64_expand_atomic_op): Likewise.
11602 (ia64_arg_type): Likewise.
11603 (ia64_mode_to_int): Likewise.
11604 (ia64_scalar_mode_supported_p): Likewise.
11605 (ia64_vector_mode_supported_p): Likewise.
11606 (expand_vec_perm_broadcast): Likewise.
11607 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
11608 (iq2000_function_arg_advance): Likewise.
11609 (iq2000_function_arg): Likewise.
11610 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
11611 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
11612 (m68k_libcall_value): Likewise.
11613 (m68k_function_value): Likewise.
11614 (sched_attr_op_type): Likewise.
11615 * config/mcore/mcore.c (mcore_output_move): Likewise.
11616 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
11618 (microblaze_function_arg): Likewise.
11619 * config/mips/mips.c (mips16_build_call_stub): Likewise.
11620 (mips_print_operand): Likewise.
11621 (mips_mode_ok_for_mov_fmt_p): Likewise.
11622 (mips_vector_mode_supported_p): Likewise.
11623 (mips_preferred_simd_mode): Likewise.
11624 (mips_expand_vpc_loongson_even_odd): Likewise.
11625 (mips_expand_vec_unpack): Likewise.
11626 (mips_expand_vi_broadcast): Likewise.
11627 (mips_expand_vector_init): Likewise.
11628 (mips_expand_vec_reduc): Likewise.
11629 (mips_expand_msa_cmp): Likewise.
11630 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
11631 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
11632 (cc_flags_for_mode): Likewise.
11633 * config/msp430/msp430.c (msp430_print_operand): Likewise.
11634 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
11635 (nds32_output_casesi_pc_relative): Likewise.
11636 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11637 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
11638 (nvptx_gen_unpack): Likewise.
11639 (nvptx_gen_pack): Likewise.
11640 (nvptx_gen_shuffle): Likewise.
11641 (nvptx_gen_wcast): Likewise.
11642 (nvptx_preferred_simd_mode): Likewise.
11643 * config/pa/pa.c (pa_secondary_reload): Likewise.
11644 * config/pa/predicates.md (base14_operand): Likewise.
11645 * config/powerpcspe/powerpcspe-c.c
11646 (altivec_resolve_overloaded_builtin): Likewise.
11647 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
11649 (rs6000_preferred_simd_mode): Likewise.
11650 (output_vec_const_move): Likewise.
11651 (rs6000_expand_vector_extract): Likewise.
11652 (rs6000_split_vec_extract_var): Likewise.
11653 (reg_offset_addressing_ok_p): Likewise.
11654 (rs6000_legitimate_offset_address_p): Likewise.
11655 (rs6000_legitimize_address): Likewise.
11656 (rs6000_emit_set_const): Likewise.
11657 (rs6000_const_vec): Likewise.
11658 (rs6000_emit_move): Likewise.
11659 (spe_build_register_parallel): Likewise.
11660 (rs6000_darwin64_record_arg_recurse): Likewise.
11661 (swap_selector_for_mode): Likewise.
11662 (spe_init_builtins): Likewise.
11663 (paired_init_builtins): Likewise.
11664 (altivec_init_builtins): Likewise.
11665 (do_load_for_compare): Likewise.
11666 (rs6000_generate_compare): Likewise.
11667 (rs6000_expand_float128_convert): Likewise.
11668 (emit_load_locked): Likewise.
11669 (emit_store_conditional): Likewise.
11670 (rs6000_output_function_epilogue): Likewise.
11671 (rs6000_handle_altivec_attribute): Likewise.
11672 (rs6000_function_value): Likewise.
11673 (emit_fusion_gpr_load): Likewise.
11674 (emit_fusion_p9_load): Likewise.
11675 (emit_fusion_p9_store): Likewise.
11676 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
11677 (fusion_gpr_mem_load): Likewise.
11678 (fusion_addis_mem_combo_load): Likewise.
11679 (fusion_addis_mem_combo_store): Likewise.
11680 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
11681 (fusion_gpr_mem_load): Likewise.
11682 (fusion_addis_mem_combo_load): Likewise.
11683 (fusion_addis_mem_combo_store): Likewise.
11684 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11686 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
11687 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
11688 (rs6000_preferred_simd_mode): Likewise.
11689 (output_vec_const_move): Likewise.
11690 (rs6000_expand_vector_extract): Likewise.
11691 (rs6000_split_vec_extract_var): Likewise.
11692 (reg_offset_addressing_ok_p): Likewise.
11693 (rs6000_legitimate_offset_address_p): Likewise.
11694 (rs6000_legitimize_address): Likewise.
11695 (rs6000_emit_set_const): Likewise.
11696 (rs6000_const_vec): Likewise.
11697 (rs6000_emit_move): Likewise.
11698 (rs6000_darwin64_record_arg_recurse): Likewise.
11699 (swap_selector_for_mode): Likewise.
11700 (paired_init_builtins): Likewise.
11701 (altivec_init_builtins): Likewise.
11702 (rs6000_expand_float128_convert): Likewise.
11703 (emit_load_locked): Likewise.
11704 (emit_store_conditional): Likewise.
11705 (rs6000_output_function_epilogue): Likewise.
11706 (rs6000_handle_altivec_attribute): Likewise.
11707 (rs6000_function_value): Likewise.
11708 (emit_fusion_gpr_load): Likewise.
11709 (emit_fusion_p9_load): Likewise.
11710 (emit_fusion_p9_store): Likewise.
11711 * config/rx/rx.c (rx_gen_move_template): Likewise.
11712 (flags_from_mode): Likewise.
11713 * config/s390/predicates.md (s390_alc_comparison): Likewise.
11714 (s390_slb_comparison): Likewise.
11715 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
11716 (s390_vector_mode_supported_p): Likewise.
11717 (s390_cc_modes_compatible): Likewise.
11718 (s390_match_ccmode_set): Likewise.
11719 (s390_canonicalize_comparison): Likewise.
11720 (s390_emit_compare_and_swap): Likewise.
11721 (s390_branch_condition_mask): Likewise.
11722 (s390_rtx_costs): Likewise.
11723 (s390_secondary_reload): Likewise.
11724 (__SECONDARY_RELOAD_CASE): Likewise.
11725 (s390_expand_cs): Likewise.
11726 (s390_preferred_simd_mode): Likewise.
11727 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
11728 * config/sh/sh.c (sh_print_operand): Likewise.
11729 (dump_table): Likewise.
11730 (sh_secondary_reload): Likewise.
11731 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11732 * config/sh/sh.md (casesi_worker_1): Likewise.
11733 (casesi_worker_2): Likewise.
11734 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
11735 (fcc_comparison_operator): Likewise.
11736 * config/sparc/sparc.c (sparc_expand_move): Likewise.
11737 (emit_soft_tfmode_cvt): Likewise.
11738 (sparc_preferred_simd_mode): Likewise.
11739 (output_cbranch): Likewise.
11740 (sparc_print_operand): Likewise.
11741 (sparc_expand_vec_perm_bmask): Likewise.
11742 (vector_init_bshuffle): Likewise.
11743 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11744 (spu_vector_mode_supported_p): Likewise.
11745 (spu_expand_insv): Likewise.
11746 (spu_emit_branch_or_set): Likewise.
11747 (spu_handle_vector_attribute): Likewise.
11748 (spu_builtin_splats): Likewise.
11749 (spu_builtin_extract): Likewise.
11750 (spu_builtin_promote): Likewise.
11751 (spu_expand_sign_extend): Likewise.
11752 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11753 (tilegx_simd_int): Likewise.
11754 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
11755 (tilepro_simd_int): Likewise.
11756 * config/v850/v850.c (const_double_split): Likewise.
11757 (v850_print_operand): Likewise.
11758 (ep_memory_offset): Likewise.
11759 * config/vax/vax.c (vax_rtx_costs): Likewise.
11760 (vax_output_int_move): Likewise.
11761 (vax_output_int_add): Likewise.
11762 (vax_output_int_subtract): Likewise.
11763 * config/visium/predicates.md (visium_branch_operator): Likewise.
11764 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
11765 (visium_print_operand_address): Likewise.
11766 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11767 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
11768 (xtensa_expand_conditional_branch): Likewise.
11769 (xtensa_copy_incoming_a7): Likewise.
11770 (xtensa_output_literal): Likewise.
11771 * dfp.c (decimal_real_maxval): Likewise.
11772 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
11774 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11775 Alan Hayward <alan.hayward@arm.com>
11776 David Sherwood <david.sherwood@arm.com>
11778 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
11779 (mode_nunits_inline): Likewise.
11780 (mode_inner_inline): Likewise.
11781 (mode_unit_size_inline): Likewise.
11782 (mode_unit_precision_inline): Likewise.
11783 (emit_insn_modes_h): Likewise. Also emit a #define of the
11785 (emit_mode_wider): Add an E_ prefix to mode names.
11786 (emit_mode_complex): Likewise.
11787 (emit_mode_inner): Likewise.
11788 (emit_mode_adjustments): Likewise.
11789 (emit_mode_int_n): Likewise.
11790 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
11791 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
11792 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
11793 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
11794 (CRC32_BUILTIN, ENTRY): Likewise.
11795 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
11796 (aarch64_pop_regs): Likewise.
11797 (aarch64_process_components): Likewise.
11798 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
11799 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
11800 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
11801 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
11802 * config/arm/arm.c (arm_init_libfuncs): Likewise.
11803 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
11805 * config/i386/i386-builtin.def (pcmpestr): Likewise.
11806 (pcmpistr): Likewise.
11807 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
11808 * config/mmix/mmix.c (mmix_output_condition): Likewise.
11809 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
11811 * config/rl78/rl78.c (mduc_regs): Likewise.
11812 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
11813 (htm_expand_builtin): Likewise.
11814 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
11815 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
11816 * config/xtensa/xtensa.c (print_operand): Likewise.
11817 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
11818 (NUM_MODE_VECTOR_INT): Likewise.
11819 * genoutput.c (null_operand): Likewise.
11820 (output_operand_data): Likewise.
11821 * genrecog.c (print_parameter_value): Likewise.
11822 * lra.c (debug_operand_data): Likewise.
11824 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11826 * dwarf2out.c (scompare_loc_descriptor_wide)
11827 (scompare_loc_descriptor_narrow): New functions, split out from...
11828 (scompare_loc_descriptor): ...here.
11829 * expmed.c (emit_store_flag_int): New function, split out from...
11830 (emit_store_flag): ...here.
11832 2017-08-30 Richard Biener <rguenther@suse.de>
11834 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
11835 (dwarf2out_early_finish): Move setting of AT_pubnames from
11836 early debug output to early finish.
11838 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11840 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
11841 and -mdata-region to the assembler.
11843 2017-08-30 Richard Biener <rguenther@suse.de>
11845 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
11847 (gen_subprogram_die): Add DW_AT_object_pointer only early.
11848 (dwarf2out_early_global_decl): Only generate a DIE for the
11849 abstract origin if it doesn't already exist or is a declaration DIE.
11850 (resolve_addr): Do not add the linkage name twice when
11851 generating a stub DIE for the DW_TAG_GNU_call_site target.
11853 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11855 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11856 Use machine_mode rather than int for arg1_mode.
11858 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
11861 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
11862 that the second argument of the built-in functions to unpack
11863 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
11864 switch statement instead a lot of if statements.
11865 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
11866 Allow 64-bit values to be in Altivec registers as well as
11867 traditional floating point registers.
11868 (pack<mode>, FMOVE128_VSX iterator): Likewise.
11870 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
11872 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
11873 MAX_REGS_PER_ADDRESS == 1.
11875 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
11877 * config/i386/i386.opt (flag_fentry): Do not init to -1.
11878 * config/i386/i386.c (ix86_option_override_internal): Simplify
11879 setting of opts->x_flag_entry.
11881 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11882 Jakub Jelinek <jakub@redhat.com>
11883 Richard Biener <rguenther@suse.de>
11885 PR tree-optimization/81503
11886 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
11887 folded constant fits in the target type; reorder tests for clarity.
11889 2017-08-29 Martin Liska <mliska@suse.cz>
11891 * passes.def: Include pass_lower_switch.
11892 * stmt.c (dump_case_nodes): Remove and move to
11893 tree-switch-conversion.
11894 (case_values_threshold): Likewise.
11895 (expand_switch_as_decision_tree_p): Likewise.
11896 (emit_case_decision_tree): Likewise.
11897 (expand_case): Likewise.
11898 (balance_case_nodes): Likewise.
11899 (node_has_low_bound): Likewise.
11900 (node_has_high_bound): Likewise.
11901 (node_is_bounded): Likewise.
11902 (emit_case_nodes): Likewise.
11903 (struct simple_case_node): New struct.
11904 (add_case_node): Remove.
11905 (emit_case_dispatch_table): Use vector instead of case_list.
11906 (reset_out_edges_aux): Remove.
11907 (compute_cases_per_edge): Likewise.
11908 (expand_case): Build list of simple_case_node.
11909 (expand_sjlj_dispatch_table): Use it.
11910 * tree-switch-conversion.c (struct case_node): Moved from
11911 stmt.c and adjusted.
11912 (emit_case_nodes): Likewise.
11913 (node_has_low_bound): Likewise.
11914 (node_has_high_bound): Likewise.
11915 (node_is_bounded): Likewise.
11916 (case_values_threshold): Likewise.
11917 (reset_out_edges_aux): Likewise.
11918 (compute_cases_per_edge): Likewise.
11919 (add_case_node): Likewise.
11920 (dump_case_nodes): Likewise.
11921 (balance_case_nodes): Likewise.
11922 (expand_switch_as_decision_tree_p): Likewise.
11923 (emit_jump): Likewise.
11924 (emit_case_decision_tree): Likewise.
11925 (try_switch_expansion): Likewise.
11926 (do_jump_if_equal): Likewise.
11927 (emit_cmp_and_jump_insns): Likewise.
11928 (fix_phi_operands_for_edge): New function.
11929 (record_phi_operand_mapping): Likewise.
11930 (class pass_lower_switch): New pass.
11931 (pass_lower_switch::execute): New function.
11932 (make_pass_lower_switch): Likewise.
11933 (conditional_probability):
11934 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
11935 * tree-pass.h: Add make_pass_lower_switch.
11937 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11940 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
11943 2017-08-29 Richard Biener <rguenther@suse.de>
11945 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
11946 we do not add a DW_AT_inline attribute twice.
11947 (gen_subprogram_die): Remove code setting DW_AT_inline on
11948 DECL_ABSTRACT_P nodes.
11950 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
11952 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
11953 calls to internal functions.
11954 (gimplify_modify_expr): Likewise.
11955 * tree-call-cdce.c (use_internal_fn): Likewise.
11956 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
11957 (convert_to_divmod): Set the nothrow flag.
11958 * tree-if-conv.c (predicate_mem_writes): Likewise.
11959 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
11960 (vectorizable_call): Likewise.
11961 (vectorizable_store): Likewise.
11962 (vectorizable_load): Likewise.
11963 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
11964 (vect_recog_mask_conversion_pattern): Likewise.
11966 2017-08-29 Martin Liska <mliska@suse.cz>
11969 * gcc.c (driver_handle_option): Add new argument.
11970 * opts-common.c (handle_option): Pass
11971 target_option_override_hook.
11972 * opts-global.c (lang_handle_option): Add new option.
11973 (set_default_handlers): Add new argument.
11974 (decode_options): Likewise.
11975 * opts.c (target_handle_option): Likewise.
11976 (common_handle_option): Call target_option_override_hook.
11977 * opts.h (struct cl_option_handler_func): Add hook for
11978 target option override.
11979 (struct cl_option_handlers): Likewise.
11980 (set_default_handlers): Add new argument.
11981 (decode_options): Likewise.
11982 (common_handle_option): Likewise.
11983 (target_handle_option): Likewise.
11984 * toplev.c (toplev::main): Pass targetm.target_option.override
11987 2017-08-29 Richard Biener <rguenther@suse.de>
11988 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
11990 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
11991 life to the active subtree.
11993 2017-08-28 Jeff Law <law@redhat.com>
11995 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
11996 derive_equivalences.
11997 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
11998 Code moved into....
11999 (edge_info::derive_equivalences): New private member function
12001 * tree-ssa-dom.c (class edge_info): Changed from a struct
12002 to a class. Add ctor/dtor, methods and data members.
12003 (edge_info::edge_info): Renamed from allocate_edge_info.
12004 Initialize additional members.
12005 (edge_info::~edge_info): New.
12006 (free_dom_edge_info): Delete the edge info.
12007 (record_edge_info): Use new class & associated member functions.
12008 Tighten forms for testing for edge equivalences.
12009 (record_temporary_equivalences): Iterate over the simple
12010 equivalences rather than assuming there's only one per edge.
12011 (cprop_into_successor_phis): Iterate over the simple
12012 equivalences rather than assuming there's only one per edge.
12013 (optimize_stmt): Use operand_equal_p rather than pointer
12014 equality for mini-DSE code.
12016 2017-08-28 Nathan Sidwell <nathan@acm.org>
12018 * gcc.c (execute): Fold SIGPIPE handling into switch
12019 statement. Adjust internal error message.
12021 2017-08-28 Richard Biener <rguenther@suse.de>
12024 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
12025 Do nothing for removed DIEs.
12027 2017-08-28 Richard Biener <rguenther@suse.de>
12029 PR tree-optimization/81977
12030 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
12033 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
12036 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
12037 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
12038 using targetm.gen_mem_thread_fence.
12040 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
12043 * config/i386/i386.md (*<btsc><mode>): Change operand 2
12044 predicate to register_operand. Reorder operands.
12045 (*btr<mode>): Ditto.
12046 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
12047 (*btr<mode>_mask): Ditto.
12049 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
12051 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
12052 * config/rs6000/xmmintrin.h: New file.
12053 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
12055 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12058 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
12059 parameter and_insn and return it.
12060 (recombine_lvx_pattern): Insert a copy to ensure availability of
12061 the base register of the copied masking operation at the point of
12062 the instruction replacement.
12063 (recombine_stvx_pattern): Likewise.
12065 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
12067 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
12068 undocumented switches.
12069 (-mpower9-dform-vector): Likewise.
12070 (-mpower9-dform): Likewise.
12071 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
12072 comments to delete references to -mpower9-dform* switches.
12073 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
12074 Delete reference to -mpower9-dform* switches, test for
12075 -mpower9-vector instead.
12076 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
12077 (OTHER_P9_VECTOR_MASKS): Likewise.
12078 (POWERPC_MASKS): Likewise.
12079 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
12080 tests against -mpower9-dform* to -mpower9-vector. Delete code
12081 that checked for -mpower9-dform* consistancy with other options.
12082 Add test for -mpower9-misc to enable other power9 switches.
12083 (rs6000_init_hard_regno_mode_ok): Likewise.
12084 (rs6000_option_override_internal): Likewise.
12085 (rs6000_emit_prologue): Likewise.
12086 (rs6000_emit_epilogue): Likewise.
12087 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
12088 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
12089 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
12091 (emit_fusion_p9_store): Likewise.
12092 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
12093 resetting these macros if the assembler does not support ISA 3.0
12095 (TARGET_P9_DFORM_VECTOR): Likewise.
12096 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
12097 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
12099 2017-08-25 Alan Modra <amodra@gmail.com>
12101 PR rtl-optimization/81747
12102 * cse.c (cse_extended_basic_block): Don't attempt to record
12103 equivalences for degenerate conditional branches that branch
12104 to their fall-through.
12106 2017-08-24 Martin Sebor <msebor@redhat.com>
12108 PR middle-end/81908
12109 * gimple-fold.c (size_must_be_zero_p): New function.
12110 (gimple_fold_builtin_memory_op): Call it.
12112 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
12114 * config/rs6000/mm_malloc.h: New file.
12116 2017-08-24 Bin Cheng <bin.cheng@arm.com>
12118 PR tree-optimization/81913
12119 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
12120 analysis when either IVs in condition can wrap.
12122 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
12124 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
12125 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
12127 2017-08-24 Richard Biener <rguenther@suse.de>
12130 * targhooks.c (default_target_can_inline_p): Properly
12131 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
12132 is present and always compare.
12133 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
12134 infer -mfpmath=sse from TARGET_SSE_P.
12135 (ix86_can_inline_p): Properly use target_option_default_node when
12136 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
12138 2017-08-24 Richard Biener <rguenther@suse.de>
12141 * dwarf2out.c (output_die): Handle flag_generate_offload like
12143 (output_comp_unit): Likewise.
12144 (gen_array_type_die): Likewise.
12145 (dwarf2out_early_finish): Likewise.
12146 (note_variable_value_in_expr): Likewise.
12147 (dwarf2out_finish): Likewise. Adjust assert.
12148 * cgraphunit.c (symbol_table::compile): Move setting of
12149 flag_generate_offload earlier ...
12150 (symbol_table::finalize_compilation_unit): ... here, before
12151 early debug finalization.
12153 2017-08-24 Richard Biener <rguenther@suse.de>
12155 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
12156 and ipa-fnsummary.h.
12157 (ix86_can_inline_p): When ix86_fpmath flags do not match
12158 check whether the callee uses FP math at all.
12160 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
12162 PR middle-end/81931
12163 * tree-ssanames.c (get_nonzero_bits): Use element_precision
12164 instead of TYPE_PRECISION.
12166 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
12167 Alan Hayward <alan.hayward@arm.com>
12168 David Sherwood <david.sherwood@arm.com>
12170 * combine.c (make_extraction): Use subreg_offset_from_lsb.
12172 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
12174 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
12176 (ix86_frame::stack_realign_allocate): New field.
12177 (struct machine_frame_state): Modify comments.
12178 (machine_frame_state::sp_realigned_fp_end): New field.
12179 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
12180 layout calculation.
12181 (sp_valid_at): Add assertion to assure no attempt to access invalid
12182 offset of a realigned stack.
12183 (fp_valid_at): Likewise.
12184 (choose_baseaddr): Modify comments.
12185 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
12186 ix86_expand_prologue.
12187 (ix86_expand_prologue): Modify stack realignment and allocation.
12188 (ix86_expand_epilogue): Modify comments.
12189 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
12190 avx2_runtime, avx512f, and avx512f_runtime.
12192 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
12194 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
12195 (mstackrealign): Do not init to -1.
12196 * config/i386/i386.c (ix86_option_override_internal):
12197 Check opts_set, not opts when setting default value of
12198 opts->x_ix86_force_align_arg_pointer.
12200 2017-08-23 Richard Biener <rguenther@suse.de>
12202 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
12203 lang_hooks.decl_printable_name.
12204 * print-rtl-function.c (print_rtx_function): Likewise.
12205 * tree-pretty-print.c (dump_function_header): Likewise.
12207 2017-08-23 Richard Biener <rguenther@suse.de>
12210 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
12211 -g0 at compile-time.
12213 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12215 PR middle-end/19706
12216 * doc/sourcebuild.texi (Other hardware attributes):
12219 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12221 PR middle-end/19706
12222 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
12223 Fix single-use check.
12225 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
12227 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
12229 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12231 * doc/install.texi: Modify to add more details on running selected
12234 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12236 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
12237 is combined with -mabi=ms.
12238 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
12241 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12243 PR tree-optimization/81488
12244 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
12245 and cached_basis fields.
12246 (MAX_SPREAD): New constant.
12247 (alloc_cand_and_find_basis): Initialize new fields.
12248 (clear_visited): New function.
12249 (create_phi_basis_1): Rename from create_phi_basis, set visited
12250 and cached_basis fields.
12251 (create_phi_basis): New wrapper function.
12252 (phi_add_costs_1): Rename from phi_add_costs, add spread
12253 parameter, set visited field, short-circuit when limits reached.
12254 (phi_add_costs): New wrapper function.
12255 (record_phi_increments_1): Rename from record_phi_increments, set
12257 (record_phi_increments): New wrapper function.
12258 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
12259 (phi_incr_cost): New wrapper function.
12260 (all_phi_incrs_profitable_1): Rename from
12261 all_phi_incrs_profitable, set visited field.
12262 (all_phi_incrs_profitable): New wrapper function.
12264 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12265 Alan Hayward <alan.hayward@arm.com>
12266 David Sherwood <david.sherwood@arm.com>
12268 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
12269 that takes the outer and inner modes.
12270 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
12271 comparison as the canonical test for a paradoxical subreg.
12272 * combine.c (simplify_set): Use paradoxical_subreg_p.
12273 (make_extraction): Likewise.
12274 (force_to_mode): Likewise.
12275 (rtx_equal_for_field_assignment_p): Likewise.
12276 (gen_lowpart_for_combine): Likewise.
12277 (simplify_comparison): Likewise.
12278 * cse.c (equiv_constant): Likewise.
12279 * expmed.c (store_bit_field_1): Likewise.
12280 * final.c (alter_subreg): Likewise.
12281 * fwprop.c (propagate_rtx): Likewise.
12282 (forward_propagate_subreg): Likewise.
12283 * ira-conflicts.c (ira_build_conflicts): Likewise.
12284 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
12285 * lra-constraints.c (curr_insn_transform): Likewise.
12286 (split_reg): Likewise.
12287 * lra-eliminations.c (move_plus_up): Likewise.
12288 (lra_eliminate_regs_1): Likewise.
12289 * recog.c (general_operand): Likewise.
12290 * ree.c (combine_reaching_defs): Likewise.
12291 * reload.c (push_reload): Likewise.
12292 (find_reloads): Likewise.
12293 * reload1.c (elimination_effects): Likewise.
12294 (compute_reload_subreg_offset): Likewise.
12295 (choose_reload_regs): Likewise.
12296 * rtlanal.c (subreg_lsb_1): Likewise.
12297 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12298 (simplify_subreg): Likewise.
12299 * var-tracking.c (track_loc_p): Likewise.
12300 * emit-rtl.c (byte_lowpart_offset): Likewise.
12301 (paradoxical_subreg_p): Delete out-of-line definition.
12303 2017-08-22 Jeff Law <law@redhat.com>
12305 PR tree-optimization/81741
12306 PR tree-optimization/71947
12307 * tree-ssa-dom.c: Include tree-inline.h.
12308 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
12309 equivalences if one is more expensive to compute than the other.
12310 * tree-ssa-scopedtables.h (class const_or_copies): Make
12311 record_const_or_copy_raw method private.
12312 (class avail_exprs_stack): New method simplify_binary_operation.
12313 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
12314 avail_exprs_stack::simplify_binary_operation as needed.
12315 (avail_exprs_stack::simplify_binary_operation): New function.
12317 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
12319 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
12320 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
12321 (DOT_SYMBOLS): Likewise.
12322 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
12323 (RELOCATABLE_NEEDS_FIXUP): Likewise.
12324 (RS6000_ABI_NAME): Likewise.
12325 (TARGET_CMODEL): Likewise.
12326 (TOC_SECTION_ASM_OP): Likewise.
12327 (SET_CMODEL): New macro.
12328 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
12330 2017-08-22 Richard Biener <rguenther@suse.de>
12332 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
12333 to allow for free-lang-data replacements similar to verify_type_variant.
12335 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12336 Alan Hayward <alan.hayward@arm.com>
12337 David Sherwood <david.sherwood@arm.com>
12339 * config/aarch64/aarch64.md (casesi): Use DImode rather than
12340 VOIDmode for the LABEL_REF.
12342 2017-08-22 Richard Biener <rguenther@suse.de>
12344 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
12346 2017-08-22 Richard Biener <rguenther@suse.de>
12348 * common.opt (feliminate-dwarf2-dups): Ignore.
12349 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
12350 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
12351 same_die_p_wrap, compute_section_prefix,
12352 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
12353 (comdat_symbol_id, comdat_symbol_number): Likewise.
12354 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
12356 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
12357 (output_die): Mark unreachable path unreachable.
12358 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
12359 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
12360 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
12361 (dwarf2out_early_finish): Likewise.
12363 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
12365 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
12367 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
12370 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
12371 not VAR_P. Filter attribute warnings with OPT_Wattributes.
12372 (avr_attribute_table) <io, io_low, address>: Initialize
12373 .decl_required with true.
12375 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
12377 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
12378 undocumented debugging options.
12379 (-mvsx-scalar-double): Likewise.
12380 (-mallow-df-permute): Likewise.
12381 (-mvectorize-builtins): Likewise.
12382 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12383 (rs6000_builtin_vectorized_function): Likewise.
12384 (rs6000_builtin_md_vectorized_function): Likewise.
12385 (rs6000_opt_vars): Likewise.
12387 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
12390 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
12391 (*btrq_imm): Rename from *btrq.
12392 (*btcq_imm): Rename from *btcq.
12393 (btsc): New code attribute.
12394 (*<btsc><mode>): New insn pattern.
12395 (*btr<mode>): Ditto.
12396 (*<btsc><mode>_mask): New insn_and_split pattern.
12397 (*btr<mode>_mask): Ditto.
12399 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12400 Alan Hayward <alan.hayward@arm.com>
12401 David Sherwood <david.sherwood@arm.com>
12403 * function.c (pad_below): Simplify padding calculation.
12405 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12406 Alan Hayward <alan.hayward@arm.com>
12407 David Sherwood <david.sherwood@arm.com>
12409 * target.def (function_prologue): Remove frame size argument.
12410 (function_epilogue): Likewise.
12411 * doc/tm.texi: Regenerate.
12412 * final.c (final_start_function): Update call to function_prologue.
12413 (final_end_function): Update call to function_epilogue.
12414 (default_function_pro_epilogue): Remove frame size argument.
12415 * output.h (default_function_pro_epilogue): Likewise.
12416 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
12417 (arm_output_function_prologue): Likewise.
12418 * config/frv/frv.c (frv_function_prologue): Likewise.
12419 (frv_function_epilogue): Likewise.
12420 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
12421 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
12422 (ia64_output_function_epilogue): Likewise.
12423 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
12424 (m32r_output_function_epilogue): Likewise.
12425 * config/microblaze/microblaze.c (microblaze_function_prologue)
12426 (microblaze_function_epilogue): Likewise.
12427 * config/mips/mips.c (mips_output_function_prologue): Likewise.
12428 (mips_output_function_epilogue): Likewise.
12429 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
12430 (mmix_target_asm_function_epilogue): Likewise.
12431 * config/msp430/msp430.c (msp430_start_function): Likewise.
12432 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
12433 (nds32_asm_function_epilogue): Likewise.
12434 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
12435 * config/pa/pa.c (pa_output_function_prologue): Likewise.
12436 (pa_output_function_epilogue): Likewise.
12437 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
12438 (rs6000_output_function_epilogue): Likewise.
12439 * config/rl78/rl78.c (rl78_start_function): Likewise.
12440 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
12441 (rs6000_output_function_epilogue): Likewise.
12442 * config/rx/rx.c (rx_output_function_prologue): Likewise.
12443 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
12444 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
12445 (sparc_asm_function_epilogue): Likewise.
12447 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12449 * tree.h (type_has_mode_precision_p): New function.
12450 * convert.c (convert_to_integer_1): Use it.
12451 * expr.c (expand_expr_real_2): Likewise.
12452 (expand_expr_real_1): Likewise.
12453 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
12454 * match.pd: Likewise.
12455 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
12456 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
12457 * tree-tailcall.c (process_assignment): Likewise.
12458 * tree-vect-loop.c (vectorizable_reduction): Likewise.
12459 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
12460 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
12461 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12462 (vectorizable_assignment): Likewise.
12463 (vectorizable_shift): Likewise.
12464 (vectorizable_operation): Likewise.
12465 * tree-vrp.c (register_edge_assert_for_2): Likewise.
12467 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
12469 * match.pd: Add pow (C, x) simplification.
12471 2017-08-21 Richard Biener <rguenther@suse.de>
12473 PR tree-optimization/81900
12474 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
12475 for blocks with abnormal predecessors.
12476 (compute_antic): Do not set visited flag prematurely.
12478 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
12481 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
12483 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12485 * stor-layout.h (vector_type_mode): Move to...
12486 * tree.h (vector_type_mode): ...here.
12487 * stor-layout.c (vector_type_mode): Move to...
12488 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
12490 2017-08-21 Richard Biener <rguenther@suse.de>
12492 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
12493 register_external_die hooks.
12494 (debug_false_tree_charstarstar_uhwistar): Declare.
12495 (debug_nothing_tree_charstar_uhwi): Likewise.
12496 * debug.c (do_nothing_debug_hooks): Adjust.
12497 (debug_false_tree_charstarstar_uhwistar): New do nothing.
12498 (debug_nothing_tree_charstar_uhwi): Likewise.
12499 * dbxout.c (dbx_debug_hooks): Adjust.
12500 (xcoff_debug_hooks): Likewise.
12501 * sdbout.c (sdb_debug_hooks): Likewise.
12502 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
12503 * dwarf2out.c (macinfo_label_base): New global.
12504 (dwarf2out_register_external_die): New function for the
12505 register_external_die hook.
12506 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
12507 (dwarf2_debug_hooks): Use them.
12508 (dwarf2_lineno_debug_hooks): Adjust.
12509 (struct die_struct): Add with_offset flag.
12510 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
12511 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
12512 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
12513 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
12514 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
12515 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
12516 defining section names for the early LTO debug variants.
12517 (reset_indirect_string): New helper.
12518 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
12519 (print_dw_val): Add support for offsetted symbol references.
12520 (get_ultimate_context): Split out from is_cxx.
12521 (is_cxx): Use get_ultimate_context.
12522 (is_fortran): Add decl overload.
12523 (compute_comp_unit_symbol): Split out worker from
12524 compute_section_prefix.
12525 (compute_section_prefix): Call compute_comp_unit_symbol and
12526 set comdat_type_p here.
12527 (output_die): Skip DIE symbol output for the LTO added one.
12528 Handle DIE symbol references with offset.
12529 (output_comp_unit): Guard section name mangling properly.
12530 For LTO debug sections emit a symbol at the section beginning
12531 which we use to refer to its DIEs.
12532 (add_abstract_origin_attribute): For DIEs registered via
12533 dwarf2out_register_external_die directly refer to the early
12534 DIE rather than indirectly through the shadow one we created.
12535 Remove obsolete call to dwarf2out_abstract_function for
12536 non-function/block origins.
12537 (gen_array_type_die): When generating early LTO debug do
12538 not emit DW_AT_string_length.
12539 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
12540 late when in LTO. As suggested place a gcc_unreachable for
12541 the DECL_ABSTRACT_P case.
12542 (gen_subprogram_die): Avoid another specification DIE
12543 for early built declarations/definitions for the late LTO case.
12544 (gen_variable_die): Add type references for late duplicated VLA dies
12546 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
12547 we have the abstract instance already.
12548 (process_scope_var): Adjust decl DIE contexts in LTO which
12549 first puts them in limbo.
12550 (gen_decl_die): Do not generate type DIEs late apart from
12551 types for VLAs or for decls we do not yet have a DIE. Do not
12552 call dwarf2out_abstract_function late.
12553 (dwarf2out_early_global_decl): Make sure to create DIEs
12554 for abstract instances of a decl first.
12555 (dwarf2out_late_global_decl): Adjust comment.
12556 (output_macinfo_op): With multiple macro sections use
12557 macinfo_label_base to distinguish labels.
12558 (output_macinfo): Likewise. Update macinfo_label_base.
12559 Pass in the line info label.
12560 (note_variable_value_in_expr): When generating LTO resolve
12561 all variable values here by generating DIEs as needed.
12562 (init_sections_and_labels): Add early LTO debug flag parameter
12563 and generate different sections and names if set. Add generation
12564 counter for the labels so we can have multiple of them.
12565 (reset_dies): Helper to allow DIEs to be output multiple times.
12566 (dwarf2out_finish): When outputting DIEs to the fat part of an
12567 LTO object first reset DIEs.
12568 (dwarf2out_early_finish): Output early DIEs when generating LTO.
12569 (modified_type_die): Check for decl_ultimate_origin being self
12571 (gen_type_die_with_usage): Likewise.
12572 (gen_typedef_die): Allow decl_ultimate_origin being self.
12573 (set_decl_abstract_flags): Remove.
12574 (set_block_abstract_flags): Likewise.
12575 (dwarf2out_abstract_function): Treat the early generated DIEs
12576 as the abstract copy and only add DW_AT_inline and
12577 DW_AT_artificial here and call set_decl_origin_self.
12578 If the DIE has an abstract origin don't do anything.
12579 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
12580 if we have none yet (Go fails to build one, PR78628).
12581 (variably_modified_type_p): Prevent endless recursion for Ada
12582 cyclic pointer types.
12583 * lto-streamer-in.c: Include debug.h.
12584 (dref_queue): New global.
12585 (lto_read_tree_1): Stream in DIE references.
12586 (lto_input_tree): Register DIE references.
12587 (input_function): Stream DECL_DEBUG_ARGS.
12588 * lto-streamer-out.c: Include debug.h.
12589 (lto_write_tree_1): Output DIE references.
12590 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
12591 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
12592 (output_function): Stream DECL_DEBUG_ARGS.
12593 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
12594 Stream DECL_ABSTRACT_ORIGIN.
12595 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
12596 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
12597 DECL_CONTEXT for file-scope decls.
12598 * lto-streamer.h (struct dref_entry): Declare.
12599 (dref_queue): Likewise.
12600 * cfgexpand.c (pass_expand::execute): Do not call the
12601 outlining_inline_function hook here.
12602 * lto-wrapper.c (debug_obj): New global.
12603 (tool_cleanup): Unlink it if required.
12604 (debug_objcopy): New function.
12605 (run_gcc): Handle early debug sections in the IL files by
12606 extracting them to separate files, partially linkin them and
12607 feeding the result back as result to the linker.
12608 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
12609 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
12610 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
12611 sections into a separate segment.
12612 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
12614 (darwin_asm_dwarf_section): Likewise.
12615 (darwin_asm_output_dwarf_offset): Likewise.
12616 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
12618 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12619 Alan Hayward <alan.hayward@arm.com>
12620 David Sherwood <david.sherwood@arm.com>
12622 * read-md.h (md_reader::record_potential_iterator_use): Replace
12623 pointer argument with an rtx and an index.
12624 * read-rtl.c (iterator_group::apply_iterator): Likewise.
12625 (apply_mode_iterator): Likewise.
12626 (apply_code_iterator): Likewise.
12627 (apply_int_iterator): Likewise.
12628 (apply_subst_iterator): Likewise.
12629 (record_iterator_use): Likewise.
12630 (record_attribute_use): Likewise.
12631 (md_reader::record_potential_iterator_use): Likewise. Update calls
12632 to record_iterator_use and apply_iterator.
12633 (iterator_use): Replace ptr with x and index.
12634 (attribute_use): Likewise.
12635 (apply_attribute_uses): Update calls to apply_iterator.
12636 (apply_iterators): Likewise. Update initialization of iterator_use.
12637 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
12638 and record_potential_iterator_use.
12639 (rtx_reader::read_rtx_operand): Likewise.
12641 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12642 Alan Hayward <alan.hayward@arm.com>
12643 David Sherwood <david.sherwood@arm.com>
12645 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
12648 2017-08-21 Richard Biener <rguenther@suse.de>
12650 PR middle-end/81884
12651 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
12652 at struct end conservatively when comparing common bases.
12654 2017-08-21 Richard Biener <rguenther@suse.de>
12656 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
12657 (mem_ref_in_stmt): Remove.
12658 (determine_max_movement): Use ref index to get at the reference.
12659 (invariantness_dom_walker::before_dom_children): Deal with
12660 lim data already initialized.
12661 (gather_mem_refs_stmt): Initialize lim data and record ref index.
12663 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12665 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
12666 (TARGET_ISA_ROUND): Ditto.
12667 (TARGET_ROUND): Ditto.
12668 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
12669 * config/i386/i386.md: Ditto.
12670 * config/i386/sse.md: Ditto.
12671 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
12672 with OPTION_MASK_ISA_SSE4_1.
12674 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12677 * doc/extend.texi (x86 Built-in Functions): Correct the name of
12678 __builtin_ia32_lzcnt_u16.
12680 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
12683 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
12684 (rs6000_set_current_function): Rewrite function to use it.
12686 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
12689 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
12690 and TYPE_WARN_IF_NOT_ALIGN.
12691 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
12692 (handle_warn_if_not_align): New.
12693 (place_union_field): Call handle_warn_if_not_align.
12694 (place_field): Call handle_warn_if_not_align.
12695 Copy TYPE_WARN_IF_NOT_ALIGN.
12696 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
12697 (layout_type): Likewise.
12698 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
12700 (tree_decl_common): Add warn_if_not_align.
12701 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
12702 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
12703 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
12704 (DECL_WARN_IF_NOT_ALIGN): Likewise.
12705 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
12706 * doc/extend.texi: Document warn_if_not_aligned attribute.
12707 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
12709 2017-08-17 Martin Liska <mliska@suse.cz>
12712 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
12713 (get_data_dependence): Use it as pointer type.
12714 (distribute_loop): Likewise.
12716 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12718 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
12719 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
12720 (p8_vmrgow_<mode>_direct): New define_insn.
12721 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
12722 handle endianness for vmrgew and vmrgow permute patterns.
12724 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
12726 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
12727 * config/rs6000/rs6000-cpus.def: Remove comment.
12728 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
12729 (POWERPC_MASKS): Likewise.
12730 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
12731 use of TARGET_VSX_TIMODE.
12732 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
12733 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
12734 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
12735 (rs6000_option_override_internal): Remove dead code.
12736 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
12737 (rs6000_legitimize_reload_address): Likewise.
12738 (rs6000_legitimate_address_p): Likewise.
12739 (rs6000_opt_masks): Delete "vsx-timode".
12740 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
12741 from function comment.
12742 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
12743 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
12744 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
12746 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
12747 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
12748 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
12749 (VSX_TI): Likewise.
12751 (define_peephole2): Likewise.
12753 2017-08-17 Martin Sebor <msebor@redhat.com>
12756 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
12757 past the end of an array.
12758 (test_pp_format): Add test cases.
12760 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
12762 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
12763 missing ECF_NOTHROW flags.
12765 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
12768 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
12769 operands residing in integer registers.
12770 (*vsx_le_perm_load_<mode>): Likewise.
12771 (*vsx_le_perm_store_<mode>): Likewise.
12772 (define_peephole2): Add peepholes to optimize the above.
12774 2017-08-17 Marek Polacek <polacek@redhat.com>
12776 PR middle-end/81814
12777 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
12778 to mimic what shorten_compare did. Change the return type to bool.
12779 (fold_cond_expr_with_comparison): Update call to
12780 operand_equal_for_comparison_p.
12781 (fold_ternary_loc): Likewise.
12783 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
12785 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
12787 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
12789 2017-08-17 Richard Biener <rguenther@suse.de>
12791 * tree-ssa-structalias.c (solve_graph): When propagating
12792 to successors update the graphs succ edges and avoid duplicate work.
12794 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
12797 * config/i386/i386.c (ix86_option_override_internal): Save target
12798 specific options after ix86_stack_protector_guard_reg was changed.
12800 2017-08-17 Richard Biener <rguenther@suse.de>
12802 PR tree-optimization/81827
12803 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
12804 (new_var_info): Initialize it conservatively.
12805 (get_call_vi): Mark register vars.
12806 (new_scalar_tmp_constraint_exp): Likewise.
12807 (handle_rhs_call): Likewise.
12808 (handle_const_call): Likewise.
12809 (create_function_info_for): Likewise.
12810 (solve_constraints): Sort varinfos to separate register from
12811 non-register vars to pack points-to solution bitmaps during
12814 2017-08-17 Marek Polacek <polacek@redhat.com>
12816 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
12818 2017-08-17 Richard Biener <rguenther@suse.de>
12820 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
12821 to true when overflow is undefined and we saturated the result.
12823 2017-08-17 Alan Modra <amodra@gmail.com>
12826 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
12827 Don't use store multiple if only one reg needs saving.
12828 (interesting_frame_related_regno): New function.
12829 (rs6000_frame_related): Don't emit frame info for regs that
12831 (rs6000_emit_epilogue): Likewise.
12833 2017-08-16 Nathan Sidwell <nathan@acm.org>
12835 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
12836 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
12837 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
12838 * tree.c (free_lang_data_in_type): Use else-if chain. Always
12839 clear TYPE_LANG_1. Remove obsolete member-function stripping.
12840 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
12841 (verify_type): Adjust for TYPE_BINFO move.
12842 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
12843 process TYPE_BINFO directly.
12844 (hash_tree): Likewise.
12845 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
12847 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
12850 2017-08-16 David Malcolm <dmalcolm@redhat.com>
12852 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
12854 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
12857 * config/i386/i386.md (*anddi_1_btr): Change predicates of
12858 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
12859 Add ix86_binary_operator_ok to insn constraint.
12860 (*iordi_1_bts): Ditto.
12861 (*xordi_1_btc): Ditto.
12862 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
12863 Update corresponding peephole2 pattern.
12867 2017-08-16 Bin Cheng <bin.cheng@arm.com>
12869 PR tree-optimization/81832
12870 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
12871 copy loop header which has IFN_LOOP_DIST_ALIAS call.
12873 2017-08-16 Marek Polacek <polacek@redhat.com>
12876 * fold-const.c (fold_indirect_ref_1): Restore original behavior
12877 regarding size_zero_node.
12879 2017-08-16 Martin Liska <mliska@suse.cz>
12882 * config.gcc: Respect previously set extra_objs in case
12885 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
12887 PR tree-optimization/81835
12888 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
12889 the phi SSA_NAME. Check that the condition in a COND_EXPR does
12890 not depend on the phi.
12892 2017-08-16 Alan Modra <amodra@gmail.com>
12894 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
12897 2017-08-16 Alan Modra <amodra@gmail.com>
12899 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
12900 (save_reg_p): ..into this. Update all callers.
12901 (first_reg_to_save): Simplify.
12903 2017-08-16 Alan Modra <amodra@gmail.com>
12905 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
12908 2017-08-15 Joseph Myers <joseph@codesourcery.com>
12912 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
12913 constant count if that count is less than 32.
12915 2017-08-15 Nathan Sidwell <nathan@acm.org>
12917 * gcc.c (execute): Emit friendlier message if inferior is killed
12918 by an external cause.
12920 2017-08-15 Richard Biener <rguenther@suse.de>
12922 PR tree-optimization/81790
12923 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
12924 CONSTRUCTORs from simplifying and VN.
12926 2017-08-14 Martin Sebor <msebor@redhat.com>
12928 * builtin-attrs.def: Add comments.
12930 2017-08-14 Martin Sebor <msebor@redhat.com>
12933 * doc/extend.texi (attribute nonstring): Document new attribute.
12935 2017-08-14 Martin Sebor <msebor@redhat.com>
12938 * tree-diagnostic.c (default_tree_printer): Handle %G.
12939 * gimple-pretty-print.h (percent_G_format): Declare new function.
12940 * gimple-pretty-print.c (percent_G_format): Define.
12941 * tree-pretty-print.c (percent_K_format): Add argument.
12943 2017-08-14 Martin Sebor <msebor@redhat.com>
12945 PR translation/79998
12946 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
12947 Remove a stray space.
12949 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
12952 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
12953 (*iordi_1_bts): Ditto.
12954 (*xordi_1_btc): Ditto.
12956 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12959 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
12960 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12962 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
12963 quoted strings, and make more translator-friendly.
12964 (darwin_rs6000_override_options): Likewise.
12965 (rs6000_option_override_internal): Likewise.
12966 (rs6000_return_in_memory): Fix overlong line.
12967 (init_cmulative_args): Use quoted strings, and make more
12968 translator-friendly.
12969 (rs6000_pass_by_reference): Fix overlong line.
12970 (def_builtin): Use quoted strings.
12971 (altivec_expand_predicate_builtin): Use quoted strings, and make
12972 more translator-friendly.
12973 (htm_expand_builtin): Use quoted strings.
12974 (cpu_expand_builtin): Use quoted strings, and make more
12975 translator-friendly.
12976 (altivec_expand_builtin): Likewise.
12977 (paired_expand_predicate_builtin): Likewise.
12978 (rs6000_invalid_builtin): Likewise.
12979 (builtin_function_type): Use quoted strings.
12980 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
12981 more translator-friendly.
12982 (rs6000_trampoline_init): Likewise.
12983 (rs6000_handle_altivec_attribute): Likewise.
12984 (rs6000_inner_target_options): Use quoted strings.
12985 (rs6000_disable_incompatible_switches): Likewise.
12986 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
12987 strings, and make more translator-friendly.
12988 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
12990 2017-08-14 Bin Cheng <bin.cheng@arm.com>
12992 PR tree-optimization/81799
12993 * tree-loop-distribution.c (version_loop_by_alias_check): Force
12994 cond_expr to simple gimple operand.
12996 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
12998 PR middle-end/46932
12999 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
13001 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
13005 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
13006 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
13007 TARGET_GASISR_PROLOGUES.
13008 * config/avr/avr.c (avr_option_override): Same.
13009 (avr_pass_pre_proep::execute): Same.
13011 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
13014 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
13015 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
13017 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
13019 * config/i386/i386.md (*load_tp_<mode>): Redefine as
13020 define_insn_and_split. Split to a memory load from 0 in
13021 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
13022 using PTR mode iterator.
13023 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
13024 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
13025 (*add_tp_<mode>): Redefine as define_insn_and_split.
13026 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
13027 address space. Merge with *add_tp_x32 using PTR mode iterator.
13028 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
13029 Split to an add with a memory load from 0 in
13030 DEFAULT_TLS_SEG_REG address space.
13032 2017-08-12 Andrew Pinski <apinski@cavium.com>
13034 * config/aarch64/aarch64-option-extensions.def (rdma):
13035 Fix feature string to what Linux prints out in /proc/cpuinfo.
13037 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
13040 * dwarf2out.c (modified_type_die): For C typedef types that have
13041 an ultimate origin, process the ultimate origin instead of the
13043 (gen_typedef_die): Assert that input DECLs have no ultimate
13045 (gen_type_die_with_usage): For typedef variants that have an
13046 ultimate origin, just call gen_decl_die on the original DECL.
13047 (process_scope_var): Avoid creating DIEs for local typedefs and
13048 concrete static variables.
13050 2017-08-12 Alan Modra <amodra@gmail.com>
13054 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
13055 match gnu-user.h startfile.
13056 (ENDFILE_LINUX_SPEC): Similarly.
13058 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
13061 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
13063 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
13065 2017-08-11 Tamar Christina <tamar.christina@arm.com>
13066 * config/aarch64/aarch64.md (mov<mode>): Change.
13067 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
13068 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
13069 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
13071 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
13073 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
13074 for storage order barriers.
13076 2017-08-11 Martin Liska <mliska@suse.cz>
13079 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
13080 variables of void type.
13082 2017-08-11 Martin Liska <mliska@suse.cz>
13084 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
13085 TARGET_SUPPORTS_ALIASES.
13086 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
13087 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
13088 (optimize_weakref): Likewise.
13089 * symtab.c (symtab_node::noninterposable_alias): Likewise.
13090 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
13091 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
13093 2017-08-11 Martin Liska <mliska@suse.cz>
13096 * config/i386/i386.c (make_resolver_func): Do complete
13097 refactoring of the function.
13099 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13102 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
13103 * config/i386/i386.c (ix86_stack_protect_guard): Use
13104 ix86_stack_protect_guard_symbol_str to generate varible declaration.
13105 * doc/invoke.texi (x86 Options): Document
13106 -mstack-protector-guard-symbol= option.
13108 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13110 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
13111 * config/i386/i386.c (ix86_split_stack_guard): New function.
13112 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
13113 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
13114 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
13115 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
13116 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
13117 (split_stack_space_check): Call ix86_split_stack_guard.
13119 2017-08-10 Martin Sebor <msebor@redhat.com>
13121 * print-tree.c (print_node): Print location using the established
13123 Replace spaces with colons.
13124 (debug_raw, debug): Ditto.
13126 2017-08-10 Martin Sebor <msebor@redhat.com>
13129 * pretty-print.c (pp_format): Correct the handling of %s precision.
13131 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
13134 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13136 (ix86_finalize_stack_frame_flags): This. Also clear
13137 frame_pointer_needed if -fno-omit-frame-pointer is used without
13139 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13140 with ix86_finalize_stack_frame_flags.
13141 (ix86_expand_epilogue): Likewise.
13142 (ix86_expand_split_stack_prologue): Likewise.
13143 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
13145 2017-08-10 Martin Liska <mliska@suse.cz>
13148 * c-attribs.c (handle_target_attribute):
13149 Report warning for an empty string argument of target attribute.
13151 2017-08-09 Jakub Jelinek <jakub@redhat.com>
13154 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
13156 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
13157 or DECL_NONLOCAL labels.
13158 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
13159 or DECL_NONLOCAL labels here.
13161 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
13163 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
13164 to indicate when early gimple folding has been disabled.
13165 (rs6000_gimple_fold_builtin): Add debug content.
13166 (rs6000_invalid_builtin): Fix whitespace.
13167 (rs6000_expand_builtin): Fix whitespace.
13168 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
13170 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
13173 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
13174 SAVE_MULTIPLE if not all the registers that saves, should be saved.
13176 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
13178 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
13179 (qdf24xx): Likewise.
13180 * config/aarch64/aarch64.md: Include falkor.md.
13181 * config/aarch64/falkor.md: New.
13183 2017-08-09 Marek Polacek <polacek@redhat.com>
13186 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
13187 * diagnostic.c (emit_diagnostic): Add a comment.
13188 (emit_diagnostic_valist): New function.
13190 2017-08-09 Marek Polacek <polacek@redhat.com>
13193 * input.c (make_location): New overload.
13194 * input.h (make_location): Declare.
13196 2017-08-08 Alan Modra <amodra@gmail.com>
13197 H.J. Lu <hongjiu.lu@intel.com>
13200 * gcc.c (NO_PIE_SPEC): Delete.
13201 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
13203 (LINK_PIE_SPEC): ..to here.
13204 (LINK_COMMAND_SPEC): Support -no-pie.
13205 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
13206 chain of crtbegin*.o selection, update for PIE_SPEC changes and
13208 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
13209 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
13210 (ENDFILE_CRTEND_SPEC): Similarly.
13212 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
13215 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
13216 (mstack-protector-guard-offset=): Ditto.
13217 * config/i386/i386.c (ix86_option_override): Handle
13218 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
13220 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
13221 ix86_stack_protect_guard_offset variables.
13222 (TARGET_STACK_PROTECT_GUARD): Always define.
13223 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
13224 and -mstack-protector-guard-offset= options.
13226 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13228 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
13229 boundary case for the last candidate.
13231 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13233 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
13234 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
13236 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13238 PR middle-end/19706
13239 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
13240 * config/aarch64/aarch64-builtins.c
13241 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
13242 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
13243 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
13245 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13246 Andrew Pinski <pinskia@gmail.com>
13248 PR middle-end/19706
13249 * internal-fn.def (XORSIGN): New.
13250 * optabs.def (xorsign_optab): New.
13251 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
13252 (convert_expand_mult_copysign): New.
13253 (pass_optimize_widening_mul::execute): Call
13254 convert_expand_mult_copysign.
13256 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13258 PR tree-optimization/81354
13259 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
13260 Insert on edges rather than explicitly creating landing pads.
13261 (analyze_candidates_and_replace): Commit edge inserts.
13263 2017-08-08 Richard Biener <rguenther@suse.de>
13265 PR middle-end/81719
13266 * tree-ssa-loop-niter.c: Include tree-dfa.h.
13267 (expand_simple_operations): Also look through ADDR_EXPRs with
13268 MEM_REF bases treating them as POINTER_PLUS_EXPR.
13270 2017-08-08 Richard Biener <rguenther@suse.de>
13272 PR tree-optimization/81723
13273 * tree-vect-slp.c (struct bst_traits): New hash traits.
13274 (bst_fail): New global.
13275 (vect_build_slp_tree_2): New worker, split out from ...
13276 (vect_build_slp_tree): ... this now wrapping it with using
13277 bst_fail set to cache SLP tree build fails. Properly handle
13279 (vect_analyze_slp_instance): Allocate and free bst_fail.
13281 2017-08-08 Martin Liska <mliska@suse.cz>
13284 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
13285 LABEL_DECLs that can be from a different function.
13287 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13289 PR tree-optimization/81744
13290 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
13291 loop's number of iterations.
13293 2017-08-08 Martin Liska <mliska@suse.cz>
13295 * asan.c: Include header files.
13296 * attribs.c (build_decl_attribute_variant): New function moved
13298 (build_type_attribute_qual_variant): Likewise.
13299 (cmp_attrib_identifiers): Likewise.
13300 (simple_cst_list_equal): Likewise.
13301 (omp_declare_simd_clauses_equal): Likewise.
13302 (attribute_value_equal): Likewise.
13303 (comp_type_attributes): Likewise.
13304 (build_type_attribute_variant): Likewise.
13305 (lookup_ident_attribute): Likewise.
13306 (remove_attribute): Likewise.
13307 (merge_attributes): Likewise.
13308 (merge_type_attributes): Likewise.
13309 (merge_decl_attributes): Likewise.
13310 (merge_dllimport_decl_attributes): Likewise.
13311 (handle_dll_attribute): Likewise.
13312 (attribute_list_equal): Likewise.
13313 (attribute_list_contained): Likewise.
13314 * attribs.h (lookup_attribute): New function moved from tree.[ch].
13315 (lookup_attribute_by_prefix): Likewise.
13316 * bb-reorder.c: Include header files.
13317 * builtins.c: Likewise.
13318 * calls.c: Likewise.
13319 * cfgexpand.c: Likewise.
13320 * cgraph.c: Likewise.
13321 * cgraphunit.c: Likewise.
13322 * convert.c: Likewise.
13323 * dwarf2out.c: Likewise.
13324 * final.c: Likewise.
13325 * fold-const.c: Likewise.
13326 * function.c: Likewise.
13327 * gimple-expr.c: Likewise.
13328 * gimple-fold.c: Likewise.
13329 * gimple-pretty-print.c: Likewise.
13330 * gimple.c: Likewise.
13331 * gimplify.c: Likewise.
13332 * hsa-common.c: Likewise.
13333 * hsa-gen.c: Likewise.
13334 * internal-fn.c: Likewise.
13335 * ipa-chkp.c: Likewise.
13336 * ipa-cp.c: Likewise.
13337 * ipa-devirt.c: Likewise.
13338 * ipa-fnsummary.c: Likewise.
13339 * ipa-inline.c: Likewise.
13340 * ipa-visibility.c: Likewise.
13342 * lto-cgraph.c: Likewise.
13343 * omp-expand.c: Likewise.
13344 * omp-general.c: Likewise.
13345 * omp-low.c: Likewise.
13346 * omp-offload.c: Likewise.
13347 * omp-simd-clone.c: Likewise.
13348 * opts-global.c: Likewise.
13349 * passes.c: Likewise.
13350 * predict.c: Likewise.
13351 * sancov.c: Likewise.
13352 * sanopt.c: Likewise.
13353 * symtab.c: Likewise.
13354 * toplev.c: Likewise.
13355 * trans-mem.c: Likewise.
13356 * tree-chkp.c: Likewise.
13357 * tree-eh.c: Likewise.
13358 * tree-into-ssa.c: Likewise.
13359 * tree-object-size.c: Likewise.
13360 * tree-parloops.c: Likewise.
13361 * tree-profile.c: Likewise.
13362 * tree-ssa-ccp.c: Likewise.
13363 * tree-ssa-live.c: Likewise.
13364 * tree-ssa-loop.c: Likewise.
13365 * tree-ssa-sccvn.c: Likewise.
13366 * tree-ssa-structalias.c: Likewise.
13367 * tree-ssa.c: Likewise.
13368 * tree-streamer-in.c: Likewise.
13369 * tree-vectorizer.c: Likewise.
13370 * tree-vrp.c: Likewise.
13371 * tsan.c: Likewise.
13372 * ubsan.c: Likewise.
13373 * varasm.c: Likewise.
13374 * varpool.c: Likewise.
13375 * tree.c: Remove functions moved to attribs.[ch].
13376 * tree.h: Likewise.
13377 * config/aarch64/aarch64.c: Add attrs.h header file.
13378 * config/alpha/alpha.c: Likewise.
13379 * config/arc/arc.c: Likewise.
13380 * config/arm/arm.c: Likewise.
13381 * config/avr/avr.c: Likewise.
13382 * config/bfin/bfin.c: Likewise.
13383 * config/c6x/c6x.c: Likewise.
13384 * config/cr16/cr16.c: Likewise.
13385 * config/cris/cris.c: Likewise.
13386 * config/darwin.c: Likewise.
13387 * config/epiphany/epiphany.c: Likewise.
13388 * config/fr30/fr30.c: Likewise.
13389 * config/frv/frv.c: Likewise.
13390 * config/ft32/ft32.c: Likewise.
13391 * config/h8300/h8300.c: Likewise.
13392 * config/i386/winnt.c: Likewise.
13393 * config/ia64/ia64.c: Likewise.
13394 * config/iq2000/iq2000.c: Likewise.
13395 * config/lm32/lm32.c: Likewise.
13396 * config/m32c/m32c.c: Likewise.
13397 * config/m32r/m32r.c: Likewise.
13398 * config/m68k/m68k.c: Likewise.
13399 * config/mcore/mcore.c: Likewise.
13400 * config/microblaze/microblaze.c: Likewise.
13401 * config/mips/mips.c: Likewise.
13402 * config/mmix/mmix.c: Likewise.
13403 * config/mn10300/mn10300.c: Likewise.
13404 * config/moxie/moxie.c: Likewise.
13405 * config/msp430/msp430.c: Likewise.
13406 * config/nds32/nds32-isr.c: Likewise.
13407 * config/nds32/nds32.c: Likewise.
13408 * config/nios2/nios2.c: Likewise.
13409 * config/nvptx/nvptx.c: Likewise.
13410 * config/pa/pa.c: Likewise.
13411 * config/pdp11/pdp11.c: Likewise.
13412 * config/powerpcspe/powerpcspe.c: Likewise.
13413 * config/riscv/riscv.c: Likewise.
13414 * config/rl78/rl78.c: Likewise.
13415 * config/rx/rx.c: Likewise.
13416 * config/s390/s390.c: Likewise.
13417 * config/sh/sh.c: Likewise.
13418 * config/sol2.c: Likewise.
13419 * config/sparc/sparc.c: Likewise.
13420 * config/spu/spu.c: Likewise.
13421 * config/stormy16/stormy16.c: Likewise.
13422 * config/tilegx/tilegx.c: Likewise.
13423 * config/tilepro/tilepro.c: Likewise.
13424 * config/v850/v850.c: Likewise.
13425 * config/vax/vax.c: Likewise.
13426 * config/visium/visium.c: Likewise.
13427 * config/xtensa/xtensa.c: Likewise.
13429 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13432 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
13433 constraints since the -mupper-regs-* switches have been
13435 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
13436 into a vector from a double word element that was extracted from
13437 another vector, and eliminate extra XXPERMDI instructions.
13438 (vsx_concat_<mode>_2): Likewise.
13439 (vsx_concat_<mode>_3): Likewise.
13440 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
13441 concat to allow optimizing inserts from previous extracts.
13443 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
13445 * config/i386/i386.c (ix86_stack_protect_guard): Generate
13446 memory reference to a SSP offset in TLS address space.
13447 (ix86_print_operand) <case '@'>: Remove.
13448 (ix86_print_operand_punct_valid_p): Remove '@' code.
13449 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
13450 UNSPEC_SP_TLS_TEST.
13451 (stack_tls_protect_set_<mode>): Remove.
13452 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
13453 (stack_tls_protect_test_<mode>): Remove.
13454 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
13456 2017-08-07 Olivier Hainque <hainque@adacore.com>
13459 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
13461 2017-08-07 Douglas Rupp <rupp@adacore.com>
13463 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
13464 variable was referenced as multidir in command.
13466 2017-08-07 Jakub Jelinek <jakub@redhat.com>
13469 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
13472 2017-08-07 Martin Liska <mliska@suse.cz>
13474 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
13475 * config/rl78/rl78.c: Add include of attribs.h.
13476 * config/sh/sh.c: Likewise.
13477 * config/v850/v850.c: Likewise.
13479 2017-08-07 Tom de Vries <tom@codesourcery.com>
13481 PR middle-end/78266
13482 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
13484 2017-08-07 Martin Liska <mliska@suse.cz>
13486 * config/mips/mips.c: Include attribs.h.
13488 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
13491 * doc/invoke.texi: Document change in behvaior for -Ofast for
13494 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
13496 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
13498 (aarch64_pop_regs): Likewise.
13499 (aarch64_gen_load_pair): Likewise.
13500 (aarch64_save_callee_saves): Likewise.
13501 (aarch64_restore_callee_saves): Likewise.
13503 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13505 * config/i386/i386.c: Revert the last change.
13507 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13510 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13512 (ix86_finalize_stack_frame_flags): This. Also clear
13513 frame_pointer_needed if -fno-omit-frame-pointer is used without
13515 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13516 with ix86_finalize_stack_frame_flags.
13517 (ix86_expand_epilogue): Likewise.
13518 (ix86_expand_split_stack_prologue): Likewise.
13520 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13523 * config/i386/i386.c (get_builtin_code_for_version): Set priority
13524 to P_AES for Westmere.
13526 2017-08-07 Jonathan Yong <10walls@gmail.com>
13528 * config/i386/mingw.opt (fset-stack-executable): Removed.
13529 * config/i386/cygming.opt (fset-stack-executable): Moved
13531 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
13533 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
13535 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
13537 2017-08-07 Marek Polacek <polacek@redhat.com>
13539 PR middle-end/81737
13540 * fold-const.c (fold_indirect_ref_1): Check type_domain.
13542 2017-08-07 Martin Liska <mliska@suse.cz>
13544 * attribs.h (canonicalize_attr_name): New function.
13545 (cmp_attribs): Move from c-format.c and adjusted.
13546 (is_attribute_p): Moved from tree.h.
13547 * tree-inline.c: Add new includes.
13548 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
13549 (private_is_attribute_p): Remove.
13550 (private_lookup_attribute): Likewise.
13551 (private_lookup_attribute_by_prefix): Simplify.
13552 (remove_attribute): Use is_attribute_p.
13553 * tree.h: Remove removed declarations.
13555 2017-08-07 Jakub Jelinek <jakub@redhat.com>
13557 PR middle-end/81698
13558 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
13559 instead of computing it in the function. Formatting fix.
13560 (expand_case): Don't rely on default_edge being the first edge,
13561 clear it if removing it, pass default_edge to
13562 emit_case_dispatch_table.
13563 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
13566 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
13568 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
13569 insn in the function, emit NOP after the insn.
13571 2017-08-06 Tom de Vries <tom@codesourcery.com>
13573 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
13576 2017-08-06 Tom de Vries <tom@codesourcery.com>
13578 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
13581 2017-08-04 Yury Gribov <tetra2005@gmail.com>
13583 PR tree-optimization/57371
13584 * match.pd: New pattern.
13586 2017-08-04 Marek Polacek <polacek@redhat.com>
13588 PR middle-end/81695
13589 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
13590 perform the computation in offset_int.
13592 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13594 PR tree-optimization/81136
13595 * tree-vectorizer.h: Include tree-hash-traits.h.
13596 (vec_base_alignments): New typedef.
13597 (vec_info): Add a base_alignments field.
13598 (vect_record_base_alignments): Declare.
13599 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
13601 (DR_IS_CONDITIONAL_IN_STMT): New macro.
13602 (create_data_ref): Add an is_conditional_in_stmt argument.
13603 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
13604 the is_conditional_in_stmt field.
13605 (data_ref_loc): Add an is_conditional_in_stmt field.
13606 (get_references_in_stmt): Set the is_conditional_in_stmt field.
13607 (find_data_references_in_stmt): Update call to create_data_ref.
13608 (graphite_find_data_references_in_stmt): Likewise.
13609 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
13610 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
13611 (vect_record_base_alignment): New function.
13612 (vect_record_base_alignments): Likewise.
13613 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
13614 for nested statements even if we fail to compute a misalignment.
13615 Use pooled base alignments for unconditional references.
13616 (vect_find_same_alignment_drs): Compare base addresses instead
13618 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
13619 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
13621 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13623 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
13624 Add an explicit name for the enum. Use auto_vec for slp_instances
13625 and grouped_stores.
13626 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
13628 (_bb_vec_info): Add a constructor and destructor.
13629 (vinfo_for_stmt): Return NULL for uids of -1 as well.
13630 (destroy_loop_vec_info): Delete.
13631 (vect_destroy_datarefs): Likewise.
13632 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
13633 (vec_info::vec_info): New function.
13634 (vec_info::~vec_info): Likewise.
13635 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
13636 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
13637 destroy_loop_vec_info.
13638 * tree-vect-loop.c (new_loop_vec_info): Replace with...
13639 (_loop_vec_info::_loop_vec_info): ...this.
13640 (destroy_loop_vec_info): Replace with...
13641 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
13642 the stmt_vec_infos. Leave handling of vec_info information to its
13643 destructor. Remove explicit vector releases.
13644 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
13645 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
13646 * tree-vect-slp.c (new_bb_vec_info): Replace with...
13647 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
13648 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
13649 (destroy_bb_vec_info): Replace with...
13650 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
13651 information to its destructor.
13652 (vect_slp_analyze_bb_1): Use new and delete instead of
13653 new_bb_vec_info and destroy_bb_vec_info.
13654 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
13657 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13659 * tree-data-ref.h (subscript): Add access_fn field.
13660 (data_dependence_relation): Add could_be_independent_p.
13661 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
13662 (same_access_functions): Move to tree-data-ref.c.
13663 * tree-data-ref.c (ref_contains_union_access_p): New function.
13664 (access_fn_component_p): Likewise.
13665 (access_fn_components_comparable_p): Likewise.
13666 (dr_analyze_indices): Add a reference to access_fn_component_p.
13667 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
13669 (constant_access_functions): Likewise.
13670 (add_other_self_distances): Likewise.
13671 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
13672 (initialize_data_dependence_relation): Use XCNEW and remove
13673 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
13674 of access functions that have the same type. Allow the
13675 subsequence to end with different bases in some circumstances.
13676 Record the chosen access functions in SUB_ACCESS_FN.
13677 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
13678 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
13679 (subscript_dependence_tester_1): Likewise dra and drb.
13680 (build_classic_dist_vector): Update calls accordingly.
13681 (subscript_dependence_tester): Likewise.
13682 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
13683 DDR_COULD_BE_INDEPENDENT_P.
13684 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
13685 comp_alias_ddrs instead of may_alias_ddrs.
13686 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
13688 (vect_analyze_data_ref_dependence): Use it if
13689 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
13690 distance vectors if that fails.
13691 (dependence_distance_ge_vf): New function.
13692 (vect_prune_runtime_alias_test_list): Use it. Don't clear
13693 LOOP_VINFO_MAY_ALIAS_DDRS.
13695 2017-08-04 Richard Biener <rguenther@suse.de>
13697 PR middle-end/81705
13698 * fold-const.c (fold_binary_loc): Properly restrict
13699 minus_var0 && minus_var1 case when associating undefined overflow
13702 2017-08-04 Tom de Vries <tom@codesourcery.com>
13704 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
13706 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13708 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13709 Don't start diagnostic messages with a capital letter.
13710 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13712 (rs6000_invalid_builtin): Likewise.
13713 (rs6000_trampoline_init): Likewise.
13715 2017-08-03 Jakub Jelinek <jakub@redhat.com>
13718 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
13719 after setting changeable df flags.
13721 2017-08-03 Richard Biener <rguenther@suse.de>
13723 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
13724 up if the use is in USE - X.
13726 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
13728 * toplev.c (dumpfile.h): New include.
13729 (internal_error_reentered): New static function. Use it...
13730 (internal_error_function): ...here to handle reentered internal_error.
13732 2017-08-03 Richard Biener <rguenther@suse.de>
13734 PR middle-end/81148
13735 * fold-const.c (split_tree): Add minus_var and minus_con
13736 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
13737 here but always use minus_*.
13738 (associate_trees): Assert we never associate with MINUS_EXPR
13739 and NULL first operand. Do not recurse for PLUS_EXPR operands
13740 when associating as MINUS_EXPR either.
13741 (fold_binary_loc): Track minus_var and minus_con.
13743 2017-08-03 Tom de Vries <tom@codesourcery.com>
13746 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
13747 ACCEL_COMPILER, apply finish_options on
13748 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
13750 2017-08-03 Tom de Vries <tom@codesourcery.com>
13753 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
13754 function_entry_patch_area_size > 0.
13756 2017-08-03 Jakub Jelinek <jakub@redhat.com>
13759 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
13760 instead of 10??LU, perform unit multiplication in wide_int,
13761 don't change alloc_object_size_limit if the limit is larger
13764 PR tree-optimization/81655
13765 PR tree-optimization/81588
13766 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
13767 the case when ranges[i].low and high are 1 for unsigned type with
13770 PR middle-end/81052
13771 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
13772 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
13774 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13776 * tree-vrp.h: Add include guard.
13778 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
13781 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
13782 (ud2): New insn pattern.
13783 * config/i386/i386.c (ix86_expand_epilogue):
13784 For naked functions, generate ud2 instead of trap insn.
13786 2017-08-02 Marek Polacek <polacek@redhat.com>
13789 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
13791 2017-08-02 Tom de Vries <tom@codesourcery.com>
13792 Cesar Philippidis <cesar@codesourcery.com>
13794 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
13795 Add missing edge probabilities.
13797 2017-08-02 Tamar Christina <tamar.christina@arm.com>
13799 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
13800 Correct endianness.
13802 2017-08-02 Jakub Jelinek <jakub@redhat.com>
13804 PR middle-end/79499
13805 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
13806 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
13807 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
13808 prologue_seq sequences - if any.
13810 2017-08-02 Richard Biener <rguenther@suse.de>
13812 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
13813 via vectors if supported, integer extracts via punning if supported
13814 or otherwise vector extracts.
13816 2017-08-02 Richard Biener <rguenther@suse.de>
13818 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
13820 (bitmap_insert_into_set): ... this.
13822 2017-08-02 Richard Biener <rguenther@suse.de>
13824 PR tree-optimization/81633
13826 2015-08-17 Alan Hayward <alan.hayward@arm.com>
13828 PR tree-optimization/71752
13829 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
13831 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
13833 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
13834 (machine_function::call_ms2sysv_pad_out): Remove field.
13835 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
13836 (ix86_compute_frame_layout): Likewise.
13838 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
13841 * config/i386/i386.c (ix86_set_func_type): Disallow naked
13842 attribute with interrupt attribute.
13844 2017-08-01 Andrew Pinski <apinski@cavium.com>
13846 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
13847 BIT_INSERT_EXPR's operand 1
13848 to see if the types precision matches.
13850 2017-08-01 Martin Liska <mliska@suse.cz>
13852 PR middle-end/70140
13853 * builtins.c (expand_builtin_memcpy_args): Remove.
13854 (expand_builtin_memcpy): Call newly added function
13855 expand_builtin_memory_copy_args.
13856 (expand_builtin_memcpy_with_bounds): Likewise.
13857 (expand_builtin_mempcpy): Remove last argument.
13858 (expand_builtin_mempcpy_with_bounds): Likewise.
13859 (expand_builtin_memory_copy_args): New function created from
13860 expand_builtin_mempcpy_args with small modifications.
13861 (expand_builtin_mempcpy_args): Remove.
13862 (expand_builtin_stpcpy): Remove unused argument.
13863 (expand_builtin): Likewise.
13864 (expand_builtin_with_bounds): Likewise.
13866 2017-08-01 Martin Liska <mliska@suse.cz>
13869 Make mempcpy more optimal (PR middle-end/70140).
13871 2017-08-01 Jakub Jelinek <jakub@redhat.com>
13874 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
13875 __builtin_vec_cmpne verify both arguments are compatible vectors
13876 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
13877 verify arg1_type is a pointer or array type. For __builtin_vec_st,
13878 move computation of aligned to after checking the argument types.
13882 * config/rs6000/vsx.md (vextract_fp_from_shorth,
13883 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
13886 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
13887 Doug Rupp <rupp@adacore.com>
13888 Olivier Hainque <hainque@adacore.com>
13890 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
13891 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
13893 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
13894 for TARGET_OS_CPP_BUILTIN.
13895 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
13896 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
13898 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
13899 passing required abi options to the assembler for EABI configurations.
13900 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
13901 of .text.hot and .text.unlikely sections for kernel modules when
13902 using ARM style exceptions.
13903 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
13904 options. Add EXTRA_CC1_SPEC.
13905 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
13907 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
13909 (ARM_TARGET2_DWARF_FORMAT): Define.
13910 * config/arm/t-vxworks: Adjust multilib control to removal of the
13911 Diab command line options.
13913 2017-08-01 Martin Liska <mliska@suse.cz>
13915 PR gcov-profile/81561
13916 * gcov.c (unblock): Make unblocking safe as we need to preserve
13917 index correspondence of blocks and block_lists.
13919 2017-08-01 Richard Biener <rguenther@suse.de>
13921 PR tree-optimization/81181
13922 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
13923 (compute_antic): ... end of iteration here.
13925 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
13927 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
13928 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
13929 (ftree-slp-vectorize): Likewise.
13930 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
13931 can no longer be set independent of flag_tree_loop_vectorize.
13932 * omp-general.c (emp_max_vf): Likewise.
13933 * opts.c (enable_fdo_optimizations): Remove references to
13934 flag_tree_vectorize, these are now implicit.
13935 (common_handle_option): Remove handling for OPT_ftree_vectorize,
13936 and leave it for the options machinery.
13938 2017-08-01 Martin Liska <mliska@suse.cz>
13940 PR middle-end/70140
13941 * builtins.c (expand_builtin_memcpy_args): Remove.
13942 (expand_builtin_memcpy): Call newly added function
13943 expand_builtin_memory_copy_args.
13944 (expand_builtin_memcpy_with_bounds): Likewise.
13945 (expand_builtin_mempcpy): Remove last argument.
13946 (expand_builtin_mempcpy_with_bounds): Likewise.
13947 (expand_builtin_memory_copy_args): New function created from
13948 expand_builtin_mempcpy_args with small modifications.
13949 (expand_builtin_mempcpy_args): Remove.
13950 (expand_builtin_stpcpy): Remove unused argument.
13951 (expand_builtin): Likewise.
13952 (expand_builtin_with_bounds): Likewise.
13954 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
13957 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
13958 print "ds:" only for immediates in generic address space.
13960 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
13963 * config/i386/i386.c (ix86_funciton_naked): New prototype.
13964 (ix86_function_ok_for_sibcall): Return false for naked functions.
13966 2017-08-01 Richard Biener <rguenther@suse.de>
13968 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
13969 (compute_antic): Seed worklist with exit block predecessors.
13970 * cfganal.c (dfs_find_deadend): For a cycle return the source
13971 of the edge closing it.
13973 2017-08-01 Tamar Christina <tamar.christina@arm.com>
13975 * config/aarch64/aarch64.c
13976 (aarch64_can_const_movi_rtx_p): Move 0 check.
13978 2017-08-01 Bin Cheng <bin.cheng@arm.com>
13980 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
13981 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
13983 * match.pd: Ditto in address comparison pattern.
13985 2017-08-01 Bin Cheng <bin.cheng@arm.com>
13987 PR tree-optimization/81627
13988 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
13989 closed ssa form for store-store chain.
13991 2017-08-01 Bin Cheng <bin.cheng@arm.com>
13993 PR tree-optimization/81620
13994 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
13995 for store-store chain.
13997 2017-08-01 Jakub Jelinek <jakub@redhat.com>
13999 PR tree-optimization/81588
14000 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
14001 ranges[i].in_p, invert comparison code ccode. For >/>=,
14002 swap rhs1 and rhs2 and comparison code unconditionally,
14003 for </<= don't do that. Don't swap rhs1/rhs2 again if
14004 ranges[i].in_p, instead invert comparison code ccode if
14005 opcode or oe->rank is BIT_IOR_EXPR.
14008 * optabs.def (vec_extract_optab, vec_init_optab): Change from
14009 a direct optab to conversion optab.
14010 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
14011 with GET_MODE_INNER as last argument instead of optab_handler.
14012 * expmed.c (extract_bit_field_1): Likewise. Use vector from
14013 vector extraction if possible and optab is available.
14014 * expr.c (store_constructor): Use convert_optab_handler instead
14015 of optab_handler. Use vector initialization from smaller
14016 vectors if possible and optab is available.
14017 * tree-vect-stmts.c (vectorizable_load): Likewise.
14018 * doc/md.texi (vec_extract, vec_init): Document that the optabs
14019 now have two modes.
14020 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
14021 of vec_init from half-sized vectors with the same element mode.
14022 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
14023 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
14024 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
14025 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
14026 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
14027 after mode in gen_vec_extract* calls.
14028 (vec_extract<mode>): Renamed to ...
14029 (vec_extract<mode><ssescalarmodelower>): ... this.
14030 (vec_extract<mode><ssehalfvecmodelower>): New expander.
14031 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
14032 element mode after mode in gen_vec_init* calls.
14033 (VEC_INIT_HALF_MODE): New mode iterator.
14034 (vec_init<mode>): Renamed to ...
14035 (vec_init<mode><ssescalarmodelower>): ... this.
14036 (vec_init<mode><ssehalfvecmodelower>): New expander.
14037 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
14038 (vec_extractv2sfsf): ... this.
14039 (vec_initv2sf): Renamed to ...
14040 (vec_initv2sfsf): ... this.
14041 (vec_extractv2si): Renamed to ...
14042 (vec_extractv2sisi): ... this.
14043 (vec_initv2si): Renamed to ...
14044 (vec_initv2sisi): ... this.
14045 (vec_extractv4hi): Renamed to ...
14046 (vec_extractv4hihi): ... this.
14047 (vec_initv4hi): Renamed to ...
14048 (vec_initv4hihi): ... this.
14049 (vec_extractv8qi): Renamed to ...
14050 (vec_extractv8qiqi): ... this.
14051 (vec_initv8qi): Renamed to ...
14052 (vec_initv8qiqi): ... this.
14053 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
14054 (vec_init<mode>): Renamed to ...
14055 (vec_init<mode><VEC_base_l>): ... this.
14056 (vec_extract<mode>): Renamed to ...
14057 (vec_extract<mode><VEC_base_l>): ... this.
14058 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
14059 (vec_initv2sfsf): ... this.
14060 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
14061 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14062 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
14063 element mode after mode in gen_vec_init* calls.
14064 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
14065 (vec_init<mode><Vel>): ... this.
14066 (vec_extract<mode>): Renamed to ...
14067 (vec_extract<mode><Vel>): ... this.
14068 * config/aarch64/iterators.md (Vel): New mode attribute.
14069 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
14070 Add element mode after mode in gen_vec_extract* calls.
14071 * config/s390/vector.md (non_vec_l): New mode attribute.
14072 (vec_extract<mode>): Renamed to ...
14073 (vec_extract<mode><non_vec_l>): ... this.
14074 (vec_init<mode>): Renamed to ...
14075 (vec_init<mode><non_vec_l>): ... this.
14076 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
14077 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
14079 * config/arm/iterators.md (V_elem_l): New mode attribute.
14080 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
14081 (vec_extract<mode><V_elem_l>): ... this.
14082 (vec_extractv2di): Renamed to ...
14083 (vec_extractv2didi): ... this.
14084 (vec_init<mode>): Renamed to ...
14085 (vec_init<mode><V_elem_l>): ... this.
14086 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
14087 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
14088 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
14089 Add element mode after gen_vec_extract* calls.
14090 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
14091 (vec_init<mode><unitmode>): ... this.
14092 (vec_extract<mode>): Renamed to ...
14093 (vec_extract<mode><unitmode>): ... this.
14094 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
14095 (vec_init<mode><unitmode>): ... this.
14096 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
14097 (vec_initv2sfsf): ... this.
14098 (vec_extractv2sf): Renamed to ...
14099 (vec_extractv2sfsf): ... this.
14100 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
14101 Add element mode after gen_vec_extract* calls.
14102 * config/mips/mips.md (unitmode): New mode iterator.
14103 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
14104 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
14105 * config/spu/spu.md (inner_l): New mode attribute.
14106 (vec_init<mode>): Renamed to ...
14107 (vec_init<mode><inner_l>): ... this.
14108 (vec_extract<mode>): Renamed to ...
14109 (vec_extract<mode><inner_l>): ... this.
14110 * config/sparc/sparc.md (veltmode): New mode iterator.
14111 (vec_init<VMALL:mode>): Renamed to ...
14112 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
14113 * config/ia64/vect.md (vec_initv2si): Renamed to ...
14114 (vec_initv2sisi): ... this.
14115 (vec_initv2sf): Renamed to ...
14116 (vec_initv2sfsf): ... this.
14117 (vec_extractv2sf): Renamed to ...
14118 (vec_extractv2sfsf): ... this.
14119 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
14120 (vec_init<mode>): Renamed to ...
14121 (vec_init<mode><VEC_base_l>): ... this.
14122 (vec_extract<mode>): Renamed to ...
14123 (vec_extract<mode><VEC_base_l>): ... this.
14124 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
14125 (vec_initv2sfsf): ... this.
14126 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
14127 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14128 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
14129 gen_vec_init* calls.
14131 2017-08-01 Richard Biener <rguenther@suse.de>
14133 PR tree-optimization/81297
14134 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
14135 TREE_OVERFLOW from INTEGER_CSTs.
14137 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
14139 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
14141 2017-07-31 Carl Love <cel@us.ibm.com>
14143 * config/rs6000/rs6000-c: Add support for built-in functions
14144 vector signed char vec_xl_be (signed long long, signed char *);
14145 vector unsigned char vec_xl_be (signed long long, unsigned char *);
14146 vector signed int vec_xl_be (signed long long, signed int *);
14147 vector unsigned int vec_xl_be (signed long long, unsigned int *);
14148 vector signed long long vec_xl_be (signed long long, signed long long *);
14149 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
14150 vector signed short vec_xl_be (signed long long, signed short *);
14151 vector unsigned short vec_xl_be (signed long long, unsigned short *);
14152 vector double vec_xl_be (signed long long, double *);
14153 vector float vec_xl_be (signed long long, float *);
14154 * config/rs6000/altivec.h (vec_xl_be): Add #define.
14155 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
14156 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
14158 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
14159 (altivec_expand_builtin): Add switch statement to call
14160 altivec_expand_xl_be for each builtin.
14161 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
14162 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
14163 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
14164 __builtin_vsx_le_be_v16qi.
14165 * doc/extend.texi: Update the built-in documentation file for the
14166 new built-in functions.
14168 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14171 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
14173 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
14175 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14177 * config.gcc: Add z14.
14178 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
14179 CPU model numbers for z13s and z14.
14180 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
14182 * config/s390/s390-opts.h (enum processor_type): Rename
14183 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14184 * config/s390/s390.c (processor_table): Add field for CPU name to
14185 be passed to Binutils.
14186 (s390_asm_output_machine_for_arch): Use the new field in
14187 processor_table for Binutils.
14188 (s390_expand_builtin): Replace arch12 with z14.
14189 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14190 (s390_get_sched_attrmask): Likewise.
14191 (s390_get_unit_mask): Likewise.
14192 * config/s390/s390.opt: Add z14 to processor_type enum.
14194 2017-07-31 Martin Jambor <mjambor@suse.cz>
14197 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
14198 regardless of optimization level.
14200 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
14201 Martin Liska <mliska@suse.cz>
14203 * predict.def: Remove old comment and adjust probability.
14204 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
14205 PREDICT statements.
14207 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14210 * config/i386/i386.c (ix86_function_naked): New function.
14211 (ix86_can_use_return_insn_p): Return false for naked functions.
14212 (ix86_expand_prologue): Skip prologue for naked functions.
14213 (ix86_expand_epilogue): Skip epilogue for naked functions
14214 and emit trap instruction.
14215 (ix86_warn_func_return): New function.
14216 (ix86_attribute_table): Add "naked" attribute specification.
14217 (TARGET_WARN_FUNC_RETURN): Define.
14218 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
14220 2017-07-31 Martin Liska <mliska@suse.cz>
14222 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
14223 (dump_gimple_bb_header): Always dump BB info.
14224 (pp_cfg_jump): Do not append info about BB when dumping a jump.
14226 2017-07-31 Martin Liska <mliska@suse.cz>
14229 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
14230 also with current_function_decl non-null equality.
14232 2017-07-31 Jakub Jelinek <jakub@redhat.com>
14235 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
14236 change type to the element type, instead add eltype variable and
14237 use it where we are interested in the element type.
14239 PR tree-optimization/81603
14240 * ipa-polymorphic-call.c
14241 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
14242 offset arithmetic in offset_int, bail out if the resulting bit offset
14243 doesn't fit into shwi.
14245 2017-07-31 Martin Liska <mliska@suse.cz>
14247 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
14248 (gimplify_save_expr): Fix comment.
14250 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
14253 * config/i386/i386.c (ix86_function_arg): Update arguments for
14255 (ix86_compute_frame_layout): Set the initial stack offset to
14256 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
14257 INCOMING_FRAME_SP_OFFSET.
14258 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
14259 stack before exception handler returns.
14260 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
14261 the 'ERROR_CODE' for exception handler.
14263 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
14265 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
14266 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
14267 (ASM_OUTPUT_REG_POP): Ditto.
14268 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
14269 instead of asm_fprintf to output pure string.
14271 2017-07-29 Jakub Jelinek <jakub@redhat.com>
14273 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
14274 to imported_module_or_decl hook.
14275 (debug_nothing_tree_tree_tree_bool): Remove.
14276 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
14277 * debug.c (do_nothing_debug_hooks): Use
14278 debug_nothing_tree_tree_tree_bool_bool instead of
14279 debug_nothing_tree_tree_tree_bool.
14280 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14281 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
14282 * sdbout.c (sdb_debug_hooks): Likewise.
14283 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
14284 (gen_namespace_die): Add DW_AT_export_symbols attribute if
14286 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
14287 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
14288 attribute, don't add anything.
14290 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14292 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
14293 (fold_build2_stat_loc): Likewise.
14294 (fold_build3_stat_loc): Likewise.
14295 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
14296 (fold_build1_loc): Remove macro.
14297 (fold_build2_loc): Likewise.
14298 (fold_build3_loc): Likewise.
14300 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14302 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
14303 (gimple_build_debug_bind_source_stat): Likewise.
14304 * gimple.h (gimple_build_debug_bind): Remove macro.
14305 (gimple_build_debug_bind_source): Likewise.
14307 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14309 * bitmap.c (bitmap_alloc): Adjust.
14310 (bitmap_gc_alloc): Likewise.
14311 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
14313 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14315 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
14316 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
14317 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
14318 (bitmap_gc_alloc_stat): Likewise.
14319 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
14321 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14323 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
14324 * rtl.h (shallow_copy_rtx): Remove macro.
14326 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14328 * emit-rtl.c (gen_raw_REG): Adjust.
14329 * gengenrtl.c (gendef): Likewise.
14330 * rtl.c (rtx_alloc_stat): Remove _stat from name.
14331 * rtl.h (rtx_alloc): Remove macro.
14333 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14335 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
14336 (build_tree_list_stat): Likewise.
14337 * tree.h (build_tree_list): Remove macro.
14338 (build_tree_list_vec): Likewise.
14340 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14342 * tree.c (make_vector_stat): Remove _stat from name.
14343 (build_vector_stat): Likewise.
14344 * tree.h (make_vector_stat): Remove macro.
14345 (build_vector_stat): Likewise.
14347 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14349 * tree.h (build_var_debug_value): Remove prototype.
14351 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14353 * tree.c (tree_cons_stat): Remove _stat from name.
14354 * tree.h (tree_cons): Remove macro.
14356 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14358 * tree.c (build_vl_exp_stat): Remove _stat from name.
14359 * tree.h (build_vl_exp): Remove macro.
14361 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14363 * tree.c (build_decl_stat): Remove _stat from name.
14364 * tree.h (build_decl): Remove macro.
14366 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14368 * gimple.c (gimple_build_with_ops_stat): Adjust.
14369 (gimple_alloc_stat): Remove _stat from name.
14370 * gimple.h (gimple_alloc): Remove macro.
14372 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14374 * tree.c (make_tree_vec_stat): Remove _stat from name.
14375 (grow_tree_vec_stat): Likewise.
14376 * tree.h (make_tree_vec_stat): Adjust prototype.
14377 (grow_tree_vec_stat): Likewise.
14378 (make_tree_vec): Remove macro.
14379 (grow_tree_vec): Likewise.
14381 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14383 * fold-const.c (fold_build1_stat_loc): Adjust.
14384 (fold_build2_stat_loc): Likewise.
14385 (fold_build3_stat_loc): Likewise.
14386 * tree.c (build0_stat): Remove _stat from name.
14387 (build1_stat): Likewise.
14388 (build2_stat): Likewise.
14389 (build3_stat): Likewise.
14390 (build4_stat): Likewise.
14391 (build5_stat): Likewise.
14392 * tree.h (build1_loc): Remove macro, and rename _stat function
14394 (build2_loc): Likewise.
14395 (build3_loc): Likewise.
14396 (build4_loc): Likewise.
14397 (build5_loc): Likewise.
14399 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14401 * tree.c (make_int_cst_stat): Remove _stat from name.
14402 * tree.h (make_int_cst_stat): Adjust prototype.
14403 (make_int_cst): Remove macro.
14405 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14407 * tree.c (make_tre_binfo_stat): Remove _stat from name.
14408 * tree.h (make_tree_binfo_stat): Adjust prototype.
14409 (make_tree_binfo): Remove.
14411 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14413 * tree.c (copy_node_stat): Rename to copy_node.
14414 (build_distinct_type_copy): Adjust.
14415 * tree.h (copy_node_stat): Adjust prototype.
14416 (copy_node): Remove macro.
14418 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14420 * tree.c (make_node_stat): rename to make_node.
14421 (build_tree_list_stat): Adjust.
14422 (build0_stat): Likewise.
14423 (build2_stat): Likewise.
14424 (build3_stat): Likewise.
14425 (build4_stat): Likewise.
14426 (build5_stat): Likewise.
14427 (build_decl_stat): Likewise.
14428 * tree.h (make_node_stat): Adjust prototype.
14429 (make_node): remove macro.
14431 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
14433 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
14434 (PPC_FEATURE2_SCV): Likewise.
14435 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
14437 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14439 * config/aarch64/aarch64.c
14440 (aarch64_internal_mov_immediate): Add new special pattern.
14441 * config/aarch64/aarch64.md (*movdi_aarch64):
14442 Add reg/32bit const mov case.
14444 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14445 Richard Sandiford <richard.sandiford@linaro.org>
14447 * config/aarch64/aarch64.md (mov<mode>): Generalize.
14448 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14449 Add integer and movi cases.
14450 (movi-split-hf-df-sf split, fp16): New.
14451 (enabled): Added TARGET_FP_F16INST.
14452 * config/aarch64/iterators.md (GPF_HF): New.
14453 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
14455 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14457 * config/aarch64/aarch64.c
14458 (aarch64_simd_container_mode): Add prototype.
14459 (aarch64_expand_mov_immediate): Add HI support.
14460 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
14461 (aarch64_can_const_movi_rtx_p): New.
14462 (aarch64_preferred_reload_class):
14463 Remove restrictions of using FP registers for certain SIMD operations.
14464 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
14465 (aarch64_valid_floating_const): Add integer move validation.
14466 (aarch64_simd_imm_scalar_p): Remove.
14467 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
14468 (aarch64_legitimate_constant_p): Expand list of supported cases.
14469 * config/aarch64/aarch64-protos.h
14470 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
14471 (aarch64_reinterpret_float_as_int): New.
14472 (aarch64_simd_imm_scalar_p): Remove.
14473 * config/aarch64/constraints.md (Uvi): New.
14474 (Dd): Split into Ds and new Dd.
14475 * config/aarch64/aarch64.md (*movsi_aarch64):
14477 (*movdi_aarch64): Add SIMD mov case.
14479 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14481 * tree-predcom.c: (struct chain): Handle store-store chain in which
14482 stores for elimination only store loop invariant values.
14483 (execute_pred_commoning_chain): Ditto.
14484 (prepare_initializers_chain_store_elim): Ditto.
14485 (prepare_finalizers): Ditto.
14486 (is_inv_store_elimination_chain): New function.
14487 (initialize_root_vars_store_elim_1): New function.
14489 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14491 * tree-predcom.c: Revise general description of the pass.
14492 (enum chain_type): New enum type for store elimination.
14493 (struct chain): New field supporting store elimination.
14494 (struct component): Ditto.
14495 (dump_chain): Dump store-stores chain.
14496 (release_chain): Release resources.
14497 (split_data_refs_to_components): Compute and create component
14498 contains only stores for elimination.
14499 (get_chain_last_ref_at): New function.
14500 (make_invariant_chain): Initialization.
14501 (make_rooted_chain): Specify chain type in parameter and record it.
14502 (add_looparound_copies): Skip for store-stores chain.
14503 (determine_roots_comp): Compute type of chain and pass it to
14505 (initialize_root_vars_store_elim_2): New function.
14506 (finalize_eliminated_stores): New function.
14507 (remove_stmt): Handle store for elimination.
14508 (execute_pred_commoning_chain): Execute predictive commoning on
14509 store-store chains.
14510 (determine_unroll_factor): Skip unroll for store-stores chain.
14511 (prepare_initializers_chain_store_elim): New function.
14512 (prepare_initializers_chain): Hanlde store-store chain.
14513 (prepare_finalizers_chain, prepare_finalizers): New function.
14514 (tree_predictive_commoning_loop): Return integer value indicating
14515 if loop is unrolled or lcssa form is corrupted.
14516 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
14518 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14520 * tree-predcom.c (initialize_root): Delete.
14521 (execute_pred_commoning_chain): Initialize root vars and replace
14522 reference of non-combined chain directly, rather than call above
14525 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14527 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
14528 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
14530 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14532 * tree-predcom.c (struct chain): New field init_seq.
14533 (release_chain): Release init_seq.
14534 (prepare_initializers_chain): Record intialization stmts in above
14536 (insert_init_seqs): New function.
14537 (tree_predictive_commoning_loop): Call insert_init_seqs.
14539 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14541 * tree-predcom.c (determine_roots_comp): Skip trivial components.
14543 2017-07-28 Richard Biener <rguenther@suse.de>
14545 * match.pd: Remove superfluous :c.
14546 * genmatch.c (simplify::id): Add member.
14547 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
14549 (current_id): New global.
14550 (dt_node::parent): Move from ...
14551 (dt_operand::parent): ... here. Add for_id member.
14552 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
14553 (decision_tree::find_node): Relax order requirement when
14554 merging DT_TRUE nodes to ones inbetween the current simplify
14555 and the one we try to merge with. Add diagnostic whenever
14556 we need to enforce pattern order by not merging.
14557 (decision_tree::insert): Set current_id.
14558 (decision_tree::print_node): Dump parent node and for_id.
14559 (parser::last_id): Add member.
14560 (parser::push_simplify): Assign unique id.
14561 (parser::parser): Initialize last_id.
14563 2017-07-28 Martin Liska <mliska@suse.cz>
14566 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
14567 gimple_build_debug_bind.
14569 2017-07-28 Richard Biener <rguenther@suse.de>
14571 PR tree-optimization/81502
14572 * match.pd: Add pattern combining BIT_INSERT_EXPR with
14574 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
14576 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
14577 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
14578 for BIT_FIELD_REF args.
14579 * fold-const.c (make_bit_field_ref): Likewise.
14580 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
14582 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14585 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
14586 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
14587 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
14588 Or it into SANITIZER_UNDEFINED.
14589 * ubsan.c: Include gimple-fold.h and varasm.h.
14590 (ubsan_expand_ptr_ifn): New function.
14591 (instrument_pointer_overflow): New function.
14592 (maybe_instrument_pointer_overflow): New function.
14593 (instrument_object_size): Formatting fix.
14594 (pass_ubsan::execute): Call instrument_pointer_overflow
14595 and maybe_instrument_pointer_overflow.
14596 * internal-fn.c (expand_UBSAN_PTR): New function.
14597 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
14598 * sanitizer.def (__ubsan_handle_pointer_overflow,
14599 __ubsan_handle_pointer_overflow_abort): New builtins.
14600 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
14601 * internal-fn.def (UBSAN_PTR): New internal function.
14602 * opts.c (sanitizer_opts): Add pointer-overflow.
14603 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
14604 * fold-const.c (build_range_check): Compute pointer range check in
14605 integral type if pointer arithmetics would be needed. Formatting
14608 2017-07-28 Martin Liska <mliska@suse.cz>
14611 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
14612 parameters that are of a variable-length.
14614 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
14616 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
14618 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
14619 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
14620 (CRT_CALL_STATIC_FUNCTION): Likewise.
14621 (ASM_DEFAULT_SPEC): New define.
14622 (ASM_SPEC32): Likewise.
14623 (ASM_SPEC64): Likewise.
14624 (ASM_SPEC_COMMON): Likewise.
14625 (ASM_SPEC): Likewise.
14626 (INVALID_64BIT): Likewise.
14627 (LINK_OS_DEFAULT_SPEC): Likewise.
14628 (LINK_OS_SPEC32): Likewise.
14629 (LINK_OS_SPEC64): Likewise.
14630 (POWERPC_LINUX): Likewise.
14631 (PTRDIFF_TYPE): Likewise.
14632 (RESTORE_FP_PREFIX): Likewise.
14633 (RESTORE_FP_SUFFIX): Likewise.
14634 (SAVE_FP_PREFIX): Likewise.
14635 (SAVE_FP_SUFFIX): Likewise.
14636 (SIZE_TYPE): Likewise.
14637 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
14638 (TARGET_64BIT): Likewise.
14639 (TARGET_64BIT): Likewise.
14640 (TARGET_AIX): Likewise.
14641 (WCHAR_TYPE_SIZE): Likewise.
14642 (WCHAR_TYPE): Undefine.
14643 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
14644 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
14645 (CPP_OS_RTEMS_SPEC): Delete.
14646 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
14647 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
14649 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
14651 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14653 PR tree-optimization/81578
14654 * tree-parloops.c (build_new_reduction): Bail out if
14655 reduction_code isn't one of the standard OpenMP reductions.
14656 Move the details printing after that decision.
14658 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14660 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
14661 related to reload_in_progress.
14662 (splat_input_operand): Likewise.
14663 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
14665 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
14667 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
14668 (TARGET_INSTANTIATE_DECLS): Likewise.
14669 (legitimate_indexed_address_p): Delete reload_in_progress code.
14670 (rs6000_debug_legitimate_address_p): Likewise.
14671 (rs6000_eliminate_indexed_memrefs): Likewise.
14672 (rs6000_emit_le_vsx_store): Likewise.
14673 (rs6000_emit_move_si_sf_subreg): Likewise.
14674 (rs6000_emit_move): Likewise.
14675 (register_to_reg_type): Likewise.
14676 (rs6000_pre_atomic_barrier): Likewise.
14677 (rs6000_machopic_legitimize_pic_address): Likewise.
14678 (rs6000_allocate_stack_temp): Likewise.
14679 (rs6000_address_for_fpconvert): Likewise.
14680 (rs6000_address_for_altivec): Likewise.
14681 (rs6000_secondary_memory_needed_rtx): Delete function.
14682 (rs6000_check_sdmode): Likewise.
14683 (rs6000_alloc_sdmode_stack_slot): Likewise.
14684 (rs6000_instantiate_decls): Likewise.
14685 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
14686 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
14687 Delete reload_in_progress.
14688 (*vec_reload_and_plus_<mptrsize>): Likewise.
14689 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
14690 (vsx_div_v2di): Likewise.
14691 (vsx_udiv_v2di): Likewise.
14693 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14695 * config/rs6000/rs6000.opt (mlra): Replace with stub.
14696 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
14697 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
14698 (rs6000_debug_reg_global): Delete print of LRA status.
14699 (rs6000_option_override_internal): Delete dead LRA related code.
14700 (rs6000_lra_p): Delete function.
14701 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
14703 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
14705 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
14706 * config/riscv/rtems.h: New file.
14708 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14709 Sudakshina Das <sudi.das@arm.com>
14711 * config/aarch64/aarch64.md
14712 (define_split for and<mode>3nr_compare): Move
14713 non aarch64_logical_operand to a register.
14714 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
14715 register immediate operand to a register.
14716 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
14718 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14720 PR middle-end/81564
14721 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
14723 2017-07-27 Richard Biener <rguenther@suse.de>
14725 PR tree-optimization/81573
14726 PR tree-optimization/81494
14727 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
14728 multi defuse cycle case.
14730 2017-07-27 Richard Biener <rguenther@suse.de>
14732 PR tree-optimization/81571
14733 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
14736 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
14738 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
14739 earlier and only if MASK_FPU is set. Adjust formatting.
14741 2017-07-27 Martin Liska <mliska@suse.cz>
14743 * opt-functions.awk: Add validation of value of Init.
14744 * optc-gen.awk: Pass new argument.
14746 2017-07-27 Martin Liska <mliska@suse.cz>
14748 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
14749 Fix wrong condition.
14751 2017-07-27 Martin Liska <mliska@suse.cz>
14753 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
14754 BBs and edges seen by autoFDO.
14756 2017-07-27 Richard Biener <rguenther@suse.de>
14758 PR tree-optimization/81502
14759 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
14760 with incompatible but same sized type.
14761 (execute_update_addresses_taken): Likewise.
14763 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
14765 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
14766 flag_tree_loop_vectorize rather than flag_tree_vectorize.
14768 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14771 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
14772 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
14773 Change s_operand to memory_operand.
14775 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
14777 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
14778 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
14779 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
14780 Emit instructions rather than returning an expression. Handle TFmode
14781 and KFmode by casting to TImode.
14782 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
14783 (rs6000_emit_le_vsx_store): Likewise.
14784 * config/rs6000/vsx.md (VSX_TI): New iterator.
14785 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
14786 (*vsx_le_undo_permute_<mode>): Likewise.
14787 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
14788 emit the split sequence.
14789 (*vsx_le_perm_store_<mode>): Likewise.
14791 2017-07-27 Jakub Jelinek <jakub@redhat.com>
14793 PR tree-optimization/81555
14794 PR tree-optimization/81556
14795 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
14796 if true, force CHANGED for the recursive invocation.
14797 (reassociate_bb): Remember original length of ops array, pass
14798 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
14800 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
14801 attributes for noipa attribute. For naked attribute use
14802 lookup_attribute first before lookup_attribute_spec.
14803 * final.c (rest_of_handle_final): Disable IPA RA for functions with
14805 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
14806 for functions with noipa attribute.
14807 (cgraph_externally_visible_p): Return true for functions with noipa
14809 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
14810 for functions with noipa attribute.
14811 * doc/extend.texi: Document noipa function attribute.
14812 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
14813 also for functions with noipa attribute.
14814 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
14816 2017-07-26 Andrew Pinski <apinski@cavium.com>
14818 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
14819 vec_unalign_load_cost and vec_unalign_store_cost.
14821 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14823 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
14824 -mvsx-small-integer option.
14825 (ISA_3_0_MASKS_IEEE): Likewise.
14826 (OTHER_VSX_VECTOR_MASKS): Likewise.
14827 (POWERPC_MASKS): Likewise.
14828 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
14829 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
14830 code, only testing for DImode being allowed in non-VSX floating
14832 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
14833 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
14835 (rs6000_option_override_internal): Delete -mvsx-small-integer.
14836 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
14837 TARGET_P8_VECTOR test.
14838 (rs6000_secondary_reload_simple_move): Likewise.
14839 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
14840 since TARGET_P9_VECTOR was already tested.
14841 (rs6000_opt_masks): Remove -mvsx-small-integer.
14842 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
14843 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14845 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
14846 test for TARGET_VEXTRACTUB was used, and that uses
14848 (p9 extract splitter): Likewise.
14849 (vsx_extract_<mode>_di_p9): Likewise.
14850 (vsx_extract_<mode>_store_p9): Likewise.
14851 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
14852 for TARGET_P9_VECTOR was used. Delete code that is now dead with
14853 the elimination of TARGET_VSX_SMALL_INTEGER.
14854 (vsx_extract_<mode>_p8): Likewise.
14855 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
14856 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
14857 (vsx_set_<mode>_p9): Likewise.
14858 (vsx_set_v4sf_p9): Likewise.
14859 (vsx_set_v4sf_p9_zero): Likewise.
14860 (vsx_insert_extract_v4sf_p9): Likewise.
14861 (vsx_insert_extract_v4sf_p9_2): Likewise.
14862 * config/rs6000/rs6000.md (sign extend splitter): Change
14863 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
14864 (floatsi<mode>2_lfiwax_mem): Likewise.
14865 (floatunssi<mode>2_lfiwzx_mem): Likewise.
14866 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
14867 since a test for TARGET_P9_VECTOR was used.
14868 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14869 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
14870 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14871 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
14872 TARGET_P8_VECTOR test.
14873 (fix_trunc<mode>si2_stfiwx): Likewise.
14874 (fix_trunc<mode>si2_internal): Likewise.
14875 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
14876 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14878 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
14879 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
14880 TARGET_P8_VECTOR test.
14881 (fixuns_trunc<mode>si2_stfiwx): Likewise.
14882 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
14883 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14885 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
14886 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
14887 since a test for TARGET_P9_VECTOR was used.
14888 (splitter for loading small constants): Likewise.
14890 2017-07-26 Andrew Pinski <apinski@cavium.com>
14892 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
14895 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
14898 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
14899 (fp_valid_at): Likewise.
14901 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
14903 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
14904 (qdf24xx_addrcost_table): Likewise.
14905 (cortexa57_tunings): Update to use generic_branch_cost.
14906 (cortexa72_tunings): Likewise.
14907 (cortexa73_tunings): Likewise.
14908 (qdf24xx_tunings): Likewise.
14910 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
14912 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
14913 (thunderx2t99_branch_cost): Likewise.
14914 (cortexa35_tunings): Update to use generic_branch_cost.
14915 (cortexa53_tunings): Likewise.
14916 (cortexa57_tunings): Likewise.
14917 (cortexa72_tunings): Likewise.
14918 (cortexa73_tunings): Likewise.
14919 (thunderx2t99_tunings): Likewise.
14921 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
14923 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
14924 (sparc_option_override): Honour MASK_FSMULD.
14925 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
14926 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
14927 * config/sparc/sparc.opt (mfsmuld): New option.
14928 * doc/invoke.texi (mfsmuld): Document option.
14930 2017-07-26 Marek Polacek <polacek@redhat.com>
14932 PR middle-end/70992
14933 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
14935 2017-07-26 Richard Biener <rguenther@suse.de>
14937 * gimple-match-head.c (do_valueize): Return OP if valueize
14939 (get_def): New helper to get at the def stmt of a SSA name
14940 if valueize allows.
14941 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
14942 do_valueize to get at the def stmt.
14943 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
14945 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
14947 PR middle-end/46932
14948 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
14950 2017-07-26 Martin Liska <mliska@suse.cz>
14953 * function.c (expand_function_start): Make expansion of
14954 nonlocal_goto_save_area after parm_birth_insn.
14956 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
14958 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
14959 from all CPU target flags enable members.
14961 2017-07-26 Richard Biener <rguenther@suse.de>
14963 * genmatch.c (dt_simplify::gen): Make iterator vars const.
14964 (decision_tree::gen): Make 'type' const.
14965 (write_predicate): Likewise.
14967 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
14969 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
14970 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
14971 (rs6000_option_override_internal): Likewise.
14972 (rs6000_expand_vector_set): Likewise.
14973 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
14974 (TARGET_UPPER_REGS_SF): Likewise.
14975 (TARGET_UPPER_REGS_DI): Likewise.
14976 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
14977 (TARGET_DIRECT_MOVE_64BIT): Likewise.
14978 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
14979 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14980 (Splitters for DI constants in Altivec registers): Likewise.
14981 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
14982 (vsx_set_v4sf_p9): Likewise.
14983 (vsx_set_v4sf_p9_zero): Likewise.
14984 (vsx_insert_extract_v4sf_p9): Likewise.
14985 (vsx_insert_extract_v4sf_p9_2): Likewise.
14987 2017-07-25 Carl Love <cel@us.ibm.com>
14989 * doc/extend.texi: Update the built-in documentation file for the
14990 existing built-in functions
14991 vector signed char vec_cnttz (vector signed char);
14992 vector unsigned char vec_cnttz (vector unsigned char);
14993 vector signed short vec_cnttz (vector signed short);
14994 vector unsigned short vec_cnttz (vector unsigned short);
14995 vector signed int vec_cnttz (vector signed int);
14996 vector unsigned int vec_cnttz (vector unsigned int);
14997 vector signed long long vec_cnttz (vector signed long long);
14998 vector unsigned long long vec_cnttz (vector unsigned long long);
15000 2017-07-25 Andrew Pinski <apinski@cavium.com>
15002 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
15003 accesses where the use is for the first operand of a BIT_INSERT.
15005 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
15008 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
15009 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
15011 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
15013 * config/i386/gstabs.h: Delete.
15014 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
15016 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
15018 * config/i386/i386.c (ix86_decompose_address): Do not check for
15019 register RTX when looking at index_reg or base_reg.
15020 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
15022 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
15024 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
15025 to update EH info here.
15027 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15029 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
15031 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15033 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
15035 2017-07-25 Torsten Duwe <duwe@suse.de>
15037 * common.opt: Introduce -fpatchable-function-entry
15038 command line option, and its variables function_entry_patch_area_size
15039 and function_entry_patch_area_start.
15040 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
15041 including a two-value parser.
15042 * target.def (print_patchable_function_entry): New target hook.
15043 * targhooks.h (default_print_patchable_function_entry): New function.
15044 * targhooks.c (default_print_patchable_function_entry): Likewise.
15045 * toplev.c (process_options): Switch off IPA-RA if
15046 patchable function entries are being generated.
15047 * varasm.c (assemble_start_function): Look at the
15048 patchable-function-entry command line switch and current
15049 function attributes and maybe generate NOP instructions by
15050 calling the print_patchable_function_entry hook.
15051 * doc/extend.texi: Document patchable_function_entry attribute.
15052 * doc/invoke.texi: Document -fpatchable_function_entry
15053 command line option.
15054 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
15056 * doc/tm.texi: Re-generate.
15058 2017-07-25 Jakub Jelinek <jakub@redhat.com>
15061 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
15062 TARGET_AVX512DQ rather than TARGET_AVX512BW.
15064 2017-07-25 Tamar Christina <tamar.christina@arm.com>
15066 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
15068 2017-07-25 Richard Biener <rguenther@suse.de>
15070 PR tree-optimization/81455
15071 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
15072 not walk in cycles when looking for guards.
15074 2017-07-25 Richard Biener <rguenther@suse.de>
15076 PR tree-optimization/81529
15077 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
15078 when optimizing backedge uses.
15080 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
15082 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
15084 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
15085 to dl_section_ref. On AIX, append an expression to subtract
15086 the size of the section length to dl_section_ref.
15088 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
15090 * configure.ac: If any of the config.* scripts fail, exit 1.
15091 * configure: Regenerate.
15093 2017-07-25 Richard Biener <rguenther@suse.de>
15095 PR middle-end/81546
15096 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
15097 of immediate uses, be more verbose on errors.
15099 2017-07-25 Richard Biener <rguenther@suse.de>
15101 PR tree-optimization/81510
15102 * tree-vect-loop.c (vect_is_simple_reduction): When the
15103 reduction stmt is not inside the loop bail out.
15105 2017-07-25 Richard Biener <rguenther@suse.de>
15107 PR tree-optimization/81303
15108 * tree-vect-loop-manip.c (vect_loop_versioning): Build
15109 profitability check against LOOP_VINFO_NITERSM1.
15111 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15113 * domwalk.c (cmp_bb_postorder): Simplify.
15114 (sort_bbs_postorder): New function. Use it...
15115 (dom_walker::walk): ...here to optimize common cases.
15117 2017-07-25 Martin Liska <mliska@suse.cz>
15120 * ipa-visibility.c (function_and_variable_visibility): Make the
15121 redirection just on target that supports aliasing.
15122 Fix GNU coding style.
15124 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
15127 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
15129 * config/arm/rtems.h: Likewise.
15130 * config/bfin/rtems.h: Likewise.
15131 * config/i386/rtemself.h: Likewise.
15132 * config/lm32/rtems.h: Likewise.
15133 * config/m32c/rtems.h: Likewise.
15134 * config/m68k/rtemself.h: Likewise.
15135 * config/microblaze/rtems.h: Likewise.
15136 * config/mips/rtems.h: Likewise.
15137 * config/moxie/rtems.h: Likewise.
15138 * config/nios2/rtems.h: Likewise.
15139 * config/powerpcspe/rtems.h: Likewise.
15140 * config/rs6000/rtems.h: Likewise.
15141 * config/rtems.h: Likewise.
15142 * config/sh/rtems.h: Likewise.
15143 * config/sh/rtemself.h: Likewise.
15144 * config/sparc/rtemself.h: Likewise.
15146 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15149 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
15150 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
15151 * tree-ssa-structalias.c (alias_get_name): Same.
15153 2017-07-25 Bin Cheng <bin.cheng@arm.com>
15156 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
15157 instructions if no du chain is found.
15159 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15161 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
15163 2017-07-25 Richard Biener <rguenther@suse.de>
15165 PR middle-end/81505
15166 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
15169 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
15171 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
15172 upper-regs options.
15173 (ISA_2_7_MASKS_SERVER): Likewise.
15174 (ISA_3_0_MASKS_IEEE): Likewise.
15175 (OTHER_P8_VECTOR_MASKS): Likewise.
15176 (OTHER_VSX_VECTOR_MASKS): Likewise.
15177 (POWERPC_MASKS): Likewise.
15178 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
15179 duplicate list of options.
15180 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
15181 explicit -mupper-regs options.
15182 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
15183 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
15184 alias for -mupper-regs-df.
15185 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
15186 (rs6000_init_hard_regno_mode_ok): Likewise.
15187 (rs6000_option_override_internal): Likewise.
15188 (rs6000_opt_masks): Likewise.
15189 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
15190 options in terms of whether -mvsx or -mpower8-vector was used.
15191 (TARGET_UPPER_REGS_DI): Likewise.
15192 (TARGET_UPPER_REGS_SF): Likewise.
15193 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
15194 -mupper-regs-* options.
15196 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15198 * passes.c (emergency_dump_function): Print some empty lines and a
15199 header before the RTL dump.
15201 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15203 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
15205 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
15208 * config/aarch64/aarch64.c (aarch64_classify_symbol):
15209 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
15211 2017-07-24 Carl Love <cel@us.ibm.com>
15213 * config/rs6000/rs6000-c.c: Add support for built-in functions
15214 vector float vec_extract_fp32_from_shorth (vector unsigned short);
15215 vector float vec_extract_fp32_from_shortl (vector unsigned short);
15216 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
15217 vec_extract_fp_from_shortl): Add defines for the two builtins.
15218 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
15219 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
15221 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
15222 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
15223 * doc/extend.texi: Update the built-in documentation file for the
15224 new built-in function.
15226 2017-07-24 Jakub Jelinek <jakub@redhat.com>
15229 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
15231 * doc/generic.texi: Likewise.
15232 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
15233 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
15235 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
15237 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
15238 (aarch64_mls_elt_merge<mode>): Likewise.
15240 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
15242 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
15243 having __cxa_atexit.
15245 2017-07-23 Michael Collison <michael.collison@arm.com>
15247 * config/arm/arm.c (arm_option_override): Deprecate
15248 use of -mstructure-size-boundary.
15249 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
15250 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
15252 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15255 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
15256 Reduce cost estimate for direct moves.
15258 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
15261 * config/i386/i386.c (ix86_option_override_internal): Disable
15262 BMI, BMI2 and TBM instructions for -m16.
15264 2017-07-21 Carl Love <cel@us.ibm.com>
15266 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15267 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15268 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15269 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15270 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15271 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15272 VMULOSW): New enum "unspec" values.
15273 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15274 altivec_vmulosw): New patterns.
15275 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15276 VMULOSW): Add definitions.
15278 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
15280 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
15281 (qdf24xx): Likewise.
15282 * config/aarch64/aarch64-options-extensions.def (rdma); New.
15283 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
15284 (AARCH64_FL_V8_1): Renumber.
15285 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
15286 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
15287 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
15288 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
15289 rdma to feature modifiers list.
15291 2017-07-21 Yury Gribov <tetra2005@gmail.com>
15293 PR middle-end/56727
15294 * ipa-visibility (function_and_variable_visibility): Convert
15295 recursive PLT call to direct call if appropriate.
15297 2017-07-21 Andrew Pinski <apinski@cavium.com>
15299 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
15300 operand 1 to see if the types precision matches.
15301 * fold-const.c (operand_equal_p): Likewise.
15303 2017-07-21 Richard Biener <rguenther@suse.de>
15305 PR tree-optimization/81303
15306 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
15307 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
15308 (vect_peeling_hash_get_lowest_cost): Adjust.
15309 (vect_enhance_data_refs_alignment): Likewise. Use
15310 vect_get_peeling_costs_all_drs to compute the penalty for no
15311 peeling to match up costs.
15313 2017-07-21 Richard Biener <rguenther@suse.de>
15315 PR tree-optimization/81500
15316 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
15317 we didn't identify a reduction path.
15319 2017-07-21 Tom de Vries <tom@codesourcery.com>
15320 Cesar Philippidis <cesar@codesourcery.com>
15322 PR gcov-profile/81442
15323 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
15326 2017-07-21 Tom de Vries <tom@codesourcery.com>
15329 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
15331 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
15332 nvptx_override_options_after_change.
15334 2017-07-21 Ulrich Drepper <drepper@redhat.com>
15336 * dwarf2out.c (output_file_names): Avoid double testing for
15337 dwarf_version >= 5.
15339 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
15341 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
15343 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15345 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
15347 (try_crossjump_to_edge): Do not punt on partitioned functions.
15349 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15351 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
15352 Put all BBs reachable only via paths crossing cold region to cold
15354 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
15356 2016-07-21 Richard Biener <rguenther@suse.de>
15358 PR tree-optimization/81303
15359 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
15360 into account prologue and epilogue iterations when raising
15361 min_profitable_iters to sth at least covering one vector iteration.
15363 2017-07-21 Tamar Christina <tamar.christina@arm.com>
15365 * config/arm/arm.c (arm_test_cpu_arch_dat):
15368 2017-07-20 Nathan Sidwell <nathan@acm.org>
15370 Remove TYPE_METHODS.
15371 * tree.h (TYPE_METHODS): Delete.
15372 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
15373 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
15374 (dbxout_type_methods): Scan TYPE_FIELDS.
15375 (dbxout_type): Don't check TYPE_METHODS here.
15376 * function.c (use_register_for_decl): Always ignore register for
15377 class types when not optimizing.
15378 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
15379 * tree.c (free_lang_data_in_type): Stitch out member functions and
15380 templates from TYPE_FIELDS.
15381 (build_distinct_type_copy, verify_type_variant,
15382 verify_type): Member fns are on TYPE_FIELDS.
15383 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
15384 * tree-pretty-print.c (dump_generic_node): Likewise.
15386 2017-07-20 Jakub Jelinek <jakub@redhat.com>
15389 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
15390 V2TImode and V4TImode.
15391 (ix86_expand_vector_extract): Likewise.
15392 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
15393 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
15394 (ssescalarmode): Handle V4TImode and V2TImode.
15395 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
15396 (*vec_extractv2ti, *vec_extractv4ti): New insns.
15397 (VEXTRACTI128_MODE): New mode iterator.
15398 (splitter for *vec_extractv?ti first element): New.
15399 (VEC_INIT_MODE): New mode iterator.
15400 (vec_init<mode>): Consolidate 3 expanders into one using
15401 VEC_INIT_MODE mode iterator.
15403 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
15405 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
15406 non_spilled_static_chain_regno_p.
15408 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
15410 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
15412 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
15414 * bb-reorder.c (connect_traces): Allow copying of blocks within
15417 2017-07-20 Richard Biener <rguenther@suse.de>
15419 * gimple.h (gimple_phi_result): Add gphi * overload.
15420 (gimple_phi_result_ptr): Likewise.
15421 (gimple_phi_arg): Likewise. Adjust index assert to only
15422 allow actual argument accesses rather than all slots available
15424 (gimple_phi_arg_def): Add gphi * overload.
15425 * tree-phinodes.c (make_phi_node): Initialize only actual
15427 (resize_phi_node): Clear memory not covered by old node,
15428 do not initialize excess argument slots.
15429 (reserve_phi_args_for_new_edge): Initialize new argument slot
15432 2017-07-20 Bin Cheng <bin.cheng@arm.com>
15434 PR tree-optimization/81388
15436 2016-07-21 Bin Cheng <bin.cheng@arm.com>
15438 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
15439 by removing computation of may_be_zero.
15441 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15442 Tom de Vries <tom@codesourcery.com>
15444 PR middle-end/81030
15445 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
15446 when gimple level profile disagrees with what RTL expander did.
15448 2017-07-20 Richard Biener <rguenther@suse.de>
15450 PR tree-optimization/61171
15451 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
15452 (vect_analyze_stmt): Add slp instance parameter.
15453 (vectorizable_reduction): Likewise.
15454 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
15455 (vect_is_simple_reduction): Deal with chains not detected
15456 as SLP reduction chain, specifically not properly associated
15457 chains containing a mix of plus/minus.
15458 (get_reduction_op): Remove.
15459 (get_initial_defs_for_reduction): Simplify, pass in whether
15460 this is a reduction chain, pass in the SLP node for the PHIs.
15461 (vect_create_epilog_for_reduction): Get the SLP instance as
15463 (vectorizable_reduction): Get the SLP instance as arg.
15464 During analysis remember the SLP node with the PHIs in the
15465 instance. Simplify getting at the vectorized reduction PHIs.
15466 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
15467 through SLP instance.
15468 (vect_slp_analyze_operations): Likewise.
15469 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
15470 (vect_transform_stmt): Likewise.
15472 2017-07-20 Tom de Vries <tom@codesourcery.com>
15474 PR tree-optimization/81489
15475 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
15476 read of phi arg location to before loop that modifies phi.
15478 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15480 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
15483 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15485 PR middle-end/81331
15486 * except.c (execute): Fix ordering issue.
15488 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
15490 PR rtl-optimization/81423
15491 * combine.c (make_compound_operation_int): Don't try to optimize
15492 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
15494 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
15496 PR rtl-optimization/81423
15497 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
15498 with a constant that is -1 in the truncated to mode.
15500 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15502 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
15503 (determine_unlikely_bbs): ... here.
15504 * predict.h (propagate_unlikely_bbs_forward): Declare.
15505 * cfgexpand.c (pass_expand::execute): Use it.
15506 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
15508 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
15509 propagate_unlikely_bbs_forward.
15511 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15513 PR middle-end/81331
15514 * except.c (maybe_add_nop_after_section_switch): New function.
15517 2017-07-19 Tom de Vries <tom@codesourcery.com>
15519 * gimple.h (gimple_phi_set_arg): Make assert more strict.
15521 2017-07-19 Tom de Vries <tom@codesourcery.com>
15523 * gimple.h (gimple_phi_arg): Make assert more strict.
15525 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
15527 * config.gcc (powerpc*-*-*): Add mmintrin.h.
15528 * config/rs6000/mmintrin.h: New file.
15529 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
15531 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15533 PR tree-optimization/81346
15534 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
15536 2017-07-19 Tom de Vries <tom@codesourcery.com>
15538 * config/nvptx/nvptx.md (VECIM): Add V2DI.
15540 2017-07-19 Tom de Vries <tom@codesourcery.com>
15542 * config/nvptx/nvptx-modes.def: Add V2DImode.
15543 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
15544 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
15545 (nvptx_output_mov_insn): Handle lack of mov.b128.
15546 (nvptx_print_operand): Handle 'H' and 'L' codes.
15547 (nvptx_vector_mode_supported): Allow V2DImode.
15548 (nvptx_preferred_simd_mode): New function.
15549 (nvptx_data_alignment): New function.
15550 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
15551 nvptx_preferred_simd_mode.
15552 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
15554 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
15556 2017-07-19 Tom de Vries <tom@codesourcery.com>
15558 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
15559 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
15560 (nvptx_vector_mode_supported): New function. Allow V2SImode.
15561 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
15562 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
15563 (mov<VECIM>_insn): New define_insn.
15564 (define_expand "mov<VECIM>): New define_expand.
15566 2017-07-19 Tom de Vries <tom@codesourcery.com>
15568 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
15570 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15572 PR tree-optimization/81346
15573 * fold-const.h (fold_div_compare, range_check_type): Declare.
15574 * fold-const.c (range_check_type): New function.
15575 (build_range_check): Use range_check_type.
15576 (fold_div_compare): No longer static, rewritten into
15577 a match.pd helper function.
15578 (fold_comparison): Don't call fold_div_compare here.
15579 * match.pd (X / C1 op C2): New optimization using fold_div_compare
15580 as helper function.
15582 2017-07-19 Nathan Sidwell <nathan@acm.org>
15584 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
15585 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
15586 * tree.c (find_decls_types_r, verify_type): Use
15587 TYPE_{MIN,MAX}_VALUE_RAW.
15588 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
15589 (hash_tree): Likewise.
15590 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
15592 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
15595 2017-07-18 Tom de Vries <tom@codesourcery.com>
15597 PR middle-end/81464
15598 * omp-expand.c (expand_omp_for_static_chunk): Handle
15599 equal-argument loop exit phi.
15601 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
15604 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
15605 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
15606 operand 2 predicate.
15607 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
15608 operand 2 predicate.
15609 (ror,rol -> rorx splitters): Use const_int_operand as
15610 operand 2 predicate.
15612 2017-06-18 Richard Biener <rguenther@suse.de>
15614 PR tree-optimization/81410
15615 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
15616 the gap in the ! slp_perm SLP case after each group.
15618 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15620 PR middle-end/81463
15621 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
15624 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15626 PR middle-end/81462
15627 * predict.c (set_even_probabilities): Cleanup; do not affect
15628 probabilities that are already known.
15629 (combine_predictions_for_bb): Call even when count is set.
15631 2017-07-18 Nathan Sidwell <nathan@acm.org>
15633 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
15636 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15639 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
15640 optimization for loop niter analysis.
15642 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
15645 * config/avr/avr.c (avr_optimize_casesi): Don't use
15646 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
15648 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
15650 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
15651 body_cost_vec from _vect_peel_extended_info.
15652 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
15653 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
15656 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15658 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
15660 2017-07-18 Richard Biener <rguenther@suse.de>
15662 PR tree-optimization/80620
15663 PR tree-optimization/81403
15664 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
15665 info when re-using a VN table entry.
15667 2017-07-18 Richard Biener <rguenther@suse.de>
15669 PR tree-optimization/81418
15670 * tree-vect-loop.c (vectorizable_reduction): Properly compute
15671 vectype_in. Verify that with lane-reducing reduction operations
15672 we have a single def-use cycle.
15674 2017-07-17 Carl Love <cel@us.ibm.com>
15676 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
15678 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15679 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15680 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15681 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15682 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15683 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15684 VMULOSW): New enum "unspec" values.
15685 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
15686 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
15687 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15688 altivec_vmulosw): New patterns.
15689 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15690 VMULOSW): Add definitions.
15691 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
15693 * config/alpha/alpha.c: Include predict.h.
15695 2017-07-17 Yury Gribov <tetra2005@gmail.com>
15697 * tree-vrp.c (compare_assert_loc): Fix comparison function
15698 to return predictable results.
15700 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15702 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
15704 (subdi3): Likewise.
15705 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
15706 * doc/invoke.texi (mexpand-adddi): Update text.
15708 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15710 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
15711 that also clobbers the CC register. The old expand code is moved
15713 (*arc_clzsi2): ... here.
15714 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
15715 the CC register. The old expand code is moved to ...
15716 (arc_ctzsi2): ... here.
15718 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15720 * config/arc/arc.opt (mindexed-loads): Use initial value
15721 TARGET_INDEXED_LOADS_DEFAULT.
15722 (mauto-modify-reg): Use initial value
15723 TARGET_AUTO_MODIFY_REG_DEFAULT.
15724 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
15725 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
15726 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
15727 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
15729 2017-07-17 Martin Liska <mliska@suse.cz>
15732 * opts.c (finish_options): Do not allow -fgnu-tm
15733 w/ -fsanitize={kernel-,}address. Say sorry.
15735 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15738 * tree-loop-distribution.c (classify_partition): Only assert on
15739 numer of iterations.
15740 (merge_dep_scc_partitions): Delete prameter. Update function call.
15741 (distribute_loop): Remove code handling loop with unknown niters.
15742 (pass_loop_distribution::execute): Skip loop with unknown niters.
15744 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15747 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
15748 function sort_partitions_by_post_order.
15750 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15752 PR tree-optimization/81374
15753 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
15754 the max index of basic blocks, rather than number of basic blocks.
15756 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15758 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
15760 (arc_legitimate_pic_operand_p): Likewise.
15761 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
15763 (arc_needs_pcl_p): Likewise.
15764 (arc_legitimate_pc_offset_p): Likewise.
15765 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
15766 function is also used in constrains.md.
15767 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
15768 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
15769 PLUS. Only return true/false in known cases, otherwise assert.
15770 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
15771 is already called in arc_legitimate_constant_p.
15772 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
15774 (LEGITIMATE_PIC_OPERAND_P): Use
15775 arc_raw_symbolic_reference_mentioned_p function.
15776 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
15781 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15782 Andrew Burgess <andrew.burgess@embecosm.com>
15784 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
15785 (arc_return_address_register): New function.
15786 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
15787 (arc_handle_fndecl_attribute): Add naked attribute.
15788 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
15789 (TARGET_WARN_FUNC_RETURN): Likewise.
15790 (arc_allocate_stack_slots_for_args): New function.
15791 (arc_warn_func_return): Likewise.
15792 (machine_function): Change type fn_type.
15793 (arc_compute_function_type): Consider new naked function type,
15794 change function return type.
15795 (arc_must_save_register): Adapt to handle new
15796 arc_compute_function_type's return type.
15797 (arc_expand_prologue): Likewise.
15798 (arc_expand_epilogue): Likewise.
15799 (arc_return_address_regs): Delete.
15800 (arc_return_address_register): New function.
15801 (arc_epilogue_uses): Use above function.
15802 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
15803 (arc_function_type): Change encoding, add naked type.
15804 (ARC_INTERRUPT_P): Change to handle the new encoding.
15805 (ARC_FAST_INTERRUPT_P): Likewise.
15806 (ARC_NORMAL_P): Define.
15807 (ARC_NAKED_P): Likewise.
15808 (arc_compute_function_type): Delete prototype.
15809 * config/arc/arc.md (in_ret_delay_slot): Use
15810 arc_return_address_register function.
15811 (simple_return): Likewise.
15812 (p_return_i): Likewise.
15814 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15816 PR tree-optimization/81428
15817 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
15818 can't be built for those types.
15820 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
15822 Remove stuff dead since r239246.
15824 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
15825 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
15826 (avr_inform_devices): Remove dead stuff.
15828 2017-07-17 Tamar Christina <tamar.christina@arm.com>
15830 * config/arm/arm_neon.h: Fix softp typo.
15832 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15834 PR tree-optimization/81365
15835 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
15836 aggregate moves onto bb predecessor edges, make sure there are no
15837 loads that could alias the lhs in between the start of bb and the
15840 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
15843 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
15844 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
15845 [LSHIFTRT, outer_code = TRUNCATE]: Same.
15847 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15849 PR tree-optimization/81396
15850 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
15851 (init_symbolic_number): Initialize it to 1.
15852 (perform_symbolic_merge): Add n_ops from both operands into the new
15854 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
15855 without base_addr as useless if they need more than one operation.
15856 (bswap_replace): Handle !bswap case for NULL base_addr.
15858 2017-07-17 Tom de Vries <tom@codesourcery.com>
15861 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
15864 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
15866 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
15867 conditional builtin define __FIX_LEON3FT_B2BST.
15869 2017-07-17 Daniel Cederman <cederman@gaisler.com>
15871 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
15872 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
15875 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
15877 PR rtl-optimization/81424
15878 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
15879 to remove potential trapping from operands if -fnon-call-exceptions.
15881 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15883 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
15884 profile_proability for scalling.
15885 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
15887 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15889 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
15891 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15893 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
15894 fixpoint arithmetics.
15896 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15898 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
15899 fixpoint arithmetics.
15901 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15903 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
15904 fixpoint arithmetics.
15906 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15908 * profile-count.h (profile_probability::from_reg_br_prob_note,
15909 profile_probability::to_reg_br_prob_note): New functions.
15910 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
15911 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
15912 * predict.c (probability_reliable_p): Update.
15913 (edge_probability_reliable_p): Update.
15914 (br_prob_note_reliable_p): Update.
15915 (invert_br_probabilities): Update.
15916 (add_reg_br_prob_note): New function.
15917 (combine_predictions_for_insn): Update.
15918 * asan.c (asan_clear_shadow): Update.
15919 * cfgbuild.c (compute_outgoing_frequencies): Update.
15920 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
15921 (update_br_prob_note): Update.
15922 (rtl_verify_edges): Update.
15923 (purge_dead_edges): Update.
15924 (fixup_reorder_chain): Update.
15925 * emit-rtl.c (try_split): Update.
15926 * ifcvt.c (cond_exec_process_insns): Update.
15927 (cond_exec_process_if_block): Update.
15928 (dead_or_predicable): Update.
15929 * internal-fn.c (expand_addsub_overflow): Update.
15930 (expand_neg_overflow): Update.
15931 (expand_mul_overflow): Update.
15932 * loop-doloop.c (doloop_modify): Update.
15933 * loop-unroll.c (compare_and_jump_seq): Update.
15934 * optabs.c (emit_cmp_and_jump_insn_1): Update.
15935 * predict.h: Update.
15936 * reorg.c (mostly_true_jump): Update.
15938 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
15939 * config/alpha/alpha.c (emit_unlikely_jump): Update.
15940 * config/arc/arc.c: (emit_unlikely_jump): Update.
15941 * config/arm/arm.c: (emit_unlikely_jump): Update.
15942 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
15943 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
15944 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
15945 (ix86_print_operand): Update.
15946 (ix86_split_fp_branch): Update.
15947 (predict_jump): Update.
15948 * config/ia64/ia64.c (ia64_print_operand): Update.
15949 * config/mmix/mmix.c (mmix_print_operand): Update.
15950 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
15951 (rs6000_expand_split_stack_prologue): Update.
15952 * config/rs6000/rs6000.c: Update.
15953 * config/s390/s390.c (s390_expand_vec_strlen): Update.
15954 (s390_expand_vec_movstr): Update.
15955 (s390_expand_cs_tdsi): Update.
15956 (s390_expand_split_stack_prologue): Update.
15957 * config/sh/sh.c (sh_print_operand): Update.
15958 (expand_cbranchsi4): Update.
15959 (expand_cbranchdi4): Update.
15960 * config/sparc/sparc.c (output_v9branch): Update.
15961 * config/spu/spu.c (get_branch_target): Update.
15962 (ea_load_store_inline): Update.
15963 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
15964 * config/tilepro/tilepro.c: Update.
15966 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
15968 * gimplify.c (mostly_copy_tree_r): Revert latest change.
15969 (gimplify_save_expr): Likewise.
15971 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15973 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
15975 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15977 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
15979 * timevar.def (TV_IPA_FNSUMMARY): Define.
15981 2017-07-16 Daniel Cederman <cederman@gaisler.com>
15983 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
15984 to back store errata sensitive sequence from being generated.
15985 (sqrtdf2_fix): Likewise.
15987 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15989 * tree-ssa-threadupdate.c (compute_path_counts,
15990 update_joiner_offpath_counts): Use profile_probability.
15992 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
15995 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15997 * config/arm/arm-c.c (arm_cpu_builtins): Define
15998 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
16000 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
16002 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16003 array entries to represent __ieee128 versions of the
16004 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
16005 scalar_extract_sig, and scalar_insert_exp built-in functions.
16006 (altivec_resolve_overloaded_builtin): Add special case handling
16007 for the __builtin_scalar_insert_exp function, as represented by
16008 the P9V_BUILTIN_VEC_VSIEDP constant.
16009 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
16010 exponent support for __ieee128 argument.
16011 (VSESQP): Add scalar extract signature support for __ieee128
16013 (VSTDCNQP): Add scalar test negative support for __ieee128
16015 (VSIEQP): Add scalar insert exponent support for __int128 argument
16016 with __ieee128 result.
16017 (VSIEQPF): Add scalar insert exponent support for __ieee128
16018 argument with __ieee128 result.
16019 (VSTDCQP): Add scalar test data class support for __ieee128
16021 (VSTDCNQP): Add overload support for scalar test negative with
16022 __ieee128 argument.
16023 (VSTDCQP): Add overload support for scalar test data class
16024 __ieee128 argument.
16025 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
16026 UNSPEC_VSX_SXSIGDP.
16027 (UNSPEC_VSX_SIEXPQP): New constant.
16028 (xsxexpqp): New insn for VSX scalar extract exponent quad
16030 (xsxsigqp): New insn for VSX scalar extract significand quad
16032 (xsiexpqpf): New insn for VSX scalar insert exponent quad
16033 precision with floating point argument.
16034 (xststdcqp): New expand for VSX scalar test data class quad
16036 (xststdcnegqp): New expand for VSX scalar test negative quad
16038 (xststdcqp): New insn to match expansions for VSX scalar test data
16039 class quad precision and VSX scalar test negative quad precision.
16040 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
16041 special case operand checking to enforce that second operand of
16042 VSX scalar test data class with quad precision argument is a 7-bit
16044 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
16045 prototypes and descriptions of __ieee128 versions of
16046 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
16047 scalar_test_data_class, and scalar_test_neg built-in functions.
16049 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16051 PR tree-optimization/81162
16052 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
16053 replace a negate with an add.
16055 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
16057 * doc/invoke.texi (arm/-mcpu): Document +crypto.
16059 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16061 * config/arm/arm-c.c (arm_cpu_builtins): Define
16062 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
16064 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16066 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
16067 (armv8-r): Set ARM Cortex-R52 as default CPU.
16068 * config/arm/arm-tables.opt: Regenerate.
16069 * config/arm/arm-tune.md: Regenerate.
16070 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
16072 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
16073 extension for -mcpu=cortex-r52.
16075 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16077 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
16078 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
16079 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
16080 (fp-armv8): Define it as FP_ARMv8 only.
16081 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
16082 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
16084 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
16085 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
16086 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
16087 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
16088 than TARGET_FPU_ARMV8.
16089 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
16090 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
16091 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
16093 * config/arm/neon.md (neon_vrint): Likewise.
16094 (neon_vcvt): Likewise.
16095 (neon_<fmaxmin_op><mode>): Likewise.
16096 (<fmaxmin><mode>3): Likewise.
16097 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
16098 * config/arm/predicates.md (arm_cond_move_operator): Check against
16099 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
16101 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
16103 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
16104 to top of function.
16106 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16108 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
16109 loop in comment with memset.
16111 2017-07-14 Martin Liska <mliska@suse.cz>
16113 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
16114 * dwarf2out.c (is_java): Remove the function.
16115 (output_pubname): Remove usage of the function.
16116 (lower_bound_default): Remove usage of DW_LANG_Java.
16117 (gen_compile_unit_die): Likewise.
16118 * gcc.c: Remove compiler defaults for .java and .zip files.
16119 * gimple-expr.c (remove_suffix): Change as there's no longer
16120 extension than 4-letter one.
16121 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
16122 (gimplify_save_expr): Likewise.
16123 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
16124 as it's possible even for other languages than Java.
16125 * langhooks.h (struct lang_hooks): Remove Java from a comment.
16126 * lto-opts.c (lto_write_options): Remove reference to Java.
16127 * opts.c (strip_off_ending): Update file extension handling.
16128 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
16129 * tree-eh.c (lower_resx): Likewise.
16130 * tree.c (free_lang_data_in_type): Remove dead code.
16131 (find_decls_types_r): Likewise.
16132 (build_common_builtin_nodes): Remove Java from a comment.
16133 (verify_type): Remove dead code.
16134 * varasm.c (assemble_external): Remove Java from a comment.
16136 2017-07-14 Martin Liska <mliska@suse.cz>
16138 * opts.c (finish_options): Add quotes.
16139 (common_handle_option): Likewise.
16141 2017-07-14 Martin Liska <mliska@suse.cz>
16143 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
16144 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
16145 Remove N_SO_PASCAL.
16146 * dwarf2out.c (lower_bound_default): Do not handle
16148 (gen_compile_unit_die): Likewise.
16149 * gcc.c: Remove default extension binding for GNU Pascal.
16150 * stmt.c: Remove Pascal language from a comment.
16151 * xcoffout.c: Likewise.
16153 2017-07-13 David Malcolm <dmalcolm@redhat.com>
16156 * diagnostic-show-locus.c (fixit_cmp): New function.
16157 (layout::layout): Sort m_fixit_hints.
16158 (column_range::column_range): Assert that the values are valid.
16159 (struct char_span): New struct.
16160 (correction::overwrite): New method.
16161 (struct source_line): New struct.
16162 (line_corrections::add_hint): Add assertions. Reimplement memcpy
16163 calls in terms of classes source_line and char_span, and
16164 correction::overwrite.
16165 (selftest::test_overlapped_fixit_printing_2): New function.
16166 (selftest::diagnostic_show_locus_c_tests): Call it.
16168 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
16170 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
16171 early if there is no lhs.
16173 2017-07-13 Martin Liska <mliska@suse.cz>
16175 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
16176 (gen_reference_type_die): Likewise.
16177 * stor-layout.c: Remove Pascal-related comment.
16179 2017-07-13 Martin Liska <mliska@suse.cz>
16181 * opts.c (finish_options): Add quotes to error messages.
16182 (parse_sanitizer_options): Likewise.
16184 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
16186 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
16188 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
16190 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
16192 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
16194 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
16196 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
16198 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
16201 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
16202 provides the hardware capability bits, define the macro
16203 __BUILTIN_CPU_SUPPORTS__.
16204 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
16205 if GLIBC does not provide the hardware capability bits. Add a
16206 gcc_unreachable call if the built-in cpu function is neither
16207 __builtin_cpu_is nor __builtin_cpu_supports.
16208 (rs6000_get_function_versions_dispatcher): Change the warning
16209 that an old GLIBC is used which does not export the capability
16210 bits to be an error.
16211 * doc/extend.texi (target_clones attribute): Document the
16212 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
16213 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
16214 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
16215 the macros defined by GCC if the newer GLIBC is available.
16217 2017-07-12 Jeff Law <law@redhat.com>
16219 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
16220 remaining includes slightly.
16221 * config/riscv/riscv-builtins.c: Include profile-count.h.
16223 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16226 * config/avr/avr.c (avr_set_current_function): In diagnostic
16227 messages: Quote keywords and (parts of) identifiers.
16228 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
16231 2017-07-12 Carl Love <cel@us.ibm.com>
16233 * config/rs6000/rs6000-c.c: Add support for built-in functions
16234 vector bool char vec_revb (vector bool char);
16235 vector bool short vec_revb (vector short char);
16236 vector bool int vec_revb (vector bool int);
16237 vector bool long long vec_revb (vector bool long long);
16238 * doc/extend.texi: Update the built-in documentation file for the
16239 new built-in functions.
16241 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16243 * config/s390/s390.md: Remove movcc splitter.
16245 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16247 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
16248 load/store on condition.
16250 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16253 * config/avr/avr.c (avr_encode_section_info)
16254 [progmem && !TREE_READONLY]: Error if progmem object needs
16257 2017-07-11 Michael Collison <michael.collison@arm.com>
16259 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
16262 2017-07-11 Carl Love <cel@us.ibm.com>
16264 * config/rs6000/rs6000-c.c: Add support for builtins
16265 vector unsigned int vec_parity_lsbb (vector signed int);
16266 vector unsigned int vec_parity_lsbb (vector unsigned int);
16267 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
16268 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
16269 vector unsigned long long vec_parity_lsbb (vector signed long long);
16270 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
16271 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
16272 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
16273 * doc/extend.texi: Update the built-in documentation file for the
16274 new built-in functions.
16276 2017-07-11 David Malcolm <dmalcolm@redhat.com>
16278 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
16279 (layout::m_primary_loc): New field.
16280 (layout::layout): Initialize new field. Move location filtering
16281 logic from here to...
16282 (layout::maybe_add_location_range): ...this new method. Add
16283 support for filtering to just the lines already specified by other
16285 (layout::will_show_line_p): New method.
16286 (gcc_rich_location::add_location_if_nearby): New method.
16287 (selftest::test_add_location_if_nearby): New test function.
16288 (selftest::diagnostic_show_locus_c_tests): Call it.
16289 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
16292 2017-07-11 Tom de Vries <tom@codesourcery.com>
16294 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
16295 (bb_first_real_insn): New function.
16296 (nvptx_single): Add extra initialization of broadcasted condition
16299 2017-07-11 Nathan Sidwell <nathan@acm.org>
16301 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
16303 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
16305 * doc/extend.texi (AVR Function Attributes): Remove weblink to
16306 Binutils doc as TEXI will mess them up.
16307 * doc/invoke.texi (AVR Options): Same here.
16309 2017-07-11 Daniel Cederman <cederman@gaisler.com>
16311 * config/sparc/sparc.opt (mfix-ut700): New option.
16312 (mfix-gr712rc): Likewise.
16313 (sparc_fix_b2bst): New variable.
16314 * doc/invoke.texi (SPARC options): Document them.
16315 (ARM options): Fix warnings.
16316 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
16317 instructions to prevent sequences that can trigger the store-store
16318 errata for certain LEON3FT processors.
16319 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
16320 (sparc_option_override): Set sparc_fix_b2bst appropriately.
16321 * config/sparc/sparc.md (fix_b2bst): New attribute.
16322 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
16324 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
16327 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
16329 (*rsqrtsf2_sse): Ditto.
16331 (div<mode>3): Macroize insn from divdf3 and divsf3
16332 using MODEF mode iterator.
16334 2017-07-10 Martin Sebor <msebor@redhat.com>
16336 PR tree-optimization/80397
16337 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
16338 instead of testing for equality to INTEGER_TYPE.
16340 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
16342 * config.gcc: Remove uclibc from arc target spec.
16344 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
16346 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
16348 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16351 * lto-wrapper.c (remove_option): New function.
16352 (merge_and_complain): Merge PIC/PIE options more realistically.
16354 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16356 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
16360 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
16361 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
16362 * config.in: Regenerate.
16363 * configure: Regenerate.
16364 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
16365 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
16366 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
16367 (TARGET_GASISR_PROLOGUES): ...target mask.
16368 * common/config/avr/avr-common.c
16369 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
16370 Set -mgas-isr-prologues.
16371 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
16372 INSERT_PASS_BEFORE for it.
16373 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
16374 * config/avr/avr.c (avr_option_override)
16375 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
16376 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
16377 (avr_attribute_table) <no_gccisr>: Add new function attribute.
16378 (avr_set_current_function) <is_no_gccisr>: Init machine field.
16379 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
16381 (make_avr_pass_pre_proep): New function.
16382 (emit_push_sfr) <treg>: Add argument to function and use it
16383 instead of TMP_REG.
16384 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
16385 and set machine->gasisr.yes.
16386 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
16387 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
16388 __gcc_isr.n_pushed to .L__stack_usage.
16389 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
16390 (avr_asm_final_postscan_insn): ...this new static function.
16391 * config/avr/avr.h (machine_function)
16392 <is_no_gccisr, use_L__stack_usage>: New fields.
16393 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
16394 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
16395 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
16396 (gasisr, *gasisr): New expander and insn.
16397 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
16398 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
16399 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
16401 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
16403 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
16406 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16408 Move jump-tables out of .text again.
16411 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
16412 (ASM_OUTPUT_ADDR_VEC): New function.
16413 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
16414 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
16415 INSN_ADDRESSes as asm comment.
16416 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
16417 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
16418 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
16419 * config/avr/avr.md (*tablejump): Adjust comment.
16420 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
16421 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
16423 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
16424 (avr_output_addr_vec): New proto.
16425 (avr_log_t) <insn_addresses>: New field.
16427 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
16430 * config/i386/i386.c (ix86_function_arg_advance): Set
16431 outgoing_args_on_stack to true if there are outgoing arguments
16433 (ix86_function_arg): Likewise.
16434 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
16435 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
16436 * config/i386/i386.h (machine_function): Add
16437 outgoing_args_on_stack.
16439 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
16441 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
16442 supporting pthreds.
16443 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
16445 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
16447 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
16448 (REAL_H): Remove $(MACHMODE_H).
16449 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
16451 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
16452 $(MACHMODE_H) and double-int.h.
16453 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
16455 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
16458 2017-07-07 Andrew Pinski <apinski@cavium.com>
16460 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
16461 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
16463 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16465 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
16466 Add warning if GCC was not configured to link against a GLIBC that
16467 exports the hardware capability bits.
16468 (make_resolver_func): Make resolver function private and not a
16469 COMDAT function. Create the name with clone_function_name instead
16470 of make_unique_name.
16473 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
16474 correct operand in doing the split.
16476 2017-07-07 Carl Love <cel@us.ibm.com>
16478 * config/rs6000/rs6000-c: Add support for built-in function
16479 vector unsigned short vec_pack_to_short_fp32 (vector float,
16481 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
16482 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
16483 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
16484 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
16485 (convert_4f32_8i16): Add define_expand.
16486 * doc/extend.texi: Update the built-in documentation file for the
16487 new built-in function.
16489 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16491 * config/sparc/m8.md: New file.
16492 * config/sparc/sparc.md: Include m8.md.
16494 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16496 * config/sparc/sparc.opt: New option -mvis4b.
16497 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
16498 (sparc_option_override): Handle VIS4B.
16499 (enum sparc_builtins): Define
16500 SPARC_BUILTIN_DICTUNPACK{8,16,32},
16501 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
16502 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
16503 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
16504 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
16505 (check_constant_argument): New function.
16506 (sparc_vis_init_builtins): Define builtins
16507 __builtin_vis_dictunpack{8,16,32},
16508 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
16509 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
16510 __builtin_vis_fpcmpde{8,16,32}shl and
16511 __builtin_vis_fpcmpur{8,16,32}shl.
16512 (sparc_expand_builtin): Check that the constant operands to
16513 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
16514 constant and in range.
16515 * config/sparc/sparc-c.c (sparc_target_macros): Handle
16517 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
16518 (SPARC_IMM5_P): Likewise.
16519 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
16520 (enabled): Handle vis4b.
16521 (UNSPEC_DICTUNPACK): New unspec.
16522 (UNSPEC_FPCMPSHL): Likewise.
16523 (UNSPEC_FPUCMPSHL): Likewise.
16524 (UNSPEC_FPCMPDESHL): Likewise.
16525 (UNSPEC_FPCMPURSHL): Likewise.
16526 (cpu_feature): New CPU feature `vis4b'.
16527 (dictunpack{8,16,32}): New insns.
16528 (FPCSMODE): New mode iterator.
16529 (fpcscond): New code iterator.
16530 (fpcsucond): Likewise.
16531 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
16532 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
16533 (fpcmpde{8,16,32}{si,di}shl): Likewise.
16534 (fpcmpur{8,16,32}{si,di}shl): Likewise.
16535 * config/sparc/constraints.md: Define constraints `q' for unsigned
16536 2-bit integer constants and `t' for unsigned 5-bit integer
16538 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
16540 (imm5_operand_dictunpack16): Likewise.
16541 (imm5_operand_dictunpack32): Likewise.
16542 (imm2_operand): Likewise.
16543 * doc/invoke.texi (SPARC Options): Document -mvis4b.
16544 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
16545 ditunpack* and fpcmp*shl builtins.
16547 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16549 * config.gcc: Handle m8 in --with-{cpu,tune} options.
16550 * config.in: Add HAVE_AS_SPARC6 define.
16551 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
16553 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
16555 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
16556 (CPP_CPU_SPEC): Handle m8.
16557 (ASM_CPU_SPEC): Likewise.
16558 * config/sparc/sparc-opts.h (enum processor_type): Add
16560 * config/sparc/sparc.c (m8_costs): New struct.
16561 (sparc_option_override): Handle TARGET_CPU_m8.
16562 (sparc32_initialize_trampoline): Likewise.
16563 (sparc64_initialize_trampoline): Likewise.
16564 (sparc_issue_rate): Likewise.
16565 (sparc_register_move_cost): Likewise.
16566 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
16567 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
16568 (ASM_CPU64_DEFAULT_SPEC): Likewise.
16569 (CPP_CPU_SPEC): Handle M8.
16570 (ASM_CPU_SPEC): Likewise.
16571 (AS_M8_FLAG): Define.
16572 * config/sparc/sparc.md: Add m8 to the cpu attribute.
16573 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
16574 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
16576 * configure: Regenerate.
16577 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
16580 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16582 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
16584 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
16585 ("*movdi_insn_sp32"): Do not set v3pipe.
16586 ("*movsi_insn"): Likewise.
16587 ("*movdi_insn_sp64"): Likewise.
16588 ("*movsf_insn"): Likewise.
16589 ("*movdf_insn_sp32"): Likewise.
16590 ("*movdf_insn_sp64"): Likewise.
16591 ("*zero_extendsidi2_insn_sp64"): Likewise.
16592 ("*sign_extendsidi2_insn"): Likewise.
16593 ("*mov<VM32:mode>_insn"): Likewise.
16594 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16595 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16596 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16597 ("<vlop:code><VL:mode>3"): Likewise.
16598 ("*not_<vlop:code><VL:mode>3"): Likewise.
16599 ("*nand<VL:mode>_vis"): Likewise.
16600 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
16601 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
16602 ("one_cmpl<VL:mode>2"): Likewise.
16603 ("faligndata<VM64:mode>_vis"): Likewise.
16604 ("alignaddrsi_vis"): Likewise.
16605 ("alignaddrdi_vis"): Likweise.
16606 ("alignaddrlsi_vis"): Likewise.
16607 ("alignaddrldi_vis"): Likewise.
16608 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16609 ("bmaskdi_vis"): Likewise.
16610 ("bmasksi_vis"): Likewise.
16611 ("bshuffle<VM64:mode>_vis"): Likewise.
16612 ("cmask8<P:mode>_vis"): Likewise.
16613 ("cmask16<P:mode>_vis"): Likewise.
16614 ("cmask32<P:mode>_vis"): Likewise.
16615 ("pdistn<P:mode>_vis"): Likewise.
16616 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16618 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16620 * config/sparc/sparc.md ("subtype"): New insn attribute.
16621 ("*wrgsr_sp64"): Set insn subtype.
16622 ("*rdgsr_sp64"): Likewise.
16623 ("alignaddrsi_vis"): Likewise.
16624 ("alignaddrdi_vis"): Likewise.
16625 ("alignaddrlsi_vis"): Likewise.
16626 ("alignaddrldi_vis"): Likewise.
16627 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16628 ("fexpand_vis"): Likewise.
16629 ("fpmerge_vis"): Likewise.
16630 ("faligndata<VM64:mode>_vis"): Likewise.
16631 ("bshuffle<VM64:mode>_vis"): Likewise.
16632 ("cmask8<P:mode>_vis"): Likewise.
16633 ("cmask16<P:mode>_vis"): Likewise.
16634 ("cmask32<P:mode>_vis"): Likewise.
16635 ("fchksm16_vis"): Likewise.
16636 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
16637 ("fmean16_vis"): Likewise.
16638 ("fp<plusminus_insn>64_vis"): Likewise.
16639 ("<plusminus_insn>v8qi3"): Likewise.
16640 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16641 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
16642 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
16643 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
16644 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
16645 ("*movqi_insn"): Likewise.
16646 ("*movhi_insn"): Likewise.
16647 ("*movsi_insn"): Likewise.
16648 ("movsi_pic_gotdata_op"): Likewise.
16649 ("*movdi_insn_sp32"): Likewise.
16650 ("*movdi_insn_sp64"): Likewise.
16651 ("movdi_pic_gotdata_op"): Likewise.
16652 ("*movsf_insn"): Likewise.
16653 ("*movdf_insn_sp32"): Likewise.
16654 ("*movdf_insn_sp64"): Likewise.
16655 ("*zero_extendhisi2_insn"): Likewise.
16656 ("*zero_extendqihi2_insn"): Likewise.
16657 ("*zero_extendqisi2_insn"): Likewise.
16658 ("*zero_extendqidi2_insn"): Likewise.
16659 ("*zero_extendhidi2_insn"): Likewise.
16660 ("*zero_extendsidi2_insn_sp64"): Likewise.
16661 ("ldfsr"): Likewise.
16662 ("prefetch_64"): Likewise.
16663 ("prefetch_32"): Likewise.
16664 ("tie_ld32"): Likewise.
16665 ("tie_ld64"): Likewise.
16666 ("*tldo_ldub_sp32"): Likewise.
16667 ("*tldo_ldub1_sp32"): Likewise.
16668 ("*tldo_ldub2_sp32"): Likewise.
16669 ("*tldo_ldub_sp64"): Likewise.
16670 ("*tldo_ldub1_sp64"): Likewise.
16671 ("*tldo_ldub2_sp64"): Likewise.
16672 ("*tldo_ldub3_sp64"): Likewise.
16673 ("*tldo_lduh_sp32"): Likewise.
16674 ("*tldo_lduh1_sp32"): Likewise.
16675 ("*tldo_lduh_sp64"): Likewise.
16676 ("*tldo_lduh1_sp64"): Likewise.
16677 ("*tldo_lduh2_sp64"): Likewise.
16678 ("*tldo_lduw_sp32"): Likewise.
16679 ("*tldo_lduw_sp64"): Likewise.
16680 ("*tldo_lduw1_sp64"): Likewise.
16681 ("*tldo_ldx_sp64"): Likewise.
16682 ("*mov<VM32:mode>_insn"): Likewise.
16683 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16684 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16686 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16688 * config/sparc/sparc.md ("type"): New insn type viscmp.
16689 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
16691 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
16692 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
16693 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16694 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
16696 ("n7_vis_logical_11cycle"): Likewise.
16697 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
16698 * config/sparc/niagara2.md ("niag3_vis": Likewise.
16699 * config/sparc/niagara.md ("niag_vis"): Likewise.
16700 * config/sparc/ultra3.md ("us3_fga"): Likewise.
16701 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
16703 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16705 * config/sparc/sparc.md: New instruction type `bmask'.
16706 (bmaskdi_vis): Use the `bmask' type.
16707 (bmasksi_vis): Likewise.
16708 * config/sparc/ultra3.md (us3_array): Likewise.
16709 * config/sparc/niagara7.md (n7_array): Likewise.
16710 * config/sparc/niagara4.md (n4_array): Likewise.
16711 * config/sparc/niagara2.md (niag2_vis): Likewise.
16712 (niag3_vis): Likewise.
16713 * config/sparc/niagara.md (niag_vis): Likewise.
16715 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16717 * ipa-comdats.c: Remove optimize check from gate.
16718 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
16719 for functions not optimized.
16720 (ipa_fn_summary_read): Skip optimize check.
16721 (ipa_fn_summary_write): Likewise.
16722 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
16724 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
16726 (can_inline_edge_p): Check flag_pcc_struct_return for match.
16727 (check_callers): Give up on caller which is not optimized.
16728 (inline_small_functions): Likewise.
16729 (ipa_inline): Do not give up when not optimizing.
16730 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
16731 away unoptimizes cdtors.
16732 (whole_program_function_and_variable_visibility): Do
16733 ipa_discover_readonly_nonaddressable_vars in LTO mode.
16734 * ipa.c (process_references): Do not check optimize.
16735 (symbol_table::remove_unreachable_nodes): Update optimize check.
16736 (set_writeonly_bit): Update optimize check.
16737 (pass_ipa_cdtor_merge::gate): Do not check optimize.
16738 (pass_ipa_single_use::gate): Remove.
16740 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16742 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
16743 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
16744 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
16745 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
16746 permute_load, permute_store, adjust_extract, adjust_splat,
16747 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
16748 replace_swap_with_copy, dump_swap_insn_table,
16749 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
16750 recombine_lvx_pattern, recombine_stvx_pattern,
16751 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
16752 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
16753 to file rs6000-p8swap.c.
16754 * config/rs6000/rs6000-p8swap.c: New file.
16755 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
16756 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
16757 and rs6000*-*-* targets.
16759 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16761 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
16763 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16765 * lto-wrapper.c (merge_and_complain): Do not merge
16766 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
16767 fsigned_zeros, ftrapping_math, fwrapv.
16768 (append_compiler_options): Do not track these options.
16769 (append_linker_options): Likewie
16771 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16773 * cgraphunit.c (cgraph_node::finalize_function): When
16774 !flag_toplevel_reorde set no_reorder flag.
16775 (varpool_node::finalize_decl): Likewise.
16776 (symbol_table::compile): Drop no toplevel reorder path.
16778 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16780 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
16781 edges; zero probability is not better than uninitialized.
16783 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
16785 * asan.h (asan_sanitize_allocas_p): Declare.
16786 * asan.c (asan_sanitize_allocas_p): New function.
16787 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
16788 (handle_builtin_alloca): Likewise.
16789 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
16790 if !asan_sanitize_allocas_p.
16791 * params.def (asan-instrument-allocas): Add new option.
16792 * params.h (ASAN_PROTECT_ALLOCAS): Define.
16793 * opts.c (common_handle_option): Disable allocas sanitization for
16796 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
16798 * asan.c: Include gimple-fold.h.
16799 (get_last_alloca_addr): New function.
16800 (handle_builtin_stackrestore): Likewise.
16801 (handle_builtin_alloca): Likewise.
16802 (asan_emit_allocas_unpoison): Likewise.
16803 (get_mem_refs_of_builtin_call): Add new parameter, remove const
16804 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
16805 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
16806 (instrument_builtin_call): Pass gimple iterator to
16807 get_mem_refs_of_builtin_call.
16808 (last_alloca_addr): New global.
16809 * asan.h (asan_emit_allocas_unpoison): Declare.
16810 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
16811 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
16812 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
16813 if function calls alloca.
16814 * gimple-fold.c (replace_call_with_value): Remove static keyword.
16815 * gimple-fold.h (replace_call_with_value): Declare.
16816 * internal-fn.c: Include asan.h.
16817 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
16818 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
16820 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16822 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
16823 (C_SELFTEST_FLAGS): New.
16824 (CPP_SELFTEST_FLAGS): New.
16825 (SELFTEST_DEPS): New, from deps of s-selftest.
16826 (C_SELFTEST_DEPS): New, from deps of s-selftest.
16827 (CPP_SELFTEST_DEPS): New.
16828 (selftest): Add dependency on s-selftest-c++.
16829 (s-selftest): Rename to...
16830 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
16831 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
16832 than SELFTEST_FLAGS.
16833 (selftest-gdb): Rename to...
16834 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
16836 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
16837 (selftest-valgrind): Rename to...
16838 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
16840 (selftest-valgrind): Reintroduce as an alias for
16841 selftest-c-valgrind.
16842 (s-selftest-c++): New.
16843 (selftest-c++-gdb): New.
16844 (selftest-c++-valgrind): New.
16846 2017-07-06 Olivier Hainque <hainque@adacore.com>
16848 * gcc.c (process_command): When deciding if undefined variables
16849 should be ignored when processing specs, accept "gcc -v" as well.
16851 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16853 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
16854 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
16856 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
16858 * config/arm/arm-cpus.in (armv8-r): Add new entry.
16859 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
16860 * config/arm/arm-tables.opt: Regenerate.
16861 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
16863 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
16865 2017-07-06 Carl Love <cel@us.ibm.com>
16867 * ChangeLog: Clean up from mid air collision
16869 2017-07-06 Carl Love <cel@us.ibm.com>
16871 * config/rs6000/rs6000-c.c: Add support for built-in functions
16872 vector signed int vec_subc (vector signed int, vector signed int);
16873 vector signed __int128 vec_subc (vector signed __int128,
16874 vector signed __int128);
16875 vector unsigned __int128 vec_subc (vector unsigned __int128,
16876 vector unsigned __int128);
16877 vector signed int vec_sube (vector signed int, vector signed int,
16878 vector signed int);
16879 vector unsigned int vec_sube (vector unsigned int,
16880 vector unsigned int,
16881 vector unsigned int);
16882 vector signed __int128 vec_sube (vector signed __int128,
16883 vector signed __int128,
16884 vector signed__int128);
16885 vector unsigned __int128 vec_sube (vector unsigned __int128,
16886 vector unsigned __int128,
16887 vector unsigned __int128);
16888 vector signed int vec_subec (vector signed int, vector signed int,
16889 vector signed int);
16890 vector unsigned int vec_subec (vector unsigned int,
16891 vector unsigned int,
16892 vector unsigned int);
16893 vector signed __int128 vec_subec (vector signed __int128,
16894 vector signed __int128,
16895 vector signed__int128);
16896 vector unsigned __int128 vec_subec (vector unsigned __int128,
16897 vector unsigned __int128,
16898 vector unsigned __int128);
16899 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
16900 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
16901 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
16902 BU_ALTIVEC_OVERLOAD_X definitions.
16903 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
16904 * doc/extend.texi: Update the built-in documentation file for the new
16905 built-in functions.
16907 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16910 * diagnostic-show-locus.c (layout::layout): Use start and finish
16911 spelling location for the start and finish of each range.
16912 * genmatch.c (linemap_client_expand_location_to_spelling_point):
16913 Add unused aspect param.
16914 * input.c (expand_location_1): Add "aspect" param, and use it
16915 to access the correct part of the location.
16916 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
16918 (expand_location_to_spelling_point): Likewise.
16919 (linemap_client_expand_location_to_spelling_point): Add "aspect"
16920 param, and pass it to expand_location_1.
16922 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
16924 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
16925 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
16926 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
16927 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
16928 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
16929 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
16930 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
16931 _mm_maskz_getmant_ss): New intrinsics.
16932 (__builtin_ia32_getexpss128_mask): Changed to ...
16933 __builtin_ia32_getexpss128_round ... this.
16934 (__builtin_ia32_getexpsd128_mask): Changed to ...
16935 __builtin_ia32_getexpsd128_round ... this.
16936 * config/i386/i386-builtin-types.def
16937 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
16938 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
16939 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
16940 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
16941 __builtin_ia32_getmantss_mask_round): New builtins.
16942 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
16943 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
16944 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
16945 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
16946 * config/i386/sse.md
16947 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
16948 avx512f_sgetexp<mode><mask_scalar_name>
16949 <round_saeonly_scalar_name> ... this.
16950 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
16951 %0, %1, %2<round_saeonly_op3>}): Changed to ...
16952 vgetexp<ssescalarmodesuffix>
16953 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16954 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
16955 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
16956 avx512f_vgetmant<mode><mask_scalar_name>
16957 <round_saeonly_scalar_name> ... this.
16958 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
16959 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
16960 vgetmant<ssescalarmodesuffix>
16961 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
16962 %0<mask_scalar_operand4>, %1, %2
16963 <round_saeonly_scalar_mask_op4>, %3} ... this.
16964 * config/i386/subst.md (mask_scalar_operand4,
16965 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
16966 round_saeonly_scalar_nimm_predicate): New subst attributes.
16968 2017-07-06 Julia Koval <julia.koval@intel.com>
16970 * config/i386/i386.c (ix86_erase_embedded_rounding):
16971 Remove code for old rounding pattern.
16973 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
16975 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
16977 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
16979 * doc/sourcebuild.texi (Test Directives, Variants of
16980 dg-require-support): Add documentation for dg-require-stack-check.
16982 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
16984 * config/i386/subst.md (mask_scalar, round_scalar,
16985 round_saeonly_scalar): New meta-templates.
16986 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
16987 round_scalar_mask_operand3, round_scalar_mask_op3,
16988 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
16989 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
16990 round_saeonly_scalar_constraint,
16991 round_saeonly_scalar_prefix): New subst attribute.
16992 * config/i386/sse.md
16993 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
16994 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
16995 <round_scalar_name> ... this.
16996 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
16997 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
16998 <round_scalar_name> ... this.
16999 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
17000 <sse>_vm<code><mode>3<mask_scalar_name>
17001 <round_saeonly_scalar_name> ... this.
17002 (v<plusminus_mnemonic><ssescalarmodesuffix>
17003 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
17004 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
17005 v<plusminus_mnemonic><ssescalarmodesuffix>
17006 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17007 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
17008 (v<multdiv_mnemonic><ssescalarmodesuffix>
17009 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
17010 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
17011 v<multdiv_mnemonic><ssescalarmodesuffix>
17012 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17013 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
17014 (v<maxmin_float><ssescalarmodesuffix>
17015 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
17016 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
17017 v<maxmin_float><ssescalarmodesuffix>
17018 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17019 %0<mask_scalar_operand3>, %1, %<iptr>2
17020 <round_saeonly_scalar_mask_op3>} ... this.
17022 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
17024 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
17025 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
17027 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
17028 Alan Hayward <alan.hayward@arm.com>
17029 David Sherwood <david.sherwood@arm.com>
17031 * combine.c (simplify_if_then_else): Remove "enum" before
17033 * compare-elim.c (can_eliminate_compare): Likewise.
17034 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
17036 (aarch64_lookup_simd_builtin_type): Likewise.
17037 (aarch64_simd_builtin_type): Likewise.
17038 (aarch64_init_simd_builtin_types): Likewise.
17039 (aarch64_simd_expand_args): Likewise.
17040 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
17042 (aarch64_reverse_mask): Likewise.
17043 (aarch64_simd_emit_reg_reg_move): Likewise.
17044 (aarch64_gen_adjusted_ldpstp): Likewise.
17045 (aarch64_ccmp_mode_to_code): Likewise.
17046 (aarch64_operands_ok_for_ldpstp): Likewise.
17047 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17048 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
17050 (aarch64_min_divisions_for_recip_mul): Likewise.
17051 (aarch64_reassociation_width): Likewise.
17052 (aarch64_get_condition_code_1): Likewise.
17053 (aarch64_simd_emit_reg_reg_move): Likewise.
17054 (aarch64_simd_attr_length_rglist): Likewise.
17055 (aarch64_reverse_mask): Likewise.
17056 (aarch64_operands_ok_for_ldpstp): Likewise.
17057 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17058 (aarch64_gen_adjusted_ldpstp): Likewise.
17059 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
17061 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
17062 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
17063 (arm_lookup_simd_builtin_type): Likewise.
17064 (arm_simd_builtin_type): Likewise.
17065 (arm_init_simd_builtin_types): Likewise.
17066 (arm_expand_builtin_args): Likewise.
17067 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
17068 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
17069 (ft32_setup_incoming_varargs): Likewise.
17070 (ft32_function_arg): Likewise.
17071 (ft32_function_arg_advance): Likewise.
17072 (ft32_pass_by_reference): Likewise.
17073 (ft32_arg_partial_bytes): Likewise.
17074 (ft32_valid_pointer_mode): Likewise.
17075 (ft32_addr_space_pointer_mode): Likewise.
17076 (ft32_addr_space_legitimate_address_p): Likewise.
17077 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
17079 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
17080 (ix86_emit_outlined_ms2sysv_restore): Likewise.
17081 (iamcu_alignment): Likewise.
17082 (canonicalize_vector_int_perm): Likewise.
17083 (ix86_noce_conversion_profitable_p): Likewise.
17084 (ix86_mpx_bound_mode): Likewise.
17085 (ix86_operands_ok_for_move_multiple): Likewise.
17086 * config/microblaze/microblaze-protos.h
17087 (microblaze_expand_conditional_branch_reg): Likewise.
17088 * config/microblaze/microblaze.c
17089 (microblaze_expand_conditional_branch_reg): Likewise.
17090 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
17092 (rs6000_reassociation_width): Likewise.
17093 (rs6000_invalid_binary_op): Likewise.
17094 (fusion_p9_p): Likewise.
17095 (emit_fusion_p9_load): Likewise.
17096 (emit_fusion_p9_store): Likewise.
17097 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
17099 (riscv_hard_regno_mode_ok_p): Likewise.
17100 (riscv_address_insns): Likewise.
17101 (riscv_split_symbol): Likewise.
17102 (riscv_legitimize_move): Likewise.
17103 (riscv_function_value): Likewise.
17104 (riscv_hard_regno_nregs): Likewise.
17105 (riscv_expand_builtin): Likewise.
17106 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
17107 (riscv_build_integer): Likewise.
17108 (riscv_split_integer): Likewise.
17109 (riscv_legitimate_constant_p): Likewise.
17110 (riscv_cannot_force_const_mem): Likewise.
17111 (riscv_regno_mode_ok_for_base_p): Likewise.
17112 (riscv_valid_base_register_p): Likewise.
17113 (riscv_valid_offset_p): Likewise.
17114 (riscv_valid_lo_sum_p): Likewise.
17115 (riscv_classify_address): Likewise.
17116 (riscv_legitimate_address_p): Likewise.
17117 (riscv_address_insns): Likewise.
17118 (riscv_load_store_insns): Likewise.
17119 (riscv_force_binary): Likewise.
17120 (riscv_split_symbol): Likewise.
17121 (riscv_force_address): Likewise.
17122 (riscv_legitimize_address): Likewise.
17123 (riscv_move_integer): Likewise.
17124 (riscv_legitimize_const_move): Likewise.
17125 (riscv_legitimize_move): Likewise.
17126 (riscv_address_cost): Likewise.
17127 (riscv_subword): Likewise.
17128 (riscv_output_move): Likewise.
17129 (riscv_canonicalize_int_order_test): Likewise.
17130 (riscv_emit_int_order_test): Likewise.
17131 (riscv_function_arg_boundary): Likewise.
17132 (riscv_pass_mode_in_fpr_p): Likewise.
17133 (riscv_pass_fpr_single): Likewise.
17134 (riscv_pass_fpr_pair): Likewise.
17135 (riscv_get_arg_info): Likewise.
17136 (riscv_function_arg): Likewise.
17137 (riscv_function_arg_advance): Likewise.
17138 (riscv_arg_partial_bytes): Likewise.
17139 (riscv_function_value): Likewise.
17140 (riscv_pass_by_reference): Likewise.
17141 (riscv_setup_incoming_varargs): Likewise.
17142 (riscv_print_operand): Likewise.
17143 (riscv_elf_select_rtx_section): Likewise.
17144 (riscv_save_restore_reg): Likewise.
17145 (riscv_for_each_saved_reg): Likewise.
17146 (riscv_register_move_cost): Likewise.
17147 (riscv_hard_regno_mode_ok_p): Likewise.
17148 (riscv_hard_regno_nregs): Likewise.
17149 (riscv_class_max_nregs): Likewise.
17150 (riscv_memory_move_cost): Likewise.
17151 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
17152 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
17153 (rl78_addr_space_address_mode): Likewise.
17154 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17156 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17157 (rs6000_reassociation_width): Likewise.
17158 (rs6000_invalid_binary_op): Likewise.
17159 (fusion_p9_p): Likewise.
17160 (emit_fusion_p9_load): Likewise.
17161 (emit_fusion_p9_store): Likewise.
17162 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
17163 (ok_for_simple_move_operands): Likewise.
17164 (ok_for_simple_move_strict_operands): Likewise.
17165 (ok_for_simple_arith_logic_operands): Likewise.
17166 (visium_legitimize_reload_address): Likewise.
17167 (visium_select_cc_mode): Likewise.
17168 (output_cbranch): Likewise.
17169 (visium_split_double_move): Likewise.
17170 (visium_expand_copysign): Likewise.
17171 (visium_expand_int_cstore): Likewise.
17172 (visium_expand_fp_cstore): Likewise.
17173 * config/visium/visium.c (visium_pass_by_reference): Likewise.
17174 (visium_function_arg): Likewise.
17175 (visium_function_arg_advance): Likewise.
17176 (visium_libcall_value): Likewise.
17177 (visium_setup_incoming_varargs): Likewise.
17178 (visium_legitimate_constant_p): Likewise.
17179 (visium_legitimate_address_p): Likewise.
17180 (visium_legitimize_address): Likewise.
17181 (visium_secondary_reload): Likewise.
17182 (visium_register_move_cost): Likewise.
17183 (visium_memory_move_cost): Likewise.
17184 (prepare_move_operands): Likewise.
17185 (ok_for_simple_move_operands): Likewise.
17186 (ok_for_simple_move_strict_operands): Likewise.
17187 (ok_for_simple_arith_logic_operands): Likewise.
17188 (visium_function_value_1): Likewise.
17189 (rtx_ok_for_offset_p): Likewise.
17190 (visium_legitimize_reload_address): Likewise.
17191 (visium_split_double_move): Likewise.
17192 (visium_expand_copysign): Likewise.
17193 (visium_expand_int_cstore): Likewise.
17194 (visium_expand_fp_cstore): Likewise.
17195 (visium_split_cstore): Likewise.
17196 (visium_select_cc_mode): Likewise.
17197 (visium_split_cbranch): Likewise.
17198 (output_cbranch): Likewise.
17199 (visium_print_operand_address): Likewise.
17200 * expmed.c (flip_storage_order): Likewise.
17201 * expmed.h (emit_cstore): Likewise.
17202 (flip_storage_order): Likewise.
17203 * genrecog.c (validate_pattern): Likewise.
17204 * hsa-gen.c (gen_hsa_addr): Likewise.
17205 * internal-fn.c (expand_arith_overflow): Likewise.
17206 * ira-color.c (allocno_copy_cost_saving): Likewise.
17207 * lra-assigns.c (find_hard_regno_for_1): Likewise.
17208 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
17209 (process_invariant_for_inheritance): Likewise.
17210 * lra-eliminations.c (move_plus_up): Likewise.
17211 * omp-low.c (lower_oacc_reductions): Likewise.
17212 * simplify-rtx.c (simplify_subreg): Likewise.
17213 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
17214 (TARGET_CHKP_BOUND_MODE): Likewise..
17215 * targhooks.c (default_chkp_bound_mode): Likewise.
17216 (default_setup_incoming_vararg_bounds): Likewise.
17217 * targhooks.h (default_chkp_bound_mode): Likewise.
17218 (default_setup_incoming_vararg_bounds): Likewise.
17219 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
17220 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
17221 (have_whole_vector_shift): Likewise.
17222 * tree-vect-stmts.c (vectorizable_load): Likewise.
17223 * doc/tm.texi: Regenerate.
17225 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17227 Graceful degrade if Binutils PR21472 is not available.
17230 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
17231 .rodata in flash test fails.
17232 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
17233 * confgure: Regenerate.
17234 * config.in: Regenerate.
17235 * config/avr/avr.c (avr_asm_named_section)
17236 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
17237 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
17238 (avr_asm_init_sections): Same.
17240 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17242 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
17243 (fma<VH:mode>4_intrinsic): Likewise.
17244 (*fmsub<VCVTF:mode>4): Likewise.
17245 (*fmsub<VH:mode>4_intrinsic): Likewise.
17247 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17250 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
17251 Don't depend on "optimize > 0".
17252 (out_movhi_r_mr, out_movqi_mr_r): Same.
17253 (out_movhi_mr_r, out_movqi_r_mr): Same.
17254 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
17255 io_address_operand on "optimize > 0".
17257 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17259 * tree-loop-distribution.c: Add general explanantion on the pass.
17260 (generate_loops_for_partition): Mark distributed loop.
17261 (pg_add_dependence_edges): New parameter. Handle alias data
17262 dependence specially and record it in the parameter if asked.
17263 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
17264 (init_partition_graph_vertices, add_partition_graph_edge): New.
17265 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
17266 (free_partition_graph_vdata, build_partition_graph): New.
17267 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
17268 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
17269 (data_ref_segment_size, latch_dominated_by_data_ref): New.
17270 (compute_alias_check_pairs, version_loop_by_alias_check): New.
17271 (version_for_distribution_p, finalize_partitions): New.
17272 (distribute_loop): Handle alias data dependence specially. Factor
17273 out loop fusion code as functions and call these functions.
17275 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17277 * tree-loop-distribution.c (classify_partition): New parameter and
17278 better handle reduction statement.
17279 (rdg_build_partitions): Revise comment.
17280 (distribute_loop): Compute statements in all partitions and pass it
17281 to classify_partition.
17283 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17285 * tree-loop-distribution.c (enum partition_type): New.
17286 (struct partition): New field type.
17287 (partition_merge_into): Add parameter. Update partition type.
17288 (data_dep_in_cycle_p, update_type_for_merge): New functions.
17289 (build_rdg_partition_for_vertex): Compute partition type.
17290 (rdg_build_partitions): Dump partition type.
17291 (distribute_loop): Update calls to partition_merge_into.
17293 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17295 * tree-loop-distribution.c (struct ddr_hasher): New.
17296 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
17298 (classify_partition): Call get_data_dependence.
17299 (pg_add_dependence_edges): Ditto.
17300 (distribute_loop): Release data dependence hash table.
17302 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17304 * tree-loop-distribution.c (ref_base_address): Delete.
17305 (similar_memory_accesses): Rename ...
17306 (share_memory_accesses): ... to this. Check if partitions access
17307 the same memory reference.
17308 (distribute_loop): Call share_memory_accesses.
17310 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17312 * tree-loop-distribution.c (struct partition): New field recording
17313 its data reference.
17314 (partition_alloc, partition_free): Init and release data refs.
17315 (partition_merge_into): Merge data refs.
17316 (build_rdg_partition_for_vertex): Collect data refs for partition.
17317 (pg_add_dependence_edges): Change parameters from vector to bitmap.
17319 (distribute_loop): Remve data refs from vertice data of partition
17322 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17324 * tree-loop-distribution.c (params.h): Include header file.
17325 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
17326 (datarefs_vec): New global var.
17327 (create_rdg_vertices): Use datarefs_vec directly.
17328 (free_rdg): Don't free data references.
17329 (build_rdg): Update use. Don't free data references.
17330 (distribute_loop): Compute global variable for data references.
17331 Bail out if there are too many data references.
17333 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17335 * tree-loop-distribution.c (loop_nest): New global var.
17336 (build_rdg): Use loop directly, rather than loop nest.
17337 (pg_add_dependence_edges): Remove loop nest parameter. Use global
17339 (distribute_loop): Compute global variable loop nest. Update use.
17341 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17343 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
17344 (partition_merge_into): New parameter. Dump reason for fusion.
17345 (distribute_loop): Update use of partition_merge_into.
17347 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17349 * tree-loop-distribution.c (bb_top_order_index): New.
17350 (bb_top_order_index_size, bb_top_order_cmp): New.
17351 (stmts_from_loop): Use topological order.
17352 (pass_loop_distribution::execute): Compute and release topological
17353 order for basic blocks.
17355 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17357 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
17360 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17362 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
17363 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
17364 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
17365 * internal-fn.def (LOOP_DIST_ALIAS): New.
17366 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
17367 (fold_loop_internal_call): ... this.
17368 (vect_loop_dist_alias_call): New function.
17369 (set_uid_loop_bbs): Call fold_loop_internal_call.
17370 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
17373 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
17376 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
17377 Require dead FLAGS_REG at the beginning of a peephole.
17379 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
17382 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
17383 arguments in the call to __builtin_ia32_sbb_u32.
17384 (_subborrow_u64): Swap _X and _Y arguments in the call to
17385 __builtin_ia32_sbb_u64.
17387 2017-07-04 Jakub Jelinek <jakub@redhat.com>
17390 * tree-vrp.c (compare_assert_loc): Turn into a function template
17391 with stable template parameter. Only test if a->e is NULL,
17392 !a->e == !b->e has been verified already. Use e == NULL or
17393 e != NULL instead of e or ! e tests. If stable is true, don't use
17394 iterative_hash_expr, on the other side allow a or b or both NULL
17395 and sort the NULLs last.
17396 (process_assert_insertions): Sort using compare_assert_loc<false>
17397 instead of compare_assert_loc, later sort using
17398 compare_assert_loc<true> before calling process_assert_insertions_for
17399 in a loop. Use break instead of continue once seen NULL pointer.
17401 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
17403 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17404 Cortex-R7 and Cortex-R8 processors.
17406 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17408 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
17409 uninitialized while src is not.
17411 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
17413 * common/config/arm/arm-common.c: Adjust include path for
17415 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
17416 (arm-cpu.h): Create in build directory. Adjust dependency rules.
17417 (arm-cpu-data.h): Likewise.
17418 (arm-cpu-cdata.h): Likewise.
17419 * config/arm/arm-cpu.h: Delete.
17420 * config/arm/arm-cpu-cdata.h: Delete.
17421 * config/arm/arm-cpu-data.h: Delete.
17423 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
17425 * config/arm/arm-cpus.in (cortex-a55): New.
17426 (cortex-a75): Likewise.
17427 (cortex-a75.cortex-a55): Likewise.
17428 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
17430 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
17431 * config/arm/arm-cpu-cdata.h: Regenerate.
17432 * config/arm/arm-cpu-data.h: Regenerate.
17433 * config/arm/arm-cpu.h: Regenerate.
17434 * config/arm/arm-tables.opt: Regenerate.
17435 * config/arm/arm-tune.md: Regenerate.
17437 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17439 * haifa-sched.c (sched_create_recovery_edges): Update profile.
17441 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17443 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
17446 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
17448 PR tree-optimization/81292
17449 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
17450 full_string_p, also call adjust_related_strinfos if the adjustment
17451 is simple, otherwise invalidate related strinfos.
17453 2017-07-04 Martin Liska <mliska@suse.cz>
17456 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
17457 newly created variable as DECL_IGNORED_P.
17459 2017-07-04 Martin Liska <mliska@suse.cz>
17462 * ipa-inline.c (inline_small_functions):
17463 Use xstrdup_for_dump.
17465 2017-07-04 Tom de Vries <tom@codesourcery.com>
17467 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
17469 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
17472 * config/darwin.c (darwin_function_switched_text_sections):
17475 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
17477 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
17479 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17481 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
17483 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17485 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
17486 min_profitable_iters, and th as inclusive lower bounds.
17487 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
17488 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
17489 for min_profitable_iters and min_profitable_estimate.
17490 (vect_transform_loop): Treat th as an inclusive lower bound.
17491 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
17493 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
17496 * config/darwin.c (darwin_function_switched_text_sections):
17497 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
17498 in two pieces, and suppress the use of buf.
17500 2017-07-03 Nathan Sidwell <nathan@acm.org>
17502 * hash-table.h (hash_table_mod1): Fix indentation.
17504 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17506 PR middle-end/81290
17507 * predict.c (force_edge_cold): Be more careful about propagation
17509 * profile-count.h (profile_probability::guessed,
17510 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
17512 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
17514 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
17516 * doc/invoke.texi (rcpc architecture extension): Document it.
17518 2017-07-03 Richard Biener <rguenther@suse.de>
17520 PR tree-optimization/60510
17521 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
17522 the scalar reduction PHI and use it.
17523 (vectorizable_reduction): Properly guard the single_defuse_cycle
17524 path for non-SLP reduction chains where we cannot use it.
17525 Rework reduc_def/index and vector type deduction. Rework
17526 vector operand gathering during reduction op code-gen.
17527 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
17528 chains dissolve the chain and leave it to non-SLP reduction
17531 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17533 * tree-data-ref.h (dr_alignment): Declare.
17534 * tree-data-ref.c (dr_alignment): New function.
17535 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
17536 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
17538 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
17540 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17542 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
17543 and base_misalignment fields.
17544 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
17545 * tree-data-ref.c: Include builtins.h.
17546 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
17547 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
17548 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
17549 * tree-vect-data-refs.c: Include tree-cfg.h.
17550 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
17551 fields instead of calculating an alignment here.
17552 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
17553 innermost_loop_behavior fields.
17555 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17557 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
17559 (DR_STEP_ALIGNMENT): New macro.
17560 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
17561 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
17562 (create_data_ref): Print it.
17563 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
17564 to tell whether the step preserves vector (mis)alignment.
17565 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17566 Move the check for an integer step and generalise to all INTEGER_CST.
17567 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
17568 Print the outer step alignment.
17570 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17572 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
17573 with offset_alignment.
17574 (DR_ALIGNED_TO): Delete.
17575 (DR_OFFSET_ALIGNMENT): New macro.
17576 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
17577 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
17578 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
17579 (create_data_ref): Likewise.
17580 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17581 (vect_analyze_data_refs): Likewise.
17582 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
17583 creating dummy innermost behavior.
17585 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17587 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
17588 with a "innermost_loop_behavior *" and refeence tree.
17589 * tree-data-ref.c (dr_analyze_innermost): Likewise.
17590 (create_data_ref): Update call accordingly.
17591 * tree-predcom.c (find_looparound_phi): Likewise.
17593 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17595 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
17596 fields with dr_wrt_vec_loop.
17597 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
17598 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
17599 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
17600 (vect_dr_behavior): New function.
17601 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17602 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
17603 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
17604 track whether the step preserves the misalignment.
17605 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17606 Use vect_dr_behavior.
17607 (vect_setup_realignment): Update call accordingly.
17608 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
17609 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
17610 call to vect_create_addr_base_for_vector_ref.
17611 (vect_create_cond_for_align_checks): Likewise.
17612 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
17613 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
17614 (vect_recog_mask_conversion_pattern): Likewise.
17615 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
17616 (new_stmt_vec_info): Remove redundant zeroing.
17618 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17620 * common/config/arm/arm-common.c (arm_be8_option): New function.
17621 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
17622 (ISA_ARMv6): Add isa_bit_be8.
17623 * config/arm/arm.h (arm_be8_option): Add prototype.
17624 (BE8_SPEC_FUNCTION): New define.
17625 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
17626 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
17627 (mlittle-endian): Similarly.
17628 (mbe8, mbe32): New options.
17629 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
17630 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
17632 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17634 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
17636 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17638 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
17639 (cleanup_tree_cfg_bb): Use it.
17640 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
17642 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
17644 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17647 * loop-doloop.c (add_test): Update profile.
17649 2017-07-03 Martin Liska <mliska@suse.cz>
17652 * sanopt.c (rewrite_usage_of_param): New function.
17653 (sanitize_rewrite_addressable_params): Likewise.
17654 (pass_sanopt::execute): Call rewrite_usage_of_param.
17656 2017-07-03 Richard Biener <rguenther@suse.de>
17658 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
17659 back to using VIEW_CONVERT_EXPR.
17661 2017-07-03 Martin Liska <mliska@suse.cz>
17664 * doc/extend.texi: Document when a resolver function is
17665 generated for target_clones.
17667 2017-07-03 Martin Liska <mliska@suse.cz>
17669 * asan.c (asan_emit_stack_protection): Unpoison just red zones
17670 and shadow memory of auto variables which are subject of
17671 use-after-scope sanitization.
17672 (asan_expand_mark_ifn): Add do set only when is_poison.
17674 2016-07-03 Richard Biener <rguenther@suse.de>
17676 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
17678 (vect_force_simple_reduction): Record reduction def -> phi mapping.
17679 (vectorizable_reduction): Perform reduction PHI creation when
17680 visiting a reduction PHI and adjust and simplify code generation
17681 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
17682 (vect_transform_loop): Visit reduction PHIs.
17683 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
17684 defs into the SLP tree.
17685 (vect_build_slp_tree): Reduction defs terminate the recursion.
17686 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
17688 (vect_get_vec_defs_for_stmt_copy): Export.
17689 (vect_get_vec_defs): Likewise.
17690 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
17692 (vect_get_vec_defs_for_stmt_copy): Declare.
17693 (vect_get_vec_defs): Likewise.
17695 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17697 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
17698 parameter with a "loop" parameter and use it instead of the
17699 loop containing DR_STMT. Don't check simple_iv when doing
17700 BB analysis. Describe the two analysis modes in the comment.
17702 2017-07-03 Tom de Vries <tom@codesourcery.com>
17704 PR tree-optimization/69468
17705 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
17706 (find_same_succ_bb): Handle ignore_edge_flags.
17708 2017-07-03 Tom de Vries <tom@codesourcery.com>
17710 PR tree-optimization/81192
17711 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
17713 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
17715 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
17717 2017-07-03 Tom de Vries <tom@codesourcery.com>
17719 PR tree-optimization/81192
17720 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
17721 BB_SAME_SUCC (bb) == NULL.
17723 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17725 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
17728 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17730 * dumpfile.c: Include profile-count.h
17731 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
17733 (insert_cond_bb): Update profile.
17734 * tree-cfg.h (insert_cond_bb): Update prototype.
17735 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
17736 * tree-dump.c: Do not include tree-cfg.
17738 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17740 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
17742 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17744 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
17747 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17749 * tree-complex.c (expand_complex_div_wide): update profile.
17751 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17752 Alan Hayward <alan.hayward@arm.com>
17753 David Sherwood <david.sherwood@arm.com>
17755 * Makefile.in (MACHMODE_H): Remove insn-modes.h
17756 (CORETYPES_H): New define.
17757 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
17758 (insn-modes-inline.h, s-modes-inline-h): New rules.
17759 (generated_files): Add insn-modes-inline.h.
17760 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
17761 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
17762 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
17763 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
17764 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
17765 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
17766 (build/gencodes.o, build/genconditions.o): Likewise.
17767 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
17768 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
17769 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
17770 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
17771 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
17772 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
17773 * coretypes.h: Include everything up to real.h for generators.
17774 Include insn-modes.h first. Include wide-int-print.h after
17775 wide-int.h. Include insn-modes-inline.h and then machmode.h.
17776 * machmode.h: Don't include insn-modes.h here.
17777 * function-tests.c: Remove includes of signop.h, machmode.h,
17778 double-int.h and wide-int.h.
17780 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
17782 * optc-save-gen.awk: Likewise.
17783 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
17784 * godump.c: Remove include of wide-int-print.h.
17785 * pretty-print.h: Likewise.
17786 * wide-int-print.cc: Likewise.
17787 * wide-int.cc: Likewise.
17788 * hash-map-tests.c: Remove include of signop.h.
17789 * hash-set-tests.c: Likewise.
17790 * rtl-tests.c: Likewise.
17791 * mkconfig.sh: Remove include of machmode.h.
17792 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
17794 (emit_insn_modes_inline_h): ...this new function. Emit the code
17795 into an insn-modes-inline.h header file, adding appropriate
17796 include guards and end comments.
17797 (emit_insn_modes_c_header): Remove include of machmode.h.
17798 (emit_min_insn_modes_c_header): Include coretypes.h rather than
17800 (main): Handle -i flag and call emit_insn_modes_inline_h when
17803 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17805 * tree-ssa-strlen.c (strinfo): Rename the length field to
17806 nonzero_chars. Add a full_string_p field.
17807 (compare_nonzero_chars, zero_length_string_p): New functions.
17808 (get_addr_stridx): Add an offset_out parameter.
17809 Use compare_nonzero_chars.
17810 (get_stridx): Update accordingly. Use compare_nonzero_chars.
17811 (new_strinfo): Update after above changes to strinfo.
17812 (set_endptr_and_length): Set full_string_p.
17813 (get_string_length): Update after above changes to strinfo.
17814 (unshare_strinfo): Update call to new_strinfo.
17815 (maybe_invalidate): Likewise.
17816 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
17817 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
17818 as a uhwi instead of an shwi. Update after above changes to
17819 strinfo and new_strinfo.
17820 (zero_length_string): Assert that chainsi contains full strings.
17821 Use zero_length_string_p. Update call to new_strinfo.
17822 (adjust_related_strinfos): Update after above changes to strinfo.
17823 Copy full_string_p from origsi.
17824 (adjust_last_stmt): Use zero_length_string_p.
17825 (handle_builtin_strlen): Update after above changes to strinfo and
17826 new_strinfo. Install the lhs as the string length if the previous
17827 entry didn't describe a full string.
17828 (handle_builtin_strchr): Update after above changes to strinfo
17830 (handle_builtin_strcpy): Likewise.
17831 (handle_builtin_strcat): Likewise.
17832 (handle_builtin_malloc): Likewise.
17833 (handle_pointer_plus): Likewise.
17834 (handle_builtin_memcpy): Likewise. Track nonzero characters
17835 that aren't necessarily followed by a nul terminator.
17836 (handle_char_store): Likewise.
17838 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17840 PR tree-optimization/80769
17841 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
17842 for malloc and calloc. Document the new invariant that all related
17843 strinfos have delayed lengths or none do.
17844 (verify_related_strinfos): Move earlier in file.
17845 (set_endptr_and_length): New function, split out from...
17846 (get_string_length): ...here. Also set the lengths of related
17848 (zero_length_string): Assert that chainsi has known (rather than
17850 (adjust_related_strinfos): Likewise.
17852 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17854 PR tree-optimization/81136
17855 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
17856 assert that two references with the same misalignment have the same
17857 compile-time misalignment if those compile-time misalignments
17860 2017-07-01 Andi Kleen <ak@linux.intel.com>
17862 * print-tree.c (print_node): Print all attributes.
17864 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17866 * cfg.c (scale_bbs_frequencies): New function.
17867 * cfg.h (scale_bbs_frequencies): Declare it.
17868 * cfgloopanal.c (single_likely_exit): Cleanup.
17869 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
17871 (scale_loop_profile): Likewise.
17872 (loop_version): Likewise.
17873 (create_empty_loop_on_edge): Update.
17874 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
17875 scale_loop_frequencies, scale_loop_profile, loopify,
17876 loop_version): Update prototypes.
17877 * modulo-sched.c (sms_schedule): Update.
17878 * predict.c (unlikely_executed_edge_p): Also check probability.
17879 (probably_never_executed_edge_p): Fix typo.
17880 * tree-if-conv.c (version_loop_for_if_conversion): Update.
17881 * tree-parloops.c (gen_parallel_loop): Update.
17882 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
17883 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
17884 * tree-ssa-loop-split.c (split_loop): Update.
17885 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
17886 * tree-vect-loop-manip.c (vect_do_peeling): Update.
17887 (vect_loop_versioning): Update.
17888 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
17890 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17892 * trans-mem.c (split_bb_make_tm_edge): Update profile.
17894 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17896 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
17897 to keep profile consistent.
17899 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17901 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
17902 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
17903 * profile-count.h (max_safe_multiplier): Make unsigned.
17904 (profile_count::guessed_zero): New.
17906 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17908 * bb-reorder.c (fix_up_crossing_landing_pad,
17909 fix_crossing_conditional_branches): Use make_single_succ_edge
17910 to keep profile consistent.
17912 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17914 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
17917 2017-07-01 Jakub Jelinek <jakub@redhat.com>
17920 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
17921 the right scopes, make sure cond_jump isn't preserved between multiple
17922 iterations. Search for fallthru edge whenever there are 3+ edges and
17923 use find_fallthru_edge for it.
17925 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17927 Patch by Alexander Monakov <amonakov@ispras.ru>
17928 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
17929 probabilities consistently.
17931 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17933 * pa.c (pa_expand_compare_and_swap_loop): Update call of
17934 emit_cmp_and_jump_insns.
17936 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17939 * tree-inline.c (expand_call_inline): Combine profile statuses.
17941 2017-06-30 Andrew Pinski <apinski@cavium.com>
17943 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
17944 fold_stmt returned true.
17946 2017-06-30 Nathan Sidwell <nathan@acm.org>
17948 * ggc.h (empty_string): Delete.
17949 * cfgexpand.c (expand_asm_stmt): Use plain "".
17950 * optabs.c (expand_asm_memory_barrier): Likewise.
17951 * stringpool.c (empty_string): Delete.
17952 (digit_vector, digit_string): Delete.
17953 (ggc_alloc_string): Use plain "", don't optimize single digit
17954 strings. Use ggc_alloc_atomic.
17956 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
17958 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
17959 comparison set and one other set, use the cost of the non-comparison
17962 2017-06-30 Nathan Sidwell <nathan@acm.org>
17964 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
17967 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
17969 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
17970 loops. Remove now unneeded calls to gimple_switch_set_label() that
17971 just set removed labels to NULL_TREE.
17973 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
17975 * tree-ssanames.c (set_range_info_raw): Abstract from ...
17976 (set_range_info): ...here. Only call set_range_info_raw if domain
17978 (set_nonzero_bits): Call set_range_info_raw.
17979 * tree-ssanames.h (set_range_info_raw): New.
17981 2017-06-30 Jakub Jelinek <jakub@redhat.com>
17984 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
17985 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
17986 of nonimmediate_operand and <store_mask_constraint> instead of m
17987 for the input operand. For V8FI iterator, always split if input
17988 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
17989 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
17990 <store_mask_predicate> instead of register_operand and
17991 <store_mask_constraint> instead of v for the input operand. Make
17992 sure both operands aren't MEMs for if not <mask_applied>.
17994 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
17996 * lto-wrapper.c (copy_file) Close both file descriptors before
17999 2017-06-30 Martin Liska <mliska@suse.cz>
18002 * multiple_target.c (create_dispatcher_calls): Make ifunc
18003 also for function that don't have calls or are not referenced.
18005 2017-06-30 Richard Biener <rguenther@suse.de>
18007 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
18008 analyze the first scalar stmt. Move vector type computation
18009 for the BB case here from ...
18010 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
18011 live operation processing in the SLP case properly.
18013 2017-06-30 Richard Biener <rguenther@suse.de>
18015 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
18017 2017-06-30 Martin Liska <mliska@suse.cz>
18020 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
18021 before BUILT_IN_UNWIND_RESUME when ASAN is used.
18023 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
18025 * doc/invoke.texi (AArch64): Add missing options and remove redundant
18028 2017-06-30 Richard Biener <rguenther@suse.de>
18030 PR tree-optimization/81249
18031 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
18032 condition reduction result to original scalar type.
18034 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18036 * profile-count.h (enum profile_quality): Fix typos and whitespace
18039 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18041 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
18042 type for branch probabilities.
18044 2017-06-29 Julian Brown <julian@codesourcery.com>
18045 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18047 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
18048 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
18049 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
18050 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
18052 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18054 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
18055 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
18056 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
18057 CC usage from generic code to here.
18058 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
18059 CC usage into the target macros.
18061 2017-06-29 Maya Rashish <coypu@sdf.org>
18063 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
18066 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18068 * arm/arm-builtins.c: Include profile-count.h
18069 * except.c (sjlj_emit_function_enter): Use
18070 profile_probability::unlikely.
18072 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18074 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
18075 and tocrel_offset be pointer args rather than implicitly using
18077 (legitimate_constant_pool_address_p, rs6000_emit_move,
18078 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
18079 tocrel_offset and use in toc_relative_expr_p call.
18080 (print_operand, print_operand_address): Use static tocrel_base_oac
18081 and tocrel_offset_oac.
18082 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
18085 2017-06-29 Maya Rashish <coypu@sdf.org>
18087 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
18089 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
18091 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
18092 objects, take into account only the alignment of 'op0' and 'mode1' if
18095 2017-06-29 Steve Ellcey <sellcey@cavium.com>
18097 * ccmp.c (ccmp_tree_comparison_p): New function.
18098 (ccmp_candidate_p): Update to use above function.
18099 (get_compare_parts): New function.
18100 (expand_ccmp_next): Update to use new functions.
18101 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
18103 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
18104 take mode as argument.
18105 * ccmp.h (expand_ccmp_expr): Add mode as argument.
18106 * expr.c (expand_expr_real_1): Pass mode as argument.
18108 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
18110 * combine.c (combine_instructions): Print insns to dump_file, together
18113 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18115 * asan.c (asan_emit_stack_protection): Update.
18116 (create_cond_insert_point): Update.
18117 * auto-profile.c (afdo_propagate_circuit): Update.
18118 * basic-block.h (struct edge_def): Turn probability to
18119 profile_probability.
18120 (EDGE_FREQUENCY): Update.
18121 * bb-reorder.c (find_traces_1_round): Update.
18122 (better_edge_p): Update.
18123 (sanitize_hot_paths): Update.
18124 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
18125 (make_single_succ_edge): Update.
18126 (check_bb_profile): Update.
18127 (dump_edge_info): Update.
18128 (update_bb_profile_for_threading): Update.
18129 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
18130 probabilitycount to 0.
18131 * cfgbuild.c (compute_outgoing_frequencies): Update.
18132 * cfgcleanup.c (try_forward_edges): Update.
18133 (outgoing_edges_match): Update.
18134 (try_crossjump_to_edge): Update.
18135 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
18136 (expand_gimple_tailcall): Update.
18137 (construct_init_block): Use make_single_succ_edge.
18138 (construct_exit_block): Use make_single_succ_edge.
18139 * cfghooks.c (verify_flow_info): Update.
18140 (redirect_edge_succ_nodup): Update.
18141 (split_edge): Update.
18142 (account_profile_record): Update.
18143 * cfgloopanal.c (single_likely_exit): Update.
18144 * cfgloopmanip.c (scale_loop_profile): Update.
18145 (set_zero_probability): Remove.
18146 (duplicate_loop_to_header_edge): Update.
18147 * cfgloopmanip.h (loop_version): Update prototype.
18148 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
18149 (force_nonfallthru_and_redirect): Update.
18150 (update_br_prob_note): Update.
18151 (rtl_verify_edges): Update.
18152 (purge_dead_edges): Update.
18153 (rtl_lv_add_condition_to_bb): Update.
18154 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
18155 * cgraphunit.c (init_lowered_empty_function): Update.
18156 (cgraph_node::expand_thunk): Update.
18157 * cilk-common.c: Include profile-count.h
18158 * dojump.c (inv): Remove.
18159 (jumpifnot): Update.
18160 (jumpifnot_1): Update.
18161 (do_jump_1): Update.
18163 (do_jump_by_parts_greater_rtx): Update.
18164 (do_compare_rtx_and_jump): Update.
18165 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
18166 do_jump_1. do_compare_rtx_and_jump): Update prototype.
18167 * dwarf2cfi.c: Include profile-count.h
18168 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
18169 (sjlj_emit_dispatch_table): Likewise.
18170 * explow.c: Include profile-count.h
18171 * expmed.c (emit_store_flag_force): Update.
18172 (do_cmp_and_jump): Update.
18173 * expr.c (compare_by_pieces_d::generate): Update.
18174 (compare_by_pieces_d::finish_mode): Update.
18175 (emit_block_move_via_loop): Update.
18176 (store_expr_with_bounds): Update.
18177 (store_constructor): Update.
18178 (expand_expr_real_2): Update.
18179 (expand_expr_real_1): Update.
18180 * expr.h (try_casesi, try_tablejump): Update prototypes.
18181 * gimple-pretty-print.c (dump_probability): Update.
18182 (dump_profile): New.
18183 (dump_gimple_label): Update.
18184 (dump_gimple_bb_header): Update.
18185 * graph.c (draw_cfg_node_succ_edges): Update.
18186 * hsa-gen.c (convert_switch_statements): Update.
18187 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
18188 (find_if_case_1): Update.
18189 (find_if_case_2): Update.
18190 * internal-fn.c (expand_arith_overflow_result_store): Update.
18191 (expand_addsub_overflow): Update.
18192 (expand_neg_overflow): Update.
18193 (expand_mul_overflow): Update.
18194 (expand_vector_ubsan_overflow): Update.
18195 * ipa-cp.c (good_cloning_opportunity_p): Update.
18196 * ipa-split.c (split_function): Use make_single_succ_edge.
18197 * ipa-utils.c (ipa_merge_profiles): Update.
18198 * loop-doloop.c (add_test): Update.
18199 (doloop_modify): Update.
18200 * loop-unroll.c (compare_and_jump_seq): Update.
18201 (unroll_loop_runtime_iterations): Update.
18202 * lra-constraints.c (lra_inheritance): Update.
18203 * lto-streamer-in.c (input_cfg): Update.
18204 * lto-streamer-out.c (output_cfg): Update.
18205 * mcf.c (adjust_cfg_counts): Update.
18206 * modulo-sched.c (sms_schedule): Update.
18207 * omp-expand.c (expand_omp_for_init_counts): Update.
18208 (extract_omp_for_update_vars): Update.
18209 (expand_omp_ordered_sink): Update.
18210 (expand_omp_for_ordered_loops): Update.
18211 (expand_omp_for_generic): Update.
18212 (expand_omp_for_static_nochunk): Update.
18213 (expand_omp_for_static_chunk): Update.
18214 (expand_cilk_for): Update.
18215 (expand_omp_simd): Update.
18216 (expand_omp_taskloop_for_outer): Update.
18217 (expand_omp_taskloop_for_inner): Update.
18218 * omp-simd-clone.c (simd_clone_adjust): Update.
18219 * optabs.c (expand_doubleword_shift): Update.
18220 (expand_abs): Update.
18221 (emit_cmp_and_jump_insn_1): Update.
18222 (expand_compare_and_swap_loop): Update.
18223 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
18224 * predict.c (predictable_edge_p): Update.
18225 (edge_probability_reliable_p): Update.
18226 (set_even_probabilities): Update.
18227 (combine_predictions_for_insn): Update.
18228 (combine_predictions_for_bb): Update.
18229 (propagate_freq): Update.
18230 (estimate_bb_frequencies): Update.
18231 (force_edge_cold): Update.
18232 * profile-count.c (profile_count::dump): Add missing space into dump.
18233 (profile_count::debug): Add newline.
18234 (profile_count::differs_from_p): Explicitly convert to unsigned.
18235 (profile_count::stream_in): Update.
18236 (profile_probability::dump): New member function.
18237 (profile_probability::debug): New member function.
18238 (profile_probability::differs_from_p): New member function.
18239 (profile_probability::differs_lot_from_p): New member function.
18240 (profile_probability::stream_in): New member function.
18241 (profile_probability::stream_out): New member function.
18242 * profile-count.h (profile_count_quality): Rename to ...
18243 (profile_quality): ... this one.
18244 (profile_probability): New.
18245 (profile_count): Update.
18246 * profile.c (compute_branch_probabilities): Update.
18247 * recog.c (peep2_attempt): Update.
18248 * sched-ebb.c (schedule_ebbs): Update.
18249 * sched-rgn.c (find_single_block_region): Update.
18250 (compute_dom_prob_ps): Update.
18251 (schedule_region): Update.
18252 * sel-sched-ir.c (compute_succs_info): Update.
18253 * stmt.c (struct case_node): Update.
18254 (do_jump_if_equal): Update.
18255 (get_outgoing_edge_probs): Update.
18256 (conditional_probability): Update.
18257 (emit_case_dispatch_table): Update.
18258 (expand_case): Update.
18259 (expand_sjlj_dispatch_table): Update.
18260 (emit_case_nodes): Update.
18261 * targhooks.c: Update.
18262 * tracer.c (better_p): Update.
18263 (find_best_successor): Update.
18264 * trans-mem.c (expand_transaction): Update.
18265 * tree-call-cdce.c: Update.
18266 * tree-cfg.c (gimple_split_edge): Upate.
18267 (move_sese_region_to_fn): Upate.
18268 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
18269 * tree-eh.c (lower_resx): Upate.
18270 (cleanup_empty_eh_move_lp): Upate.
18271 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18272 * tree-inline.c (copy_edges_for_bb): Update.
18273 (copy_cfg_body): Update.
18274 * tree-parloops.c (gen_parallel_loop): Update.
18275 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
18276 (gimple_gen_time_profiler): Update.
18277 * tree-ssa-dce.c (remove_dead_stmt): Update.
18278 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
18279 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
18280 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
18281 (unloop_loops): Update.
18282 (try_peel_loop): Update.
18283 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18284 * tree-ssa-loop-split.c (connect_loops): Update.
18285 (split_loop): Update.
18286 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
18287 (hoist_guard): Update.
18288 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
18289 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
18290 (value_replacement): Update.
18291 * tree-ssa-reassoc.c (branch_fixup): Update.
18292 * tree-ssa-tail-merge.c (replace_block_by): Update.
18293 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
18294 (create_edge_and_update_destination_phis): Update.
18295 (compute_path_counts): Update.
18296 (recompute_probabilities): Update.
18297 (update_joiner_offpath_counts): Update.
18298 (freqs_to_counts_path): Update.
18299 (duplicate_thread_path): Update.
18300 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
18301 (struct switch_conv_info): Update.
18302 (gen_inbound_check): Update.
18303 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
18304 (vect_do_peeling): Update.
18305 (vect_loop_versioning): Update.
18306 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18307 (optimize_mask_stores): Update.
18308 * ubsan.c (ubsan_expand_null_ifn): Update.
18309 * value-prof.c (gimple_divmod_fixed_value): Update.
18310 (gimple_divmod_fixed_value_transform): Update.
18311 (gimple_mod_pow2): Update.
18312 (gimple_mod_pow2_value_transform): Update.
18313 (gimple_mod_subtract): Update.
18314 (gimple_mod_subtract_transform): Update.
18315 (gimple_ic): Update.
18316 (gimple_stringop_fixed_value): Update.
18317 (gimple_stringops_transform): Update.
18318 * value-prof.h: Update.
18320 2017-06-29 Carl Love <cel@us.ibm.com>
18322 * config/rs6000/rs6000-c.c: Add support for built-in functions
18323 vector signed int vec_signed (vector float);
18324 vector signed long long vec_signed (vector double);
18325 vector signed int vec_signed2 (vector double, vector double);
18326 vector signed int vec_signede (vector double);
18327 vector signed int vec_signedo (vector double);
18328 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
18329 instruction generator.
18330 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18331 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
18332 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
18334 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
18335 vunsignede_v2df): Add define_expands.
18336 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
18337 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
18338 VEC_UNSIGNEDO): Add definitions.
18339 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18340 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
18341 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
18342 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
18343 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
18344 * config/rs6000/altivec.h (vec_signed, vec_signed2,
18345 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
18346 vec_unsignede, vec_unsignedo): Add builtin defines.
18347 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
18349 * doc/extend.texi: Update the built-in documentation file for the
18350 new built-in functions.
18352 2017-06-29 Richard Biener <rguenther@suse.de>
18354 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
18355 reduction chains to LOOP_VINFO_REDUCTIONS.
18356 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
18357 SLP reductions after processing reduction chains.
18359 2017-06-29 Nathan Sidwell <nathan@acm.org>
18361 * builtins.c (fold_builtin_FUNCTION): Use
18362 lang_hooks.decl_printable_name.
18364 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
18366 PR middle-end/81194
18367 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
18368 with only one label.
18369 * stmt.c (expand_case): Assert NCASES is greater than one.
18371 2017-06-29 Richard Biener <rguenther@suse.de>
18373 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
18375 (group_case_labels): Likewise.
18376 (find_taken_edge): Push sanity checking on val to workers...
18377 (find_taken_edge_cond_expr): ... here
18378 (find_taken_edge_switch_expr): ... and here, handle cases
18379 with just a default label.
18380 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
18381 (group_case_labels): Likewise.
18382 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
18383 group_case_labels does anything cleanup the CFG again.
18385 2017-06-29 Bin Cheng <bin.cheng@arm.com>
18387 PR tree-optimization/81196
18388 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
18389 exit condition comparing two IVs.
18391 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
18393 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
18394 profile to the dummy entry at the end of the list of architectures.
18395 * config/arm/arm-cpu-cdata.h: Regenerated.
18397 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18398 Michael Collison <michael.collison@arm.com>
18401 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
18403 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
18404 (*aarch64_reg_<mode>3_minus_mask): New pattern.
18405 (*aarch64_<optab>_reg_di3_mask2): New pattern.
18406 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
18407 of shift when the shift amount is masked with constant equal to
18408 the size of the mode.
18409 * config/aarch64/predicates.md (subreg_lowpart_operator): New
18412 2017-06-29 Martin Liska <mliska@suse.cz>
18414 * config/i386/i386.opt: Change range from [1,5] to [0,5].
18416 2017-06-29 Yury Gribov <tetra2005@gmail.com>
18419 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
18422 (edge_growth_cache_entry::edge_growth_cache_entry): New
18424 (reset_edge_growth_cache): Update to use constructor.
18426 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18428 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
18429 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
18430 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
18432 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
18434 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
18435 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
18437 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
18439 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
18440 (*-linux-uclibc*): Add t-uclibc tmake_file.
18441 * config/t-musl: New.
18442 * config/t-uclibc: New.
18444 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
18446 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
18448 (gen_comm_data): Emit architectural setting of arch_prof.
18449 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
18451 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
18452 (armv8-m.base, armv8-m.main): Likewise.
18453 * arm-protos.h (arm_build_target): Add profile field.
18454 (arch_option): Likewise.
18455 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
18457 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
18458 arm_active_target.profile.
18460 2017-06-28 Richard Biener <rguenther@suse.de>
18462 PR middle-end/81227
18463 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
18464 TYPE_OVERFLOW_WRAPS.
18465 * match.pd (negate_expr_p): Likewise.
18466 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
18467 fold_build2, not fold_binary.
18469 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
18471 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
18472 Convert memory address to Pmode.
18473 (aarch64_print_operand): Assert MEM operands are always Pmode.
18475 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
18478 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
18479 Remove redundant if.
18480 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
18481 * config/arm/aarch-common-protos.h
18482 (aarch_forward_to_shift_is_not_shifted_re): Remove.
18483 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
18485 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
18488 * multiple_target.c (create_dispatcher_calls): Set the default
18489 clone to be static, not public.
18491 2017-06-28 Richard Biener <rguenther@suse.de>
18493 * tree-vect-loop.c (vectorizable_reduction): Move special
18494 cond reduction IV var creation ...
18495 (vect_create_epilog_for_reduction): ... here. Remove induction_index
18496 parameter. Use STMT_VINFO_VECTYPE.
18497 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
18500 2017-06-28 Martin Liska <mliska@suse.cz>
18503 * ipa-visibility.c (non_local_p): Handle visibility.
18505 2017-06-28 Martin Liska <mliska@suse.cz>
18508 * common.opt: Add IntegerRange to various options.
18509 * opt-functions.awk (integer_range_info): New function.
18510 * optc-gen.awk: Add integer_range_info to cl_options struct.
18511 * opts-common.c (decode_cmdline_option): Handle
18512 CL_ERR_INT_RANGE_ARG.
18513 (cmdline_handle_error): Likewise.
18514 * opts.c (print_filtered_help): Show valid interval in
18515 when --help is provided.
18516 * opts.h (struct cl_option): Add range_min and range_max fields.
18517 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
18519 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
18521 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
18522 (x * C EQ/NE y * C): New transformation.
18524 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
18526 * genmultilib (combination_space): Accept '+' in option names.
18528 2017-06-28 Martin Liska <mliska@suse.cz>
18531 * asan.c (instrument_derefs): Bail out inner references
18532 that are hard register variables.
18534 2017-06-28 Jakub Jelinek <jakub@redhat.com>
18537 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
18538 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
18540 2017-06-28 Richard Biener <rguenther@suse.de>
18542 * tree-vectorizer.h (vect_get_vec_defs): Remove.
18543 (vect_get_slp_defs): Adjust.
18544 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
18546 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
18548 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
18549 get_initial_defs_for_reduction instead of vect_get_vec_defs.
18550 (vectorizable_reduction): Adjust.
18551 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
18553 (vect_get_slp_defs): Likewise.
18554 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
18555 (vectorizable_bswap): Adjust.
18556 (vectorizable_call): Likewise.
18557 (vectorizable_conversion): Likewise.
18558 (vectorizable_assignment): Likewise.
18559 (vectorizable_shift): Likewise.
18560 (vectorizable_operation): Likewise.
18561 (vectorizable_store): Likewise.
18562 (vectorizable_condition): Likewise.
18563 (vectorizable_comparison): Likewise.
18565 2017-06-28 Michael Collison <michael.collison@arm.com>
18568 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
18570 (arm_gen_movmemqi): Removed unused variable 'i'.
18571 Convert 'for' loop into 'while' loop.
18572 (arm_expand_prologue): Remove last unnecessary set of insn.
18573 (thumb_pop): Remove unused variable 'pushed_words'.
18574 (thumb_exit): Remove last unnecessary set of regs_to_pop.
18576 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18578 * config/s390/predicates.md: Use s390_rel_address_ok_p.
18579 * config/s390/s390-protos.h: Add prototype of
18580 s390_rel_address_ok_p.
18581 * config/s390/s390.c (s390_got_symbol): New function.
18582 (s390_rel_address_ok_p): New function.
18583 (legitimize_pic_address): Use s390_rel_address_ok_p.
18584 (s390_load_got): Use s390_got_symbol.
18585 (s390_option_override): Issue error if
18586 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
18587 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
18589 * config/s390/s390.opt: New option mpic-data-is-text-relative.
18591 2017-06-27 Andrew Pinski <apinski@cavium.com>
18593 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
18594 (X * copysign (1.0, X)): New pattern.
18595 (X * copysign (1.0, -X)): New pattern.
18596 (copysign (-1.0, CST)): New pattern.
18598 2017-06-27 Joseph Myers <joseph@codesourcery.com>
18600 * genmultilib (combination_space): Remove variable.
18601 Validate reuse rules against regular expression for any sequence
18602 of multilib options in any order.
18604 2017-06-27 Michael Collison <michael.collison@arm.com>
18606 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
18607 call aarch64_split_simd_combine.
18608 * (aarch64_combine_internal<mode>): Delete pattern.
18609 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
18610 Allow register and subreg operands.
18612 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18614 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
18615 specific need, just fallback on defaults.
18616 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
18618 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18619 Olivier Hainque <hainque@adacore.com>
18621 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
18623 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
18624 targets. Pick a default if no particular attempt applied.
18625 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
18628 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18630 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
18631 (x86_64-wrs-vxworks7): Likewise.
18633 2017-06-27 Marek Polacek <polacek@redhat.com>
18636 * ubsan.c (instrument_null): Check get_base_address's result for null.
18638 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18640 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
18642 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18644 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
18645 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
18646 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
18647 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
18648 New function types.
18649 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
18650 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
18651 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
18652 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
18653 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
18654 BUILT_IN_FEUPDATEENV): New builtins.
18655 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
18656 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
18657 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
18658 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
18660 (builtin_structptr_types): Adjust size.
18661 * tree.c (builtin_structptr_types): Add four entries.
18663 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18664 Olivier Hainque <hainque@adacore.com>
18666 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
18667 (TLS_SYM): New local macro, forcing reference to __tls__ on
18668 link command lines for VxWorks 7 RTPs, triggering initialization
18670 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
18671 OS features TLS support, true for RTPs on VxWorks 7.
18672 * config/vxworks.c (vxworks_override_options): Setup emutls
18675 2017-06-27 Jakub Jelinek <jakub@redhat.com>
18677 * predict.c (test_prediction_value_range): Use -1U instead of -1
18678 to avoid narrowing conversion warning.
18679 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
18680 to avoid narrowing conversion warning.
18681 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
18683 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
18685 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18687 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
18688 64bit configurations.
18689 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
18690 (SIZE_TYPE): Likewise.
18691 * config/vxworks.c (vxworks_emutls_var_fields): Use
18692 long_unsigned_type_node instead of unsigned_type_node as the offset
18693 field type, which is "pointer" mode in emutls.c.
18695 2017-06-27 Jakub Jelinek <jakub@redhat.com>
18698 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
18700 PR middle-end/81207
18701 * gimple-fold.c (replace_call_with_call_and_fold): Handle
18702 gimple_vuse copying separately from gimple_vdef copying.
18704 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18706 * value-prof.c (free_hist): Remove call to memset and the enclosing if
18709 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
18710 Olivier Hainque <hainque@adacore.com>
18712 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
18713 for all vxworks7 targets.
18714 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
18715 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
18716 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
18717 variations for VX6/VX7 and 32/64bits later on in ...
18718 (VXWORKS_LIB_SPEC): Leverage new macros.
18719 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
18720 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
18722 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
18724 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
18725 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
18727 2017-06-26 Carl Love <cel@us.ibm.com>
18729 * config/rs6000/rs6000-c.c: Add support for built-in functions
18730 vector bool char vec_reve (vector bool char);
18731 vector signed char vec_reve (vector signed char);
18732 vector unsigned char vec_reve (vector unsigned char);
18733 vector bool int vec_reve (vector bool int);
18734 vector signed int vec_reve (vector signed int);
18735 vector unsigned int vec_reve (vector unsigned int);
18736 vector bool long long vec_reve (vector bool long long);
18737 vector signed long long vec_reve (vector signed long long);
18738 vector unsigned long long vec_reve (vector unsigned long long);
18739 vector bool short vec_reve (vector bool short);
18740 vector signed short vec_reve (vector signed short);
18741 vector double vec_reve (vector double);
18742 vector float vec_reve (vector float);
18743 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
18744 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
18745 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
18746 (altivec_vreve): New pattern.
18747 * config/rs6000/altivec.h (vec_reve): New define.
18748 * doc/extend.texi (vec_rev): Update the built-in documentation file
18749 for the new built-in functions.
18751 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18753 PR tree-optimization/71815
18754 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
18756 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
18758 (slsr_process_phi): Likewise.
18759 (replace_uncond_cands_and_profitable_phis): Don't replace a
18760 multiply candidate with a stride of 1 (copy or cast).
18761 (phi_incr_cost): Call uses_consumed_by_stmt rather than
18763 (lowest_cost_path): Likewise.
18764 (total_savings): Likewise.
18766 2017-06-26 Richard Biener <rguenther@suse.de>
18769 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
18770 Use def_builtin_pure for all gather builtins.
18772 2017-06-26 Richard Biener <rguenther@suse.de>
18774 PR tree-optimization/81203
18775 * tree-tailcall.c (find_tail_calls): Do not move stmts into
18776 non-dominating BBs.
18778 2017-06-26 Marek Polacek <polacek@redhat.com>
18781 * doc/invoke.texi: Document -Wmultistatement-macros.
18783 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
18785 * doc/sourcebuild.texi (ARM-specific attributes): Document new
18786 arm_neon_ok_no_float_abi effective target.
18788 2017-06-26 Richard Biener <rguenther@suse.de>
18790 PR tree-optimization/80928
18791 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
18792 (copy_bbs): Set BB_DUPLICATED flag early.
18793 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
18795 (execute_on_shrinking_pred): Likewise.
18796 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
18797 BB_DUPLICATED blocks.
18798 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
18799 iterate over all PHIs considering removal of *gsi.
18801 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
18803 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
18806 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18808 * config/rs6000/rs6000-string.c: (expand_block_clear,
18809 do_load_for_compare, select_block_compare_mode,
18810 compute_current_alignment, expand_block_compare,
18811 expand_strncmp_align_check, expand_strn_compare,
18812 expand_block_move, rs6000_output_load_multiple)
18813 Move functions related to string/block move/compare
18814 to a separate file.
18815 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
18816 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
18817 for this function which is now used in two files.
18818 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
18819 * config.gcc: Add rs6000-string.o to extra_objs for
18820 targets powerpc*-*-* and rs6000*-*-*.
18822 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
18825 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
18826 32-bit, since indexed is not valid for DImode.
18827 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
18828 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
18829 (define_peephole2 for Altivec d-form load): Add 32-bit support.
18830 (define_peephole2 for Altivec d-form store): Likewise.
18833 * multiple_target.c (create_dispatcher_calls): Only create the
18834 dispatcher call if the function is the default clone of a
18835 versioned function.
18837 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
18839 PR middle-end/80902
18840 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
18841 a call, force the call to not be a tail call.
18843 2017-06-23 Jeff Law <law@redhat.com>
18845 * doc/contrib.texi: Add entry for Steven Pemberton's work on
18848 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
18850 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
18851 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
18852 handling for early expansion of vector shifts (sl,sr,sra,rl).
18853 (builtin_function_type): Add vector shift right instructions
18854 to the unsigned argument list.
18856 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
18858 rtl-optimizatoin/79286
18859 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
18860 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
18861 trap. PIC register plus a const unspec without offset can never trap.
18863 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
18865 * tree.h (builtin_structptr_type): New type.
18866 (builtin_structptr_types): Declare new array.
18867 * tree.c (builtin_structptr_types): New array.
18868 (free_lang_data, build_common_tree_nodes): Use it.
18870 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
18873 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
18876 2017-06-22 Matt Turner <mattst88@gmail.com>
18878 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
18879 Lake models to skylake case. Assume skylake for unknown
18880 models with clflushopt.
18882 2017-06-22 Jeff Law <law@redhat.com>
18884 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
18885 frame sizes that do not satisfy aarch64_uimm12_shift.
18887 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
18889 * profile-count.h (apply_probability,
18890 apply_scale, probability_in): Fix checks for zero.
18892 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18894 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
18895 * doc/cppdiropts.texi (-I @var{dir}): Document it.
18897 2016-06-22 Richard Biener <rguenther@suse.de>
18899 * tree-vect-loop.c (vect_model_reduction_cost): Handle
18900 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
18901 REDUC_MAX_EXPR support.
18902 (vectorizable_reduction): Likewise.
18903 (vect_create_epilog_for_reduction): Likewise.
18905 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
18907 * match.pd (A / (1 << B) -> A >> B): New.
18908 * generic-match-head.c: Include optabs-tree.h.
18909 * gimple-match-head.c: Likewise.
18910 * optabs-tree.h (target_supports_op_p): New.
18911 * optabs-tree.c (target_supports_op_p): New.
18913 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18915 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
18916 $gcc_cv_ld --help output.
18917 (gcc_cv_ld_demangle): Likewise.
18918 (gcc_cv_ld_eh_frame_hdr): Likewise.
18919 (gcc_cv_ld_pie): Likewise.
18920 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
18921 (gcc_cv_ld_buildid): Likewise.
18922 (gcc_cv_ld_sysroot): Likewise.
18923 (ld_bndplt_support): Likewise.
18924 (ld_pushpopstate_support): Likewise.
18925 * configure: Regenerate.
18926 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
18928 2017-06-21 Jakub Jelinek <jakub@redhat.com>
18931 * config/i386/sse.md (round<mode>2): Renumber match_dup and
18932 operands indexes to avoid gap between operands and match_dups.
18934 2017-06-21 Andrew Pinski <apinski@cavium.com>
18936 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
18937 Increment Arith_shift and Arith_shift_reg by 1.
18938 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
18940 * config/aarch64/aarch64.c (thunderx_tunings): Enable
18941 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
18942 (aarch64_strip_extend): Add new argument and test for it.
18943 (aarch64_cheap_mult_shift_p): New function.
18944 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
18945 add a cost if it is true.
18946 Update calls to aarch64_strip_extend.
18947 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
18949 2017-06-21 Andrew Pinski <apinski@cavium.com>
18951 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
18953 (thunderxt88): Likewise.
18954 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
18955 (thunderx_prefetch_tune): New variable.
18956 (thunderx2t99_prefetch_tune): Update for the correct values.
18957 (thunderxt88_tunings): New variable.
18958 (thunderx_tunings): Use thunderx_prefetch_tune instead of
18959 generic_prefetch_tune.
18960 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
18962 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18964 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
18965 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
18966 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
18967 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
18968 (aarch64_atomic_cas<mode>, GPI): Likewise.
18970 2017-06-21 Martin Liska <mliska@suse.cz>
18972 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
18973 statements on cold and hot labels.
18974 * predict.c (tree_estimate_probability_bb): Remove the
18975 prediction from this place.
18977 2017-06-21 Martin Liska <mliska@suse.cz>
18979 PR tree-optimization/79489
18980 * gimplify.c (maybe_add_early_return_predict_stmt): New
18982 (gimplify_return_expr): Call the function.
18983 * predict.c (tree_estimate_probability_bb): Remove handling
18985 * predict.def: Update comment about early return predictor.
18986 * gimple-predict.h (is_gimple_predict): New function.
18987 * predict.def: Change default value of early return to 66.
18988 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
18990 * passes.def: Put pass_strip_predict_hints to the beginning of
18993 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
18995 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
18996 FUNCTION_DECL declarations.
18997 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
18999 (dwaf2out_decl): Likewise.
19000 * godump.c (go_early_global_decl): Skip call to the real debug hook
19001 for FUNCTION_DECL declarations.
19002 * passes.c (rest_of_decl_compilation): Skip call to the
19003 early_global_decl debug hook for FUNCTION_DECL declarations, unless
19004 -fdump-go-spec is passed.
19006 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
19008 * config/i386/i386.c (struct builtin_isa): New field pure_p.
19009 Reorder for compactness.
19010 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
19011 (def_builtin_pure, def_builtin_pure2): New functions.
19012 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
19014 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
19016 * match.pd (nop_convert): New predicate.
19017 ((A +- CST1) +- CST2): Allow some NOP conversions.
19019 2017-06-21 Jakub Jelinek <jakub@redhat.com>
19022 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
19023 with ctors/dtors if GOVD_SHARED is set.
19025 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19027 * config/aarch64/aarch64.md (movti_aarch64):
19028 Emit mov rather than orr.
19029 (movtf_aarch64): Likewise.
19030 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
19031 Emit mov rather than orr.
19033 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19035 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
19036 Swap alternatives, make integer dup more expensive.
19038 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19040 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
19041 Return true for non-tls symbols.
19043 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
19045 * config/aarch64/aarch64-cores.def (cortex-a55): New.
19046 (cortex-a75): Likewise.
19047 (cortex-a75.cortex-a55): Likewise.
19048 * config/aarch64/aarch64-tune.md: Regenerate.
19049 * doc/invoke.texi (-mtune): Document new values for -mtune.
19051 2017-06-21 Tom de Vries <tom@codesourcery.com>
19053 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
19054 stack_size feature.
19055 (Effective-Target Keywords, Other attributes): Suggest using
19056 dg-add-options stack_size feature to get stack limit in stack_size
19057 effective target documentation.
19059 2017-06-21 Julian Brown <julian@codesourcery.com>
19060 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19062 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
19063 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
19064 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
19066 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
19067 attribute type list for neon_multiply.
19068 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
19069 attribute type list for neon_multiply.
19070 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
19071 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
19072 attribute type list for neon_multiply.
19073 * config/arm/types.md (crypto_pmull): Add.
19074 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
19075 attribute type list.
19077 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
19079 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
19082 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19084 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
19085 to make sure not to dereference a NULL cost_classes_ptr pointer.
19087 2017-06-20 Carl Love <cel@us.ibm.com>
19089 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19090 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19091 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
19092 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
19093 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
19094 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
19095 VMULOSW): New enum "unspec" values.
19096 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
19097 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
19098 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
19099 altivec_vmulosw): New patterns.
19100 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
19101 VMULOSW): Add definitions.
19103 2017-06-20 Julia Koval <julia.koval@intel.com>
19105 * config/i386/i386.c: Fix rounding expand for new pattern.
19106 * config/i386/subst.md: Fix pattern (parallel -> unspec).
19108 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19110 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
19111 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
19113 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19115 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
19118 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19120 * config/aarch64/aarch64-cores.def: Rearrange to sort by
19121 architecture, then by implementer ID.
19122 * config/aarch64/aarch64-tune.md: Regenerate.
19124 2017-06-20 Richard Biener <rguenther@suse.de>
19126 PR middle-end/81097
19127 * fold-const.c (split_tree): Fold to type before negating.
19129 2017-06-20 David Malcolm <dmalcolm@redhat.com>
19131 * diagnostic-show-locus.c
19132 (selftest::test_fixit_deletion_affecting_newline): New function.
19133 (selftest::diagnostic_show_locus_c_tests): Call it.
19135 2017-06-20 Andreas Schwab <schwab@suse.de>
19138 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
19141 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19143 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
19144 __ARM_FEATURE_COPROC according to support.
19146 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19148 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
19149 Rewritten to avoid overflow for > 32-bit pointers.
19152 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
19153 by removing enum keyword.
19154 (ubsan_type_descriptor): Likewise. Formatting fix.
19157 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
19158 splitter): Require TARGET_SSE2 in the condition.
19160 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
19163 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
19164 for doing vector set of SFmode on ISA 3.0.
19165 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
19166 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
19168 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
19169 SFmode value into a V4SF variable that was extracted from another
19170 V4SF variable without converting the element to double precision
19171 and back to single precision vector format.
19172 (vsx_insert_extract_v4sf_p9_2): Likewise.
19174 2017-06-19 Jakub Jelinek <jakub@redhat.com>
19176 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
19177 in UWHI to avoid undefined overflow.
19180 * ubsan.h (enum ubsan_encode_value_phase): New.
19181 (ubsan_encode_value): Change second argument to
19182 enum ubsan_encode_value_phase with default value of
19183 UBSAN_ENCODE_VALUE_GENERIC.
19184 * ubsan.c (ubsan_encode_value): Change second argument to
19185 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
19186 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
19187 create_tmp_var_raw instead of create_tmp_var and use a
19189 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
19190 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
19191 ubsan_encode_value callers.
19194 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
19195 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
19196 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
19198 2017-06-19 Richard Biener <rguenther@suse.de>
19200 PR middle-end/81118
19201 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
19202 estimates if we changed anything.
19204 2017-06-19 Richard Biener <rguenther@suse.de>
19206 PR tree-optimization/80887
19207 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
19208 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
19209 simplified lookups, then reset mprts_hook.
19210 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
19212 (try_to_simplify): Likewise.
19214 2017-06-19 Martin Liska <mliska@suse.cz>
19217 * gimplify.c (gimplify_switch_expr):
19218 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
19220 2017-06-19 Martin Liska <mliska@suse.cz>
19222 * doc/install.texi: Document that PGO runs in 4 stages.
19224 2017-06-19 Martin Liska <mliska@suse.cz>
19227 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
19228 to dispatcher function name.
19229 * multiple_target.c (replace_function_decl): New function.
19230 (create_dispatcher_calls): Redirect both edges and references.
19232 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
19234 * profile-count.c (profile_count::dump): Dump quality.
19235 (profile_count::differs_from_p): Update for unsigned val.
19236 * profile-count.h (profile_count_quality): New enum.
19237 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
19239 2017-06-19 Richard Biener <rguenther@suse.de>
19241 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
19242 struct function as arg.
19243 (estimate_numbers_of_iterations): Export overload with loop arg.
19244 (free_numbers_of_iterations_estimates_loop): Use an overload of
19245 free_numbers_of_iterations_estimates instead.
19246 * tree-cfg.c (remove_bb): Adjust.
19247 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
19248 * tree-parloops.c (gen_parallel_loop): Likewise.
19249 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
19251 (tree_unroll_loops_completely): Likewise.
19252 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
19253 Use an overload instead and export.
19254 (estimated_loop_iterations): Adjust.
19255 (max_loop_iterations): Likewise.
19256 (likely_max_loop_iterations): Likewise.
19257 (estimate_numbers_of_iterations): Take struct function as arg
19259 (loop_exits_before_overflow): Adjust.
19260 (free_numbers_of_iterations_estimates_loop): Use an overload.
19261 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
19262 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
19264 2017-06-19 Richard Biener <rguenther@suse.de>
19267 * ipa-prop.c (find_constructor_constant_at_offset): Handle
19268 RANGE_EXPR conservatively.
19270 2017-06-16 Carl Love <cel@us.ibm.com>
19272 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19273 definitions for vec_float, vec_float2, vec_floato,
19274 vec_floate built-ins.
19275 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
19276 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
19278 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
19279 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
19280 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
19281 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
19282 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
19283 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
19284 vec_floato): Add builtin defines.
19285 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
19286 Update the built-in documentation file for the new built-in
19289 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19291 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
19292 (mthumb): Mark as the negative of -marm.
19294 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19296 * doc/invoke.texi (ARM Options, -mcpu): Document supported
19298 (ARM Options, -mtune): Document that this accepts the same
19299 extension options as -mcpu.
19300 (ARM Options, -mfpu): Document addition of -mfpu=auto.
19302 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19304 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
19305 permitted extensions.
19307 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19309 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
19310 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
19311 (armv8-m.main): Add option +nodsp.
19312 * config/arm/arm-cpu-cdata.h: Regenerated.
19314 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19316 * config/arm/t-fuchsia: New file.
19317 * config.gcc (arm*-*-fuchsia*): Use it.
19319 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19321 * config/arm/t-symbian: Rewrite for new option infrastructure.
19323 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19325 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
19326 (MULTILIB_REQUIRED): Likewise.
19328 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19330 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
19331 (MULTILIB_RESUE): Likewise.
19332 (MULTILIB_MATCHES): Likewise.
19333 (MULTLIB_REQUIRED): Likewise.
19335 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19337 * config/arm/t-rtems: Rewrite for new option framework.
19339 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19341 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
19342 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
19343 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
19344 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
19345 * config/arm/t-multilib: ... here.
19346 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
19347 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
19348 armv7-a and armv8*-a when A-profile libraries have not been built.
19349 * config/arm/t-rmprofile: Rewrite.
19351 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19353 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
19354 with a backslash. Remove the backslash after substituting unescaped
19356 * doc/fragments.texi (MULTILIB_REUSE): Document it.
19358 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19360 * config.gcc: (arm*-*-*): When building a-profile libraries, force
19361 the driver to pass through the default setting of -mfloat-abi.
19362 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
19364 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
19365 (all_feat_combs): New rule.
19366 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
19368 * config/arm/t-aprofile: Rewrite.
19370 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19372 * config/arm/arm.h (FPUTYPE_AUTO): Define.
19373 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
19374 fpu is not specified by the user/command-line.
19375 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
19376 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
19377 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
19378 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
19379 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
19380 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
19382 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19384 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
19385 * config/arm/t-arm-elf: Rewritten.
19387 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19389 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
19390 have some floating-point instructions.
19391 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
19392 (TARGET_MAYBE_HARD_FLOAT): New macro.
19393 * config/arm/arm-builtins.c (arm_init_builtins): Use
19394 TARGET_MAYBE_HARD_FLOAT.
19395 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
19397 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19399 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
19400 (configargs.h): Include it.
19401 (arm_print_hint_for_fpu_option): New function.
19402 (arm_parse_fpu_option): New function.
19403 (candidate_extension): New class.
19404 (arm_canon_for_multilib): New function.
19405 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
19406 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
19407 (ARCH_CANONICAL_SPECS): New macro.
19408 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
19410 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19412 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
19413 are set after handling multilib fragments. Set target_cpu_default2
19416 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19418 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
19420 (arm*-*-*): Set target_cpu_default2 to a quoted string.
19421 * config/arm/parsecpu.awk (check_cpu): Validate any extension
19423 (check_arch): Likewise.
19424 * config/arm/arm.c (arm_configure_build_target): Handle
19425 TARGET_CPU_DEFAULT being a string constant. Scan any feature
19426 options in the default.
19428 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19430 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
19431 when an option is an alias of another.
19432 * config/arm/parsecpu.awk (optalias): New parser token.
19433 (gen_comm_data): Mark non-alias options as such. Emit entries
19434 for extension aliases.
19435 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
19436 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
19437 (armv6kz, armv6zk, armv6t2): Likewise.
19438 (armv7): Make vfpv3-d16 an alias.
19439 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
19441 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
19442 Sort in canonical order.
19443 (armv8-a): Sort in canonical order.
19444 (armv8.1-a, armv8.2-a): Likewise.
19445 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
19447 (cortex-a9): Sort in canonical order.
19448 * config/arm/arm.c (selftests.h): Include it.
19449 (arm_test_cpu_arch_data): New function.
19450 (arm_run_self_tests): New function.
19451 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
19452 (targetm): Move declaration to the end of the file.
19453 * arm-cpu-cdata.h: Regenerated.
19455 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19457 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
19458 call to target_mode_check describing the type of option passed.
19459 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
19460 (arm_target_thumb_only): Use arm_parse_arch_option_name or
19461 arm_parse_cpu_option_name to match parameters against list of
19463 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
19464 arm_arch_core_flags data structure.
19465 * config/arm/arm-cpu_cdata.h: Regenerated.
19467 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19469 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
19471 (arm_print_hint_for_cpu_option): Likewise.
19472 (arm_print_hint_for_arch_option): Likewise.
19473 (arm_parse_cpu_option_name): Likewise.
19474 (arm_parse_arch_option_name): Likewise.
19475 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
19476 of entries in the all_fpus list.
19477 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
19478 (arm_parse_cpu_option_name): Declare.
19479 (arm_parse_arch_option_name): Declare.
19480 (arm_parse_option_features): Declare.
19481 (arm_intialize_isa): Declare.
19482 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
19484 (gen_comm_data): ... here. Make definitions non-static.
19485 * config/arm/arm-cpu-data.h: Regenerated.
19486 * config/arm/arm-cpu-cdata.h: Regenerated.
19488 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19490 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
19491 (cpu_arch_extension): New structure.
19492 (cpu_arch_option, arch_option, cpu_option): New structures.
19493 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
19494 architecture types.
19495 (gen_data): Generate new format data tables.
19496 * config/arm/arm.c (cpu_tune): New structure.
19497 (cpu_option, processors): Delete.
19498 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
19499 (arm_print_hint_for_cpu_option): ... this and ...
19500 (arm_print_hint_for_arch_option): ... this.
19501 (arm_parse_arch_cpu_name): Delete. Replace with ...
19502 (arm_parse_cpu_option_name): ... this and ...
19503 (arm_parse_arch_option_name): ... this.
19504 (arm_unrecognized_feature): Change type of target parameter to
19506 (arm_parse_arch_cpu_features): Delete. Replace with ...
19507 (arm_parse_option_features): ... this.
19508 (arm_configure_build_target): Rework to use new configuration data
19510 (arm_print_tune_info): Rework for new configuration data tables.
19511 * config/arm/arm-cpu-data.h: Regenerated.
19512 * config/arm/arm-cpu.h: Regenerated.
19514 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19516 * Makefile.in (OBJS): Move sbitmap.o from here ...
19517 (OBJS-libcommon): ... to here.
19519 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19521 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
19522 (ISA_ALL_CRYPTO): New macro.
19523 (ISA_ALL_SIMD): New macro
19524 (ISA_ALL_FP): New macro.
19525 * config/arm/arm.c (fpu_bitlist): Update initializer.
19526 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
19528 (arm9e): Add fpu. Add option for nofp
19529 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
19530 (arm926ej-s, arm1026ej-s): Likewise.
19531 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
19532 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
19533 neon-fp16, neon-vfpv4, nofp and nosimd.
19534 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
19535 (cortex-a8): Add fpu. Add option for nofp.
19536 (cortex-a9): Add fpu. Add options for nosimd and nofp.
19537 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
19538 (cortex-r4f): Add fpu.
19539 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
19540 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
19542 (cortex-r8): Likewise.
19543 (cortex-m4): Add fpu. Add option for nofp.
19544 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
19545 (cortex-a17.cortex-a7): Likewise.
19546 (cortex-a32): Add fpu. Add options for crypto and nofp.
19547 (cortex-a35, cortex-a53): Likewise.
19548 (cortex-a57): Add fpu. Add option for crypto.
19549 (cortex-a72, cortex-a73): Likewise.
19550 (exynos-m1): Likewise.
19551 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
19552 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
19553 (cortex-m33): Add fpu. Add option for nofp.
19554 * config/arm/arm-cpu-cdata.h: Regenerated
19555 * config/arm/arm-cpu-data.h: Regenerated.
19557 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19559 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
19560 (armv5te, armv5tej): Likewise.
19561 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
19562 (armv7): Add options fp and vfpv3-d16.
19563 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
19564 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
19566 (armv7ve): Likewise.
19567 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
19568 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
19569 (armv8-a): Add nocrypto option.
19570 (armv8.1-a, armv8.2-a): Likewise.
19571 (armv8-m.main): add options fp, fp.dp and nofp.
19573 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19575 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
19577 (armv8-a+crc): Delete.
19578 (armv8.1-a): Add options simd, crypto and nofp.
19579 (armv8.2-a): Add options fp16, simd, crypto and nofp.
19580 (armv8.2-a+fp16): Delete.
19581 (armv8-m.main): Add option dsp.
19582 (armv8-m.main+dsp): Delete.
19583 (cortex-a8): Add fpu. Add nofp option.
19584 (cortex-a9): Add fpu. Add nofp and nosimd options.
19585 * config/arm/parsecpu.awk (gen_data): Generate option tables and
19586 link to main cpu and architecture data structures.
19587 (gen_comm_data): Only put isa attributes from the main architecture
19589 (option): New statement for architecture and CPU entries.
19590 * arm.c (struct cpu_option): New structure.
19591 (struct processors): Add entry for options.
19592 (arm_unrecognized_feature): New function.
19593 (arm_parse_arch_cpu_name): Ignore any characters after the first
19595 (arm_parse_arch_cpu_feature): New function.
19596 (arm_configure_build_target): Separate out any CPU and architecture
19597 features and parse separately. Don't error out if -mfpu=auto is
19598 used with only an architecture string.
19599 (arm_print_asm_arch_directives): New function.
19600 (arm_file_start): Call it.
19601 * config/arm/arm-cpu-cdata.h: Regenerated.
19602 * config/arm/arm-cpu-data.h: Likewise.
19603 * config/arm/arm-tables.opt: Likewise.
19605 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19607 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
19608 assembler when it is not -mfpu=auto.
19610 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19612 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
19613 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
19614 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
19615 (ASM_CPU_SPEC): Rewrite.
19616 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
19617 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
19618 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
19619 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
19620 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
19621 copied string is NUL-terminated. Also strip any characters prefixed
19623 (arm_rewrite_selected_arch): New function.
19624 (arm_rewrite_march): New function.
19626 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19628 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
19629 (x_arm_cpu_string, x_arm_tune_string): Likewise.
19630 (march, mcpu, mtune): Convert to string-based options.
19631 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
19632 (arm_parse_arch_cpu_name): New function.
19633 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
19634 identify selected architecture or CPU.
19635 (arm_option_save): New function.
19636 (TARGET_OPTION_SAVE): Redefine.
19637 (arm_option_restore): Restore string options.
19638 (arm_option_print): Print string options.
19640 2017-06-16 Martin Sebor <msebor@redhat.com>
19642 PR tree-optimization/80933
19643 PR tree-optimization/80934
19644 * builtins.c (fold_builtin_3): Do not handle bcmp here.
19645 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
19646 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
19647 (gimple_fold_builtin): Call them.
19649 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19651 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
19652 as unlikely; update profile.
19654 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19656 * predict.c (force_edge_cold): Handle declaring edges impossible
19659 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19661 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
19663 (try_unroll_loop_completely): Fix reporting.
19665 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19667 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
19669 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
19672 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
19673 if given a non-constant argument for an intrinsic which requires a
19676 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19678 * profile.c (compare_freqs): New function.
19679 (branch_prob): Sort edge list.
19680 (find_spanning_tree): Assume that the list is priority sorted.
19682 2017-06-16 Richard Biener <rguenther@suse.de>
19684 PR tree-optimization/81090
19685 * passes.def (pass_record_bounds): Remove.
19686 * tree-pass.h (make_pass_record_bounds): Likewise.
19687 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
19688 make_pass_record_bounds): Likewise.
19689 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
19690 not free niter estimates at the beginning but at the end.
19691 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
19693 2017-06-16 Richard Biener <rguenther@suse.de>
19695 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
19696 initializer to workaround ICE in host GCC 4.8.
19698 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19700 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
19702 (clone_inlined_nodes): Update.
19704 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19706 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
19707 prefetch settings, and enable prefetching by default at -O3.
19709 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19711 * config/aarch64/aarch64.c (aarch64_override_options_internal):
19712 Set flag_prefetch_loop_arrays according to tuning data.
19714 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19716 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
19717 New tune structure.
19718 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
19719 [Unrelated to main purpose of the patch] Place the pointer field last
19720 to enable type checking errors when tune structure are wrongly merged.
19721 * config/aarch64/aarch64.c (generic_prefetch_tune,)
19722 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
19723 (thunderx2t99_prefetch_tune): New tune constants.
19724 (tune_params *_tunings): Update all tunings (no functional change).
19725 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
19726 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
19727 from tunings structures.
19729 2017-06-16 Jakub Jelinek <jakub@redhat.com>
19732 * ubsan.c (instrument_null): Add T argument, use it instead
19733 of computing it based on IS_LHS.
19734 (instrument_object_size): Likewise.
19735 (pass_ubsan::execute): Adjust instrument_null and
19736 instrument_object_size callers to pass gimple_get_lhs or
19737 gimple_assign_rhs1 result to it. Use instrument_null instead of
19738 calling get_base_address and instrument_mem_ref. Handle
19739 aggregate call arguments for object-size sanitization.
19741 2017-06-16 Yury Gribov <tetra2005@gmail.com>
19743 PR tree-optimization/81089
19744 * tree-vrp.c (is_masked_range_test): Validate operands of
19747 2017-06-15 Martin Sebor <msebor@redhat.com>
19750 * dumpfile.c (dump_register): Avoid calling memset to initialize
19751 a class with a default ctor.
19752 * gcc.c (struct compiler): Remove const qualification.
19753 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
19754 * hash-table.h: Ditto.
19755 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
19757 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
19758 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
19760 * params.h (struct param_info): Make struct members non-const.
19761 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
19762 with default initialization.
19763 * vec.h (vec_copy_construct, vec_default_construct): New helper
19765 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
19766 with vec_copy_construct.
19767 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
19768 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
19769 * doc/invoke.texi (-Wclass-memaccess): Document.
19771 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19773 * emit-rtl.h (is_leaf): Update comment about local
19774 register allocator.
19776 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19779 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
19780 for a variable to have a section before checking if the section has a
19782 Set section to.persistent if persistent attribute is set.
19783 Warn if .persistent attribute is used on an automatic variable.
19785 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
19787 PR rtl-optimization/80474
19788 * reorg.c (update_block): Do not ignore instructions in a delay slot.
19790 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
19792 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
19795 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
19797 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
19798 (casesi): Emit bounds checking as RTL.
19799 (casesi_internal_mips16_<mode>): Remove bounds checking.
19801 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
19803 * config/xtensa/xtensa.c (xtensa_option_override): Append
19804 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
19805 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
19806 xtensa_doloop_hooks): Define unconditionally.
19807 (xtensa_reorg_loops): Only call reorg_loops in the presence of
19809 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
19810 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
19811 for it in xtensa_option_override.
19812 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
19813 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
19815 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
19817 * doc/cppopts.texi: Document '-' special value to -MF.
19819 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
19821 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
19822 (cortex_a53_fconst): Likewise.
19823 (cortex_a53_fpmul): Likewise.
19824 (cortex_a53_f_load_64): Likewise.
19825 (cortex_a53_f_load_many): Likewise.
19826 (cortex_a53_advsimd_alu): Likewise.
19827 (cortex_a53_advsimd_alu_q): Likewise.
19828 (cortex_a53_advsimd_mul): Likewise.
19829 (cortex_a53_advsimd_mul_q): Likewise.
19830 (fpmac bypass): Add new bypass for fpmac-fpmac case.
19831 Add missing fmul, r2f_cvt and fconst cases.
19833 2017-06-14 Richard Biener <rguenther@suse.de>
19835 PR middle-end/81088
19836 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
19838 (fold_binary_loc): When associating do not treat pre-existing
19839 TREE_OVERFLOW on literal constants as a reason to allow
19840 TREE_OVERFLOW on associated literal constants.
19842 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19844 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
19845 (MASK_FEATURES): New macro.
19846 * config/sparc/sparc.c (sparc_option_override): Remove the special
19847 handling of -mfpu and generalize it to all MASK_FEATURES switches.
19849 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19851 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
19852 a division of 0 if non-call exceptions are enabled.
19854 2017-06-14 Andrew Pinski <apinski@cavium.com>
19855 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19858 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
19859 Improve vector initialization code gen for only variable case.
19861 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19863 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
19865 2017-06-14 Richard Biener <rguenther@suse.de>
19867 PR tree-optimization/81083
19868 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
19871 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19873 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
19874 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
19875 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
19876 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
19877 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
19878 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
19880 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19882 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
19883 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
19885 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19887 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
19889 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19891 * config/rs6000/t-rtems: Don't handle SPE.
19893 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19895 * config/rs6000/t-linux: Don't handle SPE.
19897 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19899 * config/rs6000/eabispe.h: Delete file.
19901 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19903 * config/rs6000/t-spe: Delete file.
19905 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19907 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
19908 (rs6000_legitimate_offset_address_p): Return false for anything in
19909 V2SImode or V2SFmode.
19911 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19913 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
19914 except V2SF and V2SI. Rearrange the vector modes, and add comments.
19915 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
19917 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
19918 (rs6000_legitimate_offset_address_p): Ditto.
19919 (rs6000_emit_move): Ditto.
19920 (rs6000_init_builtins): Remove V4HI_type_node.
19922 2017-06-13 Martin Liska <mliska@suse.cz>
19925 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
19926 (gate_asan): Likewise.
19927 * asan.h (asan_no_sanitize_address_p): Remove the function.
19928 (sanitize_flags_p): New function.
19929 * builtins.def: Fix coding style.
19930 * common.opt: Use renamed enum value.
19931 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
19932 * doc/extend.texi: Document no_sanitize attribute.
19933 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
19934 to SANITIZE_UNDEFINED_NONDEFAULT.
19935 * gcc.c (sanitize_spec_function): Use the renamed enum value.
19936 * gimple-fold.c (optimize_atomic_compare_exchange_p):
19937 Use sanitize_flags_p.
19938 * gimplify.c (gimplify_function_tree): Likewise.
19939 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
19940 * opts.c (parse_no_sanitize_attribute): New function.
19941 (common_handle_option): Use renamed enum value.
19942 * opts.h (parse_no_sanitize_attribute): Declare.
19943 * tree.c (sanitize_flags_p): New function.
19944 * tree.h: Declared here.
19945 * tsan.c: Use sanitize_flags_p.
19946 * ubsan.c (ubsan_expand_null_ifn): Likewise.
19947 (instrument_mem_ref): Likewise.
19948 (instrument_bool_enum_load): Likewise.
19949 (do_ubsan_in_current_function): Remove the function.
19950 (pass_ubsan::execute): Use sanitize_flags_p.
19951 * ubsan.h: Remove do_ubsan_in_current_function
19952 * tree-cfg.c (print_no_sanitize_attr_value): New function.
19953 (dump_function_to_file): Use it here.
19955 2017-06-13 Martin Jambor <mjambor@suse.cz>
19957 PR tree-optimization/80803
19958 PR tree-optimization/81063
19959 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
19960 (propagate_subaccesses_across_link): Enqueue subtree whenever
19961 necessary instead of relying on the caller.
19963 2017-06-13 Martin Jambor <mjambor@suse.cz>
19965 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
19966 that have a first_link.
19967 (sort_and_splice_var_accesses): Do not check first_link before
19969 (subtree_mark_written_and_enqueue): Likewise.
19970 (propagate_all_subaccesses): Likewise and do not stop at first
19971 parent with a first_link.
19973 2017-06-13 Martin Jambor <mjambor@suse.cz>
19975 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
19978 2017-06-13 Yury Gribov <tetra2005@gmail.com>
19980 * match.pd: New pattern.
19982 2017-06-13 Yury Gribov <tetra2005@gmail.com>
19984 * tree-vrp.c (is_masked_range_test): New function.
19985 (register_edge_assert_for): Determine ranges for
19988 2017-06-13 Yury Gribov <tetra2005@gmail.com>
19990 PR tree-optimization/67328
19991 * fold-const.c (maskable_range_p): New function.
19992 (build_range_check): Generate bittests if possible.
19994 2017-06-13 Martin Liska <mliska@suse.cz>
19996 * gimple-pretty-print.c (dump_probability): Add new argument.
19997 (dump_edge_probability): Dump both probability and count.
19998 (dump_gimple_label): Likewise.
19999 (dump_gimple_bb_header): Likewise.
20001 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
20004 * config/avr/avr-devices.c: Fix indentation.
20005 * config/avr/gen-avr-mmcu-specs.c: Dito.
20007 2017-06-13 Richard Biener <rguenther@suse.de>
20009 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
20010 instead get vector type from stmt_info.
20011 (vectorizable_reduction): Adjust. Remove dead code.
20013 2017-06-13 Richard Biener <rguenther@suse.de>
20015 PR middle-end/81065
20016 * fold-const.c (extract_muldiv_1): Remove bogus distribution
20017 case of C * (x * C2 + C3).
20018 (fold_addr_of_array_ref_difference): Properly fold index difference.
20020 2017-06-12 David S. Miller <davem@davemloft.net>
20023 * config/sparc/sparc.md (return expander): Emit frame blockage if
20024 function uses alloca.
20026 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
20028 * combine.c (make_field_assignment): Check len rather than the mode
20029 precision when calling force_to_mode.
20031 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
20033 Support multilibs and devices that see flash in RAM address range.
20036 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
20037 (avr_mcu_t) <flash_pm_offset>: New field.
20038 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
20039 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
20040 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
20041 (AVR_TINY_PM_OFFSET): Remove macro.
20042 * config/avr/avr.opt (-mshort-calls): New option.
20043 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
20044 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
20045 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
20046 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
20047 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
20048 instead of avr_arch->have_jmp_call.
20049 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
20050 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
20051 avr_arch->flash_pm_offset to define.
20052 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
20053 new field flash_pm_offset. Add entry for avrxmega3.
20054 (avr_texinfo): Add entry for avrxmega3.
20055 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
20056 attiny212, attiny214,
20057 attiny412, attiny414, attiny416, attiny417,
20058 attiny814, attiny816, attiny817,
20059 attiny1614, attiny1616, attiny1617,
20060 attiny3214, attiny3216, attiny3217.
20061 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
20062 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
20063 (avr_print_operand_address) [AVR_TINY]: Same.
20064 (avr_asm_init_sections) <readonly_data_section>: Only patch
20065 callback if avr_arch->flash_pm_offset = 0.
20066 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
20067 for rodata if avr_arch->flash_pm_offset != 0.
20068 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
20069 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
20070 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
20071 (m_options): Append opt_rcall.
20072 (m_dirnames): Append dir_rcall.
20073 * config/avr/t-multilib: Regenerate.
20075 * configure.ac [target=avr]: Check whether avrxmega3 default
20076 linker description file works as needed.
20077 * configure: Regenerate.
20078 * doc/avr-mmcu.texi: Regenerate.
20079 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
20080 <__AVR_ARCH__>: Document avrxmega3 and 103.
20081 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
20082 <__AVR_SHORT_CALLS__>: Document it.
20083 <__AVR_PM_BASE_ADDRESS__>: Document it.
20084 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
20085 (AVR Variable Attributes) <progmem>: Document this is
20086 not needed for avrxmega3.
20087 (AVR Named Address Spaces) <__flash>: Dito.
20089 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
20091 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
20093 2017-06-12 Doug Rupp <rupp@adacore.com>
20095 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
20096 Append vxworks-stdint.h to the tm_file list.
20097 * config/vxworks-stdint.h: New file.
20099 2017-06-12 Martin Liska <mliska@suse.cz>
20101 PR tree-optimization/81041
20102 * tree-profile.c (gimple_gen_ic_func_profiler):
20103 Create an extra BB in profile-generate
20104 (gimple_gen_time_profiler): Likewise.
20106 2017-06-12 Jakub Jelinek <jakub@redhat.com>
20108 PR tree-optimization/81003
20109 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
20110 (update_range_test): Use it instead of force_gimple_operand_gsi.
20112 2017-06-12 Richard Biener <rguenther@suse.de>
20114 PR tree-optimization/81053
20115 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
20116 with backedge value not defined in loop. Simplify def stmt
20119 2017-06-11 Tom de Vries <tom@codesourcery.com>
20122 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
20124 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
20125 nvptx_cannot_force_const_mem.
20127 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20129 * opts.c (finish_options): Move test for flag_split_stack after
20130 it has been initialized.
20132 2017-06-11 Jason Merrill <jason@redhat.com>
20134 * tree.h (id_equal): New.
20135 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
20136 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
20137 instead of strcmp of IDENTIFIER_POINTER.
20139 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20141 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
20142 (mark_all_inlined_calls_cdtor): Fix formating.
20143 (inline_transform): Rescale profile before inlining.
20145 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20147 * cgraph.h (cgraph_edge::clone): Update prototype.
20148 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
20149 (cgraph_node::create_clone): Update.
20150 (cgraph_node::create_version_clone): Update.
20151 * tree-inline.c (copy_bb): Update.
20152 (expand_call_inline): Update.
20154 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20156 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
20157 factored out from ...
20158 (rs6000_emit_prologue): ... here.
20160 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20162 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
20163 factored out from ...
20164 (rs6000_emit_prologue): ... here.
20166 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20168 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
20170 (handle_missing_profiles): Fix computation of tp_first_run.
20171 (counts_to_freqs): Do not touch freqs when count is 0.
20173 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20175 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
20178 2017-06-10 Tom de Vries <tom@codesourcery.com>
20180 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
20181 attributes): Document signal effective target.
20183 2017-06-10 Tom de Vries <tom@codesourcery.com>
20185 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
20186 Document effective target stack_size.
20188 2017-06-09 David Malcolm <dmalcolm@redhat.com>
20190 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
20191 to the edit_context if they can be auto-applied.
20193 2017-06-9 Ian Lance Taylor <iant@golang.org>
20195 * opts.c (finish_options): If -fsplit-stack, disable implicit
20196 -forder-blocks-and-partition.
20197 * doc/invoke.texi (Optimize Options): Document that when using
20198 -fsplit-stack -forder-blocks-and-partition is not implicitly
20201 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20203 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
20204 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
20205 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
20206 * builtins.def (abort, trap, unreachable): Declare cold.
20207 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
20208 * tree-core.h (ECF_COLD): New.
20209 * tree.c (set_call_expr_flags): Handle ECF_COLD.
20210 (build_common_builtin_nodes): Mark unreachable and abort as cold.
20212 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20214 * predict.c (unlikely_executed_stmt_p): Cleanup.
20216 2017-06-09 Richard Biener <rguenther@suse.de>
20218 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
20219 model if the ref is always written to.
20221 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20223 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
20225 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20227 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
20230 2017-06-09 Tom de Vries <tom@codesourcery.com>
20233 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
20234 "target cannot support label values" when encountering LABEL_REF.
20236 2017-06-09 Martin Liska <mliska@suse.cz>
20238 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
20239 (gimple_gen_ic_func_profiler): Emit direct comparison
20240 of __gcov_indirect_call_callee with NULL.
20241 (gimple_gen_time_profiler): Change probability from
20242 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
20244 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20246 * profile.c (edge_gcov_counts): Turn to pointer.
20247 (compute_branch_probabilities, compute_branch_probabilities): Update.
20248 (branch_prob): Do not clear edge_gcov_count.
20249 * profile.h (edge_gcov_counts): Turn to pointer.
20250 (edge_gcov_count): Update.
20252 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20254 * gimple.h (gimple_check_failed): Mark cold.
20256 2017-06-09 Richard Biener <rguenther@suse.de>
20258 PR tree-optimization/66623
20259 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
20260 refactor check_reduction into two parts, properly computing
20261 whether we have to check reduction validity for outer loop
20264 2017-06-09 Richard Biener <rguenther@suse.de>
20266 PR tree-optimization/79483
20267 * graphite-scop-detection.c (order): New global.
20268 (get_order): Compute bb to order mapping that satisfies code
20269 generation constraints.
20270 (cmp_pbbs): New helper.
20271 (build_scops): Start domwalk at entry block, sort generated
20274 2017-06-09 Richard Biener <rguenther@suse.de>
20276 PR middle-end/81007
20277 * ipa-polymorphic-call.c
20278 (ipa_polymorphic_call_context::restrict_to_inner_class):
20279 Skip FIELD_DECLs with error_mark_node type.
20280 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
20283 2017-06-09 Martin Liska <mliska@suse.cz>
20285 * predict.c (struct branch_predictor): New struct.
20286 (test_prediction_value_range): New test.
20287 (predict_c_tests): New function.
20288 * selftest-run-tests.c (selftest::run_tests): Run the function.
20289 * selftest.h: Declare new tests.
20291 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
20294 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
20295 gen_add3_insn did not fail.
20296 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
20297 r0, construct that number in a temporary reg and add that reg to r0.
20298 If asked to put the result in r0 as well, fail.
20300 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
20302 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
20303 for early expansion of vec_eqv.
20305 2017-06-08 Jakub Jelinek <jakub@redhat.com>
20307 PR middle-end/81005
20308 * ubsan.c (instrument_null): Avoid pointless code temporary.
20309 (pass_ubsan::execute): Instrument aggregate arguments of calls.
20311 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
20315 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
20318 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
20321 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20323 * predict.c (unlikely_executed_edge_p): Move ahead.
20324 (probably_never_executed_edge_p): Use it.
20326 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20328 PR middle-end/79988
20329 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
20330 gimple_call_builtin_p call.
20332 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20334 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
20335 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
20336 rtl_check_failed_type2, rtl_check_failed_code1,
20337 rtl_check_failed_code2, rtl_check_failed_code_mode,
20338 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
20339 rtvec_check_failed_bounds, rtl_check_failed_flag,
20340 _fatal_insn_not_found, _fatal_insn): Likewise.
20341 * tree.h (tree_contains_struct_check_failed,
20342 tree_check_failed, tree_not_check_failed,
20343 tree_class_check_failed, tree_range_check_failed,
20344 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
20345 tree_vec_elt_check_failed, phi_node_elt_check_failed,
20346 tree_operand_check_failed, omp_clause_check_failed,
20347 omp_clause_operand_check_failed, omp_clause_range_check_failed):
20350 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20352 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
20353 flag_branch_probabilities.
20354 * ipa-inline.c (edge_badness): Likewise.
20355 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
20356 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
20357 * predict.c (maybe_hot_frequency_p): Likewise.
20358 (probably_never_executed): Likewise.
20359 * sched-ebb.c (schedule_ebbs): Likewise.
20360 * sched-rgn.c (find_single_block_region): Likewise.
20361 * tracer.c (tail_duplicate): Likewise.
20363 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20365 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
20366 longer requires x_flag_profile_use.
20368 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20370 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
20371 instead of flag_reorder_blocks_and_partition.
20372 * dbxout.c (dbxout_function_end): Likewise.
20373 * dwarf2out.c (gen_subprogram_die): Likewise.
20374 * haifa-sched.c (sched_create_recovery_edges): Likewise.
20375 * hw-doloop.c (reorg_loops): Likewise.
20376 * varasm.c (assemble_start_function,
20377 assemble_end_function): Likewise.
20378 (decide_function_section): Do not check for
20379 flag_reorder_blocks_and_partition.
20381 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20383 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20385 (chkp_get_hard_register_fake_addr_expr): Ditto.
20386 (chkp_build_addr_expr): Add check for hard reg case.
20387 (chkp_parse_array_and_component_ref): Ditto.
20388 (chkp_find_bounds_1): Ditto.
20389 (chkp_process_stmt): Don't generate bounds store for
20392 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20394 * predict.c (maybe_hot_bb_p): Do not check profile status.
20395 (maybe_hot_edge_p): Likewise.
20396 (probably_never_executed): Check for zero counts even if profile
20398 (unlikely_executed_edge_p): New function.
20399 (unlikely_executed_stmt_p): New function.
20400 (unlikely_executed_bb_p): New function.
20401 (set_even_probabilities): Use unlikely predicates.
20402 (combine_predictions_for_bb): Likewise.
20403 (predict_paths_for_bb): Likewise.
20404 (predict_paths_leading_to_edge): Likewise.
20405 (determine_unlikely_bbs): New function.
20406 (estimate_bb_frequencies): Use it.
20407 (compute_function_frequency): Use zero counts even if profile is
20409 * profile-count.h: Fix typo.
20411 2017-08-08 Julia Koval <julia.koval@intel.com>
20413 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
20414 _mm512_mask_cvtsepi16_storeu_epi8,
20415 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
20416 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
20417 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
20418 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
20419 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
20420 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
20421 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
20422 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
20423 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
20424 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
20425 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
20426 __builtin_ia32_pmovuswb256mem_mask,
20427 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
20428 __builtin_ia32_pmovwb512mem_mask): New builtins.
20430 2017-08-08 Julia Koval <julia.koval@intel.com>
20432 PR target/73350,80862
20433 * config/i386/subst.md (round): Fix round pattern.
20434 * config/i386/i386.c (ix86_erase_embedded_rounding):
20435 Fix erasing rounding for the fixed pattern.
20437 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20439 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
20441 2017-06-08 Martin Liska <mliska@suse.cz>
20443 PR gcov-profile/80911
20444 * gcov.c (block_info::block_info): New constructor.
20446 2017-06-07 Carl Love <cel@us.ibm.com>
20448 * config/rs6000/rs6000-c: The return type of the following
20449 built-in functions was implemented as int not long long. Fix sign
20450 of return value for the unsigned version of vec_mulo and vec_mule.
20451 vector unsigned long long vec_bperm (vector unsigned long long,
20452 vector unsigned char)
20453 vector signed long long vec_mule (vector signed int,
20455 vector unsigned long long vec_mule (vector unsigned int,
20456 vector unsigned int)
20457 vector signed long long vec_mulo (vector signed int,
20459 vector unsigned long long vec_mulo (vector unsigned int,
20460 vector unsigned int)
20461 * doc/extend.texi: Fix the documentation for the built-in
20464 2017-06-07 Carl Love <cel@us.ibm.com>
20467 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
20470 2017-06-07 Carl Love <cel@us.ibm.com>
20472 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
20473 support, Generate doublehv for signed int/float for BE case only.
20475 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
20477 * doc/invoke.texi (mcx16): Rewrite.
20479 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20481 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
20482 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
20483 *mov<mode>_softfloat, and an anonymous splitter): Use
20484 nonimmediate_operand instead of rs6000_nonimmediate_operand.
20486 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20488 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
20490 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
20491 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
20492 (rs6000_debug_reg_global): Adjust.
20493 (rs6000_init_hard_regno_mode_ok): Adjust.
20494 (rs6000_dbx_register_number): Adjust.
20495 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
20496 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
20497 Remove SPE_ACC and SPEFSCR.
20498 (REG_ALLOC_ORDER): Ditto.
20499 (FRAME_POINTER_REGNUM): Change to 111.
20500 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
20501 (REG_CLASS_NAMES): Ditto.
20502 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
20503 (REGISTER_NAMES): Ditto.
20504 (ADDITIONAL_REG_NAMES): Ditto.
20505 (rs6000_reg_names): Ditto.
20506 * config/rs6000/rs6000.md: Renumber some register number
20509 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20511 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
20513 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
20514 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
20516 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
20517 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
20518 Delete the SPE high registers.
20519 (REG_ALLOC_ORDER): Ditto.
20520 (enum reg_class): Remove SPE_HIGH_REGS.
20521 (REG_CLASS_NAMES): Ditto.
20522 (REG_CLASS_CONTENTS): Delete the SPE high registers.
20523 (REGISTER_NAMES): Ditto.
20524 (rs6000_reg_names): Ditto.
20525 * doc/tm.texi.in: Remove SPE as example.
20526 * doc/tm.texi: Regenerate.
20528 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20530 * config/rs6000/8540.md (ppc8540_brinc): Delete.
20531 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
20532 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
20533 * config/rs6000/rs6000.md (type): Remove "brinc".
20535 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20537 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
20538 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
20539 * config/rs6000/linuxspe.h: Delete file.
20540 * config/rs6000/rs6000.md: Don't include spe.md.
20541 * config/rs6000/spe.h: Delete file.
20542 * config/rs6000/spe.md: Delete file.
20543 * config/rs6000/t-rs6000: Remove spe.md.
20545 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20547 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
20548 (reg_or_none500mem_operand): Delete.
20549 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
20550 instead of reg_or_none500mem_operand.
20552 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20554 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
20555 handling of SPE flags.
20556 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
20558 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20560 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
20562 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
20563 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
20564 paired_divv2sf3): Similar.
20565 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
20566 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
20567 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
20569 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
20570 Rename the paired_* instruction patterns.
20571 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
20573 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
20574 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
20575 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
20576 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
20577 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
20578 PAIRED_VECTOR_MODE.
20579 (struct machine_function): Delete field spe_insn_chain_scanned_p.
20580 (spe_func_has_64bit_regs_p): Delete.
20581 (spe_expand_predicate_builtin): Delete.
20582 (spe_expand_evsel_builtin): Delete.
20583 (TARGET_DWARF_REGISTER_SPAN): Do not define.
20584 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
20585 (invalid_e500_subreg): Delete.
20586 (rs6000_legitimize_address): Always force_reg op2 as well, for
20587 paired single memory accesses.
20588 (rs6000_member_type_forces_blk): Delete.
20589 (rs6000_spe_function_arg): Delete.
20590 (rs6000_expand_unop_builtin): Delete SPE handling.
20591 (rs6000_expand_binop_builtin): Ditto.
20592 (spe_expand_stv_builtin): Delete.
20593 (bdesc_2arg_spe): Delete.
20594 (spe_expand_builtin): Delete.
20595 (spe_expand_predicate_builtin): Delete.
20596 (spe_expand_evsel_builtin): Delete.
20597 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
20598 (spe_init_builtins): Delete.
20599 (spe_func_has_64bit_regs_p): Delete.
20600 (savres_routine_name): Delete "info" parameter. Adjust callers.
20601 (rs6000_emit_stack_reset): Ditto.
20602 (rs6000_dwarf_register_span): Delete.
20603 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
20604 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
20605 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
20607 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
20609 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
20610 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
20611 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
20612 mulv2sf3, divv2sf3): Delete expanders.
20614 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20616 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
20618 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20620 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
20621 * config/rs6000/rs6000.c: Ditto.
20623 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20625 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
20626 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
20627 comparison_operator.
20629 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20631 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
20632 * config/rs6000/rs6000.opt: Ditto.
20633 * config/rs6000/t-rtems: Ditto.
20635 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20637 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
20638 TARGET_E500_SINGLE by 0, simplify.
20639 * config/rs6000/rs6000.c: Ditto.
20640 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
20641 (spe_build_register_parallel): Delete.
20642 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
20643 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
20644 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
20645 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
20646 (E500_CONVERT): Delete.
20647 * config/rs6000/spe.md: Remove many patterns and all define_constants.
20649 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20651 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
20652 * config/rs6000/dfp.md: Ditto.
20653 (negdd2, *negdd2_fpr): Merge.
20654 (absdd2, *absdd2_fpr): Merge.
20655 (negtd2, *negtd2_fpr): Merge.
20656 (abstd2, *abstd2_fpr): Merge.
20657 * config/rs6000/e500.h: Delete file.
20658 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
20659 TARGET_FPRS by 1 and simplify.
20660 * config/rs6000/rs6000-c.c: Ditto.
20661 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
20662 TARGET_DF_SPE by 0.
20663 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
20665 * config/rs6000/rs6000.md: Ditto.
20666 (floatdidf2, *floatdidf2_fpr): Merge.
20667 (move_from_CR_gt_bit): Delete.
20668 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
20669 (E500_CR_IOR_COMPARE): Delete.
20670 (All patterns that require !TARGET_FPRS): Delete.
20671 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
20673 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20675 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
20677 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20679 * graphds.c (add_edge): Intitialize edge's attached data.
20680 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
20681 pointer parameter. Call pointed function on each edge during
20682 graph traversing. Skip traversing the edge when the function
20684 (graphds_dfs, graphds_scc): Ditto.
20685 (for_each_edge): New parameter. Pass the new parameter to callback
20687 * graphds.h (skip_edge_callback): New function pointer type.
20688 (graphds_dfs, graphds_scc): New function pointer parameter.
20689 (graphds_edge_callback, for_each_edge): New parameter.
20691 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20693 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
20694 out code checking if runtime alias check is possible to below ...
20695 Call the new function.
20696 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
20697 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
20699 2017-06-07 Marek Polacek <polacek@redhat.com>
20702 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
20703 TYPE_OVERFLOW_WRAPS check.
20705 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20707 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
20708 if versioning is required.
20709 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
20710 peeling with the check for versioning.
20712 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20714 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
20715 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
20716 Set true to new parameter if new ssa variable is defined.
20717 (vect_gen_vector_loop_niters): Refactor. Set range information
20718 for the new vector loop bound variable.
20719 (vect_do_peeling): Ditto.
20721 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20723 * tree-affine.c (ssa.h): Include header file.
20724 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
20725 has wrapping overflow behavior.
20727 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20729 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
20731 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20733 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
20734 (tree_to_aff_combination): ... here.
20736 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20738 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
20739 reg_pressure model function.
20740 (ivopts_global_cost_for_size): Delete.
20741 (determine_set_costs, iv_ca_recount_cost): Call new model function
20742 ivopts_estimate_reg_pressure.
20744 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20746 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
20747 expensive than udiv. Remove floating point cases from mod.
20749 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20751 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
20752 Increase idiv cost.
20754 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20756 * config/aarch64/aarch64.md
20757 (copysignsf3): Fix mask generation.
20759 2017-06-07 Jakub Jelinek <jakub@redhat.com>
20761 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
20763 (class dump_manager): Add register_dumps method.
20764 * dumpfile.c: Include langhooks.h.
20765 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
20766 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
20767 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
20768 (dump_manager::dump_register): Start with 512 entries instead of 32.
20769 (dump_manager::register_dumps): New method.
20770 * toplev.c (general_init): Instead of invoking register_dumps
20771 langhook, invoke register_dumps method on the dump manager.
20772 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
20775 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
20777 * doc/md.texi: Clarify the restrictions on a define_insn condition.
20778 Say that # requires an associated define_split to exist, and that
20779 the define_split must be suitable for use after register allocation.
20781 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20783 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
20784 (compute_outgoing_frequencies): Also initialize zero counts.
20785 (find_many_sub_basic_blocks): Do not produce uninitialized profile
20786 around loops; preserve more of profile when nothing changes.
20788 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
20790 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
20792 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
20793 * config/arm/arm-cpu-cdata.h: Regenerate.
20794 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
20795 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
20796 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
20797 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
20798 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
20800 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
20802 * config/arm/t-rmprofile: Likewise.
20803 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
20805 2017-06-06 David S. Miller <davem@davemloft.net>
20808 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
20809 blockage if function uses alloca.
20811 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20813 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
20814 New "uid" fields to hold pretty-print IDs of group and ref.
20815 Memory references are now identified as <group_id>:<ref_id>
20816 instead of using [random] addresses.
20817 (dump_mem_details): Simplify, no functional change.
20818 (dump_mem_ref): Simplify and make output more concise.
20819 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
20820 (find_or_create_group): Initialize group uid.
20821 (record_ref): Initialize ref uid. Improve debug output.
20822 (prune_group_by_reuse, should_issue_prefetch_p,)
20823 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
20824 (mark_nontemporal_store, determine_loop_nest_reuse):
20825 Improve debug output.
20827 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20829 * dbgcnt.def (prefetch): New debug counter.
20830 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
20831 (schedule_prefetches): Stop issueing prefetches if debug counter
20834 2017-06-06 Tom de Vries <tom@codesourcery.com>
20836 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
20837 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
20839 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20841 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
20842 Use aarch64_reg_or_zero predicate for operand 4.
20843 (aarch64_compare_and_swap<mode> define_insn_and_split):
20844 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
20845 (aarch64_store_exclusive<mode>): Likewise for operand 2.
20847 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
20849 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
20850 (arm_compute_save_core_reg_mask): This.
20851 (thumb1_compute_save_reg_mask): Rename into ...
20852 (thumb1_compute_save_core_reg_mask): This.
20853 (arm_compute_save_reg0_reg12_mask): Adapt comment.
20854 (arm_compute_frame_layout): Likewise.
20856 2017-06-06 Richard Biener <rguenther@suse.de>
20858 PR tree-optimization/80974
20859 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
20860 keep or clear leaders SSA info.
20862 2017-06-06 Tom de Vries <tom@codesourcery.com>
20864 * config/nvptx/nvptx.c (split_mode_p): New function.
20865 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
20867 2017-06-06 Tom de Vries <tom@codesourcery.com>
20869 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
20871 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20874 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
20877 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20879 * shrink-wrap.c (handle_simple_exit): Update profile.
20880 (try_shrink_wrapping): Upate profile.
20882 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20884 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
20885 (tree_guess_outgoing_edge_probabilities): New.
20886 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
20887 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
20889 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20891 * ipa-split.c (split_function): Initialize return bb profile.
20893 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20895 * profile.c (compute_branch_probabilities): Also initialize
20896 EXIT_BLOCK profile.
20898 2017-06-06 Richard Biener <rguenther@suse.de>
20900 PR tree-optimization/80928
20901 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
20902 (vect_analyze_loop_operations): Properly guard analysis for
20904 (vect_transform_loop): Likewise.
20905 (vect_analyze_loop_2): Also reset SLP type on PHIs.
20906 (vect_model_induction_cost): Do not cost for pure SLP.
20907 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
20908 of induction in inner loop vectorization.
20909 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
20910 (vect_get_and_check_slp_defs): Handle vect_induction_def.
20911 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
20913 (vect_analyze_slp_cost_1): Cost induction.
20914 (vect_detect_hybrid_slp_stmts): Handle PHIs.
20915 (vect_get_slp_vect_defs): Likewise.
20916 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
20917 (vect_transform_stmt): Handle SLP reductions.
20918 * tree-vectorizer.h (vectorizable_induction): Adjust.
20920 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
20922 * config/rs6000/rs6000.c (make_resolver_func): Update
20923 init_lowered_empty_function call.
20925 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
20927 * doc/invoke.texi: Document the -fprofile-abs-path option.
20928 * common.opt (fprofile-abs-path): New option.
20929 * gcov-io.h (gcov_write_filename): Declare.
20930 * gcov-io.c (gcov_write_filename): New function.
20931 * coverage.c (coverage_begin_function): Use gcov_write_filename.
20932 * profile.c (output_location): Likewise.
20934 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20936 * shring-wrap.c: Revert accidental commit.
20938 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
20940 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
20942 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20944 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
20946 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
20947 profile in callgraph edge.
20948 * profile-count.h (apply_probability): If THIS is 0, then result is 0
20949 (apply_scale): Likewise.
20950 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
20951 Also scale profile when inlining function with zero profile.
20952 (initialize_cfun): Update exit block profile even when it is zero.
20953 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
20954 when profile is read.
20956 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
20958 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
20959 (CLONE_*): New constants to define the processors we can generate
20960 code for with the target_clone attribute.
20961 (rs6000_clone_map): New array to identify which clone processors
20962 the current program is running on.
20963 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
20964 target_clone attribute.
20965 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
20966 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
20967 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
20968 (cpu_expand_builtin): Add support for target_clone attribute.
20969 (rs6000_valid_attribute_p): Allow "default" attribute.
20970 (get_decl_name): New debug function to simplify printing the
20971 current function name in debugging statements.
20972 (rs6000_clone_priority): New functions to support the target_clone
20973 attribute, and be able to generate code to switch between ISA 2.05
20974 through ISA 3.0 (power6 through power9).
20975 (rs6000_compare_version_priority): Likewise.
20976 (rs6000_get_function_versions_dispatcher): Likewise.
20977 (make_resolver_func): Likewise.
20978 (add_condition_to_bb): Likewise.
20979 (dispatch_function_versions): Likewise.
20980 (rs6000_generate_version_dispatcher_body): Likewise.
20981 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
20982 (fusion_gpr_load_p): Fix a spacing issue.
20983 * doc/extend.texi (Common Function Attributes): Document that the
20984 PowerPC supports the target_clone attribute.
20986 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
20988 * config/arm/arm.h: explain F symbol found in description of ARM
20989 register allocation in its legend.
20991 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20993 * config/mips/frame-header-opt.c: Include profile-count.h.
20994 * config/riscv/riscv.c: Include profile-count.h
20996 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20998 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
21000 (sm_set_flag_if_changed): Add bbs field.
21001 (execute_sm_if_changed_flag_set): Pass BBS.
21002 (execute_sm): Update.
21004 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21006 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
21009 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21011 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
21012 (peephole2): New peephole2 to emit the above.
21013 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
21015 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21017 * config/aarch64/aarch64.c (define_peephole2 above
21018 *sub_<shift>_<mode>): New peephole.
21020 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21022 * config/i386/i386.c (make_resolver_func): Update.
21023 * Makefile.in: Add profile-count.h and profile-count.o
21024 * auto-profile.c (afdo_indirect_call): Update to new API.
21025 (afdo_set_bb_count): Update.
21026 (afdo_propagate_edge): Update.
21027 (afdo_propagate_circuit): Update.
21028 (afdo_calculate_branch_prob): Update.
21029 (afdo_annotate_cfg): Update.
21030 * basic-block.h: Include profile-count.h
21031 (struct edge_def): Turn count to profile_count.
21032 (struct basic_block_def): Likewie.
21033 (REG_BR_PROB_BASE): Move to profile-count.h
21034 (RDIV): Move to profile-count.h
21035 * bb-reorder.c (max_entry_count): Turn to profile_count.
21036 (find_traces): Update.
21037 (rotate_loop):Update.
21038 (connect_traces):Update.
21039 (sanitize_hot_paths):Update.
21040 * bt-load.c (migrate_btr_defs): Update.
21041 * cfg.c (RDIV): Remove.
21042 (init_flow): Use alloc_block.
21043 (alloc_block): Uninitialize count.
21044 (unchecked_make_edge): Uninitialize count.
21045 (check_bb_profile): Update.
21046 (dump_edge_info): Update.
21047 (dump_bb_info): Update.
21048 (update_bb_profile_for_threading): Update.
21049 (scale_bbs_frequencies_int): Update.
21050 (scale_bbs_frequencies_gcov_type): Update.
21051 (scale_bbs_frequencies_profile_count): New.
21052 * cfg.h (update_bb_profile_for_threading): Update.
21053 (scale_bbs_frequencies_profile_count): Declare.
21054 * cfgbuild.c (compute_outgoing_frequencies): Update.
21055 (find_many_sub_basic_blocks): Update.
21056 * cfgcleanup.c (try_forward_edges): Update.
21057 (try_crossjump_to_edge): Update.
21058 * cfgexpand.c (expand_gimple_tailcall): Update.
21059 (construct_exit_block): Update.
21060 * cfghooks.c (verify_flow_info): Update.
21061 (dump_bb_for_graph): Update.
21062 (split_edge): Update.
21063 (make_forwarder_block): Update.
21064 (duplicate_block): Update.
21065 (account_profile_record): Update.
21066 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
21067 (get_estimated_loop_iterations): Update.
21068 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
21069 (single_likely_exit): Update.
21070 * cfgloopmanip.c (scale_loop_profile): Update.
21072 (set_zero_probability): Update.
21073 (lv_adjust_loop_entry_edge): Update.
21074 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
21075 (purge_dead_edges): Update.
21076 (rtl_account_profile_record): Update.
21077 * cgraph.c (cgraph_node::create): Uninitialize count.
21078 (symbol_table::create_edge): Uninitialize count.
21079 (cgraph_update_edges_for_call_stmt_node): Update.
21080 (cgraph_edge::dump_edge_flags): Update.
21081 (cgraph_node::dump): Update.
21082 (cgraph_edge::maybe_hot_p): Update.
21083 * cgraph.h: Include profile-count.h
21084 (create_clone), create_edge, create_indirect_edge): Update.
21085 (cgraph_node): Turn count to profile_count.
21086 (cgraph_edge0: Likewise.
21087 (make_speculative, clone): Update.
21088 (create_edge): Update.
21089 (init_lowered_empty_function): Update.
21090 * cgraphclones.c (cgraph_edge::clone): Update.
21091 (duplicate_thunk_for_node): Update.
21092 (cgraph_node::create_clone): Update.
21093 * cgraphunit.c (cgraph_node::analyze): Update.
21094 (cgraph_node::expand_thunk): Update.
21095 * final.c (dump_basic_block_info): Update.
21096 * gimple-streamer-in.c (input_bb): Update.
21097 * gimple-streamer-out.c (output_bb): Update.
21098 * graphite.c (print_global_statistics): Update.
21099 (print_graphite_scop_statistics): Update.
21100 * hsa-brig.c: Include basic-block.h.
21101 * hsa-dump.c: Include basic-block.h.
21102 * hsa-gen.c (T sum_slice): Update.
21103 (convert_switch_statements):Update.
21104 * hsa-regalloc.c: Include basic-block.h.
21105 * ipa-chkp.c (chkp_produce_thunks): Update.
21106 * ipa-cp.c (struct caller_statistics): Update.
21107 (init_caller_stats): Update.
21108 (gather_caller_stats): Update.
21109 (ipcp_cloning_candidate_p): Update.
21110 (good_cloning_opportunity_p): Update.
21111 (get_info_about_necessary_edges): Update.
21112 (dump_profile_updates): Update.
21113 (update_profiling_info): Update.
21114 (update_specialized_profile): Update.
21115 (perhaps_add_new_callers): Update.
21116 (decide_about_value): Update.
21117 (ipa_cp_c_finalize): Update.
21118 * ipa-devirt.c (struct odr_type_warn_count): Update.
21119 (struct decl_warn_count): Update.
21120 (struct final_warning_record): Update.
21121 (possible_polymorphic_call_targets): Update.
21122 (ipa_devirt): Update.
21123 * ipa-fnsummary.c (redirect_to_unreachable): Update.
21124 * ipa-icf.c (sem_function::merge): Update.
21125 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
21126 * ipa-inline.c (compute_uninlined_call_time): Update.
21127 (compute_inlined_call_time): Update.
21128 (want_inline_small_function_p): Update.
21129 (want_inline_self_recursive_call_p): Update.
21130 (edge_badness): Update.
21131 (lookup_recursive_calls): Update.
21132 (recursive_inlining): Update.
21133 (inline_small_functions): Update.
21134 (dump_overall_stats): Update.
21135 (dump_inline_stats): Update.
21136 * ipa-profile.c (ipa_profile_generate_summary): Update.
21137 (ipa_propagate_frequency): Update.
21138 (ipa_profile): Update.
21139 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
21140 * ipa-utils.c (ipa_merge_profiles): Update.
21141 * loop-doloop.c (doloop_modify): Update.
21142 * loop-unroll.c (report_unroll): Update.
21143 (unroll_loop_runtime_iterations): Update.
21144 * lto-cgraph.c (lto_output_edge): Update.
21145 (lto_output_node): Update.
21146 (input_node): Update.
21147 (input_edge): Update.
21148 (merge_profile_summaries): Update.
21149 * lto-streamer-in.c (input_cfg): Update.
21150 * lto-streamer-out.c (output_cfg): Update.
21151 * mcf.c (create_fixup_graph): Update.
21152 (adjust_cfg_counts): Update.
21153 (sum_edge_counts): Update.
21154 * modulo-sched.c (sms_schedule): Update.
21155 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
21156 * predict.c (maybe_hot_count_p): Update.
21157 (probably_never_executed): Update.
21158 (dump_prediction): Update.
21159 (combine_predictions_for_bb): Update.
21160 (propagate_freq): Update.
21161 (handle_missing_profiles): Update.
21162 (counts_to_freqs): Update.
21163 (rebuild_frequencies): Update.
21164 (force_edge_cold): Update.
21165 * predict.h: Include profile-count.h
21166 (maybe_hot_count_p, counts_to_freqs): UPdate.
21167 * print-rtl-function.c: Do not include cfg.h
21168 * print-rtl.c: Include basic-block.h
21169 * profile-count.c: New file.
21170 * profile-count.h: New file.
21171 * profile.c (is_edge_inconsistent): Update.
21172 (correct_negative_edge_counts): Update.
21173 (is_inconsistent): Update.
21174 (set_bb_counts): Update.
21175 (read_profile_edge_counts): Update.
21176 (compute_frequency_overlap): Update.
21177 (compute_branch_probabilities): Update; Initialize and deinitialize
21179 (branch_prob): Update.
21180 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
21181 (edge_gcov_count): New.
21182 (bb_gcov_count): New.
21183 * shrink-wrap.c (try_shrink_wrapping): Update.
21184 * tracer.c (better_p): Update.
21185 * trans-mem.c (expand_transaction): Update.
21186 (ipa_tm_insert_irr_call): Update.
21187 (ipa_tm_insert_gettmclone_call): Update.
21188 * tree-call-cdce.c: Update.
21189 * tree-cfg.c (gimple_duplicate_sese_region): Update.
21190 (gimple_duplicate_sese_tail): Update.
21191 (gimple_account_profile_record): Update.
21192 (execute_fixup_cfg): Update.
21193 * tree-inline.c (copy_bb): Update.
21194 (copy_edges_for_bb): Update.
21195 (initialize_cfun): Update.
21196 (freqs_to_counts): Update.
21197 (copy_cfg_body): Update.
21198 (expand_call_inline): Update.
21199 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
21200 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
21201 (try_unroll_loop_completely): Update.
21202 (try_peel_loop): Update.
21203 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
21204 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
21205 * tree-ssa-loop-split.c (connect_loops): Update.
21206 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
21207 * tree-ssa-reassoc.c (branch_fixup): Update.
21208 * tree-ssa-tail-merge.c (replace_block_by): Update.
21209 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
21210 (compute_path_counts): Update.
21211 (update_profile): Update.
21212 (recompute_probabilities): Update.
21213 (update_joiner_offpath_counts): Update.
21214 (estimated_freqs_path): Update.
21215 (freqs_to_counts_path): Update.
21216 (clear_counts_path): Update.
21217 (ssa_fix_duplicate_block_edges): Update.
21218 (duplicate_thread_path): Update.
21219 * tree-switch-conversion.c (case_bit_test_cmp): Update.
21220 (struct switch_conv_info): Update.
21221 * tree-tailcall.c (decrease_profile): Update.
21222 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
21223 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
21224 * value-prof.c (check_counter): Update.
21225 (gimple_divmod_fixed_value): Update.
21226 (gimple_mod_pow2): Update.
21227 (gimple_mod_subtract): Update.
21228 (gimple_ic_transform): Update.
21229 (gimple_stringop_fixed_value): Update.
21230 * value-prof.h (gimple_ic): Update.
21232 2017-06-02 Carl Love <cel@us.ibm.com>
21234 * config/rs6000/rs6000-c: Add support for built-in functions
21235 vector double vec_doublee (vector signed int);
21236 vector double vec_doublee (vector unsigned int);
21237 vector double vec_doublee (vector float);
21238 vector double vec_doubleh (vector signed int);
21239 vector double vec_doubleh (vector unsigned int);
21240 vector double vec_doubleh (vector float);
21241 vector double vec_doublel (vector signed int);
21242 vector double vec_doublel (vector unsigned int);
21243 vector double vec_doublel (vector float);
21244 vector double vec_doubleo (vector signed int);
21245 vector double vec_doubleo (vector unsigned int);
21246 vector double vec_doubleo (vector float);.
21247 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
21248 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
21250 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
21251 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
21252 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
21254 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
21255 vec_doublel, vec_doubleh.
21256 * doc/extend.texi: Update the built-in documentation file for the
21257 new built-in functions.
21259 2017-06-02 David Malcolm <dmalcolm@redhat.com>
21262 * ipa-inline-analysis.c (free_growth_caches): Set
21263 edge_removal_hook_holder to NULL after removing it.
21265 2017-06-02 Sudakshina Das <sudi.das@arm.com>
21267 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
21268 comparision with zero.
21270 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21271 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21272 for early expansion of vec_min and vec_max builtins.
21273 (builtin_function_type): Add min/max unsigned variants to those
21274 identified as having unsigned arguments.
21276 2017-06-02 Olivier Hainque <hainque@adacore.com>
21278 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
21280 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21282 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
21283 Use VALL_F16 iterator rather than VALL.
21285 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21287 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
21288 Emit CBNZ inside loop when doing a strong exchange and comparing
21289 against zero. Generate the CC flags after the loop.
21291 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
21293 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
21294 (dl_section_ref): New.
21295 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
21296 On AIX, append an expression to subtract the size of the
21297 section length to dl_section_ref.
21299 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21301 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21302 for early expansion of vector absolute builtins.
21304 2017-06-02 Richard Biener <rguenther@suse.de>
21306 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
21307 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
21309 2017-06-02 Richard Biener <rguenther@suse.de>
21311 PR tree-optimization/80948
21312 * tree-tailcall.c (find_tail_calls): Track stmts to move in
21313 stmt order as well.
21315 2017-06-02 Richard Biener <rguenther@suse.de>
21317 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
21319 * tree-vect-stmts.c (process_use): Do not mark backedge defs
21320 for inductions as relevant.
21322 2017-06-02 Richard Biener <rguenther@suse.de>
21324 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
21325 (vectorizable_induction): ... this. Remove dead code.
21327 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
21329 * builtins. (expand_builtin_alloca): Remove second parameter and
21330 infer its value from the first parameter instead.
21331 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
21333 2017-06-02 Jakub Jelinek <jakub@redhat.com>
21335 PR rtl-optimization/80903
21336 * loop-doloop.c (add_test): Unshare sequence.
21338 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
21340 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
21342 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
21344 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
21346 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
21347 xlogue_layout::get_instance, logue_layout::xlogue_layout,
21348 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
21349 (xlogue_layout::get_stub_rtx): Make static.
21350 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
21351 (xlogue_layout::compute_stub_managed_regs): Rename to...
21352 (xlogue_layout::count_stub_managed_regs): ...this.
21353 (xlogue_layout::is_stub_managed_reg): New function.
21354 (xlogue_layout::m_stub_names): Rename to...
21355 (xlogue_layout::s_stub_names): ...this, make static.
21356 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
21357 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
21358 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
21359 xlogue_layout::s_stub_names): Instantiate statics.
21360 (stub_managed_regs): Remove.
21361 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
21362 (disable_call_ms2sysv_xlogues): Rename to...
21363 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
21364 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
21366 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
21367 change after reload_completed.
21368 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
21370 (ix86_expand_prologue): Likewise.
21371 (ix86_expand_epilogue): Likewise.
21372 (ix86_expand_split_stack_prologue): Likewise.
21373 (ix86_compute_frame_layout): Remove frame parameter ...
21374 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
21375 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
21377 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
21378 (ix86_frame): Move from here ...
21379 * config/i386/i386.h (ix86_frame): ... to here.
21380 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
21381 complete ix86_frame data structure instead. Remove some_ld_name.
21383 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
21385 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
21386 symbols that hold a DECL_VALUE_EXPR.
21388 2017-06-01 Martin Jambor <mjambor@suse.cz>
21390 PR tree-optimization/80898
21391 * tree-sra.c (process_subtree_disqualification): Removed.
21392 (disqualify_candidate): Do not acll
21393 process_subtree_disqualification.
21394 (subtree_mark_written_and_enqueue): New function.
21395 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
21396 RHS has been disqualified and re-queue LHS if necessary. Apart
21397 from that, ignore disqualified RHS.
21399 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21401 * config/s390/s390.c (s390_emit_epilogue): Disable early return
21402 address fetch for z10 or later.
21404 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21406 * config/arc/arc.md (tst_movb): Add guard when splitting.
21408 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21410 * config/arc/arc.c (arc_can_eliminate): Test against
21411 arc_frame_pointer_needed.
21413 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21415 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
21416 to prevent store reordering.
21417 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
21418 (type): Add block type.
21419 (stack_tie): Define special instruction to be used in
21422 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21424 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
21425 constraint. It is not valid for the pattern.
21426 (noncommutative_binary_comparison): Likewise.
21428 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21430 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
21433 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21435 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
21436 be used by the reg-alloc.
21438 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21440 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
21441 reg-alloc when having mul64 or mul32x16 instructions.
21442 (mulsidi3): Likewise.
21443 (umulsidi3): Likewise.
21444 (mulsi32x16): New pattern.
21445 (mulsi64): Likewise.
21446 (mulsidi64): Likewise.
21447 (umulsidi64): Likewise.
21448 (MUL32x16_REG): Define.
21449 (mul64_600): Use MUL32x16_REG.
21450 (mac64_600): Likewise.
21451 (umul64_600): Likewise.
21452 (umac64_600): Likewise.
21454 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21456 * config/arc/arc.md (mulsi3_700): Make it commutative.
21458 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
21460 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
21462 (*sign_extendsidi2_insn): Likewise for movstosw.
21464 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
21466 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
21467 the type of the input discriminant value. Convert the
21468 discriminant value of signedness vary.
21470 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
21472 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
21473 Add to -Wall section.
21475 2017-06-01 Richard Biener <rguenther@suse.de>
21477 PR middle-end/66313
21478 * fold-const.c (fold_plusminus_mult_expr): If the factored
21479 factor may be zero use a wrapping type for the inner operation.
21480 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
21481 and handle moved defs.
21482 (process_assignment): Properly guard the unary op case. Return a
21483 tri-state indicating that moving the stmt before the call may allow
21484 to continue. Pass through to_move.
21485 (find_tail_calls): Handle moving unrelated defs before
21488 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
21491 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
21492 splitter result in the canonical way.
21494 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
21496 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
21497 also for 32bit target. Update insn attributes.
21498 (zero-extendsidi2 splitter): Allow all registers for operand 1.
21500 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
21502 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
21503 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
21504 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
21505 (_mm_maskz_min_ss): New intrinsics.
21507 2017-05-31 Martin Liska <mliska@suse.cz>
21509 * tree-vect-loop.c (vect_create_epilog_for_reduction):
21510 Change comment style to one we normally use.
21511 (vectorizable_reduction): Likewise.
21512 (vectorizable_induction): Likewise.
21513 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
21514 (vectorizable_call): Likewise.
21515 (vectorizable_simd_clone_call): Likewise.
21516 (vectorizable_conversion): Likewise.
21517 (vectorizable_assignment): Likewise.
21518 (vectorizable_shift): Likewise.
21519 (vectorizable_operation): Likewise.
21520 (vectorizable_store): Likewise.
21521 (vectorizable_load): Likewise.
21522 * tree-vectorizer.h: Likewise.
21524 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
21526 * passes.c (emergency_dump_function): New.
21527 * tree-pass.h (emergency_dump_function): Declare.
21528 * plugin.c (plugins_internal_error_function): Remove.
21529 * plugin.h (plugins_internal_error_function): Remove declaration.
21530 * toplev.c (internal_error_function): New static function. Use it...
21531 (general_init): ...here.
21533 2017-05-31 Graham Markall <graham.markall@embecosm.com>
21535 * config/arc/arc.c (arc_print_operand): Handle constant operands.
21536 (arc_rtx_costs): Add costs for new patterns.
21537 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
21538 * config/arc/predicates.md: Add _1_2_3_operand predicate.
21540 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21542 * tree-ssa-strlen.c (get_next_strinfo): New function.
21543 (get_stridx_plus_constant): Use it.
21544 (zero_length_string): Likewise.
21545 (adjust_related_strinfos): Likewise.
21546 (adjust_last_stmt): Likewise.
21548 2017-05-31 Richard Biener <rguenther@suse.de>
21551 * config/i386/i386.c (ix86_expand_builtin): Remove assert
21552 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
21554 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21556 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
21557 loop_vinfo argument and use of dependence distance vectors.
21558 Check instead whether the two references differ only in their
21559 initial value and assume that they have the same alignment if the
21560 difference is a multiple of the vector alignment.
21561 (vect_analyze_data_refs_alignment): Update call accordingly.
21563 2017-05-31 Martin Liska <mliska@suse.cz>
21566 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
21568 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21570 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
21571 (create_intersect_range_checks): Move from ...
21572 * tree-data-ref.c (create_intersect_range_checks_index)
21573 (create_intersect_range_checks): ... to here.
21574 (create_runtime_alias_checks): New function factored from ...
21575 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
21576 here. Call above function.
21577 * tree-data-ref.h (create_runtime_alias_checks): New function.
21579 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21581 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
21582 segment length for dr_b and compute it in wide_int.
21584 2017-05-31 Richard Biener <rguenther@suse.de>
21586 PR tree-optimization/80906
21587 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
21588 and pass through iv_map.
21589 (copy_bb_and_scalar_dependences): Adjust.
21590 (translate_pending_phi_nodes): Likewise.
21591 (copy_loop_close_phi_args): Handle code-generating IVs instead
21594 2017-05-30 David Malcolm <dmalcolm@redhat.com>
21596 * diagnostic-color.c (color_dict): Add "type-diff".
21597 (parse_gcc_colors): Update comment.
21598 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
21599 -fdiagnostics-show-template-tree and -fno-elide-type.
21600 (GCC_COLORS): Add type-diff to example.
21602 (-fdiagnostics-show-template-tree): New.
21603 (-fno-elide-type): New.
21604 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
21605 the pp_format_decoder callback. Call any m_format_postprocessor's
21607 (pretty_printer::pretty_printer): Initialize
21608 m_format_postprocessor.
21609 (pretty_printer::~pretty_printer): Delete any
21610 m_format_postprocessor.
21611 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
21612 (class format_postprocessor): New class.
21613 (struct pretty_printer::format_decoder): Document the new parameters.
21614 (struct pretty_printer::m_format_postprocessor): New field.
21615 * tree-diagnostic.c (default_tree_printer): Update for new
21616 bool and const char ** params.
21617 * tree-diagnostic.h (default_tree_printer): Likewise.
21619 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
21621 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
21622 (lwa_operand): Delete rs6000_gen_cell_microcode test.
21623 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
21624 rs6000_gen_cell_microcode code.
21625 (rs6000_final_prescan_insn): Delete.
21626 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
21627 "warn-cell-microcode" entries.
21628 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
21629 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
21630 throughout. Change cc_reg_not_micro_cr0_operand to
21631 cc_reg_not_cr0_operand throughout.
21632 (*extendhi<mode>2_noload): Delete.
21633 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
21634 (mwarn-cell-microcode): Delete.
21635 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
21636 -mgen-cell-microcode and -mwarn-cell-microcode.
21638 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
21641 * config/i386/constraints.md (Yd): New constraint.
21643 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
21644 and (?Yd, r) alternatives. Update insn attributes.
21645 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
21646 and (?*Yd, r) alternatives. Update insn attributes.
21647 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
21649 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
21651 * gimplify.c (gimplify_modify_expr): Don't create a
21652 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
21655 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
21657 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
21659 2017-05-30 Richard Biener <rguenther@suse.de>
21661 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
21662 and reduc_def fields.
21663 (STMT_VINFO_REDUC_TYPE): New define.
21664 (STMT_VINFO_REDUC_DEF): Likewise.
21665 (vect_force_simple_reduction): Adjust prototype.
21666 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
21667 (vect_is_simple_reduction): Remove check_reduction argument.
21668 (vect_force_simple_reduction): Adjust and set
21669 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21670 (vectorizable_reduction): Do not re-do reduction analysis
21671 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21672 * tree-parloops.c (gather_scalar_reductions): Adjust.
21674 2017-05-30 Richard Biener <rguenther@suse.de>
21676 PR middle-end/80901
21677 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
21680 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
21682 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
21683 Introduce unknown_misalignment parameter and remove vf.
21684 (vect_peeling_hash_get_lowest_cost):
21685 Pass unknown_misalignment parameter.
21686 (vect_enhance_data_refs_alignment):
21687 Fix unsupportable data ref treatment.
21689 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21691 * tree-vect-data-refs.c (vect_get_data_access_cost):
21692 Workaround for SLP handling.
21693 (vect_enhance_data_refs_alignment):
21694 Compute costs for doing no peeling at all, compare to the best
21695 peeling costs so far and avoid peeling if cheaper.
21697 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21699 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
21700 Return peeling info and set costs to zero for unlimited cost
21702 (vect_enhance_data_refs_alignment): Also inspect all datarefs
21703 with unknown misalignment. Compute and costs for unknown
21704 misalignment, compare them to the costs for known misalignment
21705 and choose the cheapest for peeling.
21707 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21709 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
21710 (vect_get_peeling_costs_all_drs): Create function.
21711 (vect_peeling_hash_get_lowest_cost):
21712 Use vect_get_peeling_costs_all_drs.
21713 (vect_peeling_supportable): Create function.
21714 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
21716 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21718 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
21719 DR_HAS_NEGATIVE_STEP.
21720 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
21721 (vect_enhance_data_refs_alignment): Use.
21722 (vect_duplicate_ssa_name_ptr_info): Use.
21723 * tree-vectorizer.h (dr_misalignment): Use.
21724 (known_alignment_for_access_p): Use.
21726 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21729 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
21731 (has_section_name): New function.
21733 2017-05-30 Martin Liska <mliska@suse.cz>
21736 * auto-profile.c (get_function_decl_from_block): Fix
21739 2017-05-30 Richard Biener <rguenther@suse.de>
21741 PR middle-end/80876
21742 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
21744 2017-05-30 Martin Liska <mliska@suse.cz>
21746 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
21747 * dumpfile.h (struct dump_file_info): Remove ctors.
21749 2017-05-30 Martin Liska <mliska@suse.cz>
21751 * predict.def: Fix GNU coding style.
21753 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
21755 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
21756 Mark 'to' argument with ATTRIBUTE_UNUSED.
21758 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
21760 * config/xtensa/xtensa.c (xtensa_emit_call): Use
21761 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
21762 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
21765 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
21767 * doc/install.texi (Options specification): Restore entry of
21768 --enable-sjlj-exceptions.
21770 2017-05-27 Michael Eager <eager@eagercon.com>
21773 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
21775 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
21777 * config/microblaze/microblaze.h
21778 (FIXED_REGISTERS): Update in macro.
21779 (CALL_USED_REGISTERS): Update in macro.
21781 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
21783 * doc/install.texi: Add links to macOS binary distributions.
21785 2017-05-27 Jakub Jelinek <jakub@redhat.com>
21789 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21791 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
21793 2017-05-26 Martin Liska <mliska@suse.cz>
21795 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
21797 2017-05-26 Martin Liska <mliska@suse.cz>
21799 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
21800 always leading ';; '.
21801 (dump_bb_info): Likewise.
21802 (brief_dump_cfg): Likewise.
21803 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
21804 * dumpfile.c: Remove usage of TDF_VERBOSE.
21805 * dumpfile.h (enum dump_kind): Likewise.
21806 (dump_gimple_bb_header): Do not use TDF_COMMENT.
21807 * print-tree.c (debug_verbose): Remove.
21808 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
21809 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
21810 * tree-diagnostic.c (default_tree_printer): Replace
21811 TDF_DIAGNOSTIC with TDF_SLIM.
21813 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21815 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
21816 in parameter loop, rather than loop_vinfo.
21817 (create_intersect_range_checks): Ditto.
21818 (vect_create_cond_for_alias_checks): Update call to above functions.
21820 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21822 PR tree-optimization/80815
21823 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
21824 for merging runtime alias checks. Handle negative DR_STEPs.
21826 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21828 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
21830 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
21831 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
21832 out code pruning runtime alias checks.
21833 * tree-data-ref.c (prune_runtime_alias_test_list): New function
21834 factored out from above.
21835 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
21837 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
21839 (prune_runtime_alias_test_list): New decalaration.
21841 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21843 * tree-vect-data-refs.c (compare_tree): Rename and move ...
21844 * tree-data-ref.c (data_ref_compare_tree): ... to here.
21845 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
21846 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
21847 (operator==, comp_dr_with_seg_len_pair): Ditto.
21848 (vect_prune_runtime_alias_test_list): Ditto.
21850 2017-05-26 Martin Liska <mliska@suse.cz>
21853 * params.def: Bound partial-inlining-entry-probability param.
21855 2017-05-26 Marek Polacek <polacek@redhat.com>
21858 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
21861 2017-05-26 Richard Biener <rguenther@suse.de>
21863 PR tree-optimization/80842
21864 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
21867 2017-05-26 Richard Biener <rguenther@suse.de>
21869 PR tree-optimization/80844
21870 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
21872 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
21874 * doc/md.texi (Machine Constraints): Update x86 family
21875 machine constraints section to match 'config/i386/constraints.md'.
21877 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
21879 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
21881 2017-05-25 Nathan Sidwell <nathan@acm.org>
21883 * doc/invoke.texi (--enable-languages): Update documentation.
21885 2017-05-25 Martin Liska <mliska@suse.cz>
21887 * dumpfile.c: Add TDF_FOLDING.
21888 * dumpfile.h (enum dump_kind): Likewise.
21889 * genmatch.c (dt_simplify::gen_1): Use it.
21891 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21893 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
21895 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21897 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
21898 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
21900 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21902 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
21903 * match.pd (X == C): Rewrite it here.
21904 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
21905 with_certain_nonzero_bits2): New predicates.
21906 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
21908 2017-05-24 Nathan Sidwell <nathan@acm.org>
21910 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
21913 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
21916 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
21918 * config/powerpcspe: New port. Files are copied from the rs6000
21919 port, with "rs6000" in filenames replaced by "powerpcspe".
21921 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
21923 PR rtl-optimization/80754
21924 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
21926 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
21928 * config/sparc/sparc.md (length): Return the correct value for -mflat
21929 sibcalls to match output_sibcall.
21931 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
21935 * config/rs6000/rs6000.c (struct machine_function): Add new field
21937 (rs6000_get_separate_components): Init that field, use it.
21938 (rs6000_components_for_bb): Use the field.
21940 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
21942 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
21944 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
21946 PR middle-end/80823
21947 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
21949 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21952 * config/s390/s390.c (s390_check_qrst_address): Check incoming
21953 address against address_operand predicate.
21954 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
21956 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
21958 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
21959 parameters passed indirectly.
21961 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
21963 * config/i386/i386.md (*movdi_internal): Remove SSE4
21964 alternative 18 (?r, *v). Update insn attributes.
21965 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
21966 Update insn attributes.
21967 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
21968 Update insn attributes.
21969 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
21970 alternative 1 (r, v). Remove isa attribute.
21971 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
21972 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
21973 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
21975 2017-05-23 Tom de Vries <tom@codesourcery.com>
21977 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
21980 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21982 * cgraphunit.c (symbol_table::process_new_functions): Update.
21983 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
21984 (inline_generate_summary): Rename to ...
21985 (ipa_fn_summary_generate): ... this one.
21986 (inline_read_summary): Rename to ...
21987 (ipa_fn_summary_read): ... this one.
21988 (inline_write_summary): Rename to ...
21989 (ipa_fn_summary_write): ... this one.
21990 (inline_free_summary): Rename to ...
21991 (ipa_free_fn_summary): ... this one.
21992 (pass_data_local_fn_summary, pass_local_fn_summary,
21993 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
21994 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
21995 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
21996 make_pass_ipa_fn_summary): New.
21997 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
21998 inline_write_summary, inline_free_summary): Remove.
21999 (ipa_free_fn_summary) : New.
22000 * ipa-inline.c (ipa_inline): Update.
22001 (pass_ipa_inline): Do not generate summaries.
22002 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
22004 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
22005 and add pass_ipa_fn_summary.
22006 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
22008 (make_pass_inline_parameters): Remove.
22010 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
22012 * omp-low.c (struct omp_context): Remove "default_kind" member.
22015 * omp-offload.c (execute_oacc_device_lower): Remove the
22016 parallelism dimensions function attributes for unparallelized
22017 OpenACC kernels constructs.
22019 2017-05-23 Martin Liska <mliska@suse.cz>
22021 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
22023 (cgraph_edge::make_speculative): Likewise.
22024 (cgraph_edge::resolve_speculation): Likewise.
22025 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
22026 (cgraph_node::dump): Likewise.
22027 * cgraph.h: Likewise.
22028 * cgraphunit.c (analyze_functions): Likewise.
22029 (symbol_table::compile): Likewise.
22030 * ipa-cp.c (print_all_lattices): Likewise.
22031 (determine_versionability): Likewise.
22032 (initialize_node_lattices): Likewise.
22033 (ipcp_verify_propagated_values): Likewise.
22034 (estimate_local_effects): Likewise.
22035 (update_profiling_info): Likewise.
22036 (create_specialized_node): Likewise.
22037 (perhaps_add_new_callers): Likewise.
22038 (decide_about_value): Likewise.
22039 (decide_whether_version_node): Likewise.
22040 (identify_dead_nodes): Likewise.
22041 (ipcp_store_bits_results): Likewise.
22042 * ipa-devirt.c (dump_targets): Likewise.
22043 (ipa_devirt): Likewise.
22044 * ipa-icf.c (sem_item::dump): Likewise.
22045 (sem_function::equals): Likewise.
22046 (sem_variable::equals): Likewise.
22047 (sem_item_optimizer::read_section): Likewise.
22048 (sem_item_optimizer::execute): Likewise.
22049 (congruence_class::dump): Likewise.
22050 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
22051 (dump_inline_summary): Likewise.
22052 (estimate_node_size_and_time): Likewise.
22053 (inline_analyze_function): Likewise.
22054 * ipa-inline-transform.c (inline_call): Likewise.
22055 * ipa-inline.c (report_inline_failed_reason): Likewise.
22056 (want_early_inline_function_p): Likewise.
22057 (edge_badness): Likewise.
22058 (update_edge_key): Likewise.
22059 (inline_small_functions): Likewise.
22060 * ipa-profile.c (ipa_profile): Likewise.
22061 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
22062 (ipa_make_edge_direct_to_target): Likewise.
22063 (remove_described_reference): Likewise.
22064 (ipa_impossible_devirt_target): Likewise.
22065 (propagate_controlled_uses): Likewise.
22066 (ipa_print_node_params): Likewise.
22067 (ipcp_transform_function): Likewise.
22068 * ipa-pure-const.c (pure_const_read_summary): Likewise.
22069 (propagate_pure_const): Likewise.
22070 * ipa-reference.c (generate_summary): Likewise.
22071 (read_write_all_from_decl): Likewise.
22072 (propagate): Likewise.
22073 (ipa_reference_read_optimization_summary): Likewise.
22074 * ipa-utils.c (ipa_merge_profiles): Likewise.
22075 * ipa.c (walk_polymorphic_call_targets): Likewise.
22076 (symbol_table::remove_unreachable_nodes): Likewise.
22077 (ipa_single_use): Likewise.
22078 * passes.c (execute_todo): Likewise.
22079 * predict.c (drop_profile): Likewise.
22080 * symtab.c (symtab_node::get_dump_name): New function.
22081 (symtab_node::dump_name): Likewise.
22082 (symtab_node::dump_asm_name): Likewise.
22083 (symtab_node::dump_references): Likewise.
22084 (symtab_node::dump_referring): Likewise.
22085 (symtab_node::dump_base): Likewise.
22086 (symtab_node::debug_symtab): Likewise.
22087 * tree-sra.c (convert_callers_for_node): Likewise.
22088 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
22089 * value-prof.c (init_node_map): Likewise.
22091 2017-05-23 Martin Liska <mliska@suse.cz>
22093 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
22094 and symtab_node::debug_symtab to symbol_table::debug.
22095 * cgraphunit.c (analyze_functions): Use the renamed function.
22096 (symbol_table::compile): Likewise.
22097 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
22098 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
22099 * passes.c (execute_todo): Likewise.
22100 * symtab.c (symbol_table::dump): New function.
22101 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
22103 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22105 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
22106 that nonconst implies exec.
22108 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22110 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22111 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22112 (inline_edge_summary_vec): Turn into ...
22113 (ipa_call_summaries): ... this one.
22114 (redirect_to_unreachable, edge_set_predicate,
22115 evaluate_properties_for_edge, inline_summary_alloc,
22116 reset_ipa_call_summary, reset_inline_summary,
22117 inline_summary_t::duplicate): Update.
22118 (inline_edge_duplication_hook): Turn to ...
22119 (ipa_call_summary_t::duplicate): ... this one.
22120 (inline_edge_removal_hook): Turn to ...
22121 (ipa_call_summary_t::remove): ... this one.
22122 (dump_inline_edge_summary): Turn to ...
22123 (dump_ipa_call_summary): ... this one.
22124 (estimate_function_body_sizes): Update.
22125 (inline_update_callee_summaries): Update.
22126 (remap_edge_change_prob): Update.
22127 (remap_edge_summaries): Update.
22128 (inline_merge_summary): Update.
22129 (do_estimate_edge_time): Update.
22130 (inline_generate_summary): Update.
22131 (inline_read_section): Update.
22132 (inline_read_summary): Update.
22133 (inline_free_summary): Update.
22134 * ipa-inline.c (can_inline_edge_p): Update.
22135 (compute_inlined_call_time): Update.
22136 (want_inline_small_function_p): Update.
22137 (edge_badness): Update.
22138 (early_inliner): Update.
22139 * ipa-inline.h (inline_edge_summary): Turn to ...
22140 (ipa_call_summary): ... this one.
22141 (ipa_call_summary_t): New class.
22142 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22143 (ipa_call_summaries): New.
22144 (inline_edge_summary): Remove.
22145 (estimate_edge_growth): Update.
22146 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22147 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22148 * ipa-split.c (execute_split_functions): Update.
22149 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22151 2017-05-23 Tom de Vries <tom@codesourcery.com>
22153 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22154 attributes): Document rdrand effective target.
22156 2017-05-23 Tom de Vries <tom@codesourcery.com>
22158 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22159 attributes): Sort alphabetically.
22161 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
22163 * config/avr/genmultilib.awk: Use gsub instead of gensub.
22165 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
22168 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
22169 V2DF/V2DI splat into two separate patterns, one that handles
22170 registers, and the other that only handles memory. Drop support
22171 for splatting from a GPR on ISA 2.07 and then splitting the
22172 splat into direct move and splat.
22173 (vsx_splat_<mode>_reg): Likewise.
22174 (vsx_splat_<mode>_mem): Likewise.
22176 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
22178 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
22180 2017-05-22 Jakub Jelinek <jakub@redhat.com>
22182 PR middle-end/80809
22183 * omp-low.c (finish_taskreg_remap): New function.
22184 (finish_taskreg_scan): If unit size of ctx->record_type
22185 is non-constant, unshare the size expression and replace
22186 decls in it with possible outer var refs.
22188 PR middle-end/80809
22189 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
22190 GOVD_SHARED rather than GOVD_PRIVATE with it.
22191 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
22192 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
22194 PR middle-end/80853
22195 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
22196 as last argument to build_outer_var_ref for pointer bases of array
22197 section reductions.
22199 2017-05-19 Martin Sebor <msebor@redhat.com>
22201 * print-tree.c (print_node): Print DECL_READ_P flag.
22203 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22205 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
22206 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
22207 * cgraph.c: Likewise.
22208 * cgraphunit.c: Likewise.
22209 * gengtype.c: Likewise.
22210 * ipa-cp.c: Likewise.
22211 * ipa-devirt.c: Likewise.
22212 * ipa-icf.c: Likewise.
22213 * ipa-predicate.c: Likewise.
22214 * ipa-profile.c: Likewise.
22215 * ipa-prop.c: Likewise.
22216 * ipa-split.c: Likewise.
22218 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
22219 edge_predicate_pool, dump_inline_hints,
22220 inline_summary::account_size_time, redirect_to_unreachable,
22221 edge_set_predicate, set_hint_predicate,
22222 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
22223 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
22224 inline_summary_t::remove, remap_hint_predicate_after_duplication,
22225 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
22226 ipa_call_summary_t::remove, initialize_growth_caches,
22227 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
22228 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
22229 mark_modified, unmodified_parm_1, unmodified_parm,
22230 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
22231 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
22232 compute_bb_predicates, will_be_nonconstant_expr_predicate,
22233 will_be_nonconstant_predicate, record_modified_bb_info,
22234 get_minimal_bb, record_modified, param_change_prob,
22235 phi_result_unknown_predicate, predicate_for_phi_result,
22236 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
22237 estimate_function_body_sizes, compute_inline_parameters,
22238 compute_inline_parameters_for_curren, pass_data_inline_parameters,
22239 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
22240 inline_update_callee_summaries, remap_edge_change_prob,
22241 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
22242 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
22243 inline_analyze_function, inline_summary_t::insert,
22244 inline_generate_summary, read_ipa_call_summary, inline_read_section,
22245 inline_read_summary, write_ipa_call_summary, inline_write_summary,
22246 inline_free_summary): Move to ipa-fnsummary.h
22247 (predicate_t): Remove.
22248 * ipa-fnsummary.c: New file.
22249 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
22250 (enum inline_hints_vals, inline_hints, agg_position_info,
22251 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
22252 inline_summaries, ipa_call_summary, ipa_call_summary_t,
22253 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
22254 dump_inline_summary, dump_inline_hints, inline_generate_summary,
22255 inline_read_summary, inline_write_summary, inline_free_summary,
22256 inline_analyze_function, initialize_inline_failed,
22257 inline_merge_summary, inline_update_overall_summary,
22258 compute_inline_parameters): Move to ipa-fnsummary.h
22259 * ipa-fnsummary.h: New file.
22260 * ipa-inline-transform.h: Include ipa-inline.h.
22261 * ipa-inline.c: LIkewise.
22263 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22265 * ipa-inline.c (edge_badness): Use inlined_time instead of
22266 inline_summaries->get.
22268 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22270 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
22272 2017-05-22 Nathan Sidwell <nathan@acm.org>
22274 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
22275 (fdump-lang): Document 'raw' option.
22276 * dumpfile.h (TDI_tu): Delete.
22277 * dumpfile.c (dump_files): Remove translation-unit.
22278 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
22280 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
22282 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
22283 command option from $(AWK) call.
22284 * config/avr/genmultilib.awk: Simplify and rewrite so that it
22285 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
22286 [FORMAT]: Remove handling of variable.
22287 * config/avr/t-multilib: Regenerate.
22289 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22291 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
22293 (dump_inline_summary): Do not print self_time.
22294 (estimate_function_body_sizes): Do not set self_time.
22295 (compute_inline_parameters): Likewise.
22296 (inline_read_section, inline_write_summary): Do not stream self_time.
22297 * ipa-inline.h (inline_summary): Drop self_time.
22299 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22301 * ipa-inline-analysis.c (account_size_time): Rename to ...
22302 (inline_summary::account_size_time): ... this one.
22303 (reset_ipa_call_summary): Turn to ...
22304 (ipa_call_summary::reset): ... this one.
22305 (reset_inline_summary): Turn to ...
22306 (inline_summary::reset): ... this one.
22307 (inline_summary_t::remove): Update.
22308 (inline_summary_t::duplicate): Update.
22309 (ipa_call_summary_t::remove): Update.
22310 (dump_inline_summary): Update.
22311 (estimate_function_body_sizes): Update.
22312 (compute_inline_parameters): Update.
22313 (estimate_node_size_and_time): Update.
22314 (inline_merge_summary): Update.
22315 (inline_update_overall_summary): Update.
22316 (inline_read_section): Update.
22317 (inline_write_summary): Update.
22318 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
22319 add account_size_time and reset member functions.
22320 (ipa_call_summary): Add reset function.
22321 * ipa-predicate.h (predicate::operator &): Constify.
22323 2017-05-22 Richard Biener <rguenther@suse.de>
22325 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
22327 2017-05-19 Jason Merrill <jason@redhat.com>
22329 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
22331 2017-05-19 Marek Polacek <polacek@redhat.com>
22334 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
22335 TYPE_OVERFLOW_WRAPS checks.
22337 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
22339 * tree-core.h (enum omp_clause_default_kind): Add
22340 "OMP_CLAUSE_DEFAULT_PRESENT".
22341 * tree-pretty-print.c (dump_omp_clause): Handle it.
22342 * gimplify.c (enum gimplify_omp_var_data): Add
22343 "GOVD_MAP_FORCE_PRESENT".
22344 (gimplify_adjust_omp_clauses_1): Map it to
22345 "GOMP_MAP_FORCE_PRESENT".
22346 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
22348 * gimplify.c (oacc_default_clause): Clarify.
22350 2017-05-19 Nathan Sidwell <nathan@acm.org>
22352 LANG_HOOK_REGISTER_DUMPS
22353 * toplev.c (general_init): Call register dump lang hook.
22354 * doc/invoke.texi: Document -fdump-lang option family.
22355 * dumpfile.c (dump_files): Remove class dump here.
22356 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
22357 * dumpfile.h (tree_dump_index): Remove TDI_class.
22358 * langhooks-def.h (lhd_register_dumps): Declare.
22359 (LANG_HOOKS_REGISTER_DUMPS): Define.
22360 (LANG_HOOKS_INITIALIZER): Add it.
22361 * langhooks.c (lhd_register_dumps): Define.
22362 * langhooks.h (struct lang_hooks): Add register_dumps.
22364 2017-05-19 Nathan Sidwell <nathan@acm.org>
22366 * context.h (context::set_passes): New.
22367 * context.c (context::context): Do not create pass manager.
22368 * toplev.c (general_init): Create pass manager here.
22370 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
22372 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
22373 use this splitter if two add or or instructions would also work for
22374 the constant we want to generate.
22376 2017-05-19 Richard Biener <rguenther@suse.de>
22379 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
22380 predicate evaluation.
22382 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
22384 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
22386 * ipa-inline.c (want_inline_small_function_p): Do not cast to
22389 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
22391 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22392 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22393 (inline_edge_summary_vec): Turn into ...
22394 (ipa_call_summaries): ... this one.
22395 (redirect_to_unreachable, edge_set_predicate,
22396 evaluate_properties_for_edge, inline_summary_alloc,
22397 reset_ipa_call_summary, reset_inline_summary,
22398 inline_summary_t::duplicate): Update.
22399 (inline_edge_duplication_hook): Turn to ...
22400 (ipa_call_summary_t::duplicate): ... this one.
22401 (inline_edge_removal_hook): Turn to ...
22402 (ipa_call_summary_t::remove): ... this one.
22403 (dump_inline_edge_summary): Turn to ...
22404 (dump_ipa_call_summary): ... this one.
22405 (estimate_function_body_sizes): Update.
22406 (inline_update_callee_summaries): Update.
22407 (remap_edge_change_prob): Update.
22408 (remap_edge_summaries): Update.
22409 (inline_merge_summary): Update.
22410 (do_estimate_edge_time): Update.
22411 (inline_generate_summary): Update.
22412 (inline_read_section): Update.
22413 (inline_read_summary): Update.
22414 (inline_free_summary): Update.
22415 * ipa-inline.c (can_inline_edge_p): Update.
22416 (compute_inlined_call_time): Update.
22417 (want_inline_small_function_p): Update.
22418 (edge_badness): Update.
22419 (early_inliner): Update.
22420 * ipa-inline.h (inline_edge_summary): Turn to ...
22421 (ipa_call_summary): ... this one.
22422 (ipa_call_summary_t): New class.
22423 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22424 (ipa_call_summaries): New.
22425 (inline_edge_summary): Remove.
22426 (estimate_edge_growth): Update.
22427 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22428 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22429 * ipa-split.c (execute_split_functions): Update.
22430 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22432 2017-05-19 Richard Biener <rguenther@suse.de>
22434 PR middle-end/80764
22435 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
22437 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
22439 * config/rs6000/rs6000.c (struct machine_function): Add field
22440 fpr_is_wrapped_separately.
22441 (rs6000_get_separate_components): Use 64 components. Handle the
22442 new FPR components.
22443 (rs6000_components_for_bb): Handle the FPR components.
22444 (rs6000_emit_prologue_components): Handle the FPR components.
22445 (rs6000_emit_epilogue_components): Handle the FPR components.
22446 (rs6000_set_handled_components): Handle the FPR components.
22447 (rs6000_emit_prologue): Don't output prologue code for those FPRs
22448 that are already separately shrink-wrapped.
22449 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
22450 that are already separately shrink-wrapped.
22452 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
22455 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
22458 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
22459 (define_peephole2 for Altivec d-form load): Add peepholes to catch
22460 cases where the register allocator uses a move and an offsettable
22461 memory operation to/from a FPR register on ISA 2.06/2.07.
22462 (define_peephole2 for Altivec d-form store): Likewise.
22464 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
22467 * config/i386/mmx.md (*mov<mode>_internal): Enable
22468 alternatives 11, 12, 13 and 14 also for 32bit targets.
22469 Remove alternatives 15, 16, 17 and 18.
22470 * config/i386/sse.md (vec_concatv2di): Change
22471 alternative (!x, *y) to (x, ?!*Yn).
22473 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
22475 * dumpfile.h (enum dump_kind): Remove stray comma.
22477 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
22479 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
22480 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
22481 predicate::num_conditions
22482 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
22483 (CHANGED): turn into predicate::changed.
22484 (agg_position_info): Move to ipa-predicate.h
22485 (add_condition, predicate::add_clause, predicate::operator &=,
22486 predicate::or_with, predicate::evaluate, predicate::probability,
22487 dump_condition, dump_clause, predicate::dump,
22488 predicate::remap_after_duplication, predicate::remap_after_inlining,
22489 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
22490 (evaluate_conditions_for_known_args): Update.
22491 (set_cond_stmt_execution_predicate): Update.
22492 * ipa-inline.h: Include ipa-predicate.h
22493 (condition, inline_param_summary, conditions, agg_position_info,
22494 predicate): Move to ipa-predicate.h
22495 * ipa-predicate.c: New file.
22496 * ipa-predicate.h: New file.
22498 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
22500 * final.c (leaf_function_p): Check we are not in a sequence.
22502 2017-05-18 Martin Liska <mliska@suse.cz>
22504 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
22505 * dumpfile.c (dump_register): Use new enum dump_kind.
22506 (get_dump_file_name): Likewise.
22507 (dump_enable_all): Likewise.
22508 (dump_switch_p_1): Likewise.
22509 (enable_rtl_dump_file): Remove usage of TDF_RTL.
22510 * dumpfile.h (enum dump_kind): New enum type.
22511 (struct dump_file_info): Create constructor and
22512 format fields and comments.
22513 * passes.c (pass_manager::register_one_dump_file):
22515 * statistics.c (statistics_early_init): Likewise.
22516 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
22517 TDF_TREE with TDF_SLIM.
22518 (gather_memory_references_ref): Likewise.
22520 2017-05-18 Martin Liska <mliska@suse.cz>
22522 * vec.h (struct vnull): Use it.
22524 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
22526 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
22527 (true_predicate, false_predicate, true_predicate_p,
22528 false_predicate_p): Remove.
22529 (single_cond_predicate, not_inlined_predicate): Turn to member function
22531 (add_condition): Update.
22532 (add_clause): Turn to...
22533 (predicate::add_clause): ... this one; update; allow passing NULL
22535 (and_predicates): Turn to ...
22536 (predicate::operator &=): ... this one.
22537 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
22538 (or_predicates): Turn to ...
22539 (predicate::or_with): ... this one.
22540 (evaluate_predicate): Turn to ...
22541 (predicate::evaluate): ... this one.
22542 (predicate_probability): Turn to ...
22543 (predicate::probability): ... this one.
22544 (dump_condition): Update.
22545 (dump_predicate): Turn to ...
22546 (predicate::dump): ... this one.
22547 (account_size_time): Update.
22548 (edge_set_predicate): Update.
22549 (set_hint_predicate): UPdate.
22550 (evaluate_conditions_for_known_args): Update.
22551 (evaluate_properties_for_edge): Update.
22552 (remap_predicate_after_duplication): Turn to...
22553 (predicate::remap_after_duplication): ... this one.
22554 (remap_hint_predicate_after_duplication): Update.
22555 (inline_summary_t::duplicate): UPdate.
22556 (dump_inline_edge_summary): Update.
22557 (dump_inline_summary): Update.
22558 (set_cond_stmt_execution_predicate): Update.
22559 (set_switch_stmt_execution_predicate): Update.
22560 (compute_bb_predicates): Update.
22561 (will_be_nonconstant_expr_predicate): Update.
22562 (will_be_nonconstant_predicate): Update.
22563 (phi_result_unknown_predicate): Update.
22564 (predicate_for_phi_result): Update.
22565 (array_index_predicate): Update.
22566 (estimate_function_body_sizes): Update.
22567 (estimate_node_size_and_time): Update.
22568 (estimate_ipcp_clone_size_and_time): Update.
22569 (remap_predicate): Rename to ...
22570 (predicate::remap_after_inlining): ... this one.
22571 (remap_hint_predicate): Update.
22572 (inline_merge_summary): Update.
22573 (inline_update_overall_summary): Update.
22574 (estimate_size_after_inlining): Update.
22575 (read_predicate): Rename to ...
22576 (predicate::stream_in): ... this one.
22577 (read_inline_edge_summary): Update.
22578 (write_predicate): Rename to ...
22579 (predicate::stream_out): ... this one.
22580 (write_inline_edge_summary): Update.
22581 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
22582 (clause_t): Turn to uint32_t
22583 (predicate): Turn to class; implement constructor and operators
22585 (size_time_entry): Update.
22586 (inline_summary): Update.
22587 (inline_edge_summary): Update.
22589 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
22591 * fold-const.c (fold_binary_loc): Move transformation...
22592 * match.pd (C - X CMP X): ... here.
22594 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
22596 * config/sparc/sparc.c (sparc_option_override): Set function
22597 alignment for -mcpu=niagara7 to 64 to match the I$ line.
22598 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
22600 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
22602 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
22604 2017-05-18 Marek Polacek <polacek@redhat.com>
22607 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
22608 (pass_ubsan::execute): Call gimple_assign_single_p instead of
22609 gimple_assign_load_p.
22611 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
22613 PR middle-end/80692
22614 * real.c (do_compare): Give decimal_do_compare preference over
22615 comparing just the signs.
22617 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
22619 * doc/md.texi (Canonicalization of Instructions): Describe the
22620 canonical form of instructions that inherently set a condition
22623 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
22625 PR middle-end/80775
22626 * tree-cfg.c: Move deletion of unreachable case statements to after
22627 the merging of consecutive case labels.
22629 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
22631 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
22632 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
22633 restoring of callee-saved registers.
22635 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
22637 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
22638 * config/visium/visium.c (single_set_and_flags): Likewise.
22639 * config/visium/visium.md (Substitutions): Likewise.
22641 2017-05-17 Martin Liska <mliska@suse.cz>
22643 * cfg.c: Introduce dump_flags_t type and
22644 use it instead of int type.
22646 * cfghooks.c: Likewise.
22647 * cfghooks.h (struct cfg_hooks): Likewise.
22648 * cfgrtl.c: Likewise.
22649 * cfgrtl.h: Likewise.
22650 * cgraph.c (cgraph_node::get_body): Likewise.
22651 * coretypes.h: Likewise.
22652 * domwalk.c: Likewise.
22653 * domwalk.h: Likewise.
22654 * dumpfile.c (struct dump_option_value_info): Likewise.
22655 (dump_enable_all): Likewise.
22656 (dump_switch_p_1): Likewise.
22657 (opt_info_switch_p): Likewise.
22658 * dumpfile.h (enum tree_dump_index): Likewise.
22659 (struct dump_file_info): Likewise.
22660 * genemit.c: Likewise.
22661 * generic-match-head.c: Likewise.
22662 * gengtype.c (open_base_files): Likewise.
22663 * gimple-pretty-print.c: Likewise.
22664 * gimple-pretty-print.h: Likewise.
22665 * graph.c (print_graph_cfg): Likewise.
22666 * graphite-scop-detection.c (dot_all_sese): Likewise.
22667 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
22668 * loop-unroll.c (report_unroll): Likewise.
22669 * passes.c (pass_manager::register_one_dump_file): Likewise.
22670 * print-tree.c: Likewise.
22671 * statistics.c: Likewise.
22672 * tree-cfg.c: Likewise.
22673 * tree-cfg.h: Likewise.
22674 * tree-dfa.c: Likewise.
22675 * tree-dfa.h: Likewise.
22676 * tree-dump.c (dump_function): Likewise.
22677 * tree-dump.h (struct dump_info): Likewise.
22678 * tree-pretty-print.c: Likewise.
22679 * tree-pretty-print.h: Likewise.
22680 * tree-ssa-live.c: Likewise.
22681 * tree-ssa-live.h: Likewise.
22682 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
22683 * tree-vect-loop.c: Likewise.
22684 * tree-vect-slp.c: Likewise.
22686 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
22687 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22689 PR tree-optimization/80457
22690 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
22691 of all arguments to a statement as scalar_to_vec operations.
22692 (vectorizable_call): Adjust call to vect_model_simple_cost for
22694 (vectorizable_conversion): Likewise.
22695 (vectorizable_assignment): Likewise.
22696 (vectorizable_shift): Likewise.
22697 (vectorizable_operation): Likewise.
22698 (vectorizable_comparison): Likewise.
22699 (vect_is_simple_cond): Record the def types for operands.
22700 (vectorizable_condition): Likewise, call vect_model_simple_cost.
22701 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
22702 for statement argument count.
22704 2017-05-16 Carl Love <cel@us.ibm.com>
22706 * config/rs6000/rs6000-c: Add support for built-in functions
22707 vector unsigned long long vec_bperm (vector unsigned long long,
22708 vector unsigned char)
22709 vector signed long long vec_mule (vector signed int,
22711 vector unsigned long long vec_mule (vector unsigned int,
22712 vector unsigned int)
22713 vector signed long long vec_mulo (vector signed int,
22715 vector unsigned long long vec_mulo (vector unsigned int,
22716 vector unsigned int)
22717 vector signed char vec_sldw (vector signed char,
22718 vector signed char,
22720 vector unsigned char vec_sldw (vector unsigned char,
22721 vector unsigned char,
22723 vector signed short vec_sldw (vector signed short,
22724 vector signed short,
22726 vector unsigned short vec_sldw (vector unsigned short,
22727 vector unsigned short,
22729 vector signed int vec_sldw (vector signed int,
22732 vector unsigned int vec_sldw (vector unsigned int,
22733 vector unsigned int,
22735 vector signed long long vec_sldw (vector signed long long,
22736 vector signed long long,
22738 vector unsigned long long vec_sldw (vector unsigned long long,
22739 vector unsigned long long,
22741 * config/rs6000/rs6000-c: Add support for built-in functions
22742 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
22743 * config/rs6000/altivec.h: Add defintion for vec_sldw.
22744 * doc/extend.texi: Update the built-in documentation for the
22745 new built-in functions.
22747 2017-05-16 Marek Polacek <polacek@redhat.com>
22751 * tree.c (save_expr): Don't fold the expression.
22753 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
22755 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
22756 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
22757 and (?*y,m). Update insn attributes.
22759 2017-05-16 Martin Liska <mliska@suse.cz>
22761 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
22762 flags argument of print_gimple_stmt, print_gimple_expr,
22763 print_generic_stmt and print_generic_expr.
22764 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
22765 * coretypes.h: Likewise.
22766 * except.c (dump_eh_tree): Likewise.
22767 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
22768 * gimple-pretty-print.h: Likewise.
22769 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
22770 (backprop::push_to_worklist): Likewise.
22771 (backprop::pop_from_worklist): Likewise.
22772 (backprop::process_use): Likewise.
22773 (backprop::intersect_uses): Likewise.
22774 (note_replacement): Likewise.
22775 * gimple-ssa-store-merging.c
22776 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
22777 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
22778 (pass_store_merging::execute): Likewise.
22779 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
22780 (ssa_base_cand_dump_callback): Likewise.
22781 (dump_incr_vec): Likewise.
22782 (replace_refs): Likewise.
22783 (replace_mult_candidate): Likewise.
22784 (create_add_on_incoming_edge): Likewise.
22785 (create_phi_basis): Likewise.
22786 (insert_initializers): Likewise.
22787 (all_phi_incrs_profitable): Likewise.
22788 (introduce_cast_before_cand): Likewise.
22789 (replace_one_candidate): Likewise.
22790 * gimplify.c (gimplify_expr): Likewise.
22791 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
22792 (set_rename): Likewise.
22793 (rename_uses): Likewise.
22794 (copy_loop_phi_nodes): Likewise.
22795 (add_close_phis_to_merge_points): Likewise.
22796 (copy_loop_close_phi_args): Likewise.
22797 (copy_cond_phi_args): Likewise.
22798 (graphite_copy_stmts_from_block): Likewise.
22799 (translate_pending_phi_nodes): Likewise.
22800 * graphite-poly.c (print_pdr): Likewise.
22801 (dump_gbb_cases): Likewise.
22802 (dump_gbb_conditions): Likewise.
22803 (print_scop_params): Likewise.
22804 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
22805 (build_cross_bb_scalars_use): Likewise.
22806 (gather_bbs::before_dom_children): Likewise.
22807 * hsa-dump.c (dump_hsa_immed): Likewise.
22808 * ipa-cp.c (print_ipcp_constant_value): Likewise.
22809 (get_replacement_map): Likewise.
22810 * ipa-inline-analysis.c (dump_condition): Likewise.
22811 (estimate_function_body_sizes): Likewise.
22812 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
22813 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
22814 * ipa-prop.c (ipa_dump_param): Likewise.
22815 (ipa_print_node_jump_functions_for_edge): Likewise.
22816 (ipa_modify_call_arguments): Likewise.
22817 (ipa_modify_expr): Likewise.
22818 (ipa_dump_param_adjustments): Likewise.
22819 (ipa_dump_agg_replacement_values): Likewise.
22820 (ipcp_modif_dom_walker::before_dom_children): Likewise.
22821 * ipa-pure-const.c (check_stmt): Likewise.
22822 (pass_nothrow::execute): Likewise.
22823 * ipa-split.c (execute_split_functions): Likewise.
22824 * omp-offload.c (dump_oacc_loop_part): Likewise.
22825 (dump_oacc_loop): Likewise.
22826 * trans-mem.c (tm_log_emit): Likewise.
22827 (tm_memopt_accumulate_memops): Likewise.
22828 (dump_tm_memopt_set): Likewise.
22829 (dump_tm_memopt_transform): Likewise.
22830 * tree-cfg.c (gimple_verify_flow_info): Likewise.
22831 (print_loop): Likewise.
22832 * tree-chkp-opt.c (chkp_print_addr): Likewise.
22833 (chkp_gather_checks_info): Likewise.
22834 (chkp_get_check_result): Likewise.
22835 (chkp_remove_check_if_pass): Likewise.
22836 (chkp_use_outer_bounds_if_possible): Likewise.
22837 (chkp_reduce_bounds_lifetime): Likewise.
22838 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
22839 (chkp_mark_completed_bounds): Likewise.
22840 (chkp_register_incomplete_bounds): Likewise.
22841 (chkp_mark_invalid_bounds): Likewise.
22842 (chkp_maybe_copy_and_register_bounds): Likewise.
22843 (chkp_build_returned_bound): Likewise.
22844 (chkp_get_bound_for_parm): Likewise.
22845 (chkp_build_bndldx): Likewise.
22846 (chkp_get_bounds_by_definition): Likewise.
22847 (chkp_generate_extern_var_bounds): Likewise.
22848 (chkp_get_bounds_for_decl_addr): Likewise.
22849 * tree-chrec.c (chrec_apply): Likewise.
22850 * tree-data-ref.c (dump_data_reference): Likewise.
22851 (dump_subscript): Likewise.
22852 (dump_data_dependence_relation): Likewise.
22853 (analyze_overlapping_iterations): Likewise.
22854 * tree-inline.c (expand_call_inline): Likewise.
22855 (tree_function_versioning): Likewise.
22856 * tree-into-ssa.c (dump_defs_stack): Likewise.
22857 (dump_currdefs): Likewise.
22858 (dump_names_replaced_by): Likewise.
22859 (dump_update_ssa): Likewise.
22860 (update_ssa): Likewise.
22861 * tree-object-size.c (pass_object_sizes::execute): Likewise.
22862 * tree-parloops.c (build_new_reduction): Likewise.
22863 (try_create_reduction_list): Likewise.
22864 (ref_conflicts_with_region): Likewise.
22865 (oacc_entry_exit_ok_1): Likewise.
22866 (oacc_entry_exit_single_gang): Likewise.
22867 * tree-pretty-print.h: Likewise.
22868 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
22869 (get_scalar_evolution): Likewise.
22870 (add_to_evolution): Likewise.
22871 (get_loop_exit_condition): Likewise.
22872 (analyze_evolution_in_loop): Likewise.
22873 (analyze_initial_condition): Likewise.
22874 (analyze_scalar_evolution): Likewise.
22875 (instantiate_scev): Likewise.
22876 (number_of_latch_executions): Likewise.
22877 (gather_chrec_stats): Likewise.
22878 (final_value_replacement_loop): Likewise.
22879 (scev_const_prop): Likewise.
22880 * tree-sra.c (dump_access): Likewise.
22881 (disqualify_candidate): Likewise.
22882 (create_access): Likewise.
22883 (reject): Likewise.
22884 (maybe_add_sra_candidate): Likewise.
22885 (create_access_replacement): Likewise.
22886 (analyze_access_subtree): Likewise.
22887 (analyze_all_variable_accesses): Likewise.
22888 (sra_modify_assign): Likewise.
22889 (initialize_constant_pool_replacements): Likewise.
22890 (find_param_candidates): Likewise.
22891 (decide_one_param_reduction): Likewise.
22892 (replace_removed_params_ssa_names): Likewise.
22893 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
22894 * tree-ssa-copy.c (dump_copy_of): Likewise.
22895 (copy_prop_visit_cond_stmt): Likewise.
22896 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
22897 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
22898 (record_equivalences_from_stmt): Likewise.
22899 * tree-ssa-dse.c (compute_trims): Likewise.
22900 (delete_dead_call): Likewise.
22901 (delete_dead_assignment): Likewise.
22902 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
22903 (forward_propagate_into_cond): Likewise.
22904 (pass_forwprop::execute): Likewise.
22905 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
22906 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
22908 (move_computations_worker): Likewise.
22909 (execute_sm): Likewise.
22910 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
22911 (remove_exits_and_undefined_stmts): Likewise.
22912 (remove_redundant_iv_tests): Likewise.
22913 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
22914 (adjust_iv_update_pos): Likewise.
22915 * tree-ssa-math-opts.c (bswap_replace): Likewise.
22916 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
22917 (value_replacement): Likewise.
22918 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
22919 * tree-ssa-pre.c (print_pre_expr): Likewise.
22920 (get_representative_for): Likewise.
22921 (create_expression_by_pieces): Likewise.
22922 (insert_into_preds_of_block): Likewise.
22923 (eliminate_insert): Likewise.
22924 (eliminate_dom_walker::before_dom_children): Likewise.
22925 (eliminate): Likewise.
22926 (remove_dead_inserted_code): Likewise.
22927 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
22928 * tree-ssa-reassoc.c (get_rank): Likewise.
22929 (eliminate_duplicate_pair): Likewise.
22930 (eliminate_plus_minus_pair): Likewise.
22931 (eliminate_not_pairs): Likewise.
22932 (undistribute_ops_list): Likewise.
22933 (eliminate_redundant_comparison): Likewise.
22934 (update_range_test): Likewise.
22935 (optimize_range_tests_var_bound): Likewise.
22936 (optimize_vec_cond_expr): Likewise.
22937 (rewrite_expr_tree): Likewise.
22938 (rewrite_expr_tree_parallel): Likewise.
22939 (linearize_expr): Likewise.
22940 (break_up_subtract): Likewise.
22941 (linearize_expr_tree): Likewise.
22942 (attempt_builtin_powi): Likewise.
22943 (attempt_builtin_copysign): Likewise.
22944 (transform_stmt_to_copy): Likewise.
22945 (transform_stmt_to_multiply): Likewise.
22946 (dump_ops_vector): Likewise.
22947 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
22948 (print_scc): Likewise.
22949 (set_ssa_val_to): Likewise.
22950 (visit_reference_op_store): Likewise.
22951 (visit_use): Likewise.
22952 (sccvn_dom_walker::before_dom_children): Likewise.
22953 (run_scc_vn): Likewise.
22954 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
22956 (expr_hash_elt::print): Likewise.
22957 (const_and_copies::pop_to_marker): Likewise.
22958 (const_and_copies::record_const_or_copy_raw): Likewise.
22959 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
22960 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
22961 (dump_predicates): Likewise.
22962 (find_uninit_use): Likewise.
22963 (warn_uninitialized_phi): Likewise.
22964 (pass_late_warn_uninitialized::execute): Likewise.
22965 * tree-ssa.c (verify_vssa): Likewise.
22966 (verify_ssa): Likewise.
22967 (maybe_optimize_var): Likewise.
22968 * tree-vrp.c (dump_value_range): Likewise.
22969 (dump_all_value_ranges): Likewise.
22970 (dump_asserts_for): Likewise.
22971 (register_edge_assert_for_2): Likewise.
22972 (vrp_visit_cond_stmt): Likewise.
22973 (vrp_visit_switch_stmt): Likewise.
22974 (vrp_visit_stmt): Likewise.
22975 (vrp_visit_phi_node): Likewise.
22976 (simplify_cond_using_ranges_1): Likewise.
22977 (fold_predicate_in): Likewise.
22978 (evrp_dom_walker::before_dom_children): Likewise.
22979 (evrp_dom_walker::push_value_range): Likewise.
22980 (evrp_dom_walker::pop_value_range): Likewise.
22981 (execute_early_vrp): Likewise.
22983 2017-05-16 Richard Biener <rguenther@suse.de>
22985 * dwarf2out.c (loc_list_from_tree_1): Do not create
22986 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
22988 2017-05-16 Richard Biener <rguenther@suse.de>
22990 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
22992 (note_variable_value_in_expr): If we resolved the decl ref
22993 do not push to the stack.
22995 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
22997 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
22998 operations in fast-math mode.
22999 (vaddq_f16): Likewise.
23000 (vmul_f16): Likewise.
23001 (vmulq_f16): Likewise.
23002 (vsub_f16): Likewise.
23003 (vsubq_f16): Likewise.
23004 * config/arm/neon.md (add<mode>3): New.
23006 (fma:<VH:mode>3): New. Also remove outdated comment.
23009 2017-05-16 Martin Liska <mliska@suse.cz>
23013 * ipa-devirt.c (warn_types_mismatch): Fix typo.
23014 (odr_types_equivalent_p): Likewise.
23016 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
23018 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
23020 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
23023 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
23024 non-interunit SSE move alternatives with '?'.
23025 (zero-extendsidi peephole2): New peephole to skip intermediate
23026 general register in SSE zero-extend sequence.
23028 2017-05-15 Jeff Law <law@redhat.com>
23030 * reorg.c (relax_delay_slots): Create a new variable to hold
23031 the temporary target rather than clobbering TARGET_LABEL.
23033 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
23034 missing argument to extract_bit_field call.
23035 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
23037 2017-05-15 Martin Liska <mliska@suse.cz>
23040 * gcc.c (process_command): Do not allow empty argument of -o option.
23042 2017-05-15 Renlin Li <renlin.li@arm.com>
23044 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
23045 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
23046 * config/aarch64/constraints.md (Usf): Add long call check.
23047 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
23048 (call_value): Likewise.
23049 (sibcall): Likewise.
23050 (sibcall_value): Likewise.
23052 (call_value_insn): New.
23053 (sibcall_insn): Update rtx pattern.
23054 (sibcall_value_insn): Likewise.
23055 (call_internal): Remove.
23056 (call_value_internal): Likewise.
23057 (sibcall_internal): Likewise.
23058 (sibcall_value_internal): Likewise.
23059 (call_reg): Likewise.
23060 (call_symbol): Likewise.
23061 (call_value_reg): Likewise.
23062 (call_value_symbol): Likewise.
23064 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
23067 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
23069 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
23071 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
23072 compatible with CCGOCmode and with CCZmode.
23074 2017-05-14 Martin Sebor <msebor@redhat.com>
23076 PR middle-end/77671
23077 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
23078 (gimple_fold_builtin_snprintf): Same.
23079 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
23080 (gimple_fold_builtin_snprintf): Same.
23081 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
23082 of character types.
23083 (is_call_safe): New function.
23084 (try_substitute_return_value): Call it.
23085 (try_simplify_call): New function.
23086 (pass_sprintf_length::handle_gimple_call): Call it.
23088 2017-05-14 Martin Sebor <msebor@redhat.com>
23090 PR middle-end/80669
23091 * builtins.c (expand_builtin_stpncpy): Simplify.
23093 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
23095 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
23096 * config/i386/i386.h
23097 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
23098 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
23099 (struct machine_function): Add new members call_ms2sysv,
23100 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
23101 (struct machine_frame_state): New fields sp_realigned and
23102 sp_realigned_offset.
23103 * config/i386/i386.c
23104 (enum xlogue_stub): New enum.
23105 (enum xlogue_stub_sets): New enum.
23106 (class xlogue_layout): New class.
23107 (struct ix86_frame): New fields stack_realign_allocate_offset,
23108 stack_realign_offset and outlined_save_offset. Modify comments to
23109 detail stack layout when using out-of-line stubs.
23110 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
23111 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
23112 -mcall-ms2sysv-xlogues.
23113 (stub_managed_regs): New static variable.
23114 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
23115 registers managed by out-of-line stub.
23116 (disable_call_ms2sysv_xlogues): New function.
23117 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
23118 m->call_ms2sysv when appropriate and compute frame layout for
23120 (sp_valid_at, fp_valid_at): New inline functions.
23121 (choose_basereg): New function.
23122 (choose_baseaddr): Add align parameter, use choose_basereg and modify
23124 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
23125 Use align parameter of choose_baseaddr to generated aligned SSE movs
23127 (pro_epilogue_adjust_stack): Modify to track
23128 machine_frame_state::sp_realigned.
23129 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
23130 (ix86_nsaved_sseregs): Likewise.
23131 (ix86_emit_save_regs): Likewise.
23132 (ix86_emit_save_regs_using_mov): Likewise.
23133 (ix86_emit_save_sse_regs_using_mov): Likewise.
23134 (get_scratch_register_on_entry): Likewise.
23135 (gen_frame_set): New function.
23136 (gen_frame_load): Likewise.
23137 (gen_frame_store): Likewise.
23138 (emit_outlined_ms2sysv_save): Likewise.
23139 (emit_outlined_ms2sysv_restore): Likewise.
23140 (ix86_expand_prologue): Modify stack re-alignment code and call
23141 emit_outlined_ms2sysv_save when appropriate.
23142 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
23143 parameter rtx_insn *insn, which allows the function to be used to only
23144 generate the notes.
23145 (ix86_expand_epilogue): Modify validity checks of frame and stack
23146 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
23147 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
23148 * config/i386/predicates.md
23149 (save_multiple): New predicate.
23150 (restore_multiple): Likewise.
23151 * config/i386/sse.md
23152 (save_multiple<mode>): New pattern.
23153 (save_multiple_realign<mode>): Likewise.
23154 (restore_multiple<mode>): Likewise.
23155 (restore_multiple_and_return<mode>): Likewise.
23156 (restore_multiple_leave_return<mode>): Likewise.
23157 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
23159 2017-05-14 Julia Koval <julia.koval@intel.com>
23161 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
23162 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
23163 (__builtin_ia32_xsetbv): New builtins.
23164 * config/i386/i386.c (ix86_expand_special_args_builtin):
23166 (ix86_expand_builtin): Special expand for new intrinsics.
23167 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
23168 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
23169 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
23171 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23173 * cfganal.c (inverted_post_order_compute): Change argument type
23175 * cfganal.h (inverted_post_order_compute): Adjust prototype.
23176 * df-core.c (rest_of_handle_df_initialize): Adjust.
23177 (rest_of_handle_df_finish): Likewise.
23178 (df_analyze_1): Likewise.
23179 (df_analyze): Likewise.
23180 (loop_inverted_post_order_compute): Change argument to be a vec *.
23181 (df_analyze_loop): Adjust.
23182 (df_get_n_blocks): Likewise.
23183 (df_get_postorder): Likewise.
23184 * df.h (struct df_d): Change field to be a vec.
23185 * lcm.c (compute_laterin): Adjust.
23186 (compute_available): Likewise.
23187 * lra-lives.c (lra_create_live_ranges_1): Likewise.
23188 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
23189 * tree-ssa-pre.c (compute_antic): Likewise.
23191 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23193 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
23194 (depth_first_search::depth_first_search): Change structure init
23195 function to this constructor.
23196 (depth_first_search::add_bb): Rename function to this member.
23197 (depth_first_search::execute): Likewise.
23198 (flow_dfs_compute_reverse_finish): Adjust.
23200 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23202 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
23203 (longest_simple_path): Likewise.
23204 * shrink-wrap.c (spread_components): Likewise.
23205 (disqualify_problematic_components): Likewise.
23206 (emit_common_heads_for_components): Likewise.
23207 (emit_common_tails_for_components): Likewise.
23208 (insert_prologue_epilogue_for_components): Likewise.
23210 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23212 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
23215 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23217 * df-core.c (df_set_blocks): Start using auto_bitmap.
23218 (df_compact_blocks): Likewise.
23219 * df-problems.c (df_rd_confluence_n): Likewise.
23220 * df-scan.c (df_insn_rescan_all): Likewise.
23221 (df_process_deferred_rescans): Likewise.
23222 (df_update_entry_block_defs): Likewise.
23223 (df_update_exit_block_uses): Likewise.
23224 (df_entry_block_bitmap_verify): Likewise.
23225 (df_exit_block_bitmap_verify): Likewise.
23226 (df_scan_verify): Likewise.
23227 * lra-constraints.c (lra_constraints): Likewise.
23228 (undo_optional_reloads): Likewise.
23229 (lra_undo_inheritance): Likewise.
23230 * lra-remat.c (calculate_gen_cands): Likewise.
23231 (do_remat): Likewise.
23232 * lra-spills.c (assign_spill_hard_regs): Likewise.
23233 (spill_pseudos): Likewise.
23234 * tree-ssa-pre.c (bitmap_set_and): Likewise.
23235 (bitmap_set_subtract_values): Likewise.
23237 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23239 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
23240 management with auto_bitmap.
23241 (fix_inter_tick): Likewise.
23242 (fix_recovery_deps): Likewise.
23243 * ira.c (add_store_equivs): Likewise.
23244 (find_moveable_pseudos): Likewise.
23245 (split_live_ranges_for_shrink_wrap): Likewise.
23246 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
23247 (rtx_reuse_manager::seen_def_p): Likewise.
23248 (rtx_reuse_manager::set_seen_def): Likewise.
23249 * print-rtl.h (class rtx_reuse_manager): Likewise.
23251 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23253 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
23255 (migrate_btr_def): Likewise.
23256 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
23257 * df-core.c (loop_post_order_compute): Likewise.
23258 (loop_inverted_post_order_compute): Likewise.
23259 * hsa-common.h: Likewise.
23260 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
23261 * init-regs.c (initialize_uninitialized_regs): Likewise.
23262 * ipa-inline.c (resolve_noninline_speculation): Likewise.
23263 (inline_small_functions): Likewise.
23264 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
23265 * ira.c (combine_and_move_insns): Likewise.
23266 (build_insn_chain): Likewise.
23267 * loop-invariant.c (find_invariants): Likewise.
23268 * lower-subreg.c (propagate_pseudo_copies): Likewise.
23269 * predict.c (tree_predict_by_opcode): Likewise.
23270 (predict_paths_leading_to): Likewise.
23271 (predict_paths_leading_to_edge): Likewise.
23272 (estimate_loops_at_level): Likewise.
23273 (estimate_loops): Likewise.
23274 * shrink-wrap.c (try_shrink_wrapping): Likewise.
23275 (spread_components): Likewise.
23276 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
23277 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
23278 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
23279 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
23280 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
23281 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
23282 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
23283 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
23284 (mark_threaded_blocks): Likewise.
23285 (thread_through_all_blocks): Likewise.
23286 * tree-ssa.c (verify_ssa): Likewise.
23287 (execute_update_addresses_taken): Likewise.
23288 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
23290 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23292 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
23294 (post_order_compute): Likewise.
23295 (inverted_post_order_compute): Likewise.
23296 (pre_and_rev_post_order_compute_fn): Likewise.
23298 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23300 * genrecog.c (int_set::int_set): Explicitly construct our
23301 auto_vec base class.
23302 * vec.h (auto_vec::auto_vec): New constructor.
23304 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23306 * bitmap.h (class auto_bitmap): New constructor taking
23307 bitmap_obstack * argument.
23309 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23311 * bitmap.h (class auto_bitmap): Change type of m_bits to
23312 bitmap_head, and adjust ctor / dtor and member operators.
23314 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
23316 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
23317 when returned register mode doesn't match original mode.
23319 2017-05-12 Jeff Law <law@redhat.com>
23320 Jakub Jelinek <jakub@redhat.com>
23322 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
23323 we look for cc setter after the compare-elim changes.
23324 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
23325 within the vector to match what compare-elim now expects.
23326 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
23327 (xorsi3_flags, one_cmplsi2_flags): Likewise.
23329 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
23330 after the compare-elim changes.
23331 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
23332 the vector to match what compare-elim now expects.
23333 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
23334 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
23335 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
23336 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
23337 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
23339 * config/visium/visium.c (single_set_and_flags): Fix where
23340 we look for cc setter after the compare-elim changes.
23341 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
23342 with the vector to match what compare-elim now expects.
23343 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
23344 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
23345 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
23346 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
23347 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
23348 (neg<mode>2_insn_set_overflow): Likewise.
23350 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
23352 PR middle-end/79794
23353 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
23354 maybe_expand_insn call, set ops[0].target. If still set after call,
23355 set alt_rtl. Add extra arg to recursive calls.
23356 (extract_bit_field): Add alt_rtl argument. Pass to
23358 * expmed.h (extract_bit_field): Fix prototype.
23359 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
23360 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
23361 to extract_bit_field_calls.
23362 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
23363 Pass alt_rtl to extract_bit_field calls.
23364 * calls.c (store_unaligned_arguments_into_psuedos)
23365 load_register_parameters): Pass extra NULL to extract_bit_field calls.
23366 * optabs.c (maybe_legitimize_operand): Clear op->target when call
23368 * optabs.h (struct expand_operand): Add target bitfield.
23370 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
23372 * compare-elim.c (try_eliminate_compare): Canonicalize
23373 operation with embedded compare to
23374 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
23375 (set (reg) (operation)].
23377 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
23379 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
23382 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
23383 cost of adding a carry flag for ADC instruction.
23384 [case MINUS]: Ignore the cost of subtracting a carry flag
23385 for SBB instruction.
23387 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
23389 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
23391 * config/rs6000/bmiintrin.h: New file.
23392 * config/rs6000/bmi2intrin.h: New file.
23393 * config/rs6000/x86intrin.h: New file.
23395 2017-05-12 Jeff Law <law@redhat.com>
23397 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
23400 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
23402 PR middle-end/80707
23403 * tree-cfg.c: Remove cfg edges of unreachable case statements.
23405 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23407 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
23408 early expansion of vector divide builtins.
23409 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
23410 builtins identified as having unsigned arguments.
23412 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23414 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
23415 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
23416 expansion of vector logical operations (and, andc, or, xor,
23419 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23421 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
23422 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
23424 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23426 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
23427 early GIMPLE expansion of vector multiplies.
23429 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23431 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
23432 TARGET_HAVE_MOVT conditional.
23433 (movt splitter): Likewise.
23435 2017-05-12 Richard Biener <rguenther@suse.de>
23437 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
23438 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
23439 Fold all stmts not inplace.
23441 2017-05-12 Richard Biener <rguenther@suse.de>
23443 PR tree-optimization/80713
23444 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
23445 inserted_exprs bit for not removed stmts.
23447 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
23449 PR middle-end/69921
23450 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
23451 parallelized" attribute for parallelized OpenACC kernels.
23452 * omp-offload.c (execute_oacc_device_lower): Use it.
23454 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
23455 Set "oacc kernels" attribute.
23456 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
23457 parameter. Adjust all users.
23458 (oacc_fn_attrib_kernels_p): Remove function.
23459 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
23460 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
23461 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
23462 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
23463 assert "oacc kernels" attribute is set.
23465 2017-05-11 Carl Love <cel@us.ibm.com>
23467 * config/rs6000/rs6000-c: Add support for built-in functions
23468 vector unsigned char vec_popcnt (vector signed char)
23469 vector unsigned char vec_popcnt (vector unsigned char)
23470 vector unsigned short vec_popcnt (vector signed short)
23471 vector unsigned short vec_popcnt (vector unsigned short)
23472 vector unsigned int vec_popcnt (vector signed int)
23473 vector unsigned int vec_popcnt (vector unsigned int)
23474 vector unsigned long long vec_popcnt (vector signed long long)
23475 vector unsigned long long vec_popcnt (vector unsigned long long)
23476 vector signed long long vec_slo (vector signed long long,
23477 vector signed char)
23478 vector signed long long vec_slo (vector signed long long,
23479 vector unsigned char)
23480 vector unsigned long long vec_slo (vector unsigned long long,
23481 vector signed char)
23482 vector unsigned long long vec_slo (vector unsigned long long,
23483 vector unsigned char)
23484 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
23485 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
23486 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
23487 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
23488 * doc/extend.texi: Update the built-in documentation file for the
23489 new built-in functions.
23491 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
23493 * attribs.h (sorted_attr_string): Move machine independent
23494 functions for target clone support from the i386 port to common
23495 code. Rename ix86_function_versions to common_function_versions.
23496 Rename make_name to make_unique_name.
23497 (common_function_versions): Likewise.
23498 (make_unique_name): Likewise.
23499 (make_dispatcher_decl): Likewise.
23500 (is_function_default_version): Likewise.
23501 * attribs.c (attr_strcmp): Likewise.
23502 (sorted_attr_string): Likewise.
23503 (common_function_versions): Likewise.
23504 (make_unique_name): Likewise.
23505 (make_dispatcher_decl): Likewise.
23506 (is_function_default_version): Likewise.
23507 * config/i386/i386.c (attr_strcmp): Likewise.
23508 (sorted_attr_string): Likewise.
23509 (ix86_function_versions): Likewise.
23510 (make_name): Likewise.
23511 (make_dispatcher_decl): Likewise.
23512 (is_function_default_version): Likewise.
23513 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
23515 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23518 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23519 Account for direct move costs for vec_construct of integer
23522 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
23525 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
23526 (UNSPEC_STX_ATOMIC): Ditto.
23527 (loaddi_via_sse): New insn.
23528 (storedi_via_sse): Ditto.
23529 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
23530 Update corresponding peephole2 patterns.
23531 (atomic_storedi_fpu): Ditto.
23533 2017-05-11 Julia Koval <julia.koval@intel.com>
23535 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
23536 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
23538 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
23539 (__builtin_ia32_rsqrt14ss_mask): New builtins.
23540 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
23542 2017-05-11 Nathan Sidwell <nathan@acm.org>
23544 * graphite-poly.c: Include dumpfile.h.
23546 * dumpfle.h (dump_function): Declare here ...
23547 * tree-dump.h (dump_function): ... not here.
23548 * dumpfile.c: #include tree-cfg.h.
23549 (dump_function): Move here from ...
23550 * tree-dump.c (dump_function): ... here.
23551 * gimplify.c: #include splay-tree.h, not tree-dump.h.
23552 * graphite-poly.c: Don't include tree-dump.h.
23553 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
23554 * print-tree.c: Likewise.
23555 * stor-layout.c: Likewise.
23556 * tree-nested.c: Likewise.
23558 * dumpfile.c (dump_start): Use TDF_FLAGS.
23559 (dump_enable_all): Fix TDF_KIND check thinko.
23561 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
23563 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23564 array entries to represent two legal parameterizations of the
23565 overloaded __builtin_cmpb function, as represented by the
23566 P6_OV_BUILTIN_CMPB constant.
23567 (altivec_resolve_overloaded_builtin): Add special case handling
23568 for the __builtin_cmpb function, as represented by the
23569 P6_OV_BUILTIN_CMPB constant.
23570 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
23571 (BU_P6_64BIT_2): New macro.
23572 (BU_P6_OVERLOAD_2): New macro
23573 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
23574 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
23575 (CMPB): Add overload support to represent both 32-bit and 64-bit
23576 compare-bytes function.
23577 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
23578 support for TARGET_CMPB.
23579 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
23580 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23581 documentation of the __builtin_cmpb overloaded built-in function.
23583 2017-05-11 Richard Biener <rguenther@suse.de>
23585 PR tree-optimization/80705
23586 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
23587 bases are not vectorizable.
23589 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23591 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
23592 when counting register pressure.
23594 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23596 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
23597 (iv_ca_more_deps): Renamed to ...
23598 (iv_ca_compare_deps): ... this.
23599 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
23601 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23603 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
23605 (determine_group_iv_costs): ... here.
23606 (find_inv_vars_cb): Record inv var if it's not recorded before.
23608 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23610 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
23611 (get_shiftadd_cost): Ditto.
23613 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23615 * tree-ssa-address.c: Include header file.
23616 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
23618 (add_to_parts): Refactor.
23619 (addr_to_parts): New parameter. Update use of move_hint_to_base.
23620 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
23623 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23625 PR tree-optimization/53090
23626 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
23628 (extract_cond_operands): Detect condition with IV on both sides
23629 and return COMP_IV_EXPR_2.
23630 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
23631 (rewrite_use_compare): Simplify by removing call to function
23632 extract_cond_operands.
23634 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23636 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
23637 (extract_cond_operands): Detect condition comparing against non-
23638 invariant bound and return appropriate enum value.
23639 (find_interesting_uses_cond): Update use of extract_cond_operands.
23640 Handle its return value accordingly.
23641 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
23643 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23645 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
23646 nonlinear iv_use computation in loop invariant sensitive way.
23648 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23650 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
23651 (find_iv_candidates): Call relate_compare_use_with_all_cands.
23653 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23655 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
23656 (dump_cand): Support iv_cand.inv_exprs.
23657 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
23659 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
23662 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23664 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
23666 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
23667 as local function. Include necessary header files.
23668 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
23670 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23672 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
23674 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23676 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
23677 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
23678 RSHIFT_EXPR and BIT_NOT_EXPR.
23680 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23682 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
23683 (adjust_setup_cost): New parameter supporting round up adjustment.
23684 (struct address_cost_data): Delete.
23685 (force_expr_to_var_cost): Don't bound cost with spill_cost.
23686 (split_address_cost, ptr_difference_cost): Delete.
23687 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
23688 (struct ainc_cost_data): New struct.
23689 (get_address_cost_ainc): New function.
23690 (get_address_cost, get_computation_cost): Reimplement.
23691 (determine_group_iv_cost_address): Record inv_expr for all uses of
23693 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
23694 (iv_ca_has_deps): Reimplemented to ...
23695 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
23697 (iv_ca_extend): Call iv_ca_more_deps.
23699 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23701 * tree-ssa-address.c (struct mem_address): Move to header file.
23702 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
23703 * tree-ssa-address.h (struct mem_address): Move from C file.
23704 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
23706 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23708 * tree-affine.h (aff_combination_type): New interface.
23709 (aff_combination_zero_p): Remove static.
23710 (aff_combination_const_p): New interface.
23711 (aff_combination_singleton_var_p): New interfaces.
23713 2017-05-11 Richard Biener <rguenther@suse.de>
23715 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
23716 Skip unreachable blocks and destinations.
23717 (eliminate): Move stmt removal and fixup ...
23718 (fini_eliminate): ... here. Skip inserted exprs.
23719 (pass_pre::execute): Move fini_pre after fini_eliminate.
23720 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
23721 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
23722 PRE to get rid of dead code that has invalid SSA form and
23723 split critical edges again.
23725 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23727 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
23729 2017-05-11 Richard Biener <rguenther@suse.de>
23731 * passes.c (execute_function_todo): Verify loops if they are
23732 said to be up-to-date.
23733 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
23734 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
23736 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
23739 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
23740 handle calling assemble_external ourself.
23743 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
23744 modes with zero size. Enhance comment.
23746 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23748 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
23749 built-ins for vec_xl and vec_xst with short and char pointer
23752 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
23754 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
23755 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
23756 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
23757 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
23758 (_mm_maskz_min_round_ss): New intrinsics.
23759 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
23760 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
23761 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
23762 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
23763 (__builtin_ia32_minss_mask_round): New builtins.
23764 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23765 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23766 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
23768 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
23769 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
23771 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
23774 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
23776 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
23777 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
23778 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
23779 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
23780 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
23781 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
23782 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
23783 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23784 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
23785 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
23786 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
23787 (__builtin_ia32_mulss_mask_round): New builtins.
23788 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23789 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23790 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
23792 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
23793 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
23795 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
23798 2017-05-10 Julia Koval <julia.koval@intel.com>
23800 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
23801 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
23802 (_mm256_setr_m128i): New intrinsics.
23804 2017-05-10 Julia Koval <julia.koval@intel.com>
23806 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
23807 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
23808 (_mm_maskz_rcp14_ss): New intrinsics.
23809 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
23810 (__builtin_ia32_rcp14ss_mask): New builtins.
23811 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
23813 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
23815 PR tree-optimization/51513
23816 * tree-cfg.c (gimple_seq_unreachable_p): New function.
23817 (assert_unreachable_fallthru_edge_p): Use it.
23818 (group_case_labels_stmt): Likewise.
23819 * tree-cfg.h: Prototype it.
23820 * stmt.c: Include cfghooks.h and tree-cfg.h.
23821 (emit_case_dispatch_table) <gap_label>: New local variable.
23822 Use it to fill dispatch table gaps.
23823 Test for default_label before updating probabilities.
23824 (expand_case) <default_label>: Remove unneeded initialization.
23825 Test for unreachable default case statement and remove its edge.
23826 Set default_label accordingly.
23827 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
23829 2017-05-10 Carl Love <cel@us.ibm.com>
23831 * config/rs6000/rs6000-c: Add support for built-in functions
23832 vector signed char vec_neg (vector signed char)
23833 vector signed short int vec_neg (vector short int)
23834 vector signed int vec_neg (vector signed int)
23835 vector signed long long vec_neg (vector signed long long)
23836 vector float vec_neg (vector float)
23837 vector double vec_neg (vector double)
23838 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
23840 * config/rs6000/altivec.h: Add define for vec_neg
23841 * doc/extend.texi: Update the built-in documentation for the
23842 new built-in functions.
23844 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23846 PR tree-optimization/77644
23847 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
23849 2017-05-10 Nathan Sidwell <nathan@acm.org>
23851 * dumpfile.h (TDI_lang_all): New.
23852 (TDF_KIND): New. Renumber others
23853 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
23855 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
23857 (get_dump_file_name): Adjust suffix generation.
23858 (dump_enable_all): Use TDF_KIND.
23859 * doc/invoke.texi (-fdump-lang-all): Document.
23861 * dumpfile.h: Tabify.
23863 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
23866 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
23867 Move member access before delete.
23869 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
23871 * tree-inline.c (expand_call_inline): Split block at stmt
23874 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
23877 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
23878 are now unused after splitting mov{sf,sd}_hardfloat.
23879 (f32_lr2): Likewise.
23880 (f32_lm): Likewise.
23881 (f32_lm2): Likewise.
23882 (f32_li): Likewise.
23883 (f32_li2): Likewise.
23884 (f32_lv): Likewise.
23885 (f32_sr): Likewise.
23886 (f32_sr2): Likewise.
23887 (f32_sm): Likewise.
23888 (f32_sm2): Likewise.
23889 (f32_si): Likewise.
23890 (f32_si2): Likewise.
23891 (f32_sv): Likewise.
23892 (f32_dm): Likewise.
23893 (f32_vsx): Likewise.
23894 (f32_av): Likewise.
23895 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
23896 For movsf, order stores so the VSX stores occur before the GPR
23897 store which encourages the register allocator to use a traditional
23898 FPR instead of a GPR. For movsd, order the stores so that the GPR
23899 store comes before the VSX stores to allow the power6 to work.
23900 This is due to the power6 not having a 32-bit integer store
23901 instruction from a FPR.
23902 (movsf_hardfloat): Likewise.
23903 (movsd_hardfloat): Likewise.
23905 2017-05-09 Martin Sebor <msebor@redhat.com>
23907 PR translation/80280
23908 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
23911 PR translation/80280
23912 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
23913 data member added in r247778.
23914 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
23916 2017-05-09 Nathan Sidwell <nathan@acm.org>
23918 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
23920 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
23923 2017-05-09 Marek Polacek <polacek@redhat.com>
23925 * doc/invoke.texi: Fix typo.
23927 2017-05-09 Richard Biener <rguenther@suse.de>
23929 * tree-vrp.c (vrp_val_is_max): Adjust comment.
23930 (vrp_val_is_min): Likewise.
23931 (set_value_range_to_value): Likewise.
23932 (set_value_range_to_nonnegative): Likewise.
23933 (gimple_assign_nonzero_p): Likewise.
23934 (gimple_stmt_nonzero_p): Likewise.
23935 (vrp_int_const_binop): Likewise. Remove unreachable case.
23936 (adjust_range_with_scev): Adjust comments.
23937 (compare_range_with_value): Likewise.
23938 (extract_range_from_phi_node): Likewise.
23939 (test_for_singularity): Likewise.
23941 2017-05-09 Richard Biener <rguenther@suse.de>
23943 * tree-vrp.c (get_single_symbol): Add assert that we don't
23944 get overflowed constants as invariant part.
23945 (compare_values_warnv): Add comment before the TREE_NO_WARNING
23946 checks. Use wi::cmp instead of recursing for integer constants.
23947 (compare_values): Just ignore whether we assumed undefined
23948 overflow instead of failing the compare.
23949 (extract_range_for_var_from_comparison_expr): Add comment before the
23950 TREE_NO_WARNING sets.
23951 (test_for_singularity): Likewise.
23952 (extract_range_from_comparison): Do not disable optimization
23953 when we assumed undefined overflow.
23954 (extract_range_basic): Remove init of unused var.
23956 2017-05-09 Richard Biener <rguenther@suse.de>
23958 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
23959 (extract_range_from_multiplicative_op_1): Adjust.
23960 (extract_range_from_binary_expr_1): Use int_const_binop.
23962 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
23965 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
23966 rs6000_store_data_bypass_p in seven define_bypass directives and
23967 in several comments.
23968 * config/rs6000/rs6000-protos.h: Add prototype for
23969 rs6000_store_data_bypass_p function.
23970 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
23971 function implements slightly different (rs6000-specific) semantics
23972 than store_data_bypass_p, returning false rather than aborting
23973 with assertion error when arguments do not satisfy the
23974 requirements of store data bypass.
23975 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
23976 rs6000_store_data_bypass_p.
23978 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
23980 * config/xtensa/xtensa-protos.h
23981 (xtensa_initial_elimination_offset): New declaration.
23982 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
23983 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
23984 macro definition, add case for FRAME_POINTER_REGNUM when
23985 FRAME_GROWS_DOWNWARD.
23986 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
23987 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
23988 xtensa_initial_elimination_offset.
23990 2017-05-08 Nathan Sidwell <nathan@acm.org>
23992 * doc/invoke.texi: Alphabetize -fdump options.
23994 2017-05-08 Martin Sebor <msebor@redhat.com>
23996 PR translation/80280
23997 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
23999 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
24001 * target.def (compute_frame_layout): New optional target hook.
24002 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
24003 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
24004 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
24006 * reload1.c (verify_initial_elim_offsets): Likewise.
24007 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
24008 (use_simple_return_p): Call arm_compute_frame_layout if needed.
24009 (arm_get_frame_offsets): Split up into this ...
24010 (arm_compute_frame_layout): ... and this function.
24012 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
24014 * config/aarch64/constraints.md (Usa): New constraint.
24015 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
24017 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
24019 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
24020 with_multilib_list after it has been checked.
24022 2017-05-08 Richard Biener <rguenther@suse.de>
24024 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
24025 (bitmap_set_subtract_values): Likewise.
24027 2017-05-08 Richard Biener <rguenther@suse.de>
24029 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
24030 (gimple_assign_nonzero): ... this and remove strict_overflow_p
24032 (gimple_stmt_nonzero_warnv_p): Rename to ...
24033 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
24035 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
24036 (extract_range_basic): Adjust, do not disable propagation on
24037 strict overflow sensitive simplification.
24038 (vrp_visit_cond_stmt): Likewise.
24040 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
24042 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
24043 body size unconditionally.
24045 2017-05-07 Jeff Law <law@redhat.com>
24048 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 Jeff Law <law@redhat.com>
24062 PR tree-optimization/78496
24063 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24066 PR tree-optimization/78496
24067 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24068 (simplify_stmt_using_ranges): Call it.
24069 (vrp_dom_walker::before_dom_children): Extract equivalences
24070 from an ASSERT_EXPR with an equality comparison against a
24073 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24075 * lra-constraints.c (lra_copy_reg_equiv): New function.
24076 (split_reg): Use it to copy equivalence information from the
24077 original register to the spill register.
24079 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24081 PR rtl-optimization/75964
24082 * simplify-rtx.c (simplify_const_relational_operation): Remove
24083 invalid handling of comparisons of integer ABS.
24085 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
24087 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
24088 initialize to zero.
24089 (init_regs): Remove declaration.
24090 (function_arg_advance_32): Initialize error_p as boolean variable.
24092 2017-05-05 Nathan Sidwell <nathan@acm.org>
24094 * store-motion.c (remove_reachable_equiv_notes): Reformat long
24095 lines. Use for (;;).
24097 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24099 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
24100 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
24101 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
24102 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
24103 VF=2 that require versioning.
24105 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24107 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
24110 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24112 * diagnostic.h (diagnostic_override_option_index): Convert from
24113 macro to inline function.
24115 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24117 * diagnostic.c (last_module_changed_p): New function.
24118 (set_last_module): New function.
24119 (diagnostic_report_current_module): Convert macro usage to
24120 the above functions.
24121 * diagnostic.h (diagnostic_context::last_module): Strengthen
24122 from const line_map * to const line_map_ordinary *.
24123 (diagnostic_last_module_changed): Delete macro.
24124 (diagnostic_set_last_module): Delete macro.
24126 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24128 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
24129 with diagnostic_report_diagnostic.
24130 (diagnostic_n_impl_richloc): Likewise.
24131 * diagnostic.h (report_diagnostic): Delete macro.
24132 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
24133 with diagnostic_report_diagnostic.
24134 * substring-locations.c (format_warning_va): Likewise.
24136 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24138 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
24139 save/restor of format_spec. Move option-printing code to...
24140 (print_option_information): ...this new function, and
24141 reimplement by simply printing to the pretty_printer,
24142 rather than appending to the format string.
24144 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24146 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
24147 handling logic into...
24148 (update_effective_level_from_pragmas): ...this new function.
24150 2017-05-04 Andrew Waterman <andrew@sifive.com>
24152 * config/riscv/riscv.opt (mstrict-align): New option.
24153 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
24154 (SLOW_UNALIGNED_ACCESS): Define.
24155 (riscv_slow_unaligned_access): Declare.
24156 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
24158 (riscv_slow_unaligned_access): New variable.
24159 (rocket_tune_info): Set slow_unaligned_access to true.
24160 (optimize_size_tune_info): Set slow_unaligned_access to false.
24161 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
24162 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
24163 (riscv_option_override): Set riscv_slow_unaligned_access.
24164 * doc/invoke.texi: Add -mstrict-align to RISC-V.
24166 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
24168 * config/riscv/riscv.md: Unify indentation.
24170 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
24175 * config/rs6000/rs6000.md (u code attribute): Add FIX and
24177 (extendsi<mode>2): Add support for doing sign extension via
24178 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
24179 don't have ISA 3.0 instructions.
24180 (extendsi<mode>2 splitter): Likewise.
24181 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
24182 generate the normal insns since SImode can now go in vector
24183 registers. Disallow the special UNSPECs needed for previous
24184 machines to hide SImode being used. Add new insns
24185 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
24186 (fix_trunc<mode>si2_stfiwx): Likewise.
24187 (fix_trunc<mode>si2_internal): Likewise.
24188 (fixuns_trunc<mode>si2): Likewise.
24189 (fixuns_trunc<mode>si2_stfiwx): Likewise.
24190 (fctiw<u>z_<mode>_smallint): Likewise.
24191 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
24192 of floating point to 32-bit integer from doing a direct move to
24193 the GPR registers to do a store.
24194 (fctiwz_<mode>): Break long line.
24196 2017-05-05 Bin Cheng <bin.cheng@arm.com>
24198 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
24199 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
24200 (addr_list, addr_offset_valid_p): New.
24201 (split_address_groups): Check offset validity with above function.
24202 (gt-tree-ssa-loop-ivopts.h): Include header file.
24204 2017-05-05 Nathan Sidwell <nathan@acm.org>
24206 * config.gcc (arm*-*-*): Add missing 'fi'.
24208 2017-05-05 Steve Ellcey <sellcey@cavium.com>
24210 * doc/invoke.texi (-fopt-info): Explicitly say order of options
24211 included in -fopt-info does not matter.
24212 * doc/optinfo.texi (-fopt-info): Fix description of default
24213 behavour. Explicitly say order of options included in -fopt-info
24216 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
24218 * config.gcc: Allow combinations of aprofile and rmprofile values for
24219 --with-multilib-list.
24220 * config/arm/t-multilib: New file.
24221 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
24222 variables. Remove setting of ISA and floating-point ABI in
24223 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
24224 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
24225 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
24226 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
24228 * config/arm/t-rmprofile: Likewise except for the matches changes.
24229 * doc/install.texi (--with-multilib-list): Document the combination of
24230 aprofile and rmprofile values and warn about pitfalls in doing that.
24232 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24234 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
24235 (movdi_aarch64): Likewise.
24237 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24239 PR tree-optimization/80632
24240 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
24242 (build_arrays): Initialize it for virtual phis.
24243 (fix_phi_nodes): Use it for virtual phis.
24245 PR tree-optimization/80558
24246 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
24247 [x, y] op z into [x op, y op z] for op & or | if conditions
24250 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
24251 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24254 * config/arm/arm.md (use_literal_pool): Remove.
24255 (64-bit immediate split): No longer takes cost into consideration
24256 if arm_disable_literal_pool is enabled.
24257 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
24258 used when arm_disable_literal_pool is enabled.
24259 (arm_max_const_double_inline_cost): Remove use of
24260 arm_disable_literal_pool.
24261 (push_minipool_fix): Add assert.
24262 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
24263 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
24264 (no_literal_pool_sf_immediate): New.
24266 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24268 PR tree-optimization/80613
24269 * tree-ssa-dce.c (propagate_necessity): Remove cases for
24270 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
24272 2017-05-05 Richard Biener <rguenther@suse.de>
24274 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
24276 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
24278 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
24279 of this flag from insn conditions due to removal from r247495.
24281 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24283 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
24285 (arm_early_store_addr_dep_ptr): Likewise.
24286 * config/arm/aarch-common-protos.h
24287 (arm_early_load_addr_dep_ptr): Add prototype.
24288 (arm_early_store_addr_dep_ptr): Likewise.
24289 * config/arm/cortex-a53.md: Add new bypasses.
24291 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24293 * tree.c (next_type_uid): Change type to unsigned.
24294 (type_hash_canon): Decrement back next_type_uid if
24295 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
24296 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
24299 2017-05-04 Martin Sebor <msebor@redhat.com>
24301 * builtins.c: Fix a trivial typo in a comment.
24303 PR middle-end/79234
24304 * builtins.c (check_sizes): Adjust to handle reading past the end.
24305 Avoid printing excessive upper bound of ranges. Use %E to print
24306 tree nodes instead of converting them to %wu.
24307 (expand_builtin_memchr): New function.
24308 (compute_dest_size): Rename...
24309 (compute_objsize): ...to this.
24310 (expand_builtin_memcpy): Adjust.
24311 (expand_builtin_mempcpy): Adjust.
24312 (expand_builtin_strcat): Adjust.
24313 (expand_builtin_strcpy): Adjust.
24314 (check_strncat_sizes): Adjust.
24315 (expand_builtin_strncat): Adjust.
24316 (expand_builtin_strncpy): Adjust and simplify.
24317 (expand_builtin_memset): Adjust.
24318 (expand_builtin_bzero): Adjust.
24319 (expand_builtin_memcmp): Adjust.
24320 (expand_builtin): Handle memcmp.
24321 (maybe_emit_chk_warning): Check strncat just once.
24323 2017-05-04 Martin Sebor <msebor@redhat.com>
24325 PR preprocessor/79214
24326 PR middle-end/79222
24327 PR middle-end/79223
24328 * builtins.c (check_sizes): Add inlining context and issue
24329 warnings even when -Wno-system-headers is set.
24330 (check_strncat_sizes): Same.
24331 (expand_builtin_strncat): Same.
24332 (expand_builtin_memmove): New function.
24333 (expand_builtin_stpncpy): Same.
24334 (expand_builtin): Handle memmove and stpncpy.
24336 2017-05-04 Bin Cheng <bin.cheng@arm.com>
24338 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
24339 which is not used any more.
24341 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24343 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
24345 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24347 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
24348 (cortexa53_tunings): Likewise.
24349 (cortexa57_tunings): Likewise.
24350 (cortexa72_tunings): Likewise.
24351 (cortexa73_tunings): Likewise.
24353 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24355 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
24356 Set loop alignment to 8.
24358 2017-05-04 Martin Sebor <msebor@redhat.com>
24360 PR translation/80280
24361 * builtins.c (expand_builtin_object_size): Add missing quoting to
24362 %D and like directives.
24363 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
24364 (hsa_type_for_tree_type): Same.
24365 (verify_function_arguments): Same.
24366 * symtab.c (symbol_table::change_decl_assembler_name): Same.
24367 * varasm.c (get_section): Same.
24370 2017-05-04 Martin Sebor <msebor@redhat.com>
24372 PR translation/80280
24373 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
24375 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24377 * config/aarch64/aarch64.c (generic_addrcost_table):
24378 Change HI/TI mode setting.
24380 2017-05-04 Martin Jambor <mjambor@suse.cz>
24382 PR tree-optimization/80622
24383 * tree-sra.c (comes_initialized_p): New function.
24384 (build_accesses_from_assign): Only set write lazily when
24385 comes_initialized_p is false.
24386 (analyze_access_subtree): Use comes_initialized_p.
24387 (propagate_subaccesses_across_link): Assert !comes_initialized_p
24388 instead of testing for PARM_DECL.
24390 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24392 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
24393 constraint on operand 0 to allow more general addressing modes.
24394 Adjust output template.
24395 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
24397 * config/aarch64/aarch64-protos.h
24398 (aarch64_address_valid_for_prefetch_p): Declare prototype.
24399 * config/aarch64/constraints.md (Dp): New address constraint.
24400 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
24403 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
24405 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
24406 update use of estimate_ipcp_clone_size_and_time.
24407 (estimate_local_effects): Update use of
24408 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
24409 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
24410 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
24411 Return nonspecialized time.
24413 2017-05-04 Richard Biener <rguenther@suse.de>
24415 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
24416 for the last VUSE which def dominates the PHI. Directly call
24418 (get_continuation_for_phi_1): Remove.
24420 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
24422 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
24423 to explain the use of truncating division. Cap the number of
24424 iterations to the maximum given by nb_iterations_upper_bound,
24427 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24429 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
24430 * configure: Regenerate.
24431 * config.in: Regenerate.
24432 * config/i386/driver-mingw32.c: new file.
24433 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
24434 * config.host: Link driver-mingw32.o on MinGW host.
24435 * doc/install.texi: Document new --enable-mingw-wildcard configure
24438 2017-05-04 Marek Polacek <polacek@redhat.com>
24440 PR tree-optimization/80612
24441 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
24443 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24444 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
24446 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
24447 (movt splitter): Likewise.
24448 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
24449 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
24450 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
24451 block for Thumb-1 with MOVT.
24452 (thumb2_legitimate_address_p): Move code block ...
24453 (can_avoid_literal_pool_for_label_p): ... into this new function.
24454 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
24456 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
24457 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
24458 "M-profile targets with the MOVT instruction".
24460 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24462 * config/arm/arm-builtins.c (arm_init_builtins): Rename
24463 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
24464 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
24466 2017-05-04 Martin Liska <mliska@suse.cz>
24468 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
24469 variable cond_code.
24471 2017-05-04 Richard Biener <rguenther@suse.de>
24473 * tree.c (array_at_struct_end_p): Handle arrays at struct
24474 end with flexarrays more conservatively. Refactor and treat
24475 arrays of arrays or aggregates more strict. Fix
24476 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
24477 * tree.c (array_at_struct_end_p): Adjust prototype.
24478 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
24479 * gimple-fold.c (get_range_strlen): Likewise.
24480 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
24482 2017-05-04 Richard Biener <rguenther@suse.de>
24484 PR tree-optimization/31130
24485 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
24487 (supports_overflow_infinity): Likewise.
24488 (is_negative_overflow_infinity): Likewise.
24489 (is_positive_overflow_infinity): Likewise.
24490 (is_overflow_infinity): Likewise.
24491 (stmt_overflow_infinity): Likewise.
24492 (overflow_infinity_range_p): Likewise.
24493 (usable_range_p): Remove as always returning true.
24494 (make_overflow_infinity): Remove.
24495 (negative_overflow_infinity): Likewise.
24496 (positive_overflow_infinity): Likewise.
24497 (avoid_overflow_infinity): Likewise.
24498 (set_value_range): Adjust accordingly.
24499 (set_value_range_to_nonnegative): Likewise, remove now unused
24500 overflow_infinity arg.
24501 (vrp_operand_equal_p): Adjust.
24502 (update_value_range): Likewise.
24503 (range_int_cst_singleton_p): Likewise.
24504 (operand_less_p): Likewise.
24505 (compare_values_warnv): Likewise.
24506 (extract_range_for_var_from_comparison_expr): Likewise.
24507 (vrp_int_const_binop): Likewise.
24508 (zero_nonzero_bits_from_vr): Likewise.
24509 (extract_range_from_multiplicative_op_1): Likewise.
24510 (extract_range_from_binary_expr_1): Likewise.
24511 (extract_range_from_unary_expr): Likewise.
24512 (extract_range_from_comparison): Likewise.
24513 (extract_range_basic): Likewise.
24514 (adjust_range_with_scev): Likewise.
24515 (compare_ranges): Likewise.
24516 (compare_range_with_value): Likewise.
24517 (dump_value_range): Likewise.
24518 (test_for_singularity): Likewise, remove strict_overflow_p parameter
24520 (simplify_cond_using_ranges): Adjust.
24522 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
24524 * brig-builtins.def: Added a builtin for class_f64.
24525 * builtin-types.def: Added a builtin type needed by class_f64.
24527 2017-05-03 Jason Merrill <jason@redhat.com>
24529 * timevar.def: Add TV_CONSTEXPR.
24531 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24533 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
24535 2017-05-03 Martin Jambor <mjambor@suse.cz>
24537 * ipa-prop.c (ipa_update_after_lto_read): Removed.
24538 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
24539 * ipa-cp.c (ipcp_propagate_stage): Do not call
24540 ipa_update_after_lto_read.
24541 * ipa-inline.c (ipa_inline): Likewise.
24543 2017-05-03 Martin Jambor <mjambor@suse.cz>
24545 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
24546 tag. Added a default constructor and a destructor.
24547 (ipa_edge_args_sum_t): New class;
24548 (ipa_edge_args_sum): Declare.
24549 (ipa_edge_args_vector): Remove declaration.
24550 (IPA_EDGE_REF): Use ipa_edge_args_sum.
24551 (ipa_free_edge_args_substructures): Remove declaration.
24552 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
24553 (ipa_edge_args_info_available_for_edge_p): Likewise.
24554 * ipa-prop.c (ipa_edge_args_vector): Removed.
24555 (edge_removal_hook_holder): Likewise.
24556 (edge_duplication_hook_holder): Likewise.
24557 (ipa_edge_args_sum): New variable.
24558 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
24559 ipa_edge_args_vector.
24560 (ipa_free_edge_args_substructures): Likewise.
24561 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
24562 ipa_edge_args_vector.
24563 (ipa_edge_removal_hook): Turned into method
24564 ipa_edge_args_sum_t::remove.
24565 (ipa_edge_duplication_hook): Turned into method
24566 ipa_edge_args_sum_t::duplicate.
24567 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
24568 registering edge hooks.
24569 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
24570 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
24571 ipa_edge_args_sum instead of ipa_edge_args_vector.
24572 * ipa-profile.c (ipa_profile): Likewise.
24574 2017-05-03 Martin Jambor <mjambor@suse.cz>
24576 * symbol-summary.h (function_summary): New method exists.
24577 (function_summary::symtab_removal): Deallocate through release.
24578 (call_summary): New class.
24579 (gt_ggc_mx): New overload.
24580 (gt_pch_nx): Likewise.
24581 (gt_pch_nx): Likewise.
24583 2017-05-03 Jeff Law <law@redhat.com>
24585 PR tree-optimization/78496
24586 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
24587 from simplify_cond_using_ranges. Split off code to walk
24588 backwards through casts into ...
24589 (simplify_cond_using_ranges_2): New function.
24590 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
24591 (execute_vrp): After identifying jump threads, call
24592 simplify_cond_using_ranges_2.
24594 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
24597 * ipa-inline.h (inline_summary): Add ctor.
24598 (create_ggc): Do not use ggc_cleared_alloc.
24600 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
24601 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24603 * gcc.c (handle_braces): Support escaping in switch matching
24605 * doc/invoke.texi (Spec Files): Document it.
24606 Remove superfluous @code markup in items.
24608 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24610 * diagnostic-show-locus.c (struct column_range): New struct.
24611 (get_affected_columns): New function.
24612 (get_printed_columns): New function.
24613 (struct correction): New struct.
24614 (correction::ensure_capacity): New function.
24615 (correction::ensure_terminated): New function.
24616 (struct line_corrections): New struct.
24617 (line_corrections::~line_corrections): New dtor.
24618 (line_corrections::add_hint): New function.
24619 (layout::print_trailing_fixits): Reimplement in terms of the new
24621 (selftest::test_overlapped_fixit_printing): New function.
24622 (selftest::diagnostic_show_locus_c_tests): Call it.
24624 2017-05-03 Nathan Sidwell <nathan@acm.org>
24626 Canonicalize canonical type hashing
24627 * tree.h (type_hash_canon_hash): Declare.
24628 * tree.c (type_hash_list, attribute_hash_list): Move into
24629 type_hash_canon_hash.
24630 (build_type_attribute_qual_variant): Break out hash code calc into
24631 type_hash_canon_hash.
24632 (type_hash_canon_hash): New. Generic type hash computation.
24633 (build_range_type_1, build_array_type_1, build_function_type,
24634 build_method_type_directly, build_offset_type, build_complex_type,
24635 make_vector_type): Call it.
24637 2017-05-03 Richard Biener <rguenther@suse.de>
24639 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
24640 When all DRs have unknown misaligned do not always peel
24641 when there is a store but apply the same costing model as if
24642 there were only loads.
24644 2017-05-03 Richard Biener <rguenther@suse.de>
24647 PR tree-optimization/80492
24648 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
24649 compare_base_decls returning dont-know properly.
24651 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
24653 * config/arm/iterators.md (CCSI): New mode iterator.
24654 (arch): New mode attribute.
24655 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
24656 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
24657 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
24658 code iterator for success result mode.
24659 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
24660 the corresponding new insn generators.
24662 2017-05-03 Bin Cheng <bin.cheng@arm.com>
24665 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24666 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24668 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
24670 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
24671 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
24672 (DDR_A): Wrap DDR argument in brackets.
24673 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
24674 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
24675 (DDR_REVERSED_P): Likewise.
24677 2017-05-03 Jakub Jelinek <jakub@redhat.com>
24679 PR tree-optimization/79472
24680 * tree-switch-conversion.c (struct switch_conv_info): Add
24681 contiguous_range and default_case_nonstandard fields.
24682 (collect_switch_conv_info): Compute contiguous_range and
24683 default_case_nonstandard fields, don't clear final_bb if
24684 contiguous_range and only the default case doesn't have the required
24686 (check_all_empty_except_final): Set default_case_nonstandard instead
24687 of failing if contiguous_range and the default case doesn't have empty
24689 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
24690 and only the default case doesn't have the required constants. Skip
24692 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
24693 if default_case_nonstandard.
24694 (build_constructors): Build constant 1 just once. Assert that default
24695 values aren't inserted in between cases if contiguous_range. Skip
24697 (build_arrays): Skip virtual phis.
24698 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
24699 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
24700 Handle virtual phis.
24701 (gen_inbound_check): Handle default_case_nonstandard case.
24702 (process_switch): Adjust check_final_bb caller. Call
24703 gather_default_values with the first non-default case instead of
24704 default case if default_case_nonstandard.
24706 2017-05-02 Nathan Sidwell <nathan@acm.org>
24708 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
24709 check. Fix formatting.
24711 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
24713 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
24714 errors when comparing specialized and unspecialized times.
24716 2017-05-02 David Malcolm <dmalcolm@redhat.com>
24718 * diagnostic-show-locus.c
24719 (layout::should_print_annotation_line_p): Make private.
24720 (layout::print_annotation_line): Make private.
24721 (layout::annotation_line_showed_range_p): Make private.
24722 (layout::show_ruler): Make private.
24723 (layout::print_source_line): Make private. Pass in line and
24724 line_width, rather than calling location_get_source_line. Drop
24726 (layout::print_leading_fixits): New method.
24727 (layout::print_any_fixits): Rename to...
24728 (layout::print_trailing_fixits): ...this, and make private.
24729 Don't print newline fixits.
24730 (diagnostic_show_locus): Move logic for printing one row into...
24731 (layout::print_line): ...this new function. Move the
24732 location_get_source_line call and error-handling from
24733 print_source_line to here. Call print_leading_fixits, and rename
24734 print_any_fixits to print_trailing_fixits.
24735 (selftest::test_fixit_insert_containing_newline): Update now that
24736 newlines are partially supported.
24737 (selftest::test_fixit_insert_containing_newline_2): New test.
24738 (selftest::test_fixit_replace_containing_newline): Update comments.
24739 (selftest::diagnostic_show_locus_c_tests): Call the new test.
24740 * edit-context.c (class added_line): New class.
24741 (class edited_line): Describe newline handling in comment.
24742 (edited_line::actually_edited_p): New method.
24743 (edited_line::print_content): Delete redundant decl.
24744 (edited_line::m_predecessors): New field.
24745 (edited_file::print_content): Call edited_line::print_content.
24746 (edited_file::print_diff): Update to support newlines.
24747 (edited_file::print_diff_hunk): Likewise.
24748 (edited_file::print_run_of_changed_lines): New function.
24749 (edited_file::print_diff_line): Convert to...
24750 (print_diff_line): ...this.
24751 (edited_file::get_effective_line_count): New function.
24752 (edited_line::edited_line): Initialize new field m_predecessors.
24753 (edited_line::~edited_line): Clean up m_predecessors.
24754 (edited_line::apply_fixit): Handle newlines.
24755 (edited_line::get_effective_line_count): New function.
24756 (edited_line::print_content): New function.
24757 (edited_line::print_diff_lines): New function.
24758 (selftest::test_applying_fixits_insert_containing_newline): New
24760 (selftest::test_applying_fixits_replace_containing_newline): New
24762 (selftest::insert_line): New function.
24763 (selftest::test_applying_fixits_multiple_lines): Add example of
24765 (selftest::edit_context_c_tests): Call the new tests.
24767 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24769 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
24770 parameter cand. Update dump information.
24771 (get_computation_cost): Update uses.
24773 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24775 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
24776 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
24777 (get_computation_at, rewrite_use_address): Update use of
24778 get_computation_aff.
24780 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24782 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
24783 (get_computation): Delete.
24784 (get_computation_cost): Implement like get_computation_cost_at.
24785 Use get_computation_at.
24786 (get_computation_cost_at): Delete.
24787 (rewrite_use_nonlinear_expr): Use get_computation_at.
24788 (rewrite_use_compare, remove_unused_ivs): Ditto.
24790 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24792 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
24794 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24796 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
24797 (ivopts_global_cost_for_size): Rename parameter and update uses.
24798 (iv_ca_recount_cost): Update uses.
24799 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
24800 candidates seperately in n_invs and n_cands.
24801 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
24803 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24805 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
24806 (find_inv_vars_cb): New.
24807 (find_depends): Renamed to ...
24808 (find_inv_vars): ... this.
24809 (add_candidate_1, force_var_cost): Call find_inv_vars.
24810 (split_address_cost, determine_group_iv_cost_cond): Ditto.
24812 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24814 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
24815 inv_vars. Add inv_exprs.
24816 (struct iv_cand): Rename depends_on to inv_vars.
24817 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
24818 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
24819 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
24820 (dump_cand): Dump inv_vars.
24821 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
24822 (record_invariant, find_depends, add_candidate_1): Ditto.
24823 (set_group_iv_cost, force_var_cost): Ditto.
24824 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
24825 (get_computation_cost_at, get_computation_cost): Ditto.
24826 (determine_group_iv_cost_generic): Ditto.
24827 (determine_group_iv_cost_address): Ditto.
24828 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
24829 (determine_group_iv_costs): Ditto.
24830 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
24831 (iv_ca_set_remove_invariants): Renamed to ...
24832 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
24833 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
24834 (iv_ca_set_add_invariants): Renamed to ...
24835 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
24836 (iv_ca_set_cp): Use iv_ca_set_add_invs.
24837 (iv_ca_has_deps): Support inv_vars and inv_exprs.
24838 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
24839 (create_new_ivs): Remove useless dump.
24841 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24843 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
24845 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
24846 (iv_ca_set_no_cp, create_new_iv): Ditto.
24848 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24850 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24852 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
24854 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
24855 function tree_check2.
24857 2017-05-02 Martin Liska <mliska@suse.cz>
24859 * doc/gcov.texi: Add missing preposition.
24860 * gcov.c (function_info::function_info): Properly fill up
24861 all member variables.
24863 2017-05-02 Tamar Christina <tamar.christina@arm.com>
24865 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
24867 2017-05-02 Tamar Christina <tamar.christina@arm.com>
24869 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
24871 2017-05-02 Martin Liska <mliska@suse.cz>
24874 * lto-streamer-in.c (lto_read_tree_1): Remove
24875 LTO_STREAMER_DEBUG.
24876 * lto-streamer.c (struct tree_hash_entry): Likewise.
24877 (struct tree_entry_hasher): Likewise.
24878 (tree_entry_hasher::hash): Likewise.
24879 (tree_entry_hasher::equal): Likewise.
24880 (lto_streamer_init): Likewise.
24881 (lto_orig_address_map): Likewise.
24882 (lto_orig_address_get): Likewise.
24883 (lto_orig_address_remove): Likewise.
24884 * lto-streamer.h: Likewise.
24885 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
24886 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
24888 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
24890 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
24891 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
24892 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
24893 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
24894 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
24895 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
24896 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
24897 (mm_maskz_sub_ss): New intrinsics.
24898 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24899 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
24900 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
24901 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
24902 (__builtin_ia32_subss_mask_round): New builtins.
24903 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24904 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24905 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
24907 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
24908 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
24910 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
24913 2017-05-02 Martin Jambor <mjambor@suse.cz>
24915 PR tree-optimization/78687
24916 * tree-sra.c (access): New field parent.
24917 (process_subtree_disqualification): New function.
24918 (disqualify_candidate): Call it.
24919 (build_accesses_from_assign): Reset write flag if creating an
24921 (build_access_subtree): Fill in parent field and also prpagate
24922 down grp_write flag.
24923 (create_artificial_child_access): New parameter set_grp_write, set
24924 grp_write to its value.
24925 (propagate_subaccesses_across_link): Also propagate grp_write flag
24927 (propagate_all_subaccesses): Push the closest parent back to work
24928 queue if add_access_to_work_queue returned true.
24930 2017-05-02 Richard Biener <rguenther@suse.de>
24932 * common.opt (fstrict-overflow): Alias negative to fwrapv.
24933 * doc/invoke.texi (fstrict-overflow): Remove all traces of
24934 -fstrict-overflow documentation.
24935 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
24936 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
24937 flag_strict_overflow.
24938 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
24939 * lto-opts.c (lto_write_options): Do not stream it.
24940 * lto-wrapper.c (merge_and_complain): Do not handle it.
24941 * opts.c (default_options_table): Do not set -fstrict-overflow.
24942 (finish_options): Likewise do not clear it when sanitizing.
24943 * simplify-rtx.c (simplify_const_relational_operation): Do not
24944 test flag_strict_overflow.
24946 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
24948 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
24949 using enabled attribute.
24950 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
24951 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
24952 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
24953 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
24954 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
24955 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
24956 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
24957 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
24958 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
24959 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
24961 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
24963 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
24965 2017-05-02 Richard Biener <rguenther@suse.de>
24967 PR tree-optimization/80591
24969 2017-04-10 Richard Biener <rguenther@suse.de>
24971 * tree-ssa-structalias.c (find_func_aliases): Properly handle
24974 2017-05-02 Richard Biener <rguenther@suse.de>
24976 PR tree-optimization/80549
24977 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
24978 (cleanup_tree_cfg_noloop): Create forwarders to known loop
24979 headers if they do not have a preheader.
24981 2017-05-02 Martin Liska <mliska@suse.cz>
24984 * common.opt: Fix typo.
24985 * doc/invoke.texi: Likewise.
24987 2017-05-01 Jan Beulich <jbeulich@suse.com>
24989 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
24990 swapping, add (x,x,m,x,n) alternative.
24992 2017-05-01 Nathan Sidwell <nathan@acm.org>
24994 * calls.c (combine_pending_stack_adjustment_and_call): Remove
24995 unnecessary unadjusted_alignment check.
24997 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
25000 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
25002 * gimplify.c (gimplify_cilk_detach): New function.
25003 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
25004 * tree-core.h: Document EXPR_CILK_SPAWN.
25005 * tree.h (EXPR_CILK_SPAWN): Define.
25007 2017-05-01 David Malcolm <dmalcolm@redhat.com>
25009 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
25010 to use new fixit_hint representation, using the "replace" logic.
25011 (get_line_span_for_fixit_hint): Likewise.
25012 (layout::print_any_fixits): Likewise.
25013 (selftest::test_one_liner_many_fixits): Rename to...
25014 (selftest::test_one_liner_many_fixits_1): ...this, and update
25015 comment and expected output to reflect that the multiple fix-it
25016 hints are now consolidated into one insertion.
25017 (selftest::test_one_liner_many_fixits_2): New test.
25018 (selftest::test_diagnostic_show_locus_one_liner): Update for
25020 (selftest::test_fixit_consolidation): Update for fix-it API
25022 * diagnostic.c (print_parseable_fixits): Likewise.
25023 * edit-context.c (edited_line::m_line_events): Convert from
25024 auto_vec <line_event *> to auto_vec <line_event>.
25025 (class line_event): Convert from abstract base class to a concrete
25026 class, taking over the role of replace_event.
25027 (class insert_event): Delete.
25028 (class replace_event): Rename to class line_event. Convert to
25030 (edit_context::add_fixits): Reimplement.
25031 (edit_context::apply_insert): Delete.
25032 (edit_context::apply_replace): Rename to...
25033 (edit_context::apply_fixit): ...this. Convert to half-open range.
25034 (edited_file::apply_insert): Delete.
25035 (edited_file::apply_replace): Rename to...
25036 (edited_file::apply_fixit): ...this.
25037 (edited_line::~edited_line): Drop deletion of events.
25038 (edited_line::apply_insert): Delete.
25039 (edited_line::apply_replace): Rename to...
25040 (edited_line::apply_fixit): ...this. Convert to half-open range.
25041 Update for change to type of m_line_events.
25042 * edit-context.h (edit_context::apply_insert): Delete.
25043 (edit_context::apply_replace): Rename to...
25044 (edit_context::apply_fixit): ...this.
25046 2017-05-01 Martin Sebor <msebor@redhat.com>
25048 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
25051 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
25054 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
25055 __get_cpuid_max returns 0.
25056 (__get_cpuid_count): Ditto.
25058 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
25060 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
25061 replacement expression is another instance of one of its arguments.
25063 2017-05-01 Jakub Jelinek <jakub@redhat.com>
25066 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
25067 check for stack push/pop autoinc.
25068 * config/i386/i386.c (ix86_agi_dependent): Return false
25069 if the only reason why modified_in_p returned true is that
25070 addr is SP based and set_insn is a push or pop.
25072 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25074 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
25077 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25080 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
25081 (account_size_time): Use two predicates - exec_pred and
25083 (evaluate_conditions_for_known_args): Compute both clause and
25085 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
25086 (inline_summary_t::duplicate): Update.
25087 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
25089 (compute_inline_parameters): Likewise.
25090 (estimate_edge_size_and_time): Update caluclation of time.
25091 (estimate_node_size_and_time): Compute both time and nonspecialized
25093 (estimate_ipcp_clone_size_and_time): Update.
25094 (inline_merge_summary): Update.
25095 (do_estimate_edge_time): Update.
25096 (do_estimate_edge_size): Update.
25097 (do_estimate_edge_hints): Update.
25098 (inline_read_section, inline_write_summary): Stream both new predicates.
25099 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
25101 (compute_inlined_call_time): Cleanup.
25102 (big_speedup_p): Update.
25103 (edge_badness): Update.
25104 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
25105 (size_time_entry): Replace predicate by exec_predicate and
25106 nonconst_predicate.
25107 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
25108 (estimate_edge_time): Return also nonspec_time.
25109 (reset_edge_growth_cache): Update.
25111 2017-04-29 Jakub Jelinek <jakub@redhat.com>
25113 PR rtl-optimization/80491
25114 * ifcvt.c (noce_process_if_block): When looking for x setter
25115 with missing else_bb, don't check only the insn right before
25116 cond_earliest, but look for the last insn that x is modified in
25117 within the same bb.
25119 PR rtl-optimization/80491
25120 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
25122 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
25124 PR tree-optimization/80487
25125 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
25127 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25129 PR tree-optimization/79697
25130 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
25131 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
25132 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
25134 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
25135 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
25137 2017-04-28 Martin Sebor <msebor@redhat.com>
25139 PR tree-optimization/80523
25140 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
25141 (init_target_to_host_charmap, target_to_host, target_strtol10): New
25143 (maybe_warn, format_directive, parse_directive): Use new functions.
25144 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
25146 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
25148 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
25150 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
25152 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
25153 target_header_dir): Set correctly.
25154 * configure: Regenerated.
25155 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25156 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25157 instead of SYSTEM_HEADER_DIR.
25159 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
25161 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
25162 (estimate_local_effects): Likewise.
25163 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
25164 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
25165 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
25166 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
25167 do_estimate_edge_time, estimate_edge_time): Likewise.
25168 * ipa-inline-analysis.c (estimate_node_size_and_time,
25169 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
25170 (estimate_time_after_inlining): Remove.
25172 2017-04-28 Martin Liska <mliska@suse.cz>
25174 * doc/gcov.texi: Enhance documentation of gcov.
25176 2017-04-28 Martin Liska <mliska@suse.cz>
25178 * doc/gcov.texi: Sort options in alphabetic order.
25179 * doc/gcov-dump.texi: Likewise.
25180 * doc/gcov-tool.texi: Likewise.
25181 * gcov.c (print_usage): Likewise.
25182 * gcov-dump.c (print_usage): Likewise.
25183 * gcov-tool.c (print_merge_usage_message): Likewise.
25184 (print_rewrite_usage_message): Likewise.
25185 (print_overlap_usage_message): Likewise.
25187 2017-04-28 Martin Liska <mliska@suse.cz>
25189 PR gcov-profile/53915
25190 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
25192 2017-04-28 Martin Liska <mliska@suse.cz>
25194 PR gcov-profile/79891
25195 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
25196 is marked by compiler as living on a line.
25197 (get_cycles_count): Remove usage of the union.
25198 (output_intermediate_file): Likewise.
25199 (find_source): Fix GNU coding style.
25200 (accumulate_line_counts): Remove old non-all block mode.
25201 (output_lines): Remove usage of the union.
25202 * profile.c (output_location): Include all BBs, even if
25203 belonging to a same line (and file) as a previous BB.
25205 2017-04-28 Martin Liska <mliska@suse.cz>
25207 * gcov.c (process_args): Handle new argument 'w'.
25208 (read_graph_file): Assign ID to BBs.
25209 (output_branch_count): Display BB # if verbose flag is set.
25210 (output_lines): Likewise for arcs.
25211 (print_usage): Add '--verbose' option help.
25212 * doc/gcov.texi: Document --verbose (-w) option.
25214 2017-04-28 Martin Liska <mliska@suse.cz>
25216 * gcov.c (struct block_location_info): New struct.
25217 (process_file): Fill up the new structure.
25218 (read_graph_file): Replace usage of encoding by the newly added
25220 (add_line_counts): Likewise.
25221 (accumulate_line_counts): Remove usage of the union.
25222 (function_info::function_info): New function.
25223 (function_info::~function_info): Likewise.
25224 (process_file): Call delete instead of release_function.
25225 (release_function): Release the function.
25226 (release_structures): Call delete instead of release_function.
25227 (solve_flow_graph): Replace usage of num_blocks.
25228 (find_exception_blocks): Likewise.
25229 (output_lines): Fix GNU coding style.
25231 2017-04-28 Martin Liska <mliska@suse.cz>
25234 * coverage.c (coverage_remove_note_file): New function.
25235 * coverage.h: Declare the function.
25236 * toplev.c (finalize): Clean if an error has been seen.
25238 2017-04-28 Martin Liska <mliska@suse.cz>
25240 PR gcov-profile/80031
25241 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
25242 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
25243 * gcov.c (read_graph_file): Read just number of blocks.
25244 * profile.c (branch_prob): Do not stream 0 flags per a basic
25247 2017-04-28 Martin Liska <mliska@suse.cz>
25249 * gcov-dump.c (tag_*): Add new argument to declarations.
25250 (dump_gcov_file): Likewise.
25251 (tag_blocks): Add and use new argument depth.
25252 (tag_arcs): Likewise.
25253 (tag_lines): Likewise.
25254 (tag_counters): Likewise.
25255 (tag_summary): Likewise.
25256 (dump_working_sets): Use depth to do a proper indentation.
25258 2017-04-28 Jakub Jelinek <jakub@redhat.com>
25261 * cgraph.h (symtab_node::debug_symtab): No longer inline.
25262 * symtab.c (symtab_node::debug_symtab): Move definition here.
25264 2017-04-28 Richard Biener <rguenther@suse.de>
25266 * lto-streamer.h (LTO_major_version): Bump to 7.
25268 2017-04-28 Richard Biener <rguenther@suse.de>
25270 * tree-vrp.c (assert_info): New struct.
25271 (add_assert_info): New helper.
25272 (register_edge_assert_for_2): Refactor to add asserts to a vector
25274 (register_edge_assert_for_1): Likewise.
25275 (register_edge_assert_for): Likewise.
25276 (finish_register_edge_assert_for): New helper actually registering
25277 asserts where live on edge.
25278 (find_conditional_asserts): Adjust.
25279 (find_switch_asserts): Likewise.
25280 (evrp_dom_walker::try_find_new_range): Generalize.
25281 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
25283 2017-04-27 Marek Polacek <polacek@redhat.com>
25286 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
25287 arg10 and arg11 to itype.
25289 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
25291 * doc/extend.texi (Object Size Checking): Improve grammar.
25293 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
25296 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
25297 that the logic for permitting reciprocal estimates matches that
25300 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25303 * tree.c (type_cache_hasher::equal): Only compare
25304 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
25305 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
25306 non-aggregate element types.
25307 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
25308 about the flag on ARRAY_TYPEs in the comment, formatting fix.
25310 2017-04-27 Richard Biener <rguenther@suse.de>
25312 PR middle-end/80533
25313 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
25314 stripping ARRAY_REFs from MEM_EXPR make sure we're not
25315 keeping a reference to a trailing array.
25317 2017-04-27 Richard Biener <rguenther@suse.de>
25319 PR middle-end/80539
25320 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
25321 being in loop-closed SSA form conservatively.
25322 (chrec_fold_multiply_poly_poly): Likewise.
25324 2017-04-27 Tamar Christina <tamar.christina@arm.com>
25326 PR middle-end/79665
25327 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
25328 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
25330 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25333 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
25334 (aarch64_function_arg_alignment): Return unsigned int again, but still
25335 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
25336 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
25337 Don't emit -Wpsabi note.
25338 (aarch64_function_arg_boundary): Likewise.
25339 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
25342 2017-04-26 Nathan Sidwell <nathan@acm.org>
25344 * tree.h (crc32_unsigned_n): Declare.
25345 (crc32_unsigned, crc32_unsigned): Make inline.
25346 * tree.c (crc32_unsigned_bits): Replace with ...
25347 (crc32_unsigned_n): ... this.
25348 (crc32_unsigned, crc32_byte): Remove.
25349 (crc32_string): Remove unnecessary braces.
25351 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25353 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
25354 * ipa-inline-analysis.c (MAX_TIME): Remove.
25355 (account_size_time): Use sreal for time.
25356 (dump_inline_summary): Update.
25357 (estimate_function_body_sizes): Update.
25358 (estimate_edge_size_and_time): Update.
25359 (estimate_calls_size_and_time): Update.
25360 (estimate_node_size_and_time): Update.
25361 (inline_merge_summary): Update.
25362 (inline_update_overall_summary): Update.
25363 (estimate_time_after_inlining): Update.
25364 (inline_read_section): Update.
25365 (inline_write_summary): Update.
25366 * ipa-inline.c (compute_uninlined_call_time): Update.
25367 (compute_inlined_call_time): Update.
25368 (recursive_inlining): Update.
25369 (inline_small_functions): Update.
25370 (dump_overall_stats): Update.
25371 * ipa-inline.h: Include sreal.h.
25372 (size_time_entry): Turn time to sreal.
25373 (inline_summary): Turn self_time nad time to sreal.
25375 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25377 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
25379 (sreal::stream_out, sreal::stream_in): New.
25380 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
25382 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25384 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
25387 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
25390 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
25392 (dimode_scalar_chain::compute_convert_gain): Ditto.
25393 (dimode_scalar_chain::make_vector_copies): Ditto.
25394 (dimode_scalar_chain::convert_reg): Ditto.
25395 (dimode_scalar_chain::convert_insn): Ditto.
25396 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
25397 (VI248_AVX512BW_1): New mode iterator.
25398 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
25399 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
25402 2017-04-25 Martin Sebor <msebor@redhat.com>
25404 PR tree-optimization/80497
25405 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
25406 constants are representable in HOST_WIDE_INT.
25407 (parse_directive): Ditto.
25409 2017-04-25 Martin Sebor <msebor@redhat.com>
25412 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
25413 (new_zero_array): Adjust signature.
25414 (dom_info::dom_init): Used unsigned rather that size_t.
25415 (dom_info::dom_info): Same.
25417 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25418 Jakub Jelinek <jakub@redhat.com>
25421 * config/arm/arm.c: Include gimple.h.
25422 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
25423 returns negative, increment ncrn only if it returned positive.
25424 (arm_needs_doubleword_align): Return int instead of bool,
25425 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
25426 members, but if there is any such non-FIELD_DECL
25427 > PARM_BOUNDARY aligned decl, return -1 instead of false.
25428 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
25429 returns negative, increment nregs only if it returned positive.
25430 (arm_setup_incoming_varargs): Likewise.
25431 (arm_function_arg_boundary): Emit -Wpsabi note if
25432 arm_needs_doubleword_align returns negative, return
25433 DOUBLEWORD_ALIGNMENT only if it returned positive.
25435 2017-04-25 Marek Polacek <polacek@redhat.com>
25438 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
25439 first argument to type.
25441 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
25444 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
25445 type checks to test for compatibility instead of equality.
25447 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25448 Jakub Jelinek <jakub@redhat.com>
25451 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
25453 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
25454 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
25455 the alignment computation, but return their maximum in warn_alignment.
25456 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
25457 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
25459 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
25460 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
25463 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25465 * config/arc/simdext.md (dmpyh): Fix typo.
25467 2017-04-25 Richard Biener <rguenther@suse.de>
25469 PR tree-optimization/80492
25470 * alias.c (compare_base_decls): Handle registers with asm
25471 specification conservatively.
25472 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
25473 compare_base_decls returning dont-know properly.
25475 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25477 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
25478 (legitimate_offset_address_p): New function.
25479 (arc_legitimate_address_p): Use above function.
25481 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25483 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
25485 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25487 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
25488 ACCH registers whenever they are available.
25490 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25492 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
25493 double regs fix when not used.
25495 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25497 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
25499 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
25500 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
25502 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25504 * config/arc/arc.c (arc_output_addsi): Check for h-register class
25505 when emitting short ADD instructions.
25507 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25509 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
25511 (cmpsi_cc_c_insn): Likewise.
25512 (cbranchsi4_scratch): Compute proper instruction length using
25513 compact_hreg_operand.
25514 * config/arc/predicates.md (compact_hreg_operand): New predicate.
25516 2017-04-25 Richard Biener <rguenther@suse.de>
25518 PR middle-end/80509
25519 * passes.c (pass_manager::pass_manager): Initialize
25520 m_name_to_pass_map.
25522 2017-04-25 Richard Biener <rguenther@suse.de>
25524 PR tree-optimization/79201
25525 * tree-ssa-sink.c (statement_sink_location): Handle calls.
25527 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25530 * config/s390/vector.md: Split MEM->GPR vector moves for
25531 non-s_operand addresses.
25533 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25536 * config/s390/predicates.md (reload_const_wide_int_operand): New
25538 * config/s390/s390.md ("movti"): Remove d/P alternative.
25539 ("movti_bigconst"): New pattern definition.
25541 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25544 * s390-protos.h (s390_expand_cs_hqi): Removed.
25545 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
25546 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
25547 modes as well as CCZ1mode and CCZmode.
25548 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
25549 signature of s390_emit_compare_and_swap.
25550 (s390_expand_cs_hqi): Likewise, make static.
25551 (s390_expand_cs_tdsi): Generate an explicit compare before trying
25552 compare-and-swap, in some cases.
25553 (s390_expand_cs): Wrapper function.
25554 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
25556 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
25557 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
25558 patterns for small and large integers. Forbid symref memory operands.
25559 Move expander to s390.c. Require cc register.
25560 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
25561 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
25562 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
25563 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
25564 symref memory operands. Remove CC mode and call s390_match_ccmode
25566 ("atomic_exchange<mode>"): Allow and implement all integer modes.
25568 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25570 * config/s390/s390.md (define_peephole2): New peephole to help
25571 combining the load-and-test pattern with volatile memory.
25573 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25575 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
25576 with CCZmode for TARGET_Z196.
25578 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25580 PR rtl-optimization/80501
25581 * combine.c (make_compound_operation_int): Set subreg_code to SET
25582 even for AND with mask of the sign bit of mode.
25584 PR rtl-optimization/80500
25585 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
25586 sum's initial value.
25588 2017-04-25 Julian Brown <julian@codesourcery.com>
25589 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25591 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
25593 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
25595 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
25597 2017-04-25 Julian Brown <julian@codesourcery.com>
25598 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25600 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
25601 (thunderx2t99_sha): New Reservation.
25603 2017-04-25 Julian Brown <julian@codesourcery.com>
25604 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25606 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
25607 type for 1-element load.
25609 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
25611 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
25613 2017-04-24 Martin Jambor <mjambor@suse.cz>
25615 PR tree-optimization/80293
25616 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
25617 char arrays not totally scalarizable if it is false.
25618 (analyze_all_variable_accesses): Pass correct value in the new
25619 parameter. Add a statistics counter.
25621 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
25623 PR middle-end/79931
25624 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
25626 2017-04-24 Richard Biener <rguenther@suse.de>
25628 PR tree-optimization/80494
25629 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
25630 out for complex types.
25632 2017-04-24 Richard Biener <rguenther@suse.de>
25634 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
25635 * tree-ssa-sccvn.c (print_scc): Print SCC size.
25636 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
25637 (DFS): Adjust and never fail.
25638 (sccvn_dom_walker::fail): Remove.
25639 (sccvn_dom_walker::before_dom_children): Adjust.
25640 (run_scc_vn): Likewise and never fail.
25641 * tree-ssa-pre.c (pass_pre::execute): Adjust.
25642 (pass_fre::execute): Likewise.
25644 2017-04-24 Richard Biener <rguenther@suse.de>
25646 PR tree-optimization/79725
25647 * tree-ssa-sink.c (statement_sink_location): Return whether
25648 failure reason was zero uses. Move that check later.
25649 (sink_code_in_bb): Deal with zero uses by removing the stmt
25652 2017-04-24 Richard Biener <rguenther@suse.de>
25655 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
25656 pointer-based references.
25658 2017-04-24 Richard Biener <rguenther@suse.de>
25661 * pass_manager.h (pass_manager::operator new): Remove.
25662 (pass_manager::operator delete): Likewise.
25663 * passes.c (pass_manager::operator new): Remove.
25664 (pass_manager::operator delete): Likewise.
25665 (pass_manager::pass_manager): Zero individual pass members.
25667 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
25670 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
25671 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
25672 Check "XEXP (src, 1)" operand here.
25673 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
25674 Check "XEXP (src, 1)" operand here.
25675 (dimode_scalar_chain::make_vector_copies): Detect count register
25676 of a shift instruction. Zero extend count register from QImode
25677 to DImode to satisfy vector shift pattern count operand predicate.
25678 Substitute vector shift count operand with a DImode copy.
25679 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
25682 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
25684 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
25685 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
25686 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
25687 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
25688 (UNSPEC_NOREX_MEM): Remove definition.
25690 2017-04-21 Richard Biener <rguenther@suse.de>
25692 PR tree-optimization/79547
25693 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
25694 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
25695 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
25696 without any constraints.
25698 2017-04-21 Richard Biener <rguenther@suse.de>
25700 PR tree-optimization/78847
25701 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
25703 2017-04-21 Richard Biener <rguenther@suse.de>
25705 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
25706 (build_distinct_type_copy): Likewise.
25707 (build_variant_type_copy): Likewise.
25708 * tree.c (build_qualified_type): Pass down mem-stat info.
25709 (build_distinct_type_copy): Likewise.
25710 (build_variant_type_copy): Likewise.
25712 2017-04-21 Richard Biener <rguenther@suse.de>
25714 PR tree-optimization/80237
25715 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
25717 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
25718 for a simplified result.
25720 2016-04-21 Richard Biener <rguenther@suse.de>
25722 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
25723 sth as strict as a simple_iv but a chrec without symbols and an
25724 operand defined in the loop we are peeling (and not some subloop).
25725 (propagate_constants_for_unrolling): Propagate all constants.
25727 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25730 * config/i386/i386.c (print_reg): Remove assert for disalowed
25731 regno values, call output_operand_lossage instead.
25733 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25736 * config/i386/constraints.md (Yc): New register constraint.
25737 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
25738 Use Yc constraint for alternative 2 of operand 0. Remove
25739 preferred_for_speed attribute.
25741 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
25743 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
25744 lastprivate clauses in SIMT case.
25746 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
25748 * doc/invoke.texi (-Wextra-semi): Document new warning option.
25750 2017-04-20 Richard Biener <rguenther@suse.de>
25752 PR tree-optimization/57796
25753 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
25754 as N scalar stores.
25755 (vect_model_load_cost): Cost gathers as N scalar loads.
25757 2017-04-20 Richard Biener <rguenther@suse.de>
25759 * ggc-page.c (ggc_allocated_p): Rename to ...
25760 (safe_lookup_page_table_entry): ... this and return the lookup
25762 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
25764 2017-04-20 Richard Biener <rguenther@suse.de>
25766 PR tree-optimization/80453
25767 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
25768 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
25769 from the conditions.
25770 (vn_phi_eq): Pass them down.
25771 (vn_phi_lookup): Record them.
25772 (vn_phi_insert): Likewise.
25774 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
25776 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
25777 uninitialized variable warning to avoid buffer overrun.
25779 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
25782 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
25783 is suppressed for '{ 0 }' in C.
25785 2017-04-20 Jakub Jelinek <jakub@redhat.com>
25787 * BASE-VER: Set to 8.0.0.
25789 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
25791 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
25792 priority .init_array and .fini_array section with SECTION_NOTYPE
25795 2017-04-20 Jakub Jelinek <jakub@redhat.com>
25797 PR middle-end/80423
25798 * tree.h (build_array_type): Add typeless_storage default argument.
25799 * tree.c (type_cache_hasher::equal): Also compare
25800 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
25801 (build_array_type): Add typeless_storage argument, set
25802 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
25804 (build_nonshared_array_type): Adjust build_array_type_1 caller.
25805 (build_array_type): Likewise. Add typeless_storage argument.
25807 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
25808 Jakub Jelinek <jakub@redhat.com>
25810 PR tree-optimization/80426
25811 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
25812 operation on symbolic operands, also compute the overflow for the
25813 invariant part when the operation degenerates into a negation.
25815 2017-04-19 Jakub Jelinek <jakub@redhat.com>
25818 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
25819 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
25822 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
25824 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
25827 * config/avr/avr.c (tree.h): Include it.
25828 (cgraph.h): Include it.
25829 (avr_encode_section_info): Don't warn for uninitialized progmem
25830 variable if it's just an alias.
25832 2017-04-19 Richard Biener <rguenther@suse.de>
25835 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
25836 when needed by AutoPGO.
25838 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
25841 * doc/lto.texi: Remove an extra 'that'.
25843 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
25845 PR rtl-optimization/80429
25846 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
25847 are only used in debug insns.
25849 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
25850 Vladimir Makarov <vmakarov@redhat.com>
25852 * config/sparc/predicates.md (input_operand): Add comment. Return
25853 true for any memory operand when LRA is in progress.
25854 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
25856 2017-04-18 Jeff Law <law@redhat.com>
25859 * mips.md ({return,simple_return}_internal): Do not overwrite
25862 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25864 PR tree-optimization/80443
25865 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
25866 instead of adding 1, subtract -1 and similarly instead of subtracting
25869 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
25871 PR rtl-optimization/80357
25872 * haifa-sched.c (tmp_bitmap): New variable.
25873 (model_recompute): Handle duplicate use records.
25874 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
25875 (free_global_sched_pressure_data): Free it.
25877 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
25880 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
25881 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25882 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25883 instead of SYSTEM_HEADER_DIR.
25885 2017-04-18 Jeff Law <law@redhat.com>
25887 PR middle-end/80422
25888 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
25889 predecessors after walking up the insn chain.
25891 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25894 * dwarf2out.c (modified_type_die): Try harder not to emit internal
25895 sizetype type into debug info.
25897 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
25900 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
25901 unneeded test for TARGET_UPPER_REGS_SF.
25902 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
25904 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25907 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
25908 instead of gsi_after_labels.
25910 2017-04-18 Jeff Law <law@redhat.com>
25912 * regcprop.c (maybe_mode_change): Avoid creating copies of the
25916 2017-04-13 Jeff Law <law@redhat.com>
25917 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
25918 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
25920 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
25923 * config/avr/avr.c (intl.h): Include it.
25924 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
25926 2017-04-18 Martin Liska <mliska@suse.cz>
25928 PR gcov-profile/78783
25929 * gcov-tool.c (gcov_output_files): Validate that destination
25930 file is either removed by the tool or by a user.
25932 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
25933 Guy Benyei <guybe@mellanox.com>
25935 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
25936 block, and do not negate it, the stored id is already negative.
25938 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
25940 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
25942 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
25945 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
25946 masks of options that should be turned off if the VSX vector
25947 options are turned off.
25948 (OTHER_P8_VECTOR_MASKS): Likewise.
25949 (OTHER_VSX_VECTOR_MASKS): Likewise.
25950 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
25951 rs6000_disable_incompatible_switches to validate no type switches
25953 (rs6000_incompatible_switch): New function to disallow turning on
25954 other vector options if -mno-vsx, -mno-power8-vector, or
25955 -mno-power9-vector are specified.
25957 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
25959 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
25961 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
25963 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
25964 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
25965 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
25966 (ARG_POINTER_CFA_OFFSET): Likewise.
25968 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
25970 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
25971 conditions to take advantage of various optimizations.
25973 2017-04-13 Jeff Law <law@redhat.com>
25975 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
25976 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
25977 (zero_extendsidi2_dext): Likewise.
25979 2017-04-13 Jakub Jelinek <jakub@redhat.com>
25982 * fold-const.c (fold_ternary_loc): Revert
25983 use op0 instead of fold_convert_loc (loc, type, arg0) part of
25986 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
25988 PR rtl-optimization/80343
25989 * lra-remat.c (update_scratch_ops): Assign original hard reg to
25990 new scratch pseudo.
25992 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
25995 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
25996 to ubsan_encode_value.
25998 2017-04-13 Jeff Law <law@redhat.com>
26000 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
26001 appearing in DEBUG_INSNs.
26003 2017-04-13 Martin Liska <mliska@suse.cz>
26005 PR gcov-profile/80413
26006 * gcov-io.c (gcov_write_string): Copy to buffer just when
26007 allocated size is greater than zero.
26009 2017-04-13 Jakub Jelinek <jakub@redhat.com>
26012 * dwarf2out.c (decls_for_scope): Ignore declarations of
26013 current_function_decl in BLOCK_NONLOCALIZED_VARS.
26015 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
26018 * ipa-visibility.c (non_local_p): Fix typos.
26019 (localize_node): When localizing symbol in same comdat group,
26020 dissolve the group only when we know external symbols are going
26022 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
26024 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26026 PR tree-optimization/79390
26027 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
26028 order does not result in usable sequence, retry with reversed operand
26034 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
26035 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
26036 op0 instead of fold_convert_loc (loc, type, arg0).
26038 2017-04-12 Jeff Law <law@redhat.com>
26040 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
26041 has a delay slot in the generated code.
26043 * config/cris/cris.md (cris_preferred_reload_class): Return
26044 GENNONACR_REGS rather than GENERAL_REGS.
26046 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26049 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
26050 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
26051 signedness of the result type.
26053 2017-04-12 Richard Biener <rguenther@suse.de>
26054 Jeff Law <law@redhat.com>
26056 PR tree-optimization/80359
26057 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
26058 trim stores to TARGET_MEM_REFs.
26060 2017-04-12 Richard Biener <rguenther@suse.de>
26062 PR tree-optimization/79390
26063 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
26064 threading case even more.
26066 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
26069 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
26070 for quad_address_p for TImode, instead of just not indexed_address.
26072 2017-04-12 Richard Biener <rguenther@suse.de>
26073 Bernd Edlinger <bernd.edlinger@hotmail.de>
26075 PR middle-end/79671
26076 * alias.c (component_uses_parent_alias_set_from): Handle
26077 TYPE_TYPELESS_STORAGE.
26078 (get_alias_set): Likewise.
26079 * tree-core.h (tree_type_common): Add typeless_storage flag.
26080 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
26081 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
26082 for types containing members with TYPE_TYPELESS_STORAGE.
26083 (place_field): Likewise.
26084 (layout_type): Likewise for ARRAY_TYPE.
26085 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
26086 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
26087 TYPE_TYPELESS_STORAGE.
26088 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
26090 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26093 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
26094 first argument to type.
26096 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26100 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
26101 CONST0_RTX (mode) rather than const0_rtx where appropriate.
26102 (rs6000_expand_binop_builtin): Likewise.
26103 (rs6000_expand_ternop_builtin): Likewise; also add missing
26104 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
26105 vshasigma built-ins.
26106 * doc/extend.texi: Document that vec_xxpermdi's third argument
26107 must be a constant.
26109 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
26111 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
26112 Use shift_const cost parameter when calculating gain of STV shifts.
26114 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
26116 PR rtl-optimization/70478
26117 * lra-constraints.c (process_alt_operands): Check memory for
26118 disfavoring memory insn operand.
26120 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26122 PR middle-end/80100
26123 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
26124 left shift in unsigned HOST_WIDE_INT type.
26126 PR rtl-optimization/80385
26127 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
26128 (not (neg X)) into (plus X -1) for complex or non-integral modes.
26131 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
26132 if they have any depend clauses.
26134 2017-04-11 Martin Liska <mliska@suse.cz>
26137 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
26138 * ipa-split.c (split_function): Create a local comdat symbol
26139 if caller is in a comdat group.
26141 2017-04-11 Martin Liska <mliska@suse.cz>
26144 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
26147 2017-04-11 Martin Sebor <msebor@redhat.com>
26149 PR middle-end/80364
26150 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
26151 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
26153 (directive::set_width, directive::set_precision, format_character):
26155 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
26158 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
26161 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
26162 conflict, set target->arch_name instead of target->cpu_name.
26164 2017-04-11 Richard Biener <rguenther@suse.de>
26166 PR tree-optimization/80374
26167 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
26168 build_zero_cst, remove fold_convertible_p check again.
26170 2017-04-11 Martin Liska <mliska@suse.cz>
26173 * ubsan.c (instrument_object_size): Do not instrument register
26176 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26179 * config/i386/i386-builtin-types.def
26180 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
26181 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
26182 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
26183 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
26184 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
26185 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
26186 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
26187 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
26188 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
26189 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
26190 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
26191 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
26192 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
26193 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
26194 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
26195 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
26196 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
26197 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
26198 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
26199 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
26200 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
26201 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
26202 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
26203 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
26204 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
26205 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
26206 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
26207 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
26208 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
26209 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
26210 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
26211 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
26212 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
26213 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
26214 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
26215 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
26216 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
26217 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
26218 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
26219 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
26220 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
26221 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
26222 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
26223 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
26224 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
26225 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
26227 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
26228 flag to second_arg_count, handle 4 argument function type _COUNT
26229 aliases, handle second_arg_count on second argument rather than last.
26231 2017-04-10 Jeff Law <law@redhat.com>
26233 PR tree-optimization/80374
26234 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
26235 record anything if we can not convert integer_zero_node to the
26238 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
26241 * config/rs6000/rs6000.c (rs6000_option_override_internal):
26242 Enhance special handling given to the TARGET_P9_MINMAX option in
26243 relation to certain other options.
26245 2017-04-10 Bin Cheng <bin.cheng@arm.com>
26247 PR tree-optimization/80153
26248 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
26249 remove POINTER_PLUS_EXPR's base part directly, rather than through
26252 2017-04-10 Richard Biener <rguenther@suse.de>
26253 Bin Cheng <bin.cheng@arm.com>
26255 PR tree-optimization/80153
26256 * tree-affine.c (aff_combination_to_tree): Get base pointer from
26257 the first element of pointer type aff_tree. Build result expr in
26259 (add_elt_to_tree): Convert to type unconditionally. Remove other
26260 fold_convert calls.
26261 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
26262 (rewrite_use_nonlinear_expr): Check invariant using iv information.
26264 2017-04-10 Richard Biener <rguenther@suse.de>
26266 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26269 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
26271 PR rtl-optimization/70478
26272 * lra-constraints.c (curr_small_class_check): New.
26273 (update_and_check_small_class_inputs): New.
26274 (process_alt_operands): Update curr_small_class_check. Disfavor
26275 alternative insn memory operands. Check available regs for small
26278 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26281 * config/mips/mips.opt (-mvirt): Update description.
26282 * doc/invoke.texi (-mvirt): Likewise.
26284 2017-04-10 Richard Biener <rguenther@suse.de>
26286 PR middle-end/80362
26287 * fold-const.c (fold_binary_loc): Look at unstripped ops when
26288 looking for NEGATE_EXPR in -A / -B to A / B folding.
26290 2017-04-10 Martin Liska <mliska@suse.cz>
26292 PR gcov-profile/80224
26293 * gcov.c (print_usage): Fix usage string.
26294 (get_gcov_intermediate_filename): Remove.
26295 (output_gcov_file): Use both for normal and intermediate format.
26296 (generate_results): Do not initialize special file for
26297 intermediate format.
26299 2017-04-10 Richard Biener <rguenther@suse.de>
26301 PR tree-optimization/80304
26302 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
26305 2017-04-10 Nathan Sidwell <nathan@acm.org>
26308 * config/rs6000/rs6000.c (rs6000_vector_type): New.
26309 (rs6000_init_builtins): Use it.
26311 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26313 * config/arm/arm.md (<mrc>): Add mode to SET source.
26314 (<mrrc>): Likewise.
26316 2017-04-10 Richard Biener <rguenther@suse.de>
26318 PR middle-end/80344
26319 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
26321 2017-04-10 Jakub Jelinek <jakub@redhat.com>
26324 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
26325 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
26326 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
26327 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
26328 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
26329 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
26330 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
26331 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
26332 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
26333 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
26334 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
26335 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
26336 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
26337 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
26338 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
26339 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
26340 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
26341 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
26342 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
26343 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
26344 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
26345 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
26346 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
26348 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
26350 PR rtl-optimization/70478
26351 * lra-constraints.c: Reverse the last patch.
26353 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
26355 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
26356 Add comment for WCHAR_T.
26358 2017-04-08 Martin Liska <mliska@suse.cz>
26361 2017-04-07 Martin Liska <mliska@suse.cz>
26364 * ipa-split.c (split_function): Add function part to a same comdat
26367 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26370 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
26372 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
26374 * rs6000/rs6000.c (vec_load_pendulum): Rename...
26375 (vec_pairing): ...to this.
26376 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
26377 (rs6000_sched_init): Adjust for name change.
26378 (struct rs6000_sched_context): Likewise.
26379 (rs6000_init_sched_context): Likewise.
26380 (rs6000_set_sched_context): Likewise.
26382 2017-04-07 Jakub Jelinek <jakub@redhat.com>
26388 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
26390 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
26391 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
26392 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
26394 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
26396 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
26398 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
26400 PR rtl-optimization/70703
26401 * ira-color.c (update_conflict_hard_regno_costs): Use
26402 int64_t instead of HOST_WIDE_INT.
26404 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
26406 PR rtl-optimization/70478
26407 * lra-constraints.c (process_alt_operands): Disfavor alternative
26408 insn memory operands.
26410 2017-04-07 Jeff Law <law@redhat.com>
26412 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
26413 CALL and NOTE_INSN_CALL_ARG_LOCATION.
26415 2017-04-07 Martin Liska <mliska@suse.cz>
26418 * config/aarch64/aarch64.c (aarch64_process_target_attr):
26419 Show error message instead of an ICE.
26421 2017-04-07 Martin Liska <mliska@suse.cz>
26424 * ipa-split.c (split_function): Add function part to a same comdat
26427 2017-04-07 Richard Biener <rguenther@suse.de>
26429 PR middle-end/80341
26430 * tree.c (get_unwidened): Also handle ! for_type case for
26432 * convert.c (do_narrow): Split out from ...
26433 (convert_to_integer_1): ... here. Do not pass final truncation
26434 type to get_unwidened for TRUNC_DIV_EXPR.
26436 2017-04-07 Richard Biener <rguenther@suse.de>
26438 * tree-affine.c (wide_int_ext_for_comb): Take type rather
26440 (aff_combination_const): Adjust.
26441 (aff_combination_scale): Likewise.
26442 (aff_combination_add_elt): Likewise.
26443 (aff_combination_add_cst): Likewise.
26444 (aff_combination_convert): Likewise.
26445 (add_elt_to_tree): Likewise. Remove unused argument.
26446 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
26448 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
26450 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
26452 * config/arm/arm.c (arm_default_short_enums): Use
26453 ARM_DEFAULT_SHORT_ENUMS.
26454 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
26456 2017-04-06 Jakub Jelinek <jakub@redhat.com>
26459 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
26460 members with redundant out-of-class redeclaration.
26462 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
26465 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
26466 * config/i386/i386.md (*zero_extendsidi2):
26467 Add (?*x,*x) and (?*v,*v) alternatives.
26469 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
26472 * config/i386/i386.c (ix86_expand_builtin)
26473 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
26474 mode from insn data. Convert operands to insn operand mode.
26475 Copy operands that don't satisfy insn predicate to a register.
26477 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
26479 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
26482 2017-04-06 Richard Biener <rguenther@suse.de>
26484 PR tree-optimization/80334
26485 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
26486 preserve alignment of accesses.
26488 2017-04-06 Richard Biener <rguenther@suse.de>
26490 PR tree-optimization/80262
26491 * tree-sra.c (build_ref_for_offset): Preserve address-space
26493 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
26494 Drop useless address-space information on MEM_REF offsets.
26496 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
26498 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
26500 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
26502 PR rtl-optimization/70703
26503 * ira-color.c (update_conflict_hard_regno_costs): Use
26504 HOST_WIDE_INT instead of long.
26506 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
26509 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
26510 not defined for x86_64 target. Add -mmmx target option when __SSE2__
26512 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
26513 for x86_64 target. Handle -m3dnowa option.
26515 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
26517 PR rtl-optimization/70703
26518 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
26519 (update_conflict_hard_regno_costs): Use long instead of unsigned
26520 arithmetic for cost calculation.
26522 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26523 Bernd Edlinger <bernd.edlinger@hotmail.de>
26526 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
26529 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
26532 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
26533 ptr_mode with Pmode throughout.
26534 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
26535 into probe_stack_range and use DImode.
26537 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
26540 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
26541 call_eh_return is true.
26543 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26545 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
26546 Initialize last_match_fntype_index.
26548 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26551 * tree-nvr.c: Include internal-fn.h.
26552 (pass_return_slot::execute): Ignore internal calls without
26555 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26556 Richard Biener <rguenther@suse.de>
26559 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
26560 captures used multiple times, except for the last use.
26561 * generic-match-head.c: Include gimplify.h.
26563 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26565 PR tree-optimization/79390
26566 * target.h (struct noce_if_info): Declare.
26567 * targhooks.h (default_noce_conversion_profitable_p): Declare.
26568 * target.def (noce_conversion_profitable_p): New target hook.
26569 * ifcvt.h (struct noce_if_info): New type, moved from ...
26570 * ifcvt.c (struct noce_if_info): ... here.
26571 (noce_conversion_profitable_p): Renamed to ...
26572 (default_noce_conversion_profitable_p): ... this. No longer
26574 (noce_try_store_flag_constants, noce_try_addcc,
26575 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
26576 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
26577 instead of noce_conversion_profitable_p.
26578 * config/i386/i386.c: Include ifcvt.h.
26579 (ix86_option_override_internal): Don't override
26580 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
26581 (ix86_noce_conversion_profitable_p): New function.
26582 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
26583 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
26584 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
26585 * doc/tm.texi: Regenerated.
26587 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26589 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
26592 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26595 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
26596 instructions for small multiply cores.
26598 2017-04-04 Jeff Law <law@redhat.com>
26600 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
26602 (mips_expand_vec_perm_const): Initialize elements in orig_perm
26603 that are not set by the loop over the elements.
26605 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26608 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
26609 int mode, convert_modes it to mode as unsigned, otherwise use
26610 lowpart_subreg to mode rather than SImode.
26611 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
26612 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
26613 Use DImode instead of SImode for the shift count operand.
26614 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
26617 2017-04-04 Richard Biener <rguenther@suse.de>
26619 PR middle-end/80281
26620 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
26621 arithmetic done for the negate or the plus. Simplify.
26622 (A - (-B) -> A + B): Likewise.
26623 * fold-const.c (split_tree): Make sure to not negate pointers.
26625 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
26627 PR rtl-optimization/60818
26628 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
26629 a compare of comparisons with the thing compared if this results
26630 in a different machine mode.
26632 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
26634 * alias.c (base_alias_check): Fix typo in comment.
26635 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
26636 * cgraphunit.c (symbol_table::compile): Likewise.
26637 * collect2.c (maybe_run_lto_and_relink): Likewise.
26638 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
26639 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
26640 * config/avr/avr.c (avr_map_op_t): Likewise.
26641 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
26642 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
26643 * config/epiphany/epiphany.md (movcc): Likewise.
26644 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
26645 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
26647 * config/mips/mips.c (mips_save_restore_reg): Likewise.
26648 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
26649 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
26650 * config/sh/sh.c (sh_rtx_costs): Likewise.
26651 * fold-const.c (fold_truth_andor): Likewise.
26652 * genautomata.c (collapse_flag): Likewise.
26653 * gengtype.h (struct type::u::s): Likewise.
26654 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
26655 * input.c (FORMAT_AMOUNT): Likewise.
26656 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
26657 (known_aggs_to_agg_replacement_list): Likewise.
26658 * ipa-inline-analysis.c: Likewise.
26659 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
26660 * ipa-polymorphic-call.c
26661 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
26662 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
26663 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
26665 * modulo-sched.c (apply_reg_moves): Likewise.
26666 * omp-expand.c (build_omp_regions_1): Likewise.
26667 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
26668 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
26669 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
26670 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
26671 * value-prof.c: Likewise.
26672 * var-tracking.c (val_reset): Likewise.
26674 2017-04-03 Richard Biener <rguenther@suse.de>
26676 PR tree-optimization/80275
26677 * fold-const.c (split_address_to_core_and_offset): Handle
26680 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
26682 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
26683 descriptors is at least equal to that of functions.
26685 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
26687 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
26689 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
26692 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
26693 (mov<IMOD4:mode>): New expander.
26694 (*mov<IMOD4:mode>_internal): New insn and split pattern.
26696 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
26698 PR rtl-optimization/79405
26699 * fwprop.c (propagations_left): New variable.
26700 (forward_propagate_into): Decrement it.
26701 (fwprop_init): Initialize it.
26702 (fw_prop): If the variable has reached zero, stop propagating.
26703 (fwprop_addr): Ditto.
26705 2017-03-31 Jakub Jelinek <jakub@redhat.com>
26708 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
26709 a FUNCTION_DECL, pass it as decl instead of origin to
26712 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
26714 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
26717 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
26720 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
26721 TARGET_VSX_SMALL_INTEGER.
26723 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26725 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
26726 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
26728 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26730 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
26731 extraction from odd-numbered MSA register.
26733 2017-03-31 Jakub Jelinek <jakub@redhat.com>
26735 PR middle-end/80173
26736 * expmed.c (store_bit_field_1): Don't attempt to create
26737 a word subreg out of hard registers wider than word if they
26738 have HARD_REGNO_NREGS of 1 for their mode.
26740 PR middle-end/80163
26741 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
26742 conversions to integer types wider than word and pointer.
26745 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
26746 (rtx_equal_for_cselib_p): Pass 0 to it.
26747 * cselib.c (cselib_hasher::equal): Likewise.
26748 (rtx_equal_for_cselib_1): Add depth argument. If depth
26749 is 128, don't look up VALUE locs and punt. Increment
26750 depth in recursive calls when walking VALUE locs.
26752 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
26754 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
26755 (make_gcov_file_name): Use the canonical path name for generating
26757 (read_line): Fix handling of files with ascii null bytes.
26759 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
26761 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
26762 to initialise a vector register instead
26763 of using a const_int.
26765 2017-03-30 Jakub Jelinek <jakub@redhat.com>
26767 PR translation/80189
26768 * gimplify.c (omp_default_clause): Use %qs instead of %s in
26769 diagnostic messages.
26771 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
26774 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
26775 (dfp_diex_<mode>): Update mode of operand 1.
26776 * doc/extend.texi (dxex, dxexq): Document change to return type.
26777 (diex, diexq): Document change to argument type.
26779 2017-03-30 Martin Jambor <mjambor@suse.cz>
26782 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
26783 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
26784 it reflects the signature changes performed at the callee side.
26785 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
26786 to cgraph_build_function_type_skip_args.
26787 (build_function_decl_skip_args): Adjust call to the above function.
26789 2017-03-30 Jakub Jelinek <jakub@redhat.com>
26792 * config/i386/sse.md
26793 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
26794 register as dest whenever it is a MEM not rtx_equal_p to the
26795 corresponding dup operand, and when forcing into reg move the
26796 reg into the memory afterwards.
26797 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
26798 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
26799 for the force_reg mode.
26800 (avx512vl_vextractf128<mode>): Use register as dest either
26801 always when a MEM, or when it is a MEM not rtx_equal_p to the
26802 corresponding dup operand, or even not when it is a CONST_VECTOR
26803 depending on the mode and lo vs. hi.
26804 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
26806 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
26807 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
26808 Likewise. Require that operands[2] is even.
26809 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
26810 Remove extraneous parens. Require that operands[2] is a multiple
26812 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
26813 operands[0] is a MEM if <mask_applied>, the predicates/constraints
26814 disallow memory then.
26816 2017-03-30 Richard Biener <rguenther@suse.de>
26818 PR tree-optimization/77498
26819 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
26820 to non-constants over backedges.
26822 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
26824 PR rtl-optimization/80233
26825 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
26826 as last_combined_insn. Do not test for BARRIER_P separately.
26828 2017-03-29 Andreas Schwab <schwab@suse.de>
26831 * calls.c (prepare_call_address): Convert funexp to Pmode before
26832 copying to temp reg.
26834 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26836 PR tree-optimization/80158
26837 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
26838 Handle possible future case of more than one alternate
26840 (replace_rhs_if_not_dup): Likewise.
26841 (replace_one_candidate): Likewise.
26843 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
26845 PR rtl-optimization/80193
26846 * ira.c (ira): Do not check allocation for LRA.
26848 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
26850 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
26851 (nvptx_output_simt_exit): Declare.
26852 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
26853 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
26854 (init_softstack_frame): Move initialization of crtl->is_leaf to...
26855 (nvptx_declare_function_name): ...here. Emit declaration of local
26856 memory space buffer for omp_simt_enter insn.
26857 (nvptx_output_unisimt_switch): New.
26858 (nvptx_output_softstack_switch): New.
26859 (nvptx_output_simt_enter): New.
26860 (nvptx_output_simt_exit): New.
26861 * config/nvptx/nvptx.h (struct machine_function): New fields
26862 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
26863 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
26864 (UNSPECV_SIMT_EXIT): Ditto.
26865 (omp_simt_enter_insn): New insn.
26866 (omp_simt_enter): New expansion.
26867 (omp_simt_exit): New insn.
26868 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
26870 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
26871 (expand_GOMP_SIMT_ENTER_ALLOC): New.
26872 (expand_GOMP_SIMT_EXIT): New.
26873 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
26874 (GOMP_SIMT_ENTER_ALLOC): Ditto.
26875 (GOMP_SIMT_EXIT): Ditto.
26876 * target-insns.def (omp_simt_enter): New insn.
26877 (omp_simt_exit): Ditto.
26878 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
26880 (lower_rec_simd_input_clauses): Implement SIMT privatization.
26881 (lower_rec_input_clauses): Likewise.
26882 (lower_lastprivate_clauses): Handle SIMT privatization.
26884 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
26885 (ompdevlow_adjust_simt_enter): New.
26886 (find_simtpriv_var_op): New.
26887 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
26888 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
26890 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
26891 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
26892 (copy_decl_for_dup_finish): Ditto.
26894 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
26896 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
26899 * config/i386/i386.c (ix86_option_override_internal): Always
26900 allow -mpreferred-stack-boundary=3 for 64-bit targets.
26902 2017-03-28 Bin Cheng <bin.cheng@arm.com>
26904 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
26906 2017-03-28 Bin Cheng <bin.cheng@arm.com>
26908 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
26909 mark new edge's irreducible flag accordign to it.
26910 (vect_do_peeling): Check loop preheader edge's irreducible flag
26911 and pass it to function slpeel_add_loop_guard.
26913 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
26915 PR tree-optimization/80218
26916 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
26917 Update block frequencies and counts.
26919 2017-03-28 Richard Biener <rguenther@suse.de>
26921 PR tree-optimization/78644
26922 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
26923 of a simplification result we may not use it at all.
26925 2017-03-28 Richard Biener <rguenther@suse.de>
26928 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
26929 without arguments, generate default definition of a SSA name.
26931 2017-03-28 Richard Biener <rguenther@suse.de>
26933 PR middle-end/80222
26934 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
26935 TYPE_REF_CAN_ALIAS_ALL references.
26936 * fold-const.c (fold_indirect_ref_1): Likewise.
26938 2017-03-28 Martin Liska <mliska@suse.cz>
26941 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
26942 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
26944 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
26945 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
26947 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
26948 (EXTRA_SPECS): Define.
26949 (SUBTARGET_EXTRA_SPECS): Likewise.
26950 (SUBTARGET_CPP_SPEC): Likewise.
26951 * config/arc/elf.h (EXTRA_SPECS): Renamed to
26952 SUBTARGET_EXTRA_SPECS.
26953 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
26955 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
26957 * config/arc/simdext.md (vst64_insn): Update pattern.
26958 (vld32wh_insn): Likewise.
26959 (vld32wl_insn): Likewise.
26960 (vld64_insn): Likewise.
26961 (vld32_insn): Likewise.
26963 2017-03-28 Marek Polacek <polacek@redhat.com>
26966 * fold-const.c (fold_comparison): Use protected_set_expr_location
26967 instead of SET_EXPR_LOCATION.
26969 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
26971 * tree.c (add_expr): Avoid name lookup warning.
26973 2017-03-27 Jeff Law <law@redhat.com>
26975 PR tree-optimization/80216
26976 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
26977 function name. Limit recursion depth.
26978 (record_temporary_equivalences): Corresponding changes.
26980 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
26982 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
26985 2017-03-27 Jakub Jelinek <jakub@redhat.com>
26988 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
26990 * cfgcleanup.c (reg_note_cfa_p): New array.
26991 (insns_have_identical_cfa_notes): New function.
26992 (old_insns_match_p): Don't cross-jump in between /f
26993 and non-/f instructions. If both i1 and i2 are frame related,
26994 verify all CFA notes, their order and content.
26996 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
26999 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
27000 HImode and SImode with zero extend to DImode to one insn.
27001 (bswap<mode>2_extenddi): Likewise.
27002 (bswapsi2_extenddi): Likewise.
27003 (bswaphi2_extendsi): Likewise.
27004 (bswaphi2): Combine bswap HImode and SImode into one insn.
27005 Separate memory insns from swapping register.
27006 (bswapsi2): Likewise.
27007 (bswap<mode>2): Likewise.
27008 (bswaphi2_internal): Delete, no longer used.
27009 (bswapsi2_internal): Likewise.
27010 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
27011 store, and gpr<-gpr swap insns.
27012 (bswap<mode>2_store): Likewise.
27013 (bswaphi2_reg): Register only splitter, combine with the splitter.
27014 (bswaphi2 splitter): Likewise.
27015 (bswapsi2_reg): Likewise.
27016 (bswapsi2 splitter): Likewise.
27017 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
27018 the insns into load, store, and register/register insns.
27019 (bswapdi2_ldbrx): Likewise.
27020 (bswapdi2_load): Likewise.
27021 (bswapdi2_store): Likewise.
27022 (bswapdi2_reg): Likewise.
27024 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
27026 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
27027 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
27029 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
27032 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
27034 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27035 special handling for target option conflicts between dform
27036 options (-mpower9-dform, -mpower9-dform-vector,
27037 -mpower9-dform-scalar) and -mno-direct-move.
27039 2017-03-27 Richard Biener <rguenther@suse.de>
27041 PR tree-optimization/80181
27042 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
27044 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27046 * config/arc/predicates.md (move_double_src_operand): Replace the
27047 call to move_double_src_operand with a call to address_operand.
27049 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27051 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
27052 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
27053 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
27055 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27057 * config/arc/predicates.md (long_immediate_loadstore_operand):
27058 Consider scaled addresses cases.
27060 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27062 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
27063 restored when in interrupt.
27064 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
27065 doesn't have delay slot.
27067 2017-03-27 Richard Biener <rguenther@suse.de>
27070 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
27071 inlined thunk clones.
27073 2017-03-27 Jakub Jelinek <jakub@redhat.com>
27076 * asan.c (instrument_derefs): Copy over last operand from
27077 original COMPONENT_REF to the new COMPONENT_REF with
27078 DECL_BIT_FIELD_REPRESENTATIVE.
27079 * ubsan.c (instrument_object_size): Likewise.
27081 2017-03-27 Richard Biener <rguenther@suse.de>
27083 PR tree-optimization/80170
27084 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
27085 sure DR/SCEV didnt fold in constants we do not see when looking
27086 at the reference base alignment.
27088 2017-03-27 Richard Biener <rguenther@suse.de>
27090 PR middle-end/80171
27091 * gimple-fold.c (fold_ctor_reference): Properly guard against
27092 NULL return value from canonicalize_constructor_val.
27094 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
27097 * config/i386/i386.c (ix86_expand_builtin)
27098 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
27099 flags reg setting and flags reg using instructions.
27100 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
27101 clobbering instructions to zero extend op2.
27103 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
27105 * doc/install.texi (Configuration) <--with-aix-soname>:
27106 Update link to AIX ld.
27108 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
27110 PR rtl-optimization/80160
27111 PR rtl-optimization/80159
27112 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
27113 reg_alternate_class into account.
27115 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
27118 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
27119 to consider in curr_insn_transform.
27121 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27123 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
27124 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
27125 and emit_mode_inner.
27127 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27129 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
27130 argument to the overloaded builtin variants. Use the new flag to
27131 deprecate certain builtin variants.
27132 * config/s390/s390-builtin-types.def: Add new builtin types.
27133 * config/s390/s390-builtins.h: Support new flags field for
27134 overloaded builtins.
27135 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
27136 (s390_macro_to_expand): Enable vector float data type.
27137 (s390_cpu_cpp_builtins_internal): Indicate support of the new
27138 builtins by incrementing the __VEC__ version number.
27139 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
27141 (s390_resolve_overloaded_builtin): Emit error messages depending
27142 on the builtin flags.
27143 * config/s390/s390.c (s390_expand_builtin): Support additional
27144 flags argument. Change error message to match the messages
27145 emitted in s390-c.c.
27146 * config/s390/s390.md: New UNSPEC_* constants.
27147 (op_type): Add new instruction types.
27148 * config/s390/vecintrin.h: Add new builtins and test data class
27150 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
27151 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
27152 (VEC_INEXACT, VEC_NOINEXACT): New constants.
27153 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
27154 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
27155 ("vec_mergel<mode>"): V_HW -> VEC_HW.
27157 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
27158 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
27159 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
27160 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
27162 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
27163 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
27164 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
27165 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
27167 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
27168 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
27169 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
27170 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
27171 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
27172 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
27173 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
27175 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
27176 ("vec_scatter_element<V_HW_4:mode>_DI")
27177 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
27178 ("vec_fpint<mode>", "vflls")
27179 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
27180 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
27181 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
27182 ("*vec_cmphe<mode>_cc"): ... these.
27184 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
27185 mode constant instead of magic value.
27187 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27189 * config/s390/s390.c (s390_expand_vec_compare): Support other
27190 vector floating point modes than just V2DF.
27191 (s390_expand_vcond): Likewise.
27192 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
27193 (s390_cannot_change_mode_class): Prevent mode changes between TF
27194 and V1TF in vector registers.
27195 * config/s390/s390.md (DF, SF): New mode attributes.
27196 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
27197 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
27198 SFmode support for VRs.
27199 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
27201 (VFT, VF_HW): New mode iterators.
27202 (vw, sdx): New mode attributes.
27203 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
27204 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
27205 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
27206 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
27207 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
27208 also the new vector floating point modes. Renaming to ...
27210 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
27211 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
27212 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
27213 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
27214 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
27215 ("vec_unordered<mode>"): ... these.
27217 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
27218 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
27219 ("*vec_extendv2df"): New insn definitions.
27221 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27223 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
27224 ("mulditi3_2", "*muldi3_sign"): New patterns.
27225 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
27226 rename the pattern definition.
27228 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27230 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
27232 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
27234 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27236 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
27237 instruction if possible.
27238 * config/s390/vector.md (vec_halfnumelts): New mode
27240 ("*vec_vllezlf<mode>"): New pattern.
27242 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27244 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
27245 ("popcountv4si2", "popcountv2di2"): Rename to ...
27246 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
27247 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
27249 ("popcount<mode>2_vxe"): New pattern.
27251 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27253 * common/config/s390/s390-common.c (processor_flags_table): Add
27255 * config.gcc: Add arch12.
27256 * config/s390/driver-native.c (s390_host_detect_local_cpu):
27257 Default to arch12 for unknown CPU model numbers.
27258 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
27259 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
27260 PROCESSOR_max sanity check.
27261 * config/s390/s390-opts.h (enum processor_type): Add
27263 * config/s390/s390.c (processor_table): Add arch12.
27264 (s390_expand_builtin): Add check for B_VXE flag.
27265 (s390_issue_rate): Add PROCESSOR_ARCH12.
27266 (s390_get_sched_attrmask): Likewise.
27267 (s390_get_unit_mask): Likewise.
27268 (s390_sched_score): Enable z13 scheduling for arch12.
27269 (s390_sched_reorder): Likewise.
27270 (s390_sched_variable_issue): Likewise.
27271 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
27273 (s390_tune_attr): Use z13 scheduling also for arch12.
27274 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
27275 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
27276 (TARGET_VXE_P): New macros.
27277 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
27278 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
27279 * config/s390/s390.opt: Add arch12 as processor_type.
27281 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27283 * config/s390/s390.md
27284 ("fixuns_truncdddi2", "fixuns_trunctddi2")
27285 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
27286 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
27288 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
27289 Rename expanders to ...
27291 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
27292 ("fixuns_truncdddi2_emu"): ... these.
27294 ("fixuns_trunc<mode>si2_emu"): New expander.
27296 ("*fixuns_truncdfdi2_z13"): Rename to ...
27297 ("*fixuns_truncdfdi2_vx"): ... this.
27299 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27301 * config/s390/2964.md: Remove the single element vector compare
27302 instructions which are no longer used.
27303 * config/s390/s390.c (s390_select_ccmode): Remove handling of
27305 (s390_canonicalize_comparison): Remove handling of DFmode
27307 (s390_expand_vec_compare_scalar): Remove function.
27308 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
27309 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
27311 ("*cmp<mode>_ccs"): Add wfcdb instruction.
27313 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27315 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
27317 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
27318 will anyway by matched by mov<mode>_64dfp.
27320 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27322 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
27323 vlef/vstef. Add missing operand to vleif.
27325 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27327 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
27328 pair for all vector types with 64 bit elements.
27329 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
27330 * config/s390/vector.md (V_HW_64): ... here.
27331 (V_128_NOSINGLE): New mode iterator.
27332 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
27333 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
27334 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
27335 ("*vec_load_pairv2di"): Change to ...
27336 ("*vec_load_pair<mode>"): ... this one.
27338 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27340 * config/s390/constraints.md: Add comments.
27341 (jKK): Reject element sizes > 8 bytes.
27342 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
27344 * config/s390/s390.md: Add the s_operand checks formerly in
27345 s390_split_ok_p to various splitters where they are still
27347 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
27348 for 128 bit vectors. Plus two splitters.
27350 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27352 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
27355 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27358 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
27359 error if the boundary argument is not constant.
27361 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27363 PR rtl-optimization/80112
27364 * loop-doloop.c (doloop_condition_get): Don't check condition
27365 if cmp isn't SET with IF_THEN_ELSE src.
27367 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27369 PR tree-optimization/80158
27370 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
27371 replacing a candidate statement, also replace it for the
27372 candidate's alternate interpretation.
27373 (replace_rhs_if_not_dup): Likewise.
27374 (replace_one_candidate): Likewise.
27376 2017-03-24 Richard Biener <rguenther@suse.de>
27378 PR tree-optimization/80167
27379 * graphite-isl-ast-to-gimple.c
27380 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
27382 (translate_isl_ast_to_gimple::get_rename): Likewise.
27384 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
27386 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
27387 handling of certain combinations of target options, including the
27388 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
27389 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
27391 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27394 * config/arm/arm.md (*load_multiple): Add reload_completed to
27395 matching condition.
27397 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27398 Richard Biener <rguenth@suse.de>
27400 PR tree-optimization/79908
27401 PR tree-optimization/80136
27402 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27403 been cast away, gimplify_and_add suffices.
27405 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
27407 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
27409 2017-03-23 Richard Biener <rguenther@suse.de>
27411 PR tree-optimization/80032
27412 * gimplify.c (gimple_push_cleanup): Forced unconditional
27413 cleanups still have to go to the conditional_cleanups
27416 2017-03-22 Jakub Jelinek <jakub@redhat.com>
27418 PR tree-optimization/80072
27419 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
27421 (next_operand_entry_id): Change type to unsigned int.
27422 (sort_by_operand_rank): Make sure to return the right return value
27423 even if unsigned fields are bigger than INT_MAX.
27424 (struct oecount): Change cnt and id type to unsigned int.
27425 (oecount_hasher::equal): Formatting fix.
27426 (oecount_cmp): Make sure to return the right return value
27427 even if unsigned fields are bigger than INT_MAX.
27428 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
27431 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
27432 TREE_READONLY on result if writing it more than once.
27435 * doc/invoke.texi (-fsanitize=thread): Document that with
27436 -fnon-call-exceptions atomics are not able to throw
27440 * tsan.c: Include tree-eh.h.
27441 (instrument_builtin_call): Call maybe_clean_eh_stmt or
27442 maybe_clean_or_replace_eh_stmt where needed.
27443 (instrument_memory_accesses): Add cfg_changed argument.
27444 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
27445 if it returned true.
27446 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
27448 PR rtl-optimization/63191
27449 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
27450 wrapper function, moved the whole old content into ...
27451 (ix86_delegitimize_address_1): ... this. New inline function.
27452 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
27453 true as last argument instead of ix86_delegitimize_address.
27455 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
27457 * config/aarch64/aarch64.c (generic_branch_cost): Copy
27458 cortexa57_branch_cost.
27460 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
27462 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
27464 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27467 * doc/md.texi (Constraints): Document wA constraint.
27468 * config/rs6000/constraints.md (wA): New.
27469 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
27470 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
27471 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
27472 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
27474 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
27477 * gimplify.c (is_oacc_declared): New function.
27478 (oacc_default_clause): Use it to set default flags for acc declared
27479 variables inside parallel regions.
27480 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
27481 declared variables.
27482 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
27483 declare attribute to any decl as necessary.
27485 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
27488 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
27489 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
27490 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
27491 (arm_arch_lpae): This.
27492 * config/arm/arm.c (arm_arch7ve): Rename into ...
27493 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
27494 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
27497 2017-03-22 Martin Liska <mliska@suse.cz>
27500 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
27501 error message instead of an ICE.
27503 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27505 * doc/extend.texi (6.11 Additional Floating Types): Revise.
27507 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
27509 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
27511 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27514 2017-03-21 Martin Sebor <msebor@redhat.com>
27516 * doc/extend.texi: Use "cannot" instead of "can't."
27517 * doc/hostconfig.texi: Same.
27518 * doc/install.texi: Same.
27519 * doc/invoke.texi: Same.
27520 * doc/loop.texi: Same.
27521 * doc/md.texi: Same.
27522 * doc/objc.texi: Same.
27523 * doc/rtl.texi: Same.
27524 * doc/tm.texi: Same.
27525 * doc/tm.texi.in: Same.
27526 * doc/trouble.texi: Same.
27528 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
27531 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
27532 (collect_checksum_attributes): Set it.
27533 (die_checksum_ordered): Use it.
27535 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27537 PR tree-optimization/79908
27538 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
27539 change: For a VA_ARG whose LHS has been cast away, use
27540 force_gimple_operand to construct the side effects.
27542 2017-03-21 David Malcolm <dmalcolm@redhat.com>
27544 PR translation/80001
27545 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
27546 more amenable to translation.
27547 (oacc_loop_auto_partitions): Likewise.
27549 2017-03-21 Marek Polacek <polacek@redhat.com>
27550 Martin Sebor <msebor@redhat.com>
27552 PR tree-optimization/80109
27553 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
27554 on INTEGRAL_TYPE_P.
27556 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27557 Segher Boessenkool <segher@kernel.crashing.org>
27560 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
27561 check reg_used_between_p between insn and one of succ or succ2
27562 depending on if succ is artificial insn not inserted into insn
27565 2017-03-21 Martin Liska <mliska@suse.cz>
27567 PR gcov-profile/80081
27568 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
27569 * doc/gcc.texi: Include gcov-dump stuff.
27570 * doc/gcov-dump.texi: New file.
27572 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
27574 PR rtl-optimization/79150
27575 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
27576 conditional jump, if the jump is the last insn of the loop.
27578 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27579 Richard Biener <rguenth@suse.de>
27581 PR tree-optimization/79908
27582 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27583 been cast away, use force_gimple_operand to construct the side
27586 2017-03-21 Martin Liska <mliska@suse.cz>
27588 PR libfortran/79956
27589 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
27592 2017-03-21 Brad Spengler <spender@grsecurity.net>
27595 * plugin.c (htab_hash_plugin): New function.
27596 (add_new_plugin): Use it and adjust.
27597 (parse_plugin_arg_opt): Adjust.
27598 (init_one_plugin): Likewise.
27600 2017-03-21 Richard Biener <rguenther@suse.de>
27602 PR tree-optimization/80032
27603 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
27604 if set force the cleanup to happen unconditionally.
27605 (gimplify_target_expr): Push inserted clobbers with force_uncond
27606 to avoid them being removed by control-dependent DCE.
27608 2017-03-21 Richard Biener <rguenther@suse.de>
27610 PR tree-optimization/80122
27611 * tree-inline.c (copy_bb): Do not expans va-arg packs or
27612 va_arg_pack_len when the inlined call stmt requires pack
27614 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
27616 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27619 * tsan.c (instrument_builtin_call): If the memory model argument
27620 is not a constant, assume it is valid.
27623 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
27626 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27628 PR rtl-optimization/79910
27629 * combine.c (can_combine_p): Do not allow combining an I0 or I1
27630 if its dest is used by an insn before I2 (other than the combined
27631 insns themselves, which are properly handled already).
27633 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27636 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
27638 * combine.c (record_used_regs): New static function.
27639 (try_combine): Handle situations where there is an additional
27640 instruction between I2 and I3 which needs to have a LOG_LINK
27644 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27646 * combine.c (try_combine): Delete redundant i1 test. Call
27647 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27649 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27652 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
27653 alternatives 13/14.
27655 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27657 PR tree-optimization/80054
27658 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
27659 the optimization if a PHI or any of its arguments is not dominated
27660 by the candidate's basis. Use gphi* rather than gimple* as
27662 (replace_profitable_candidates): Clean up a gimple* variable that
27663 should be a gphi* variable.
27665 2017-03-20 Martin Sebor <msebor@redhat.com>
27668 * doc/extend.texi (attribute constructor): Document present limitation.
27670 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
27673 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
27674 __POWER9_VECTOR__ #ifdef control, change template definition to
27675 use Power9-specific built-in function.
27676 (vec_any_eq): Likewise.
27677 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
27678 to control outcomes from this test.
27679 (vector_ae_<mode>p): For VEC_F modes, likewise.
27681 2017-03-20 Ian Lance Taylor <iant@google.com>
27683 * config/i386/i386.c (ix86_function_regparm): Save an extra
27684 register for -fsplit-stack with DECL_STATIC_CHAIN.
27686 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
27689 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
27690 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
27692 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
27694 * config/riscv/riscv.c (riscv_print_operand): Use "fence
27696 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
27699 2017-03-20 Marek Polacek <polacek@redhat.com>
27702 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
27704 2017-03-20 Richard Biener <rguenther@suse.de>
27706 PR tree-optimization/80113
27707 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
27708 allocate extra SSA name for PHI def.
27709 (add_close_phis_to_outer_loops): Likewise.
27710 (add_close_phis_to_merge_points): Likewise.
27711 (copy_loop_close_phi_args): Likewise.
27712 (copy_cond_phi_nodes): Likewise.
27714 2017-03-20 Martin Liska <mliska@suse.cz>
27716 PR middle-end/79753
27717 * tree-chkp.c (chkp_build_returned_bound): Do not build
27718 returned bounds for a LHS that's not a BOUNDED_P type.
27720 2017-03-20 Martin Liska <mliska@suse.cz>
27724 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
27725 COMPLEX_CST and VECTOR_CST.
27727 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27730 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
27731 target operand. A new splitter adds the clobber statement in case
27732 the target operand is dead anyway.
27734 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
27736 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
27737 to age-old versions of binutils and glibc.
27739 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
27741 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
27743 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
27745 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
27747 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
27749 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
27750 requirement for binutils 2.13.
27752 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27754 * combine.c (try_combine): Delete redundant i1 test. Call
27755 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27757 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
27759 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
27760 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
27762 <riscv64-*-elf>: Re-arrange section
27763 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
27764 <riscv32-*-linux>: Likewise.
27765 <riscv64-*-elf>: Likewise
27766 <riscv64-*-linux>: Likewise.
27768 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
27771 * aarch64.opt(verbose-cost-dump): Fix typo.
27773 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
27776 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
27777 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
27779 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
27781 * reload.c (find_reloads): When reloading a nonoffsettable address,
27782 use RELOAD_OTHER for it and its address reloads.
27784 PR rtl-optimization/79910
27785 * combine.c (record_used_regs): New static function.
27786 (try_combine): Handle situations where there is an additional
27787 instruction between I2 and I3 which needs to have a LOG_LINK
27790 2017-03-17 Jeff Law <law@redhat.com>
27792 PR tree-optimization/71437
27793 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
27794 conditional in the hash table first.
27795 (vrp_dom_walker::before_dom_children): Extract condition from
27796 ASSERT_EXPR. Record condition, its inverion and any implied
27797 conditions as well.
27799 2017-03-17 Marek Polacek <polacek@redhat.com>
27800 Markus Trippelsdorf <markus@trippelsdorf.de>
27802 PR tree-optimization/80079
27803 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
27806 2017-03-17 Richard Biener <rguenther@suse.de>
27808 PR middle-end/80075
27809 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
27810 Properly verify the LHS before the RHS possibly claims to be
27812 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
27815 2017-03-17 Martin Jambor <mjambor@suse.cz>
27817 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
27818 (List of -O2 options): Likewise.
27819 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
27822 2017-03-17 Tom de Vries <tom@codesourcery.com>
27824 * gcov-dump.c (print_usage): Print bug_report_url.
27826 2017-03-17 Richard Biener <rguenther@suse.de>
27828 PR middle-end/80050
27829 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
27830 (parser::peek): Likewise.
27832 2017-03-17 Richard Biener <rguenther@suse.de>
27834 PR tree-optimization/80048
27835 * sese.c (free_sese_info): Properly release rename_map and
27836 copied_bb_map elements.
27838 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
27840 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
27841 Add linked-list forward and backlinks. Insert on
27842 construction, remove on destruction.
27843 (class pass_store_merging): Add m_stores_head field.
27844 (pass_store_merging::terminate_and_process_all_chains):
27845 Iterate over m_stores_head list.
27846 (pass_store_merging::terminate_all_aliasing_chains):
27848 (pass_store_merging::execute): Check for debug stmts first.
27849 Push new chains onto the m_stores_head stack.
27851 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
27854 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
27855 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
27856 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
27858 2017-03-16 Jeff Law <law@redhat.com>
27860 PR tree-optimization/71437
27861 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
27862 member function. Implementation moved into after_dom_children
27863 member function and into the threader's thread_outgoing_edges
27865 (dom_opt_dom_walker::after_dom_children): Simplify by moving
27866 some code into new thread_outgoing_edges.
27867 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
27868 definition. Simplify marker handling (do it here). Assume we always
27869 have the available expression and the const/copies tables.
27870 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
27872 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
27873 * tree-vrp.c (equiv_stack): No longer file scoped.
27874 (vrp_dom_walker): New class.
27875 (vrp_dom_walker::before_dom_children): New member function.
27876 (vrp_dom_walker::after_dom_children): Likewise.
27877 (identify_jump_threads): Setup domwalker. Use it rather than
27878 walking edges in a random order by hand. Simplify setup/finalization.
27879 (finalize_jump_threads): Remove.
27880 (vrp_finalize): Do not call identify_jump_threads here.
27881 (execute_vrp): Do it here instead and call thread_through_all_blocks
27884 PR tree-optimization/71437
27885 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
27887 (simplify_stmt_for_jump_threading): Add basic_block argument. All
27889 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
27890 (dom_opt_dom_walker::thread_across_edge): Remove
27891 handle_dominating_asserts argument. All callers changed.
27892 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
27893 changes. Remove calls to lhs_of_dominating_assert. Other
27894 uses of handle_dominating_asserts turn into unconditional code
27895 (simplify_control_stmt_condition_1): Likewise.
27896 (simplify_control_stmt_condition): Likewise.
27897 (thread_through_normal_block, thread_across_edge): Likewise.
27898 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
27899 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
27900 object if it is not an SSA_NAME.
27901 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
27902 before calling into the VRP specific simplifiers.
27903 (identify_jump_threads): Remove handle_dominating_asserts
27906 2017-03-16 Jakub Jelinek <jakub@redhat.com>
27909 * tree-diagnostic.c (default_tree_printer): No longer static.
27910 * tree-diagnostic.h (default_tree_printer): New prototype.
27912 2017-03-16 Tamar Christina <tamar.christina@arm.com>
27914 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
27915 Change ins into fmov.
27917 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27919 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
27920 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
27921 Use h_con constraint for operand 1.
27922 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
27923 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
27925 2017-03-15 Jeff Law <law@redhat.com>
27927 PR tree-optimization/71437
27928 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
27929 (record_temporary_equivalences): Use it.
27931 PR tree-optimization/71437
27932 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
27933 tree-ssa-scopedtables.
27934 (lookup_avail_expr, build_and_record_new_cond): Likewise.
27935 (record_conditions, record_cond, vuse_eq): Likewise.
27936 (record_edge_info): Adjust to API tweak of record_conditions.
27937 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
27938 (record_temporary_equivalences, optimize_stmt): Likewise.
27939 (eliminate_redundant_computations): Likewise.
27940 (record_equivalences_from_stmt): Likewise.
27941 * tree-ssa-scopedtables.c: Include options.h and params.h.
27942 (vuse_eq): New function, moved from tree-ssa-dom.c
27943 (build_and_record_new_cond): Likewise.
27944 (record_conditions): Likewise. Accept vector of conditions rather
27945 than edge_equivalence structure for first argument.
27946 for the first argument.
27947 (avail_exprs_stack::lookup_avail_expr): New member function, moved
27948 from tree-ssa-dom.c.
27949 (avail_exprs_stack::record_cond): Likewise.
27950 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
27951 from tree-ssa-dom.c.
27952 (avail_exprs_stack): Add new member functions lookup_avail_expr
27954 (record_conditions): Declare.
27956 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
27959 * lra-constraints.c (process_alt_operands): Increase reject for
27960 reloading an input/output operand.
27962 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
27965 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
27966 insns to convert from signed/unsigned char/short to IEEE 128-bit
27968 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
27970 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
27973 * config/i386/i386.c (ix86_vector_duplicate_value): Create
27974 subreg of inner mode for values already in registers.
27976 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
27978 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
27979 iteration reg is used after the loop.
27981 2017-03-14 Martin Sebor <msebor@redhat.com>
27983 PR tree-optimization/79800
27984 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
27985 precision in negative-positive range.
27986 (format_floating): Call non-const overload with adjusted precision.
27988 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
27991 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
27994 2017-03-14 Martin Sebor <msebor@redhat.com>
27996 PR middle-end/80020
27997 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
27998 * builtins.def (aligned_alloc): Use it.
28001 * Makefile.in (GTFILES): Add calls.c.
28002 * calls.c: Include "gt-calls.h".
28004 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
28006 PR rtl-optimization/79728
28007 * regs.h (struct target_regs): New field
28008 x_contains_allocatable_regs_of_mode.
28009 (contains_allocatable_regs_of_mode): New macro.
28010 * reginfo.c (init_reg_sets_1): Initialize it, and change
28011 contains_reg_of_mode so it includes global regs as well.
28012 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
28013 rather than contains_regs_of_mode.
28015 2017-03-14 Martin Liska <mliska@suse.cz>
28017 * doc/invoke.texi: Document options that can't be combined with
28018 -fcheck-pointer-bounds.
28020 2017-03-14 Martin Liska <mliska@suse.cz>
28022 PR middle-end/79831
28023 * doc/invoke.texi (-Wchkp): Document the option.
28025 2017-03-14 Martin Liska <mliska@suse.cz>
28027 * Makefile.in: Install gcov-dump.
28029 2017-03-14 Martin Liska <mliska@suse.cz>
28031 * multiple_target.c (expand_target_clones): Bail out for
28032 an invalid attribute.
28034 2017-03-14 Richard Biener <rguenther@suse.de>
28036 * alias.c (struct alias_set_entry): Pack properly.
28037 * cfgloop.h (struct loop): Likewise.
28038 * cse.c (struct set): Likewise.
28039 * ipa-utils.c (struct searchc_env): Likewise.
28040 * loop-invariant.c (struct invariant): Likewise.
28041 * lra-remat.c (struct cand): Likewise.
28042 * recog.c (struct change_t): Likewise.
28043 * rtl.h (struct address_info): Likewise.
28044 * symbol-summary.h (function_summary): Likewise.
28045 * tree-loop-distribution.c (struct partition): Likewise.
28046 * tree-object-size.c (struct object_size_info): Likewise.
28047 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
28048 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
28049 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
28050 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
28051 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
28052 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
28053 (struct _stmt_vec_info): Likewise.
28055 2017-03-14 Martin Liska <mliska@suse.cz>
28058 * multiple_target.c (create_dispatcher_calls): Check that
28059 a target can create a function dispatcher.
28061 2017-03-14 Martin Liska <mliska@suse.cz>
28064 * multiple_target.c (expand_target_clones): Drop local.local
28065 flag for default implementation.
28067 2017-03-14 Richard Biener <rguenther@suse.de>
28069 PR tree-optimization/80030
28070 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
28072 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
28074 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
28075 gcc_fallthrough() instead of __attribute__((fallthrough));
28077 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28079 * doc/gcc.texi: Remove "up" link to (DIR).
28080 * doc/gccint.texi: Ditto.
28082 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28084 * doc/install.texi (Specific) <avr>: Remove reference to
28087 2017-03-13 Jeff Law <law@redhat.com>
28089 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
28090 attribute rather than comments.
28092 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
28093 match_scratch operand is highest.
28095 2017-03-13 Martin Liska <mliska@suse.cz>
28097 PR middle-end/78339
28098 * ipa-pure-const.c (warn_function_noreturn): If the declarations
28099 is a CHKP clone, use original declaration.
28101 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28103 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
28104 (arc_conditional_register_usage): Use a different allocation order
28105 when optimizing for size.
28106 * common/config/arc/arc-common.c (arc_option_optimization_table):
28107 Section anchors default on when optimizing for size.
28109 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28111 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
28113 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28115 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
28116 * config/arc/arc.md (cpu_facility): Add cd variant.
28117 (*movqi_insn): Add code density variant.
28118 (*movhi_insn): Likewise.
28119 (*movqi_insn): Likewise.
28120 (*addsi3_mixed): Likewise.
28121 (subsi3_insn): Likewise.
28123 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28125 * config/arc/arc.md (movsi_cond_exec): Update constraint.
28127 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28129 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
28130 expressions with MINUS and UNARY ops.
28132 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28135 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
28137 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
28138 between vec_select and vector argument.
28139 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
28140 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
28141 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
28142 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
28143 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
28144 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
28146 2017-03-13 Richard Biener <rguenther@suse.de>
28149 * params.def (vect-max-peeling-for-alignment): Fix typo.
28151 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28153 * doc/install.texi (Specific) <mips-*-*>: Remove description of
28154 issue that only occurred with binutils below 2.18.
28156 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28158 * doc/install.texi (Specific) <cris-axis-elf>: No longer
28159 refer to binutils 2.11/2.12 minimum.
28161 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28163 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
28164 ftp.kernel.org and simplify binutils requirement.
28166 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
28168 * doc/invoke.texi (Warning Options): Fix spelling of link-time
28170 (Optimize Options): Ditto. Also remove redundancy.
28172 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28174 PR translation/79848
28175 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
28177 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
28178 to G_ to avoid double translation.
28180 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28182 PR translation/79923
28183 * auto-profile.c (get_combined_location): Convert leading
28184 character of diagnostics to lower case and remove trailing period.
28185 (read_profile): Likewise for various diagnostics.
28186 * config/arm/arm.c (arm_option_override): Remove trailing period
28187 from various diagnostics.
28188 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
28189 (msp430_expand_delay_cycles): Likewise.
28191 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28194 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
28195 full command-line argument, rather than just "str".
28196 (aarch64_validate_march): Likewise.
28197 (aarch64_validate_mtune): Likewise.
28199 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
28201 PR rtl-optimization/78911
28202 * lra-assigns.c (must_not_spill_p): New function.
28203 (spill_for): Use it.
28205 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28207 PR tree-optimization/79981
28208 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
28209 ATOMIC_COMPARE_EXCHANGE ifn result.
28210 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
28211 IFN_ATOMIC_COMPARE_EXCHANGE.
28213 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28216 * opts.c (parse_sanitizer_options): Add missing question mark to
28217 "did you mean" message.
28219 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28221 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
28223 (VMULEUH_UNS): Likewise.
28224 (VMULOUB_UNS): Likewise.
28225 (VMULOUH_UNS): Likewise.
28226 * config/rs6000/rs6000.c (builtin_function_type): Remove
28227 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
28229 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28232 * read-rtl-function.c (function_reader::read_rtx_operand): Update
28233 x with result of extra_parsing_for_operand_code_0.
28234 (function_reader::extra_parsing_for_operand_code_0): Convert
28235 return type from void to rtx, returning x. When reading
28236 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
28237 larger size containing struct block_symbol.
28239 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
28241 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
28242 -mfloat128-hardware without -m64.
28244 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
28247 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
28248 entries to the case statement that marks unsigned arguments to
28249 overloaded functions.
28251 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
28253 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
28254 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
28256 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
28259 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
28260 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
28262 2017-03-10 Martin Liska <mliska@suse.cz>
28266 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
28267 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
28270 2017-03-10 Olivier Hainque <hainque@adacore.com>
28272 * tree-switch-conversion (array_value_type): Start by resetting
28273 candidate type to it's main variant.
28275 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28277 PR rtl-optimization/79909
28278 * combine.c (try_combine): Use simplify_replace_rtx on individual
28279 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
28280 of the whole CALL_INSN_FUNCTION_USAGE.
28282 PR tree-optimization/79972
28283 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
28284 get_range_info on SSA_NAMEs. Formatting fixes.
28286 2017-03-10 Richard Biener <rguenther@suse.de>
28287 Jakub Jelinek <jakub@redhat.com>
28289 PR tree-optimization/77975
28290 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
28291 edge to be constant.
28292 (get_val_for): For constant x return it. Formatting fix.
28293 (loop_niter_by_eval): Avoid pointless looping if the next iteration
28294 would use the same bases as the current one.
28296 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28298 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
28299 instead of vec_select for V1TImode.
28300 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
28302 (VSX_LE_128): Add V1TI to this mode iterator.
28303 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
28304 (*vsx_le_perm_store_<mode>): Likewise.
28305 (pre-reload splitter for VSX stores): Likewise.
28306 (post-reload splitter for VSX stores): Likewise.
28307 (*vsx_xxpermdi2_le_<mode>): Likewise.
28308 (*vsx_lxvd2x2_le_<mode>): Likewise.
28309 (*vsx_stxvd2x2_le_<mode>): Likewise.
28311 2017-03-09 Michael Eager <eager@eagercon.com>
28313 Correct failures with --enable-checking=yes,rtl.
28315 * config/microblaze/microblaze.c (microblaze_expand_shift):
28316 Replace GET_CODE test with CONST_INT_P and INTVAL test with
28317 test for const0_rtx.
28318 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
28319 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
28321 2017-03-09 Richard Biener <rguenther@suse.de>
28323 PR tree-optimization/79977
28324 * graphite-scop-detection.c (scop_detection::merge_sese):
28325 Handle the case of extra exits to blocks dominating the entry.
28327 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
28329 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28332 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
28334 PR rtl-optimization/79949
28335 * lra-constraints.c (process_alt_operands): Check memory when
28336 trying to predict a cycle. Print about the overall increase.
28338 2017-03-09 Richard Biener <rguenther@suse.de>
28340 PR middle-end/79971
28341 * gimple-expr.c (useless_type_conversion_p): Preserve
28342 TYPE_SATURATING for fixed-point types.
28344 2017-03-09 Richard Biener <rguenther@suse.de>
28347 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
28348 alignment of BLKmode params.
28350 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28353 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
28354 (VALL_NO_V2Q): Likewise.
28356 * config/aarch64/aarch64-simd.md
28357 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
28359 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
28360 VALL_NO_V2Q mode iterator.
28361 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
28363 2017-03-09 Martin Liska <mliska@suse.cz>
28365 PR tree-optimization/79631
28366 * tree-chkp-opt.c (chkp_is_constant_addr): Call
28367 tree_int_cst_sign_bit just for INTEGER constants.
28369 2017-03-09 Martin Liska <mliska@suse.cz>
28373 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
28376 2017-03-09 Marek Polacek <polacek@redhat.com>
28379 * tree.c (inchash::add_expr): Handle TREE_VEC.
28381 2017-03-09 Martin Liska <mliska@suse.cz>
28384 (chkp_narrow_size_and_offset): New function.
28385 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
28386 (void chkp_parse_bit_field_ref): New function.
28387 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
28388 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
28390 2017-03-09 Martin Liska <mliska@suse.cz>
28393 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
28394 (chkp_find_bounds_1): Remove gcc_unreachable.
28396 2017-03-09 Jakub Jelinek <jakub@redhat.com>
28399 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
28400 BUILT_IN_SYNC*, determine the access type from the size suffix and
28401 always build a MEM_REF with that type. Handle forgotten
28402 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
28405 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
28406 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
28407 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
28408 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
28409 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
28410 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
28411 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
28412 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
28413 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
28414 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
28415 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
28416 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
28417 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
28418 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
28419 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
28420 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
28421 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
28422 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
28423 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
28424 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
28425 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
28426 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
28427 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
28428 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
28429 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
28430 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
28431 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
28432 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
28433 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
28434 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
28435 definitions outside of __OPTIMIZE__ guarded section.
28438 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
28439 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
28440 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
28441 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
28444 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28446 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
28447 ("vfenez<mode>"): Add missing constraints.
28449 2017-03-08 Martin Sebor <msebor@redhat.com>
28452 * config/nds32/nds32.c (nds32_option_override):
28453 Fix misspelled diagnostic.
28455 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28458 * gimplify.c (gimplify_omp_for): Replace index var in outer
28459 taskloop statement with an artificial variable and add
28460 OMP_CLAUSE_PRIVATE clause for it.
28462 2017-03-08 Richard Biener <rguenther@suse.de>
28464 PR tree-optimization/79955
28465 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
28466 for accesses that are completely outside of the variable.
28468 2017-03-08 Andrew Haley <aph@redhat.com>
28470 PR tree-optimization/79943
28471 * tree-ssa-loop-split.c (compute_new_first_bound): When
28472 calculating the new upper bound, (END-BEG) should be added, not
28475 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28477 * config/avr/avr.md (setmemhi): Make sure match_dup
28478 operand number comes before match_scratch.
28480 2017-03-08 Richard Biener <rguenther@suse.de>
28482 PR tree-optimization/79920
28483 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
28484 with ncopies == 1 to ...
28485 (vect_transform_slp_perm_load): ... here. Properly compute
28486 all element loads by iterating VF times over the group. Do
28487 not handle ncopies (computed in a broken way) in
28488 vect_create_mask_and_perm.
28490 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28493 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
28494 is a uniform vector, use uniform_vector_p return value instead of
28495 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
28497 2017-03-07 Marek Polacek <polacek@redhat.com>
28499 PR middle-end/79809
28500 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
28501 (alloca_call_type): Likewise.
28503 2017-03-07 Martin Liska <mliska@suse.cz>
28505 * gcov.c (process_args): Put comment to correct location.
28507 2017-03-07 Martin Liska <mliska@suse.cz>
28509 PR middle-end/68270
28510 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
28511 Use array_at_struct_end_p instead of DECL_CHAIN (field).
28512 (chkp_narrow_bounds_for_field): Likewise.
28513 (chkp_parse_array_and_component_ref): Pass one more argument to
28516 2017-03-07 Richard Biener <rguenther@suse.de>
28518 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
28521 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
28523 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
28524 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
28526 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28529 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
28530 to end of description.
28531 (PARAM_MAX_STORES_TO_MERGE): Likewise.
28533 2017-03-07 Jakub Jelinek <jakub@redhat.com>
28535 PR rtl-optimization/79901
28536 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
28538 (*avx512f_<code><mode>3<mask_name>): ... this.
28539 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
28540 iterator instead of VI8_AVX2_AVX512BW.
28542 PR rtl-optimization/79901
28543 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
28544 min/max expander, expand it using expand_vec_cond_expr.
28547 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
28550 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28553 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
28554 to void * for PCH reasons.
28555 * dwarf2out.c (output_loc_operands, output_die): Cast
28556 v.val_vec.array to unsigned char *.
28558 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
28561 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
28564 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
28566 PR rtl-optimization/79571
28567 * lra-constraints.c (process_alt_operands): Calculate static
28568 reject and subtract it from overall when only addresses will be
28571 2017-03-06 Julia Koval <julia.koval@intel.com>
28574 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
28575 incoming stack boundary to 128 for 64-bit targets.
28577 2017-03-06 Richard Biener <rguenther@suse.de>
28579 PR tree-optimization/79894
28580 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
28581 to NULL after folding it.
28583 2017-03-06 Richard Biener <rguenther@suse.de>
28585 PR tree-optimization/79824
28586 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
28587 check disabling peeling for gaps.
28589 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
28591 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28592 attributes): Document gettimeofday.
28594 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28596 * config/s390/s390.c (s390_option_override_internal): Set
28597 PARAM_MIN_VECT_LOOP_BOUND
28599 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28601 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
28602 * config/s390/s390.md: Likewise.
28604 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28607 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
28608 (<avx2_avx512>_perm<mode>): Rename to ...
28609 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
28611 (<avx512>_perm<mode>_mask): Rename to ...
28612 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
28614 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
28615 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
28616 instead of VI8F_256_512.
28617 (avx512f_perm<mode>): New define_expand.
28618 (avx512f_perm<mode>_mask): Likewise.
28619 (avx512f_perm<mode>_1<mask_name>): New define_insn.
28620 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
28622 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28624 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
28625 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
28627 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
28629 2017-03-06 Martin Liska <mliska@suse.cz>
28632 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
28633 when having a SSA NAME w/o VAR_DECL assigned to it.
28635 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28637 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
28638 msa_dpsub_<su>_d): Fix MODE for vec_select.
28640 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28642 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
28644 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
28646 2017-03-06 Richard Biener <rguenther@suse.de>
28648 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
28649 * plugin.c (register_plugin_info): Likewise.
28650 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
28652 2017-03-05 Jakub Jelinek <jakub@redhat.com>
28654 * config/i386/sse.md (sse_storehps, sse_storelps,
28655 avx_<castmode><avxsizesuffix>_<castmode>,
28656 avx512f_<castmode><avxsizesuffix>_<castmode>,
28657 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
28658 in condition that at least one operand is not a MEM.
28660 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28662 PR middle-end/79805
28663 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
28664 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
28666 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
28667 gimple_call_nothrow_p flag based on whether original builtin can throw.
28668 If it can, emit following stmts on the fallthrough edge.
28669 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
28670 don't create new bb if inserting just debug stmts on the edge, try to
28671 insert them on the fallthru bb or just reset debug stmts.
28673 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
28676 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
28677 restore recog_data (including the operand rtxes inside it) around
28678 the call to get_insn_template.
28680 2017-03-03 Martin Sebor <msebor@redhat.com>
28682 PR tree-optimization/79699
28683 * context.c (context::~context): Free MPFR caches to avoid
28684 a memory leak on program exit.
28686 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28688 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
28689 Use wide_int::ulow () instead of .elt (0).
28691 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
28693 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
28694 (*pushxf): Limit oF constraint to 32bit targets and add oC
28695 constraint for 64bit targets.
28696 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
28697 (*pushdf): Change rmF constraint to rmC.
28699 2017-03-03 Martin Liska <mliska@suse.cz>
28701 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
28702 Remove unused variable.
28704 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28707 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
28708 is a memory operand, increase num_memory.
28709 (ix86_expand_args_builtin): Likewise.
28711 2017-03-03 Jan Hubicka <jh@suse.cz>
28714 * ipa-devirt.c (maybe_record_node): Properly handle
28715 __cxa_pure_virtual visibility.
28717 2017-03-03 Martin Liska <mliska@suse.cz>
28719 PR tree-optimization/79803
28720 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
28722 (pass_loop_prefetch::execute): Disabled optimization if an
28723 assumption about L1 cache size is not met.
28725 2017-03-03 Martin Liska <mliska@suse.cz>
28727 PR rtl-optimization/79574
28728 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
28729 (hash_scan_set): Likewise.
28730 (dump_hash_table): Likewise.
28731 (hoist_code): Likewise.
28733 2017-03-03 Richard Biener <rguenther@suse.de>
28735 * fixed-value.c (fixed_from_string): Restore use of elt (1)
28736 in place of uhigh ().
28737 (fixed_convert_from_real): Likewise.
28739 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
28742 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
28744 2017-03-03 Richard Biener <rguenther@suse.de>
28746 PR middle-end/79818
28747 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
28748 TYPE_OVERFLOW_UNDEFINED check.
28750 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28752 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
28754 (vector_ae_<mode>_p): Likewise.
28755 (vector_nez_<mode>_p): Likewise.
28756 (vector_ne_v2di_p): Likewise.
28757 (vector_ae_v2di_p): Likewise.
28758 (vector_ne_<mode>_p): Likewise.
28759 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
28761 (vsx_tsqrt<mode>2_fe): Likewise.
28763 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
28766 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
28768 2017-03-02 Jakub Jelinek <jakub@redhat.com>
28770 PR rtl-optimization/79780
28771 * cprop.c (one_cprop_pass): When second and further conditional trap
28772 in a single basic block is turned into an unconditional trap, turn it
28773 into a deleted note to avoid RTL verification failures.
28775 2017-03-02 Richard Biener <rguenther@suse.de>
28777 * fold-const.c (const_binop): Use ulow () instead of elt (0).
28779 2017-03-02 Richard Biener <rguenther@suse.de>
28781 PR tree-optimization/79345
28783 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
28784 param and abort the walk, returning -1 if it is hit.
28785 (walk_aliased_vdefs): Take a limit param and pass it on.
28786 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
28787 defaulting to 0 and return a signed int.
28788 * tree-ssa-uninit.c (struct check_defs_data): New struct.
28789 (check_defs): New helper.
28790 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
28791 about uninitialized memory.
28792 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
28793 bogus uninitialized warning.
28794 (fixed_convert_from_real): Likewise.
28796 2017-03-02 Bin Cheng <bin.cheng@arm.com>
28798 PR tree-optimization/66768
28799 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
28800 iv_use if base object can't be determined.
28802 2017-03-02 Jakub Jelinek <jakub@redhat.com>
28804 PR tree-optimization/79345
28805 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
28806 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
28807 (get_pattern_stats): Initialize it.
28808 * genemit.c (gen_expand): Verify match_scratch numbers come after
28809 match_operand/match_dup numbers.
28810 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
28811 match_scratch numbers.
28812 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
28814 * config/s390/s390.md (trunctdsd2): Likewise.
28816 2017-03-02 Richard Biener <rguenther@suse.de>
28818 * wide-int.h (wide_int_storage::operator=): Implement in terms
28821 2017-03-02 Richard Biener <rguenther@suse.de>
28823 PR tree-optimization/79777
28824 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
28825 the to insert expression to sth existing.
28827 2017-03-01 Martin Sebor <msebor@redhat.com>
28829 PR middle-end/79692
28830 * gimple-ssa-sprintf.c
28831 (directive::known_width_and_precision): New function.
28832 (format_integer): Use it.
28833 (get_mpfr_format_length): Consider the full range of precision
28834 when computing %g output with the # flag. Set the likely byte
28835 count to 3 rather than 1 when precision is indeterminate.
28836 (format_floating): Correct the lower bound of precision.
28838 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28840 * doc/invoke.texi: Document default code model for 64-bit Linux.
28842 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28845 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
28846 udiv rather than div since input pattern is unsigned.
28848 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
28850 * config/i386/i386.c (print_reg): Warn for values of
28851 unsupported size in integer register.
28853 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
28856 * config/rs6000/predicates.md (current_file_function_operand): Do
28857 not allow self calls to be local if the function is replaceable.
28859 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
28862 * config/rs6000/altivec.h (vec_ctz and others): Change the
28863 preprocessor macro that controls conditional compilation from
28864 _ARCH_PWR9 to __POWER9_VECTOR__.
28865 (vec_all_ne): Change parameterization of __altivec_scalar_pred
28866 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
28867 control (instead of _ARCH_PWR9 control) so that template
28868 definition uses power9-specific function.
28869 (vec_any_eq): Likewise.
28870 (vec_all_ne): Change macro definition to use a power9-specific
28871 expansion under #ifdef __POWER9_VECTOR__ control (instead of
28872 _ARCH_PWR9 control).
28873 (vec_any_eq) Likewise.
28874 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
28875 expansion for CMPNEF to remove support for xvcmpnesp instruction.
28876 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
28877 support for xvcmpnedp instruction.
28878 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
28879 macro expansion so that Power9 implementation of vec_all_ne does
28880 not use the AltiVec predicate framework.
28881 (VCMPNEH_P): Likewise.
28882 (VCMPNEW_P): Likewise.
28883 (VCMPNED_P): Likewise.
28884 (VCMPNEFP_P): Likewise.
28885 (VCMPNEDP_P): Likewise.
28886 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
28887 implementation of vec_any_eq to not use AltiVec predicate
28889 (VCMPAEH_P): Likewise.
28890 (VCMPAEW_P): Likewise.
28891 (VCMPAED_P): Likewise.
28892 (VCMPAEFP_P): Likewise.
28893 (VCMPAEDP_P): Likewise.
28894 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
28895 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
28896 not use the AltiVec predicate framework.
28897 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
28898 of vec_any_eq to not use AltiVec predicate framework.
28899 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
28900 support for predefined __POWER9_VECTOR__ macro to indicate that
28901 Power9 instruction selection is enabled.
28902 (altivec_overloaded_builtins): Remove extraneous
28903 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
28904 function argument types RS6000_BTI_bool_V16QI and
28905 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
28906 entry for overloaded function argument types RS6000_BTI_bool_V4SI
28907 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
28908 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
28909 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
28910 Power9 for implementations of vec_cmpne. Change the signature for
28911 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
28912 (representing vec_all_ne) to remove the previously described first
28913 argument of type RS6000_BTI_INTSI, as this was an artifact of
28914 reliance on the AltiVec predicate framework, which is no longer
28915 used in the implementation of these functions. Add
28916 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
28917 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
28918 since, unlike the AltiVec predicate framework implementation, we
28919 do not share function descriptors between vec_alle and vec_anyeq.
28920 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
28921 set of modes that receive special treatment even when
28922 TARGET_P9_VECTOR is true. The special treatment emits code that
28923 does not depend on Power9 instructions.
28924 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
28925 define_expand to not rely on AltiVec predicate framework.
28926 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
28928 (vector_ne_v2di_p): Change this define_expand to not rely on
28929 AltiVec predicate framework.
28930 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
28932 (vector_ne_<mode>_p): Change this define_expand to not rely on
28933 AltiVec predicate framework.
28934 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
28936 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
28937 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
28938 define_insn pattern.
28939 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
28940 define_insn pattern because the xvcmpne<VSs>. instruction is not
28942 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
28943 instruction is not supported.
28945 2017-03-01 Jakub Jelinek <jakub@redhat.com>
28947 * config/nvptx/nvptx.c: Include intl.h.
28949 2017-03-01 Martin Jambor <mjambor@suse.cz>
28952 * ipa-prop.h (ipa_bits): Removed field known.
28953 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
28954 to pointers. Adjusted their comments to warn about their sharing.
28955 (ipcp_transformation_summary): Change bits to a vector of pointers.
28956 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
28957 (ipa_get_ipa_bits_for_value): Declare.
28958 * tree-vrp.h (value_range): Mark as GTY((for_user)).
28959 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
28960 (ipa_bits_hash_table): Likewise.
28961 (ipa_vr_ggc_hash_traits): Likewise.
28962 (ipa_vr_hash_table): Likewise.
28963 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
28964 being pointers and vr_known being removed.
28965 (ipa_set_jf_unknown): Likewise.
28966 (ipa_get_ipa_bits_for_value): New function.
28967 (ipa_set_jfunc_bits): Likewise.
28968 (ipa_get_value_range): New overloaded functions.
28969 (ipa_set_jfunc_vr): Likewise.
28970 (ipa_compute_jump_functions_for_edge): Use the above functions to
28971 construct bits and vr parts of jump functions.
28972 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
28973 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
28975 (ipcp_grow_transformations_if_necessary): Also allocate
28976 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
28978 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
28979 them. Fix too long lines.
28980 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
28981 vr_known being removed.
28982 (ipa_read_jump_function): Use new setter functions to construct bits
28983 and vr parts of jump functions or set them to NULL.
28984 (write_ipcp_transformation_info): Adjust for bits being pointers.
28985 (read_ipcp_transformation_info): Likewise.
28986 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
28988 Include gt-ipa-prop.h.
28989 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
28991 (ipcp_store_bits_results): Likewise.
28992 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
28993 Do not write to existing jump functions but use a temporary instead.
28995 2017-03-01 Jakub Jelinek <jakub@redhat.com>
28998 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
28999 attempt to use its first operand as BIT_FIELD_REF base.
29001 2017-03-01 Richard Biener <rguenther@suse.de>
29003 PR middle-end/79721
29004 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
29005 interpolating formula in wrapping arithmetic.
29006 (chrec_apply): Convert chrec_evaluate return value to wanted type.
29008 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29010 PR tree-optimization/79734
29011 * tree-vect-generic.c (expand_vector_condition): Optimize
29012 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
29013 Handle VEC_COND_EXPR where comparison has different inner width from
29014 type's inner width.
29016 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
29018 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
29019 markup, and similar issues. Remove @opindex entries for things
29020 that aren't options. Add missing -mmpy-option entries.
29022 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29024 PR tree-optimization/79737
29025 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
29026 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
29027 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
29028 instead of byte_size. Formatting fix.
29029 (shift_bytes_in_array_right): Formatting fix.
29031 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
29034 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
29035 condition on optimize for the leaf function test.
29037 2017-02-28 Martin Liska <mliska@suse.cz>
29040 * read-rtl-function.c (function_reader::handle_unknown_directive):
29041 Bail out when one uses -flto.
29043 2017-02-28 Martin Liska <mliska@suse.cz>
29045 * common.opt: Replace space with tabular for options of <number>
29047 * config/i386/i386.opt: Show <number> value for
29048 -mlarge-data-threshold.
29049 * opts.c (print_filtered_help): Do not display number in hexadecimal
29052 2017-02-28 Martin Liska <mliska@suse.cz>
29054 * common.opt: Fix --help=option -Q for options which are of
29057 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
29059 * config/i386/i386.c (print_reg): Error out for values
29060 of 8-bit size in invalid integer register.
29062 2017-02-28 Martin Sebor <msebor@redhat.com>
29064 PR tree-optimization/79691
29065 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
29067 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29070 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
29071 gcc_unreachable with output_operand_lossage.
29073 2017-02-28 Richard Biener <rguenther@suse.de>
29075 PR tree-optimization/79740
29076 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
29078 (visit_nary_op): Insert the nary into the hashtable if we
29079 pattern-matched sth.
29080 * tree-ssa-pre.c (eliminate_insert): Robustify.
29082 2017-02-28 Richard Biener <rguenther@suse.de>
29084 PR middle-end/79731
29085 * fold-const.c (decode_field_reference): Reject out-of-bound
29088 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29090 * config/i386/i386.c: Include intl.h.
29091 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
29092 instead of just cond ? "..." : "...".
29093 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
29094 * coverage.c (read_counts_file): Likewise.
29095 * omp-offload.c: Include intl.h.
29096 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
29097 of just cond ? "..." : "...".
29098 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
29099 of just cond ? "..." : "...".
29101 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
29104 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
29106 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
29107 'tune for' CPU name.
29108 * config/arm/arm-cpu-data.h: Regenerated.
29110 2017-02-28 Richard Biener <rguenther@suse.de>
29112 PR tree-optimization/79732
29113 * tree-inline.c (expand_call_inline): Do not shadow var.
29115 2017-02-28 Richard Biener <rguenther@suse.de>
29117 PR tree-optimization/79723
29118 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
29119 address-space properly.
29121 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
29123 * doc/optinfo.texi (Optimization groups): Fix option used for
29125 * doc/invoke.texi (-fopt-info): Document "omp".
29126 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
29127 (OPTGROUP_ALL): Add OPTGROUP_OMP.
29128 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
29129 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
29130 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
29132 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
29134 * dumpfile.c (optgroup_options): Instead of "openmp", associate
29135 OPTGROUP_OMP with "omp".
29137 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
29140 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
29141 for arithmetic shift of unsigned V2DI.
29143 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
29145 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
29146 arc/linux.h headers.
29147 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
29148 (LINK_SPEC): Likewise.
29149 (ARC_TLS_EXTRA_START_SPEC): Likewise.
29150 (EXTRA_SPECS): Likewise.
29151 (STARTFILE_SPEC): Likewise.
29152 (ENDFILE_SPEC): Likewise.
29153 (LIB_SPEC): Likewise.
29154 (TARGET_SDATA_DEFAULT): Likewise.
29155 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
29156 (MULTILIB_DEFAULTS): Likewise.
29157 (DWARF2_UNWIND_INFO): Likewise.
29158 * config/arc/big.h: New file.
29159 * config/arc/elf.h: Likewise.
29160 * config/arc/linux.h: Likewise.
29161 * config/arc/t-uClibc: Remove.
29163 2017-02-27 Bin Cheng <bin.cheng@arm.com>
29165 PR tree-optimization/77536
29166 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
29167 (tree_transform_and_unroll_loop): Use above function to compute the
29168 estimated niter of unrolled loop and use it when scaling profile.
29169 Also use count info rather than frequency if it's non-zero.
29170 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
29171 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
29172 (vect_transform_loop): Call above function.
29174 2017-02-27 Richard Biener <rguenther@suse.de>
29176 PR tree-optimization/45397
29177 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
29178 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
29179 (visit_nary_op): Add pattern matching for CSEing sign-changed
29180 or truncated operations with wider ones.
29182 2017-02-27 Richard Biener <rguenther@suse.de>
29184 PR tree-optimization/79690
29185 * tree-vect-stmts.c (vectorizable_store): Use vector type
29186 built from the DR with address-space.
29188 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
29190 * doc/invoke.texi (Optimize Options): Refine the description
29191 of asan-use-after-return.
29193 2017-02-25 Alan Modra <amodra@gmail.com>
29195 PR rtl-optimization/79584
29196 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
29197 base, not ad->base_term, the reg within base. Remove assertion
29198 that ad->base == ad->base_term. Replace gen_int_mode using
29199 bogus mode with const0_rtx.
29201 2017-02-25 Jakub Jelinek <jakub@redhat.com>
29203 PR middle-end/79396
29204 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
29205 FMA_EXPR like tcc_binary or tcc_unary.
29207 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
29210 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
29212 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
29213 (output_loc_operands): Handle DW_OP_call_ref and
29214 DW_OP_GNU_variable_value.
29215 (struct variable_value_struct): New type.
29216 (struct variable_value_hasher): Likewise.
29217 (variable_value_hash): New variable.
29218 (string_types): Remove.
29219 (copy_loc_descr): New function.
29220 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
29221 (prepend_loc_descr_to_each): New function.
29222 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
29223 instead of add_loc_descr_to_each if the first argument is single
29224 location list and the second has multiple.
29225 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
29226 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
29227 when looking for variable value which doesn't have other location info.
29228 (loc_list_from_tree): Formatting fix.
29229 (gen_array_type_die): Simplify DW_AT_string_length handling.
29230 (adjust_string_types): Remove.
29231 (gen_subprogram_die): Don't call adjust_string_types nor test/set
29232 string_types. Call resolve_variable_values.
29233 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
29234 (resolve_addr_in_expr): Likewise. Add A argument.
29235 (copy_deref_exprloc): Remove deref argument. Adjust for the
29236 original expression being DW_OP_GNU_variable_value with optionally
29237 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
29238 optionally after it.
29239 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
29240 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
29241 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
29242 (variable_value_hasher::hash, variable_value_hasher::equal): New
29244 (resolve_variable_value_in_expr, resolve_variable_value,
29245 resolve_variable_values, note_variable_value_in_expr,
29246 note_variable_value): New functions.
29247 (dwarf2out_early_finish): Call note_variable_value on all toplevel
29250 2017-02-24 Jakub Jelinek <jakub@redhat.com>
29253 * opts.h (handle_generated_option): Add GENERATED_P argument.
29254 * opts-common.c (handle_option): Adjust function comment.
29255 (handle_generated_option): Add GENERATED_P argument, pass it to
29257 (control_warning_option): Pass false to handle_generated_option
29259 * opts.c (maybe_default_option): Pass true to handle_generated_option
29261 * optc-gen.awk: Likewise.
29263 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29265 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
29266 a REG, look at the REG it is a SUBREG of.
29267 (splitter for cmpeqsi_t): Ditto.
29269 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29271 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
29272 the special USEs with the pattern of the insn, not the insn itself.
29274 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
29277 * doc/invoke.texi: Document -mload-store-pairs.
29279 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29280 Sandra Loosemore <sandra@codesourcery.com>
29282 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
29283 argument isn't a CONST_INT.
29284 (nios2_alternate_compare_const): Assert op is a CONST_INT.
29285 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
29286 (nios2_validate_compare): Bypass alternate compare logic if *op2
29287 is not a CONST_INT.
29288 (ldstwm_operation_p): Return false if first_base is not a REG or
29289 if first_offset is not a CONST_INT.
29291 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29293 * config/cris/cris.md: Use correct operand in a define_peephole2.
29295 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29297 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
29299 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29301 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
29302 this_insn if it is an INSN or JUMP_INSN.
29303 (force_offsettable): Look at base, not at addr.
29304 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
29305 on things that aren't necessarily CONST_INTs.
29307 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
29309 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
29310 -mfpmath=sse is the default also for x86-32 targets with SSE2
29311 instruction set when @option{-ffast-math} is enabled
29313 2017-02-24 Jeff Law <law@redhat.com>
29315 PR rtl-optimizatoin/79286
29316 * ira.c (update_equiv_regs): Drop may_trap_p exception to
29319 2017-02-24 Richard Biener <rguenther@suse.de>
29321 PR tree-optimization/79389
29322 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
29325 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
29327 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
29328 function comment to reflect reality.
29329 (loop_exits_before_overflow): Fix typo in function description.
29331 2017-02-24 Richard Biener <rguenther@suse.de>
29333 PR tree-optimization/79389
29334 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
29335 properly that a threading opportunity exists. Detect conditional
29336 copy/constant propagation opportunities.
29338 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
29340 * config/visium/visium.md (type): Add trap.
29341 (b): New mode attribute.
29342 (*btst): Rename into...
29343 (*btst<mode>): ...this and adjust.
29344 (*cbranchsi4_btst_insn): Rename into...
29345 (*cbranch<mode>4_btst_insn): ...this and adjust.
29346 (trap): New define_insn.
29348 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29350 PR tree-optimization/79389
29351 * ifcvt.c (struct noce_if_info): Add rev_cond field.
29352 (noce_reversed_cond_code): New function.
29353 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
29354 reversed_comparison_code. Formatting fix.
29355 (noce_try_store_flag): Test rev_cond != NULL in addition to
29356 reversed_comparison_code.
29357 (noce_try_store_flag_constants): Likewise.
29358 (noce_try_store_flag_mask): Likewise.
29359 (noce_try_addcc): Use rev_cond if non-NULL instead of
29360 reversed_comparison_code.
29361 (noce_try_cmove_arith): Likewise. Formatting fixes.
29362 (noce_try_minmax, noce_try_abs): Clear rev_cond.
29363 (noce_find_if_block): Initialize rev_cond.
29364 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
29365 instead of false as last argument never attempt to reverse it
29368 2017-02-23 Bin Cheng <bin.cheng@arm.com>
29370 PR tree-optimization/79663
29371 * tree-predcom.c (combine_chains): Process refs in reverse order
29372 only for ZERO length chains, and add explaining comment.
29374 2017-02-23 Jeff Law <law@redhat.com>
29376 PR tree-optimization/79578
29377 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
29378 in call to operand_equal_p.
29380 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
29383 * config/i386/cpuid.h: Fix another undefined behavior.
29385 2017-02-23 Richard Biener <rguenther@suse.de>
29387 PR tree-optimization/79683
29388 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
29389 vector types for data-refs.
29391 2017-02-23 Martin Liska <mliska@suse.cz>
29393 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
29395 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29397 PR middle-end/79665
29398 * internal-fn.c (get_range_pos_neg): Moved to ...
29399 * tree.c (get_range_pos_neg): ... here. No longer static.
29400 * tree.h (get_range_pos_neg): New prototype.
29401 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
29402 are known to be in between 0 and signed maximum inclusive, try to
29403 expand both unsigned and signed divmod and use the cheaper one from
29406 2017-02-22 Jeff Law <law@redhat.com>
29408 PR tree-optimization/79578
29409 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
29410 to compare base operands.
29412 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
29415 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
29416 gpc_reg_operand instead of fpr_reg_operand.
29418 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
29420 * config/mips/mips.c (mips_return_in_memory): Force FP
29421 vector types to be returned in memory for o32 ABI.
29423 2017-02-22 Jakub Jelinek <jakub@redhat.com>
29425 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
29426 instead of DW_TAG_member for static data member declarations and don't
29427 set no_linkage_name for static inline data members.
29428 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
29431 2017-02-22 Martin Liska <mliska@suse.cz>
29433 * doc/invoke.texi: Replace inequality signs with square brackets
29436 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
29439 * lra-constraints.c (simplify_operand_subreg): Handle
29440 WORD_REGISTER_OPERATIONS targets.
29442 2017-02-22 Jakub Jelinek <jakub@redhat.com>
29445 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
29446 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
29447 elimination by swapping fld*.
29449 2017-02-22 Richard Biener <rguenther@suse.de>
29451 PR tree-optimization/79673
29452 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
29453 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
29454 irrelevant address-space qualifiers and avoiding a
29455 ADDR_SPACE_CONVERT_EXPR from fold_convert.
29457 2017-02-22 Richard Biener <rguenther@suse.de>
29459 PR tree-optimization/79666
29460 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
29461 to not symbolically negate if that may introduce undefined
29464 2017-02-22 Martin Liska <mliska@suse.cz>
29467 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
29468 * data-streamer-out.c (streamer_write_gcov_count_stream):
29470 * value-prof.c (stream_out_histogram_value): Make assert more
29471 precise based on type of counter.
29473 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
29476 * config/i386/i386.md (standard_x87sse_constant_load splitter):
29477 Use nonimmediate_operand instead of memory_operand for operand 1.
29478 (float-extend standard_x87sse_constant_load splitter): Ditto.
29480 2017-02-21 Jeff Law <law@redhat.com>
29482 PR tree-optimization/79621
29483 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
29484 blocks with edges to themselves.
29486 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29489 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
29490 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
29491 Use gimple_call_builtin_p.
29494 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
29495 on temporarily removed DEBUG_INSNs.
29497 PR tree-optimization/79649
29498 * tree-loop-distribution.c (classify_partition): Give up on
29499 non-generic address space loads/stores.
29501 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
29503 * doc/loop.texi (Loop manipulation): Remove nonexistent
29504 tree_ssa_loop_version from the documentation.
29505 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
29507 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29510 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
29511 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
29512 * config/rs6000/rs6000.c: Include except.h.
29513 (rs6000_expand_split_stack_prologue): Call
29514 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
29516 2017-02-21 Martin Jambor <mjambor@suse.cz>
29519 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
29520 have been analyzed.
29522 2017-02-21 Martin Jambor <mjambor@suse.cz>
29524 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
29525 for backward compatibility only.
29526 * doc/invoke.texi (Option Summary): Remove all references to
29527 -fipa-cp-alignment.
29529 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
29533 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29535 * lra-constraints.c (curr_insn_transform): Handle
29536 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29538 2017-02-21 Martin Liska <mliska@suse.cz>
29540 * config/i386/i386.opt: Replace -masm-dialect with -masm.
29542 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
29544 PR translation/79638
29545 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
29547 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
29550 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
29551 (arm_function_ok_for_sibcall): Return false for an indirect call by
29552 descriptor if all the argument registers are used.
29553 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
29554 alignment of the function.
29556 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29558 PR tree-optimization/61441
29559 * simplify-rtx.c (simplify_const_unary_operation): For
29560 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
29561 the sNaN unmodified.
29563 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
29565 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29566 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29567 instead of SYSTEM_HEADER_DIR.
29569 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
29570 Martin Liška <mliska@suse.cz>
29572 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
29573 Fix typos and grammar, use active voice, and clarify.
29575 2017-02-20 Marek Polacek <polacek@redhat.com>
29577 PR middle-end/79537
29578 * gimplify.c (gimplify_expr): Handle unused *&&L;.
29581 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
29583 2017-02-20 Jakub Jelinek <jakub@redhat.com>
29586 * config/i386/i386.c (ix86_expand_builtin): Handle
29587 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
29588 ix86_builtins_isa[fcode].isa as a requirement of those
29589 flags and any other flag in the bitmask.
29590 (ix86_init_mmx_sse_builtins): Use 0 instead of
29591 ~OPTION_MASK_ISA_64BIT as mask.
29592 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
29593 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
29594 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
29595 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
29597 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29600 * lra-constraints.c (split_reg): Check requested split mode
29601 is supported by the register.
29603 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29605 * lra-constraints.c (simplify_operand_subreg): Remove early
29608 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29611 * lra-constraints.c (curr_insn_transform): Tighten condition
29612 for converting SUBREG reloads from OP_OUT to OP_INOUT.
29614 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29617 * lra-constraints.c (curr_insn_transform): Handle
29618 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29620 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
29623 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
29625 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
29627 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
29630 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
29633 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29635 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
29638 2017-02-18 Jakub Jelinek <jakub@redhat.com>
29640 * final.c (last_columnnum, override_columnnum): New variables.
29641 (final_start_function): Set last_columnnum, pass it to begin_prologue
29642 hook and pass 0 to dwarf2out_begin_prologue.
29643 (final_scan_insn): Update override_columnnum. Pass last_columnnum
29644 to source_line debug hook.
29645 (notice_source_line): Compute last_columnnum and for debug_column_info
29646 return true on column changes.
29647 * debug.h (struct gcc_debug_hooks): Add column argument to
29648 source_line and begin_prologue hooks.
29649 (debug_nothing_int_charstar_int_bool): Remove prototype.
29650 (debug_nothing_int_int_charstar,
29651 debug_nothing_int_int_charstar_int_bool): New prototypes.
29652 (dwarf2out_begin_prologue): Add column argument.
29653 * debug.c (do_nothing_debug_hooks): Adjust source_line and
29654 begin_prologue hooks.
29655 (debug_nothing_int_charstar_int_bool): Remove.
29656 (debug_nothing_int_int_charstar,
29657 debug_nothing_int_int_charstar_int_bool): New functions.
29658 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
29659 through to dwarf2out_source_line.
29660 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
29661 (dwarf2out_source_line): Add column argument, emit it if requested.
29662 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
29664 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29665 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29666 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
29667 through to dwarf2out_begin_prologue.
29668 (vmsdbgout_source_line): Add column argument, pass it through to
29669 dwarf2out_source_line.
29670 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
29671 dbxout_source_line caller.
29672 (dbxout_source_line): Add column argument.
29674 * common.opt (gno-column-info, gcolumn-info): New options.
29675 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
29676 (check_die): Also test for multiple DW_AT_decl_column attributes.
29677 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
29678 DW_AT_decl_column if requested.
29679 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
29681 (gen_variable_die): Likewise.
29682 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
29683 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
29686 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
29687 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
29688 (ix86_handle_option): Handle OPT_m3dnowa.
29689 * doc/invoke.texi (-m3dnowa): Document.
29690 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
29691 -m3dnowa instead of -m3dnow -march=athlon.
29694 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
29695 instead of gcc_assert for K, r and R code checks. Formatting fixes.
29697 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29700 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
29701 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
29702 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
29703 generator for vsx_xxpermdi_<mode>_be.
29704 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
29705 force big-endian semantics.
29706 (vsx_xxpermdi_<mode>_be): New define_expand with same
29707 implementation as previous version of vsx_xxpermdi_<mode>.
29709 2017-02-17 Jakub Jelinek <jakub@redhat.com>
29711 PR tree-optimization/79327
29712 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
29713 variable, its initialization and use.
29715 2017-02-17 Julia Koval <julia.koval@intel.com>
29717 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
29718 (OPTION_MASK_ISA_PKU_UNSET): New.
29719 (ix86_handle_option): Handle -mrdpid.
29720 * config/i386/cpuid.h (bit_RDPID): New.
29721 * config/i386/driver-i386.c (host_detect_local_cpu):
29722 Detect RDPID feature.
29723 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
29724 * config/i386/i386-c.c (ix86_target_macros_internal):
29726 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
29727 (ix86_valid_target_attribute_inner_p): Add "rdpid".
29728 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
29729 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
29730 * config/i386/i386.md (define_insn "rdpid"): New.
29731 * config/i386/i386.opt Add -mrdpid.
29732 * config/i386/immintrin.h (_rdpid_u32): New.
29734 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
29736 PR rtl-optimization/79541
29737 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
29738 instead of transforming it into USE.
29740 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
29742 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
29743 If HONOR_SNANS (SFmode) force the input to a register.
29744 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
29745 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
29746 an frsp or similar insn.
29748 2017-02-17 Martin Liska <mliska@suse.cz>
29750 PR rtl-optimization/79577
29751 * params.def (selsched-max-sched-times): Increase minimum to 1.
29753 2017-02-17 Martin Liska <mliska@suse.cz>
29755 PR rtl-optimization/79574
29756 * gcse.c (want_to_gcse_p): Prevent integer overflow.
29758 2017-02-17 Martin Liska <mliska@suse.cz>
29760 PR tree-optimization/79529
29761 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
29762 ssa_defined_default_def_p to handle cases which are implicitly
29764 * tree-ssa.c (ssa_defined_default_def_p): New function.
29765 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
29766 which are implicitly defined.
29767 * tree-ssa.h (ssa_defined_default_def_p): Declare.
29769 2017-02-17 Richard Biener <rguenther@suse.de>
29771 PR middle-end/79576
29772 * params.def (max-ssa-name-query-depth): Limit to 10.
29774 2017-02-17 Richard Biener <rguenther@suse.de>
29776 PR tree-optimization/79552
29777 * tree-ssa-structalias.c (visit_loadstore): Properly verify
29780 2017-02-17 Richard Biener <rguenther@suse.de>
29783 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
29785 2017-02-17 Marek Polacek <polacek@redhat.com>
29787 PR middle-end/79536
29788 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
29789 (fold_negate_expr): New wrapper.
29791 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
29793 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
29794 Correct terminology and de-emphasize pre-standard behavior.
29796 2017-02-16 Alan Modra <amodra@gmail.com>
29798 PR rtl-optimization/79286
29799 * ira.c (def_dominates_uses): New function.
29800 (update_equiv_regs): Don't create an equivalence for insns that
29801 may trap where the register def does not dominate the use.
29803 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
29805 PR rtl-optimization/78127
29806 * lra.c (lra): Call lra_eliminate before finish the loop after
29809 2017-02-16 Richard Biener <rguenther@suse.de>
29811 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
29813 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
29814 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
29815 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
29816 (isl_val_int_from_wi): New function.
29817 (extract_affine_gmp): Rename to ...
29818 (extract_affine_wi): ... this, take a widest_int.
29819 (extract_affine_int): Just wrap extract_affine_wi.
29820 (add_param_constraints): Use isl_val_int_from_wi.
29821 (add_loop_constraints): Likewise, and extract_affine_wi.
29823 2017-02-15 Jeff Law <law@redhat.com>
29825 PR middle-end/79521
29826 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
29827 ira_init_register_move_cost_if_necessary.
29829 2017-02-15 Martin Sebor <msebor@redhat.com>
29831 PR middle-end/32003
29832 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
29833 removed in a prior commit.
29835 2017-02-15 Bin Cheng <bin.cheng@arm.com>
29837 PR tree-optimization/79347
29838 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
29839 counters during peeling.
29841 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
29843 * Makefile.in (site.exp): Remove "set ISLVER".
29845 2017-02-15 Jakub Jelinek <jakub@redhat.com>
29848 * real.c (real_from_integer): Call real_convert even for decimal.
29850 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
29853 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
29855 2017-02-14 Andrew Pinski <apinski@cavium.com>
29857 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
29858 cores and change the partno/implementer to be correct.
29859 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
29860 the 'B" as the implementer.
29861 * config/aarch64/aarch64-tune.md: Regenerate.
29863 2017-02-14 Carl Love <cel@us.ibm.com>
29865 * config/rs6000/rs6000.c: Add case statement entry to make the
29866 xvcvuxdsp built-in argument unsigned.
29867 * config/rs6000/vsx.md: Fix the source and return operand types so they
29868 match the instruction definitions from the ISA document. Fix typo
29869 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
29872 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
29875 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
29876 member early_clobber_alts.
29877 * lra-lives.c (reg_early_clobber_p): New.
29878 (process_bb_lives): Use it.
29879 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
29880 (debug_operand_data): Initialize early_clobber_alts.
29881 (setup_operand_alternative): Set up early_clobber_alts.
29882 (collect_non_operand_hard_regs): Ditto. Pass early clobber
29883 alternatives to new_insn_reg.
29884 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
29886 (lra_update_insn_regno_info): Pass the new arg.
29888 2017-02-14 Jakub Jelinek <jakub@redhat.com>
29890 PR middle-end/79505
29891 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
29892 (new_oacc_loop_raw): Don't clear already cleared fields.
29895 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
29896 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
29897 _mm512_prefetch_i64gather_ps): New inline functions and macros.
29899 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
29902 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
29904 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
29907 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
29908 the extra instruction to the right place to store 128-bit constant
29911 2017-02-14 Martin Sebor <msebor@redhat.com>
29913 PR middle-end/79448
29914 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
29915 warning for strings of unknown length.
29917 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
29919 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
29921 2017-02-14 Jeff Law <law@redhat.com>
29924 * ira-costs.c (scan_one_insn): Initialize register move costs
29925 for pseudos seen in USE/CLOBBER insns.
29927 PR tree-optimization/79095
29928 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
29929 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
29930 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
29931 if the operands are known to be not equal, then the resulting range
29933 (intersect_ranges): If the new range is ~[0,0] and the old range is
29934 wide, then prefer ~[0,0].
29935 * tree-vrp.c (overflow_comparison_p_1): New function.
29936 (overflow_comparison_p): New function.
29937 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
29938 if NAME is used in an overflow test.
29939 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
29940 overflow check that can be expressed as an equality test, then adjust
29941 ops to be that equality test.
29943 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29945 * config/s390/s390-builtin-types.def: Remove flags argument.
29946 * config/s390/s390.c (s390_init_builtins): Likewise.
29948 2017-02-14 Martin Liska <mliska@suse.cz>
29950 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
29951 vector. Fix trailing white spaces.
29953 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
29955 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
29958 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29960 PR rtl-optimization/68664
29961 * config/arm/arm.c (arm_sched_can_speculate_insn):
29962 New function. Declare prototype.
29963 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
29965 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29967 PR rtl-optimization/68664
29968 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
29970 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
29972 2017-02-14 Amit Pawar <amit.pawar@amd.com>
29974 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
29975 max skip bytes for function, loop and jump.
29977 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29979 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
29980 ABS_EXPR for gimple dump.
29982 2017-02-14 Jakub Jelinek <jakub@redhat.com>
29985 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
29987 PR tree-optimization/79408
29988 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
29989 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
29990 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
29991 also if rhs1 is INTEGER_CST.
29993 2017-02-14 Richard Biener <rguenther@suse.de>
29995 PR middle-end/79432
29996 * tree-into-ssa.c (insert_phi_nodes): When the function can
29997 have abnormal edges rewrite SSA names with broken use-def
29998 dominance out of SSA and register them for PHI insertion.
30000 2017-02-13 Martin Sebor <msebor@redhat.com>
30002 PR middle-end/79496
30003 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
30004 clearing info.nowrite flag when snprintf size argument is a range.
30006 2017-02-13 Jakub Jelinek <jakub@redhat.com>
30008 * cprop.c (cprop_jump): Add missing space in string literal.
30009 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
30010 (get_constraint_for_component_ref): Likewise.
30011 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
30012 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
30013 * lra-constraints.c (process_alt_operands): Likewise.
30014 * ipa-inline.c (inline_small_functions): Likewise.
30015 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
30016 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
30017 * trans-mem.c (diagnose_tm_1_op): Likewise.
30018 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
30019 (grid_parallel_clauses_gridifiable): Likewise.
30021 * config/nvptx/mkoffload.c (process): Add space in between
30024 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
30025 "MOD4_SSE_REGS" and "ALL_REGS".
30027 * spellcheck.c (test_data): Add , in between "foo" and "food".
30029 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30032 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
30033 boundary crossing check and subsequent code generation agree.
30035 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30037 * config/aarch64/aarch64.c (has_memory_op): Delete.
30038 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
30041 2017-02-13 Jakub Jelinek <jakub@redhat.com>
30043 PR rtl-optimization/79388
30044 PR rtl-optimization/79450
30045 * combine.c (distribute_notes): When removing TEM_INSN for which
30046 corresponding dest has last value recorded, invalidate that last
30049 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30051 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
30052 of explicit '@'. Add missing assembly comment marker on branch costs
30055 2017-02-13 Nathan Sidwell <nathan@acm.org>
30057 * gengtype-lex.l (<in_struct>): Add '/'.
30059 2017-02-13 Martin Liska <mliska@suse.cz>
30062 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
30064 2017-02-13 Richard Biener <rguenther@suse.de>
30066 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
30068 * configure: Re-generate.
30069 * config.in: Likewise.
30070 * graphite-dependences.c: Simplify as if
30071 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
30072 * graphite-isl-ast-to-gimple.c: Likewise.
30073 * graphite-optimize-isl.c: Likewise.
30074 * graphite-poly.c: Likewise.
30075 * graphite-sese-to-poly.c: Likewise.
30076 * graphite.h: Likewise.
30077 * toplev.c: Include isl/version.h and use isl_version () for
30078 printing the ISL version.
30079 * doc/install.texi: Update ISL requirement.
30081 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30083 * doc/standards.texi (Standards): Update reference to
30086 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30088 * doc/extend.texi (Named Address Spaces): sourceware.org now
30090 * doc/install.texi (Binaries): Ditto.
30093 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30095 * doc/cpp.texi: Replace "stringify"/"stringification" with C
30096 standard terminology "stringize"/"stringizing" throughout.
30097 * doc/cppinternals.texi: Likewise.
30099 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30101 * doc/extend.texi: Fix some spelling mistakes and typos.
30102 * doc/invoke.texi: Likewise.
30104 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30107 * params.def (inline-min-speedup) Change from 10 to 8.
30109 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30111 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
30114 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30117 * ipa-inline-analysis.c (get_minimal_bb): New function.
30118 (record_modified): Use it.
30119 (remap_edge_change_prob): Handle also ancestor functions.
30121 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
30123 * doc/contrib.texi (Contributors): Remove broken link into
30124 the Mauve CVS repository.
30126 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30128 PR middle-end/79454
30129 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
30130 result computation whenever lhs doesn't have vector mode, not
30131 just when it has BLKmode.
30133 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30135 * doc/makefile.texi (profiledbootstrap): Refer to the
30136 installation instructions only in textual form.
30138 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30141 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
30143 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30145 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
30146 (Specific): Update mingw-w64 reference.
30148 (Specific): Remove broken link to Renesas RX processor.
30150 2017-02-10 Richard Biener <rguenther@suse.de>
30152 * toplev.c (process_options): Do not mention obsolete graphite
30153 options when printing sorry message about missing graphite support.
30154 Mention -floop-nest-optimize.
30156 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
30158 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
30159 (vtst_p16): Likewise.
30160 (vtstq_p8): Likewise.
30161 (vtstq_p16): Likewise.
30163 (vtstq_p64): Likewise.
30164 * config/arm/arm_neon.h (vgetq_lane_p64): New.
30165 (vset_lane_p64): New.
30166 (vsetq_lane_p64): New.
30168 2017-02-10 Jakub Jelinek <jakub@redhat.com>
30170 PR tree-optimization/79411
30171 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
30172 stmt operands are SSA_NAMEs used in abnormal phis.
30173 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
30176 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30179 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
30182 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30184 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
30186 2017-02-09 Jakub Jelinek <jakub@redhat.com>
30188 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
30192 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
30193 not arbitrary TREE_CONSTANT.
30196 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
30197 "omp declare target link" attribute unless is_global_var.
30198 * omp-offload.c (find_link_var_op): Likewise.
30200 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
30201 Chung-Lin Tang <cltang@codesourcery.com>
30203 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
30205 (gimplify_adjust_omp_clauses): Don't delete TILE.
30206 (gimplify_omp_for): Deal with TILE.
30207 * internal-fn.c (expand_GOACC_TILE): New function.
30208 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
30210 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
30211 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
30213 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
30214 avoid DIV for outermost collapse var.
30215 (expand_oacc_for): Insert tile element loop as needed. Adjust.
30216 Remove out of date comments, fix whitespace.
30217 * omp-general.c (omp_extract_for_data): Deal with tiling.
30218 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
30219 adjust OLF_DIM_BASE value.
30220 (struct omp_for_data): Add tiling field.
30221 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
30222 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
30223 for auto loops. Remove default auto determining, moved to
30224 oacc_loop_fixed_partitions.
30225 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
30226 stmts, add e_mask field.
30227 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
30228 (oacc_thread_numbers): Use oacc_dim_call.
30229 (oacc_xform_tile): New.
30230 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
30231 (finish_oacc_loop): Adjust for ifns vector.
30232 (oacc_loop_discover_walk): Append loop abstraction sites to list,
30233 add case for GOACC_TILE fns.
30234 (oacc_loop_xform_loop): Delete.
30235 (oacc_loop_process): Iterate over call list directly, and add
30236 handling for GOACC_TILE fns.
30237 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
30239 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
30240 vector partitioning to outer loops. Assign 2 partitions to loops
30241 when available. Add TILE handling.
30242 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
30243 (execite_oacc_device_lower): Process GOACC_TILE fns,
30244 ignore unknown specs.
30245 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
30246 * tree.c (omp_clause_num_ops): Adjust TILE ops.
30247 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
30249 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
30251 * configure.ac (ACX_BUGURL): Update.
30252 * configure: Regenerate.
30254 2017-02-09 Richard Biener <rguenther@suse.de>
30256 PR tree-optimization/69823
30257 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
30258 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
30260 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30262 * config/arc/arc-c.def: Add __NPS400__ definition.
30263 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
30264 (TARGET_NPS400): Define.
30266 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30268 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
30270 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
30271 pointer, arch_info.
30272 (arc_cpu_types): Fill the arch_info field with a pointer into the
30273 arc_arch_types table.
30274 (arc_selected_cpu): Declare.
30275 * config/arc/arc.c (arc_selected_cpu): Make global.
30276 (arc_selected_arch): Delete.
30277 (arc_base_cpu): Delete.
30278 (arc_override_options): Remove references to deleted variables,
30279 update access to arch information.
30280 (ARC_OPT): Update access to arch information.
30281 (ARC_OPTX): Likewise.
30282 * config/arc/arc.h (arc_base_cpu): Remove declaration.
30283 (TARGET_ARC600): Update access to arch information.
30284 (TARGET_ARC601): Likewise.
30285 (TARGET_ARC700): Likewise.
30286 (TARGET_EM): Likewise.
30287 (TARGET_HS): Likewise.
30288 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
30291 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
30294 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
30295 condition/operands for integer GE/LE/GEU/LEU operations.
30297 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
30299 PR translation/79397
30300 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
30303 2017-02-08 Martin Jambor <mjambor@suse.cz>
30306 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
30307 whether allocation happened.
30308 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
30309 nothing was allocated.
30311 2017-02-08 Jakub Jelinek <jakub@redhat.com>
30313 PR tree-optimization/79408
30314 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
30315 constant, but SSA_NAME with a known integer range, use the minimum
30316 of that range instead of op1 to determine if modulo can be replaced
30317 with its first operand.
30319 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30321 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
30323 2017-02-08 Richard Biener <rguenther@suse.de>
30325 PR tree-optimization/71824
30326 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
30327 Check all loops contained in the merged region.
30329 2017-02-07 Andrew Pinski <apinski@cavium.com>
30331 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
30333 2017-02-07 Andrew Pinski <apinski@cavium.com>
30335 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
30336 (thunderxt88): Likewise.
30337 (thunderxt81): Disable LSE and change v8.1 to v8.
30338 (thunderxt83): Likewise.
30340 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30341 Richard Biener <rguenther@suse.de>
30343 PR middle-end/79399
30344 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
30345 type from int to size_t.
30346 * ira-costs.c (struct_costs_size): Change type from int to size_t.
30348 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30350 PR rtl-optimization/79386
30351 * cprop.c (bypass_conditional_jumps): Initialize
30352 bypass_last_basic_block already before splitting bbs after
30353 unconditional traps...
30354 (bypass_conditional_jumps): ... rather than here.
30357 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
30358 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
30359 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
30360 fix -masm=intel patterns.
30362 2017-02-07 Richard Biener <rguenther@suse.de>
30364 PR tree-optimization/79256
30365 PR middle-end/79278
30366 * builtins.c (get_object_alignment_2): Use min_align_of_type
30367 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
30368 and ADJUST_FIELD_ALIGN.
30370 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
30372 * doc/tm.texi: Regenerate.
30373 * stor-layout.c (layout_decl): Adjust.
30374 (update_alignment_for_field): Likewise.
30375 (place_field): Likewise.
30376 (min_align_of_type): Likewise.
30377 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
30378 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
30379 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
30380 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
30381 * config/frv/frv.c (frv_adjust_field_align): Likewise.
30382 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
30383 * config/i386/i386.c (x86_field_alignment): Likewise.
30384 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
30385 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
30386 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
30387 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
30388 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
30389 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
30393 2017-01-30 Richard Biener <rguenther@suse.de>
30395 PR tree-optimization/79256
30396 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
30397 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
30400 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
30402 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
30403 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
30404 builtins to SImode and emit a zero-extend, if necessary.
30406 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30408 * docs/invoke.texi (RISC-V Options): Alphabetize.
30410 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30412 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
30415 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30417 * config/riscv/riscv.c: New file.
30418 * common/config/riscv/riscv-common.c: Likewise.
30419 * config.gcc: Likewise.
30420 * config/riscv/constraints.md: Likewise.
30421 * config/riscv/elf.h: Likewise.
30422 * config/riscv/generic.md: Likewise.
30423 * config/riscv/linux.h: Likewise.
30424 * config/riscv/multilib-generator: Likewise.
30425 * config/riscv/peephole.md: Likewise.
30426 * config/riscv/pic.md: Likewise.
30427 * config/riscv/predicates.md: Likewise.
30428 * config/riscv/riscv-builtins.c: Likewise.
30429 * config/riscv/riscv-c.c: Likewise.
30430 * config/riscv/riscv-ftypes.def: Likewise.
30431 * config/riscv/riscv-modes.def: Likewise.
30432 * config/riscv/riscv-opts.h: Likewise.
30433 * config/riscv/riscv-protos.h: Likewise.
30434 * config/riscv/riscv.h: Likewise.
30435 * config/riscv/riscv.md: Likewise.
30436 * config/riscv/riscv.opt: Likewise.
30437 * config/riscv/sync.md: Likewise.
30438 * config/riscv/t-elf-multilib: Likewise.
30439 * config/riscv/t-linux: Likewise.
30440 * config/riscv/t-linux-multilib: Likewise.
30441 * config/riscv/t-riscv: Likewise.
30442 * configure.ac: Likewise.
30443 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
30444 Waterman as RISC-V maintainers.
30445 * doc/install.texi: Add RISC-V entries.
30446 * doc/invoke.texi: Add RISC-V options section.
30447 * doc/md.texi: Add RISC-V constraints section.
30448 * configure: Regenerated.
30450 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
30453 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
30454 false values to be constant vectors with all 0 or all 1 bits set.
30455 (vcondu<mode><mode>): Likewise.
30456 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
30458 (fpmask_comparison_operator): Update comment.
30459 (vecint_comparison_operator): New predicate.
30460 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
30461 vector conditionals when the true and false values are constant
30462 vectors with all 0 bits or all 1 bits set.
30464 2017-02-06 Martin Sebor <msebor@redhat.com>
30466 PR tree-optimization/79376
30467 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
30469 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
30471 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
30472 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
30473 to simplify split condition.
30475 2017-02-06 Jakub Jelinek <jakub@redhat.com>
30477 * omp-expand.c (oxpand_omp_atomic_fetch_op,
30478 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
30481 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
30483 PR rtl-optimization/68664
30484 * target.def (can_speculate_insn): New hook.
30485 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
30486 * doc/tm.texi: Regenerate.
30487 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
30488 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
30489 (rs6000_sched_can_speculate_insn): New function.
30491 2017-02-06 Jakub Jelinek <jakub@redhat.com>
30493 PR tree-optimization/79284
30494 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
30495 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
30496 vectorizable_mask_load_store, vectorizable_operation,
30497 vect_is_simple_cond, get_same_sized_vectype): Use it instead
30498 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
30499 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
30500 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
30501 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
30502 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
30503 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
30504 is_gimple_assign (stmt). Replace another such test with
30505 is_gimple_assign (stmt).
30507 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
30510 * config/avr/avr.c (rtl-iter.h): Include it.
30511 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
30512 (avr_legitimate_combined_insn): ...and implementation.
30514 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
30516 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
30517 * config/s390/s390.c (s390_const_operand_ok)
30518 (s390_canonicalize_comparison, s390_extract_part)
30519 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
30520 (s390_contiguous_bitmask_p, s390_rtx_costs)
30521 (legitimize_pic_address): Likewise.
30522 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
30523 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
30524 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
30525 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
30526 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
30528 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
30530 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
30531 REGNO($0) == REGNO($1).
30533 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30535 * config/s390/linux.h(SIZE_TYPE): Add comment.
30537 2017-02-06 Julian Brown <julian@codesourcery.com>
30538 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30539 Virendra Pathak <virendra.pathak@broadcom.com>
30541 * config/aarch64/aarch64-cores.def: Change the scheduler
30543 * config/aarch64/aarch64.md: Include thunderx2t99.md.
30544 * config/aarch64/thunderx2t99.md: New file.
30546 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30548 * doc/standards.texi (Go Language): Update link to language
30551 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30553 * tree-eh.c (lower_resx): Sanitize profile.
30554 (cleanup_empty_eh_move_lp): Likewise.
30556 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30559 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
30561 * cfgloopmanip.h (loop_version): Update prototype.
30562 * modulo-sched.c (sms_schedule): Update call of loop_version.
30563 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
30564 * tree-parloops.c (gen_parallel_loop): Likewise.
30565 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
30566 * tree-ssa-loop-split.c (split_loop): Likewise.
30567 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
30568 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
30570 2017-02-05 Martin Liska <mliska@suse.cz>
30573 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
30575 (print_operand_address): Initialize a struct to zero.
30577 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30579 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
30580 garbage collector only in textual form.
30582 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30584 * doc/extend.texi (x86 specific memory model extensions for
30585 transactional memory): Simplify a phrase.
30587 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
30590 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
30591 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
30592 (atomic_storedi_1): Likewise.
30594 2017-02-04 Jakub Jelinek <jakub@redhat.com>
30596 PR tree-optimization/79338
30597 * tree-parloops.c (gather_scalar_reductions): Don't call
30598 vect_analyze_loop_form for loop->inner before destroying loop's
30601 2017-02-03 Martin Sebor <msebor@redhat.com>
30603 PR tree-optimization/79327
30604 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
30605 when precision has resulted in leading zeros.
30606 (format_integer): Adjust the likely counter to assume an unknown
30607 argument that may be zero is non-zero.
30609 2017-02-03 Jason Merrill <jason@redhat.com>
30612 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
30613 avoid copying non-taken branch.
30615 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30617 PR tree-optimization/79340
30618 * tree-vect-loop.c (vectorizable_reduction): Release
30619 vec_defs elements after safe_splicing them into other vectors.
30622 PR tree-optimization/79327
30623 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
30624 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
30626 (format_integer): Use wide_int_to_tree instead of build_int_cst
30627 + to_?hwi. If argmin is NULL, just set argmin and argmax to
30628 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
30629 of shortest and longest sequence.
30631 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
30633 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
30634 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
30636 2017-02-03 Walter Lee <walt@tilera.com>
30639 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
30640 after initial stackframe link reg save.
30641 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
30643 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30646 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
30647 wu for stxssp alternative.
30649 2017-02-03 Martin Sebor <msebor@redhat.com>
30651 PR tree-optimization/79352
30652 * gimple-fold.c (get_range_strlen): Add argument.
30653 (get_range_strlen): Change return type to bool.
30654 (get_maxval_strlen): Pass in a dummy argument.
30655 * gimple-fold.h (get_range_strlen): Change return type to bool.
30656 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
30657 * tree.h (array_at_struct_end_p): Add argument.
30658 * tree.c (array_at_struct_end_p): Handle it.
30660 2017-02-03 Martin Liska <mliska@suse.cz>
30663 * multiple_target.c (create_dispatcher_calls): Redirect edge
30664 from a caller of a dispatcher.
30665 (expand_target_clones): Make the clones local.
30666 (ipa_target_clone): Do both target clones and resolvers.
30667 (ipa_dispatcher_calls): Remove the pass.
30668 (pass_dispatcher_calls::gate): Likewise.
30669 (make_pass_dispatcher_calls): Likewise.
30670 * passes.def (pass_target_clone): Put as very first IPA early
30673 2017-02-03 Martin Liska <mliska@suse.cz>
30675 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
30676 in case of a function with ifunc attribute.
30678 2017-02-03 Martin Liska <mliska@suse.cz>
30680 * cgraph.c (cgraph_node::dump): Dump function version info.
30681 * symtab.c (symtab_node::dump_base): Add missing new line.
30683 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
30685 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
30686 (ifcombine_ifandif): Use it.
30688 2017-02-03 Martin Liska <mliska@suse.cz>
30690 * doc/invoke.texi: Document default value for
30691 use-after-scope-direct-emission-threshold.
30693 2017-02-03 Martin Liska <mliska@suse.cz>
30695 PR tree-optimization/79339
30696 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
30697 (format_floating): Likewise.
30699 2017-02-03 Martin Liska <mliska@suse.cz>
30702 * ipa-prop.c (ipa_node_params_t::insert): Remove current
30704 (ipa_node_params_t::remove): Likewise.
30705 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
30706 initialization from removed ipa_node_params_t::insert.
30707 (ipa_node_params::~ipa_node_params): Move from removed
30708 ipa_node_params_t::release.
30709 * symbol-summary.h (symbol_summary::m_released): New member.
30710 Do not release a summary twice. Do not allow to call finalizer
30711 for types of a summary that live in GGC memory.
30713 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30715 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
30718 2017-02-02 Martin Sebor <msebor@redhat.com>
30720 PR middle-end/79275
30721 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
30722 (format_string): Tighten up the range of output for non-constant
30723 strings and correct the expected range for wide non-constant strings.
30725 2017-02-02 Martin Sebor <msebor@redhat.com>
30727 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
30729 PR middle-end/32003
30730 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
30732 (-fdump-tree-@var): Add to index and document how to come up
30733 with pass-specific option and dump file names.
30734 (-fdump-passes): Clarify where to look for output.
30736 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
30738 PR middle-end/77445
30739 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
30740 statistics of the analyzed path; allow threading for speed when
30741 any of BBs along the path are optimized for speed.
30743 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
30745 PR middle-end/78468
30746 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
30747 settings of the virtual registers.
30750 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
30752 * explow.c (get_dynamic_stack_size): Take known alignment of stack
30753 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
30756 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30758 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
30759 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
30761 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30763 * config/s390/s390.md: Add missing comments with the expanded
30765 * config/s390/vector.md: Likewise.
30766 * config/s390/vx-builtins.md: Likewise.
30768 2017-02-02 Jakub Jelinek <jakub@redhat.com>
30771 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
30772 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
30773 conditions on a single line.
30775 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30777 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
30778 __S390_VX__ to __VX__.
30780 2017-02-01 Andrew Pinski <apinski@cavium.com>
30782 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
30783 stmt_info to record_stmt_cost.
30784 (vect_get_known_peeling_cost): Pass stmt_info if known to
30786 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
30787 cpu_vector_cost field into
30788 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
30789 field into vec_int_stmt_cost and vec_fp_stmt_cost.
30790 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
30791 splitting of scalar_stmt_cost and vec_stmt_cost.
30792 (thunderx_vector_cost): Likewise.
30793 (cortexa57_vector_cost): LIkewise.
30794 (exynosm1_vector_cost): Likewise.
30795 (xgene1_vector_cost): Likewise.
30796 (thunderx2t99_vector_cost): Improve after the splitting of the two
30798 (aarch64_builtin_vectorization_cost): Update for the splitting of
30799 scalar_stmt_cost and vec_stmt_cost.
30801 2017-02-01 Torvald Riegel <triegel@redhat.com>
30802 Richard Henderson <rth@redhat.com>
30804 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
30805 conditional on existance of a fast atomic load.
30806 * optabs-query.c (can_atomic_load_p): New function.
30807 * optabs-query.h (can_atomic_load_p): Declare it.
30808 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
30809 no fast atomic load is available for the particular size of access.
30810 (expand_atomic_compare_and_swap): Likewise.
30811 (expand_atomic_load): Likewise.
30812 (expand_atomic_store): Likewise.
30813 (expand_atomic_fetch_op): Likewise.
30814 * testsuite/lib/target-supports.exp
30815 (check_effective_target_sync_int_128): Remove x86 because it provides
30816 no fast atomic load.
30817 (check_effective_target_sync_int_128_runtime): Likewise.
30819 2017-02-01 Richard Biener <rguenther@suse.de>
30821 * graphite.c: Include tree-vectorizer.h for find_loop_location.
30822 (graphite_transform_loops): Provide opt-info for optimized nests.
30823 * tree-parloop.c (parallelize_loops): Provide opt-info for
30824 parallelized loops.
30826 2017-02-01 Richard Biener <rguenther@suse.de>
30828 PR middle-end/79315
30829 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
30830 was not set before.
30832 2017-02-01 Richard Biener <rguenther@suse.de>
30834 PR tree-optimization/71824
30835 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
30836 Verify the loops are valid in the merged SESE region.
30837 (scop_detection::can_represent_loop_1): Check analyzing the
30838 evolution of the number of iterations in the region succeeds.
30840 2017-01-31 Ian Lance Taylor <iant@golang.org>
30842 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
30843 REG_ARGS_SIZE note to 32-bit push insns and call insn.
30845 2017-01-31 David Malcolm <dmalcolm@redhat.com>
30847 PR preprocessor/79210
30848 * input.c (get_substring_ranges_for_loc): Replace line_width
30849 assertion with error-handling.
30851 2017-01-31 Richard Biener <rguenther@suse.de>
30853 PR tree-optimization/77318
30854 * graphite-sese-to-poly.c (extract_affine): Fix assert.
30855 (create_pw_aff_from_tree): Take loop parameter.
30856 (add_condition_to_pbb): Pass loop of the condition to
30857 create_pw_aff_from_tree.
30859 2017-01-31 Jakub Jelinek <jakub@redhat.com>
30861 * config/s390/s390.c (s390_asan_shadow_offset): New function.
30862 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
30864 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
30868 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
30870 (convert_int_to_float128): Likewise.
30871 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
30872 (convert_int_to_float128): Likewise.
30873 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
30874 (UNSPEC_IEEE128_CONVERT): Likewise.
30875 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
30876 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
30877 Use local variables for IBM extended format.
30878 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
30879 (fix_trunc<mode>si2_fprs): Likewise.
30880 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
30881 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
30882 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
30883 to know that we can now have integers of all sizes in vector
30885 (fix<uns>_<mode>di2_hw): Likewise.
30886 (float<uns>_<mode>si2_hw): Likewise.
30887 (fix_<mode>si2_hw): Likewise.
30888 (fixuns_<mode>si2_hw): Likewise.
30889 (float<uns>_<mode>di2_hw): Likewise.
30890 (float_<mode>di2_hw): Likewise.
30891 (float_<mode>si2_hw): Likewise.
30892 (floatuns_<mode>di2_hw): Likewise.
30893 (floatuns_<mode>si2_hw): Likewise.
30894 (xscvqp<su>wz_<mode>): Delete, no longer used.
30895 (xscvqp<su>dz_<mode>): Likewise.
30896 (xscv<su>dqp_<mode>): Likewise.
30897 (ieee128_mfvsrd_64bit): Likewise.
30898 (ieee128_mfvsrd_32bit): Likewise.
30899 (ieee128_mfvsrwz): Likewise.
30900 (ieee128_mtvsrw): Likewise.
30901 (ieee128_mtvsrd_64bit): Likewise.
30902 (ieee128_mtvsrd_32bit): Likewise.
30904 2017-01-31 Martin Liska <mliska@suse.cz>
30907 * ipa-prop.c (ipa_free_all_node_params): Call release method
30908 instead of ~sumbol_summary to not to trigger double times
30911 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
30913 PR tree-optimization/71691
30914 * bitmap.h (class auto_bitmap): New.
30915 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
30916 is_maybe_undefined instead of ssa_undefined_value_p.
30918 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30920 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
30921 __S390_ARCH_LEVEL__ to __ARCH__.
30923 2017-01-31 Jakub Jelinek <jakub@redhat.com>
30925 PR tree-optimization/79267
30926 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
30927 if should_remove_lhs_p is true.
30929 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
30932 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
30933 (add_alignment_attribute): New.
30934 (base_type_die): Add alignment attribute.
30935 (subrange_type_die): Likewise.
30936 (modified_type_die): Likewise.
30937 (gen_array_type_die): Likewise.
30938 (gen_descr_array_type_die: Likewise.
30939 (gen_enumeration_type_die): Likewise.
30940 (gen_subprogram_die): Likewise.
30941 (gen_variable_die): Likewise.
30942 (gen_field_die): Likewise.
30943 (gen_ptr_to_mbr_type_die): Likewise.
30944 (gen_struct_or_union_type_die): Likewise.
30945 (gen_subroutine_type_die): Likewise.
30946 (gen_typedef_die): Likewise.
30947 (base_type_cmp): Compare alignment attribute.
30949 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30952 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
30953 (setb_unsigned) New pattern for setb with CCUNS.
30954 * config/rs6000/rs6000.c (expand_block_compare): Use a different
30955 subfc./subfe sequence to avoid overflow problems. Generate a
30956 shorter sequence with cmpld/setb for power9.
30957 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
30958 for generating subfc. instruction.
30959 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
30960 now uses this instruction.
30962 2017-01-30 Ian Lance Taylor <iant@google.com>
30965 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
30966 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
30968 2017-01-30 Martin Sebor <msebor@redhat.com>
30970 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
30971 Move constant to the right of a relational operator.
30972 (get_mpfr_format_length, format_character, format_string): Ditto.
30973 (should_warn_p, maybe_warn): Same.
30975 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
30977 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
30980 * asan.c (get_translation_unit_decl): Remove function.
30981 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
30983 2017-01-30 Martin Liska <mliska@suse.cz>
30985 PR gcov-profile/79259
30986 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
30987 -fprofile-generate.
30989 2017-01-30 Martin Liska <mliska@suse.cz>
30992 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
30993 Initialize variables with NULL value.
30995 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
30998 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
31000 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
31002 2017-01-30 Richard Biener <rguenther@suse.de>
31004 PR tree-optimization/79276
31005 * tree-vrp.c (process_assert_insertions): Properly adjust common
31006 when removing a duplicate.
31008 2017-01-30 Richard Biener <rguenther@suse.de>
31010 PR tree-optimization/79256
31011 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
31012 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
31014 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
31016 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
31019 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
31020 ("*r<noxa>sbg_<mode>_sll_bitmask")
31021 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
31022 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
31023 Use contiguous_bitmask_nowrap_operand.
31025 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31028 * config/rs6000/altivec.h (vec_xl): Revise #define.
31029 (vec_xst): Likewise.
31031 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
31033 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
31035 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
31037 PR rtl-optimization/79194
31038 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
31039 traps before call to bypass_conditional_jumps.
31041 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31043 PR tree-optimization/71374
31044 * lra-constraints.c (check_conflict_input_operands): New.
31045 (match_reload): Use it.
31047 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31050 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
31051 account to calculate conflict_set.
31053 2017-01-27 Bin Cheng <bin.cheng@arm.com>
31055 PR rtl-optimization/78559
31056 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
31057 other_insn in combine.
31059 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31061 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
31062 uint16_type_node for BT_UINT16.
31064 2017-01-27 David Malcolm <dmalcolm@redhat.com>
31066 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
31067 "RTL Tests" to menu.
31068 (GIMPLE Tests): New node.
31069 (RTL Tests): New node.
31071 2017-01-27 Richard Biener <rguenther@suse.de>
31073 PR tree-optimization/79245
31074 * tree-loop-distribution.c (distribute_loop): Apply cost
31075 modeling also to detected patterns.
31077 2017-01-27 Richard Biener <rguenther@suse.de>
31079 PR tree-optimization/71433
31080 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
31081 (compare_assert_loc): New function.
31082 (process_assert_insertions): Sort and optimize assert locations
31083 to remove duplicates and push down identical assertions on
31084 edges to their destination block.
31086 2017-01-27 Richard Biener <rguenther@suse.de>
31088 PR tree-optimization/79244
31089 * tree-vrp.c (remove_range_assertions): Forcefully propagate
31090 out SSA names even if abnormal.
31092 2017-01-27 Jakub Jelinek <jakub@redhat.com>
31094 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
31095 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
31096 instead of MPFR_RNDN.
31098 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
31101 * arm.c (arm_option_override): Don't call build_target_option_node
31102 until after doing all option overrides.
31103 (arm_valid_target_attribute_tree): Likewise.
31105 2017-01-27 Martin Liska <mliska@suse.cz>
31107 * doc/invoke.texi (-fprofile-arcs): Document profiling support
31108 for {cd}tors and C++ {cd}tors.
31110 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
31112 * config/s390/s390.md ("*setmem_long_and")
31113 ("*setmem_long_and_31z"): Use zero_extend instead of and.
31115 2017-01-26 Martin Sebor <msebor@redhat.com>
31117 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
31120 2017-01-26 Martin Sebor <msebor@redhat.com>
31122 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
31123 HAVE_DFmode before using XFmode or DFmode.
31124 (parse_directive): Avoid using the z length modifier to avoid
31125 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
31127 PR middle-end/78703
31128 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
31129 to accept adjustment as an array.
31130 (get_int_range): New function.
31131 (struct directive): Make width and prec arrays.
31132 (directive::set_width, directive::set_precision): Call get_int_range.
31133 (format_integer, format_floating): Handle width and precision ranges.
31134 (format_string, parse_directive): Same.
31136 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31139 * dwarf2out.c (generate_skeleton_bottom_up): For children with
31140 comdat_type_p set, just clone them, but keep the children in the
31144 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
31145 which have direct callers with -fvar-tracking-assignments enabled
31147 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
31148 inside of type units.
31150 2017-01-26 Martin Sebor <msebor@redhat.com>
31152 PR middle-end/78703
31153 * gimple-ssa-sprintf.c (struct result_range): Add likely and
31155 (struct format_result): Replace number_chars, number_chars_min,
31156 and number_chars_max with a single member of struct result_range.
31158 (format_result::operator+=): Adjust.
31159 (struct fmtresult): Remove bounded. Handle likely and unlikely
31161 (fmtresult::adjust_for_width_or_precision): New function.
31162 (fmtresult:type_max_digits): New function.
31163 (bytes_remaining): Handle likely and unlikely counters.
31164 (min_bytes_remaining): Remove.
31165 (format_percent): Simplify.
31166 (format_integer, format_floating): Set likely and unlikely counters.
31167 (get_string_length, format_character, format_string): Same.
31168 (format_plain, should_warn_p): New function.
31169 (maybe_warn): Call should_warn_p. Update diagnostic messages
31170 and handle those for all directives, including plain strings.
31171 (format_directive): Handle likely and unlikely counters.
31172 Remove unnecessary quoting from diagnostics. Add an informational
31174 (add_bytes): Remove.
31175 (pass_sprintf_length::compute_format_length): Simplify.
31176 (try_substitute_return_value): Handle likely and unlikely counters.
31178 2017-01-26 Carl Love <cel@us.ibm.com>
31180 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
31181 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
31183 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
31186 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
31187 endianess for subregs into account.
31188 * lra-constraints.c (lra_constraints): Do risky transformations
31189 always on the first iteration.
31190 * lra-lives.c (check_pseudos_live_through_calls): Add arg
31191 last_call_used_reg_set.
31192 (process_bb_lives): Define and use last_call_used_reg_set.
31193 * lra.c (lra): Always continue after lra_constraints on the first
31196 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
31198 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
31200 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
31202 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31204 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
31205 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
31206 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
31207 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
31208 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
31209 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
31210 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
31211 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
31212 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
31214 2017-01-26 Marek Polacek <polacek@redhat.com>
31217 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
31218 for the third operand.
31220 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31222 PR middle-end/79236
31223 * omp-low.c (struct omp_context): Add simt_stmt field.
31224 (scan_omp_for): Return omp_context *.
31225 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
31226 context to the _simt_ SIMD stmt.
31227 (lower_omp_for): For combined SIMD with sibling _simt_
31228 SIMD, make sure to use the same decls in _looptemp_
31229 clauses as in the sibling.
31231 2017-01-26 David Sherwood <david.sherwood@arm.com>
31233 PR middle-end/79212
31234 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
31237 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31240 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
31241 emit fld b; fld a; if possible.
31243 * brig-builtins.def: Update copyright years.
31244 * config/arm/arm_acle_builtins.def: Update copyright years.
31246 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
31249 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
31250 constraint instead of o for the stxsd instruction.
31252 2017-01-25 Carl Love <cel@us.ibm.com>
31254 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
31255 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
31257 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
31259 * doc/invoke.texi (C++ Dialect Options): Fix typo.
31261 2017-01-25 Richard Biener <rguenther@suse.de>
31263 PR tree-optimization/69264
31264 * target.def (vector_alignment_reachable): Improve documentation.
31265 * doc/tm.texi: Regenerate.
31266 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
31268 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
31269 earlier changes with respect to TYPE_USER_ALIGN.
31270 (vector_alignment_reachable_p): Likewise. Improve dumping.
31272 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31275 * config/arm/arm.md (xordi3): Force constant operand into a register
31278 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31280 * doc/invoke.texi (-fstore-merging): Correct default optimization
31281 levels at which it is enabled.
31282 (-O): Move -fstore-merging from list to...
31285 2017-01-25 Richard Biener <rguenther@suse.de>
31288 * omp-expand.c: Include debug.h.
31289 (expand_omp_taskreg): Make sure to generate early debug before
31290 outlining anything from a function.
31291 (expand_omp_target): Likewise.
31292 (grid_expand_target_grid_body): Likewise.
31294 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
31297 * asan.c (get_translation_unit_decl): New function.
31298 (asan_add_global): Extract modules file name from globals
31299 TRANSLATION_UNIT_DECL name.
31301 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
31304 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
31305 for long calls with APCS frame and VFP.
31307 2017-01-24 David Malcolm <dmalcolm@redhat.com>
31309 * cfg.c (original_copy_tables_initialized_p): New function.
31310 * cfg.h (original_copy_tables_initialized_p): New decl.
31311 * cfgrtl.c (relink_block_chain): Guard the call to
31312 free_original_copy_tables with a call to
31313 original_copy_tables_initialized_p.
31314 * cgraph.h (symtab_node::native_rtl_p): New decl.
31315 * cgraphunit.c (symtab_node::native_rtl_p): New function.
31316 (symtab_node::needed_p): Don't assert for early assembly output
31317 for __RTL functions.
31318 (cgraph_node::finalize_function): Set "force_output" for __RTL
31320 (cgraph_node::analyze): Bail out early for __RTL functions.
31321 (analyze_functions): Update assertion to support __RTL functions.
31322 (cgraph_node::expand): Bail out early for __RTL functions.
31323 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
31325 * function.h (struct function): Update comment for field
31327 * gimple-expr.c: Include "tree-pass.h".
31328 (gimple_has_body_p): Return false for __RTL functions.
31329 * Makefile.in (OBJS): Add run-rtl-passes.o.
31330 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
31332 (gcc::pass_manager::get_clean_slate): New accessor.
31333 * passes.c: Include "insn-addr.h".
31334 (should_skip_pass_p): Add logging. Update logic for running
31335 "expand" to be compatible with both __GIMPLE and __RTL. Guard
31336 property-provider override so it is only done for gimple passes.
31338 (skip_pass): New function.
31339 (execute_one_pass): Call skip_pass when skipping passes.
31340 * read-md.c (md_reader::read_char): Support filtering
31341 the input to a subset of line numbers.
31342 (md_reader::md_reader): Initialize fields
31343 m_first_line and m_last_line.
31344 (md_reader::read_file_fragment): New function.
31345 * read-md.h (md_reader::read_file_fragment): New decl.
31346 (md_reader::m_first_line): New field.
31347 (md_reader::m_last_line): New field.
31348 * read-rtl-function.c (function_reader::create_function): Only
31349 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
31350 curr_properties. Set DECL_INITIAL to a dummy block.
31351 (read_rtl_function_body_from_file_range): New function.
31352 * read-rtl-function.h (read_rtl_function_body_from_file_range):
31354 * run-rtl-passes.c: New file.
31355 * run-rtl-passes.h: New file.
31357 2017-01-24 Jeff Law <law@redhat.com>
31359 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
31362 2017-01-24 Bin Cheng <bin.cheng@arm.com>
31364 PR tree-optimization/79159
31365 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
31366 (record_nonwrapping_iv): Improve boundary using above function if no
31367 value range information.
31369 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
31370 Martin Jambor <mjambor@suse.cz>
31372 * brig-builtins.def: New file.
31373 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
31374 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
31375 (DEF_HSAIL_SAT_BUILTIN): Likewise.
31376 (DEF_HSAIL_INTR_BUILTIN): Likewise.
31377 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
31378 * builtin-types.def (BT_INT8): New.
31379 (BT_INT16): Likewise.
31380 (BT_UINT8): Likewise.
31381 (BT_UINT16): Likewise.
31382 (BT_FN_ULONG): Likewise.
31383 (BT_FN_UINT_INT): Likewise.
31384 (BT_FN_UINT_ULONG): Likewise.
31385 (BT_FN_UINT_LONG): Likewise.
31386 (BT_FN_UINT_PTR): Likewise.
31387 (BT_FN_ULONG_PTR): Likewise.
31388 (BT_FN_INT8_FLOAT): Likewise.
31389 (BT_FN_INT16_FLOAT): Likewise.
31390 (BT_FN_UINT32_FLOAT): Likewise.
31391 (BT_FN_UINT16_FLOAT): Likewise.
31392 (BT_FN_UINT8_FLOAT): Likewise.
31393 (BT_FN_UINT64_FLOAT): Likewise.
31394 (BT_FN_UINT16_UINT32): Likewise.
31395 (BT_FN_UINT32_UINT16): Likewise.
31396 (BT_FN_UINT16_UINT16_UINT16): Likewise.
31397 (BT_FN_INT_PTR_INT): Likewise.
31398 (BT_FN_UINT_PTR_UINT): Likewise.
31399 (BT_FN_LONG_PTR_LONG): Likewise.
31400 (BT_FN_ULONG_PTR_ULONG): Likewise.
31401 (BT_FN_VOID_UINT64_UINT64): Likewise.
31402 (BT_FN_UINT8_UINT8_UINT8): Likewise.
31403 (BT_FN_INT8_INT8_INT8): Likewise.
31404 (BT_FN_INT16_INT16_INT16): Likewise.
31405 (BT_FN_INT_INT_INT): Likewise.
31406 (BT_FN_UINT_FLOAT_UINT): Likewise.
31407 (BT_FN_FLOAT_UINT_UINT): Likewise.
31408 (BT_FN_ULONG_UINT_UINT): Likewise.
31409 (BT_FN_ULONG_UINT_PTR): Likewise.
31410 (BT_FN_ULONG_ULONG_ULONG): Likewise.
31411 (BT_FN_UINT_UINT_UINT): Likewise.
31412 (BT_FN_VOID_UINT_PTR): Likewise.
31413 (BT_FN_UINT_UINT_PTR: Likewise.
31414 (BT_FN_UINT32_UINT64_PTR): Likewise.
31415 (BT_FN_INT_INT_UINT_UINT): Likewise.
31416 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
31417 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
31418 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
31419 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
31420 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
31421 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
31422 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
31423 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
31424 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
31425 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
31426 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
31427 * doc/frontends.texi: List BRIG FE.
31428 * doc/install.texi (Testing): Add BRIG tesring requirements.
31429 * doc/invoke.texi (Overall Options): Mention BRIG.
31430 * doc/standards.texi (Standards): Doucment BRIG HSA version.
31432 2017-01-24 Richard Biener <rguenther@suse.de>
31434 PR translation/79208
31435 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
31437 2017-01-24 Martin Jambor <mjambor@suse.cz>
31440 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
31441 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
31442 and known_contexts.
31444 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
31446 PR middle-end/79123
31447 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
31448 casts from signed to unsigned really don't have a range.
31450 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
31452 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
31453 GMP_RNDx for compatiblity.
31455 2017-01-24 Martin Liska <mliska@suse.cz>
31458 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
31459 that would prevent us to call alloca with -1 as argument.
31461 2017-01-24 Jakub Jelinek <jakub@redhat.com>
31463 * dwarf2out.c (output_compilation_unit_header, output_file_names):
31464 Avoid -Wformat-security warning.
31466 2017-01-23 Andrew Pinski <apinski@cavium.com>
31468 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
31471 2017-01-23 Martin Sebor <msebor@redhat.com>
31473 PR middle-end/78703
31474 * gimple-ssa-sprintf.c (warn_level): New global.
31475 (format_integer): Use it here and throughout the rest of the file.
31476 Use the same switch to compute sign as base.
31477 (maybe_warn): New function.
31478 (format_directive): Factor out warnings into maybe_warn.
31479 Add debugging output. Use warn_level.
31480 (add_bytes): Use warn_level.
31481 (pass_sprintf_length::compute_format_length): Add debugging output.
31482 (try_substitute_return_value): Same.
31483 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
31485 PR middle-end/78703
31486 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
31487 (struct fmtresult, format_integer, format_floating): Adjust.
31488 (fmtresult::fmtresult): Set max correctly in two argument ctor.
31489 (get_string_length, format_string,format_directive): Same.
31490 (pass_sprintf_length::compute_format_length): Same.
31491 (try_substitute_return_value): Simplify slightly.
31493 PR middle-end/78703
31494 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
31495 (fmtresult::operator+=): Outlined.
31496 (struct fmtresult): Add ctors.
31497 (struct conversion_spec): Rename...
31498 (struct directive): ...to this. Add and remove data members.
31499 (directive::set_width, directive::set_precision): New functions.
31500 (format_percent): Use fmtresult ctor.
31501 (get_width_and_precision): Remove.
31502 (format_integer): Make naming changes. Avoid computing width and
31504 (format_floating): Same. Adjust indentation.
31505 (format_character, format_none): New functions.
31506 (format_string): Moved character handling to format_character.
31507 (format_directive): Remove arguments, change return type.
31508 (parse_directive): New function.
31509 (pass_sprintf_length::compute_format_length): Move directive
31510 parsing to parse_directive.
31512 2017-01-23 Jakub Jelinek <jakub@redhat.com>
31514 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
31515 (assign_assembler_name_if_needed): ... this.
31516 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
31517 (assign_assembler_name_if_needed): ... this.
31518 (free_lang_data_in_cgraph): Adjust callers.
31519 * cgraphunit.c (cgraph_node::analyze): Likewise.
31520 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
31523 2017-01-23 Richard Biener <rguenther@suse.de>
31525 PR tree-optimization/79088
31526 PR tree-optimization/79188
31527 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
31528 resetting loop bounds after last path deletion. Reset loop
31529 bounds of the target loop, make code match the comments.
31530 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
31531 Make sure loops need no fixups.
31533 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
31535 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
31536 exponent support with double type for first argument.
31537 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
31538 type returned by __builtin_vec_extract_sig,
31539 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
31540 functions from "vector int" to "vector unsigned int" or from
31541 "vector long long int" to "vector unsigned long long int".
31542 Changed type returned by __builtin_vec_extract_exp,
31543 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
31544 functions from "vector int" to "vector unsigned int" or from
31545 "vector long long int" to "vector unsigned long long int".
31546 Changed return type of __builtin_vec_test_data_class,
31547 __builtin_vec_test_data_class_sp, and
31548 __builtin_vec_test_data_class_dp from "vector int" to
31549 "vector bool int" or from "vector long long int" to "vector bool
31550 long long int" and changed second argument type from "unsigned
31551 int" to "int". Added new overloaded function forms "vector float
31552 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
31553 "vector float __builtin_vec_insert_exp_sp (vector float, vector
31554 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
31555 double, vector unsigned long long int)" and "vector double
31556 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
31557 long int)". Changed return type of
31558 __builtin_scalar_test_data_class and
31559 __builtin_scalar_test_data_class_sp and
31560 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
31561 int" and changed second argument from "unsigned int" to "int".
31562 Changed type returned by __builtin_scalar_test_neg,
31563 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
31564 from "int" to "bool int". Added new overloaded function form
31565 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
31566 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
31567 exponent double-precision with floating point first argument.
31568 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
31569 documentation of scalar_test_data_class, scalar_test_neg,
31570 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
31571 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
31572 vec_test_data_class built-in functions to reflect refinements in
31573 their type signatures.
31575 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
31577 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
31579 (aarch64_elf_asm_destructor): Likewise.
31581 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
31583 PR rtl-optimization/78634
31584 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
31585 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
31586 * ifcvt.c (noce_try_cmove): Add missing cost check.
31588 PR rtl-optimization/71724
31589 * combine.c (if_then_else_cond): Look for situations where it is
31590 beneficial to undo the work of one of the recursive calls.
31592 2017-01-23 Bin Cheng <bin.cheng@arm.com>
31594 PR tree-optimization/70754
31595 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
31596 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
31597 combined stmt before it if not NULL.
31598 (combine_chains): Process refs reversely and compute dominance point
31601 2017-01-23 Martin Liska <mliska@suse.cz>
31603 PR tree-optimization/79196
31604 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
31605 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
31607 (strlen_optimize_stmt): Call the renamed function.
31609 2017-01-23 Michael Matz <matz@suse.de>
31611 PR tree-optimization/78384
31612 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
31614 2017-01-23 Richard Biener <rguenther@suse.de>
31616 PR tree-optimization/79186
31617 * tree-vrp.c (register_new_assert_for): Make sure we've seen
31618 both incoming edges before moving an assert.
31620 2017-01-23 Martin Jambor <mjambor@suse.cz>
31622 * ipa-prop.c (load_from_param_1): Removed.
31623 (load_from_unmodified_param): Bits from load_from_param_1 put back
31625 (load_from_param): Removed.
31626 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
31627 with stmt. Reverted back to use of load_from_unmodified_param.
31629 2017-01-23 Martin Jambor <mjambor@suse.cz>
31632 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
31633 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
31634 field a pointer to garbage collected vector, mark lattices and
31635 ipcp_orig_node with GTY((skip)).
31636 (ipa_get_param_count): Adjust to descriptors being a pointer.
31637 (ipa_get_param): Likewise.
31638 (ipa_get_type): Likewise.
31639 (ipa_get_param_move_cost): Likewise.
31640 (ipa_set_param_used): Likewise.
31641 (ipa_get_controlled_uses): Likewise.
31642 (ipa_set_controlled_uses): Likewise.
31643 (ipa_is_param_used): Likewise.
31644 (ipa_node_params_t): Move into garbage collector. New methods insert
31646 (ipa_node_params_sum): Annotate wth GTY(()).
31647 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
31649 (ipa_load_from_parm_agg): Adjust declaration.
31650 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
31651 * ipa-profile.c (ipa_profile): Likewise.
31652 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
31653 (ipa_populate_param_decls): Make descriptors parameter garbage
31655 (ipa_dump_param): Adjust to descriptors being a pointer.
31656 (ipa_alloc_node_params): Likewise.
31657 (ipa_initialize_node_params): Likewise.
31658 (load_from_param_1): Make descriptors parameter garbage collected.
31659 (load_from_unmodified_param): Likewise.
31660 (load_from_param): Likewise.
31661 (ipa_load_from_parm_agg): Likewise.
31662 (ipa_node_params::~ipa_node_params): Removed.
31663 (ipa_free_all_node_params): Remove call to delete operator.
31664 (ipa_node_params_t::insert): New.
31665 (ipa_node_params_t::remove): Likewise.
31666 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
31667 copy known_csts and known_contexts vectors.
31668 (ipa_read_node_info): Adjust to descriptors being a pointer.
31669 (ipcp_modif_dom_walker): Make m_descriptors field garbage
31671 (ipcp_transform_function): Make descriptors variable garbage
31674 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
31676 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
31677 * config/i386/avx512dqintrin.h: Ditto.
31678 * config/i386/avx512fintrin.h: Ditto.
31679 * config/i386/i386.c: Handle new builtins.
31680 * config/i386/i386-builtin.def: Add new builtins.
31681 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
31682 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
31684 2017-01-23 Jakub Jelinek <jakub@redhat.com>
31685 Martin Liska <mliska@suse.cz>
31687 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
31688 * asan.c (asan_expand_poison_ifn): Support stores and use
31689 appropriate ASAN report function.
31690 * internal-fn.c (expand_ASAN_POISON_USE): New function.
31691 * internal-fn.def (ASAN_POISON_USE): Declare.
31692 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
31693 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
31694 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
31695 ASAN_POISON calls w/o LHS.
31696 * tree-ssa.c (execute_update_addresses_taken): Create clobber
31697 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
31698 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
31699 * gimplify.c (asan_poison_variables): Add attribute
31700 use_after_scope_memory to variables that really needs to live
31702 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
31703 having the attribute.
31705 2017-01-23 Martin Liska <mliska@suse.cz>
31707 * asan.c (create_asan_shadow_var): New function.
31708 (asan_expand_poison_ifn): Likewise.
31709 * asan.h (asan_expand_poison_ifn): New declaration.
31710 * internal-fn.c (expand_ASAN_POISON): Likewise.
31711 * internal-fn.def (ASAN_POISON): New builtin.
31712 * sanopt.c (pass_sanopt::execute): Expand
31713 asan_expand_poison_ifn.
31714 * tree-inline.c (copy_decl_for_dup_finish): Make function
31716 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
31717 * tree-ssa.c (is_asan_mark_p): New function.
31718 (execute_update_addresses_taken): Rewrite local variables
31719 (identified just by use-after-scope as addressable) into SSA.
31721 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
31723 * doc/install.texi (Specific): opensource.apple.com uses https
31724 now. Remove trailing slash.
31726 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
31728 * README.Portability: Remove note on an Irix compatibility issue.
31730 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
31732 * gcov.c (INCLUDE_ALGORITHM): Define.
31733 (INCLUDE_VECTOR): Define.
31734 No longer include <vector> and <algorithm> directly.
31736 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
31738 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
31740 * doc/invoke.texi (Code Gen Options): Ditto.
31742 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
31745 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
31747 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
31749 rtl-optimization/79125
31750 * cprop.c (local_cprop_pass): Handle cases where we make an
31751 unconditional trap.
31753 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
31757 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
31758 read from, for big endian.
31760 2017-01-20 Jiong Wang <jiong.wang@arm.com>
31762 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
31763 register pauth builtins for LP64 only.
31765 2017-01-20 Marek Polacek <polacek@redhat.com>
31768 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
31771 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
31773 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
31775 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
31776 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
31777 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
31779 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31782 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
31783 in big-endian mode when they are not a single duplicated value.
31785 2017-01-20 Richard Biener <rguenther@suse.de>
31787 * BASE-VER: Bump to 7.0.1.
31789 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
31791 * omp-low.c (omplow_simd_context): New struct. Use it...
31792 (lower_rec_simd_input_clauses): ...here and...
31793 (lower_rec_input_clauses): ...here to hold common data. Adjust all
31794 references to idx, lane, max_vf, is_simt.
31796 2017-01-20 Graham Markall <graham.markall@embecosm.com>
31798 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
31801 2017-01-20 Martin Jambor <mjambor@suse.cz>
31803 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
31804 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
31806 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
31807 (GTFILES): Rename hsa.c to hsa-common.c.
31808 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
31809 * hsa-dump.c: Likewise.
31810 * hsa-gen.c: Likewise.
31811 * hsa-regalloc.c: Likewise.
31812 * ipa-hsa.c: Likewise.
31813 * omp-expand.c: Likewise.
31814 * omp-low.c: Likewise.
31815 * toplev.c: Likewise.
31817 2017-01-20 Marek Polacek <polacek@redhat.com>
31820 * doc/invoke.texi: Document -Wduplicated-branches.
31821 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
31822 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
31823 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
31824 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
31825 return 0 only when not OEP_LEXICOGRAPHIC.
31826 (fold_build_cleanup_point_expr): Use the expression
31827 location when building CLEANUP_POINT_EXPR.
31828 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
31829 * tree.c (add_expr): Handle error_mark_node.
31831 2017-01-20 Martin Liska <mliska@suse.cz>
31834 * tree-profile.c (init_ic_make_global_vars): Do not call
31836 (gimple_init_gcov_profiler): Likewise.
31838 2017-01-20 Martin Liska <mliska@suse.cz>
31841 * cgraph.h (maybe_create_reference): Remove argument and
31843 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
31845 * ipa-cp.c (create_specialized_node): Likewise.
31846 * symtab.c (symtab_node::maybe_create_reference): Handle
31847 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
31849 2017-01-20 Martin Liska <mliska@suse.cz>
31851 * read-rtl-function.c (function_reader::create_function): Use
31852 build_decl instread of build_decl_stat.
31854 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
31856 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
31857 * config/i386/avx512dqintrin.h: Ditto.
31858 * config/i386/avx512fintrin.h: Ditto.
31859 * config/i386/i386-builtin-types.def: Add new types.
31860 * config/i386/i386.c: Handle new types.
31861 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
31862 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
31863 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
31864 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
31865 (__builtin_ia32_kshiftridi): New.
31866 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
31868 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
31872 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
31873 define to rs6000_init_stack_protect_guard.
31874 (rs6000_init_stack_protect_guard): New function.
31876 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
31877 Yunqiang Su <yunqiang.su@imgtec.com>
31879 * config.gcc (supported_defaults): Add madd4.
31880 (with_madd4): Add validation.
31881 (all_defaults): Add madd4.
31882 * config/mips/mips.opt (mmadd4): New option.
31883 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
31885 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
31887 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
31888 (ISA_HAS_FUSED_MADD4): Likewise.
31889 * doc/invoke.texi (-mmadd4): Document the new option.
31890 * doc/install.texi (--with-madd4): Document the new option.
31892 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31894 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
31895 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
31896 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
31897 (aarch64_init_pauth_hint_builtins): New.
31898 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
31899 (aarch64_expand_builtin): Expand new builtins.
31901 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31903 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
31904 * combine-stack-adj.c (no_unhandled_cfa): Handle
31905 REG_CFA_TOGGLE_RA_MANGLE.
31906 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
31907 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
31908 info for return address signing.
31909 (aarch64_expand_epilogue): Likewise.
31911 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31913 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
31914 * config/aarch64/aarch64-protos.h
31915 (aarch64_return_address_signing_enabled): New declaration.
31916 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
31918 (aarch64_expand_prologue): Sign return address before it's pushed onto
31920 (aarch64_expand_epilogue): Authenticate return address fetched from
31922 (aarch64_override_options): Sanity check for ILP32 and ISA level.
31923 (aarch64_attributes): New function attributes for "sign-return-address".
31924 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
31925 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
31926 ("*do_return"): Generate combined instructions according to key index.
31927 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
31928 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
31930 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
31931 * config/aarch64/aarch64.opt (msign-return-address=): New.
31932 * doc/extend.texi (AArch64 Function Attributes): Documents
31933 "sign-return-address=".
31934 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
31936 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
31938 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
31939 overall option summary.
31941 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31943 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
31944 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
31945 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
31946 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
31948 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
31950 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
31951 -mpower9-minmax by default for -mcpu=power9.
31952 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
31953 128-bit floating point.
31955 2017-01-20 Alan Modra <amodra@gmail.com>
31957 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
31958 optimizing for size.
31960 2017-01-20 Alan Modra <amodra@gmail.com>
31963 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
31964 for strcmp and strncmp from corresponding builtin decl.
31966 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
31968 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
31969 instead of i386/rtems-64.h.
31970 * config/i386/rtems-64.h: Remove.
31972 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
31976 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
31978 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
31980 2017-01-19 Tamar Christina <tamar.christina@arm.com>
31982 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
31983 Change int to HOST_WIDE_INT.
31984 * config/aarch64/aarch64-protos.h
31985 (aarch64_simd_gen_const_vector_dup): Likewise.
31986 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
31988 2017-01-19 David Malcolm <dmalcolm@redhat.com>
31990 * langhooks-def.h (lhd_type_for_size): New decl.
31991 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
31992 * langhooks.c (lhd_type_for_size): New function, taken from
31995 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
31997 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
31998 define_bypass for CR latency.
31999 (power9-cracked-alu): Update bypass latency and remove power9-branch.
32000 (power9-alu2): Add define_bypass for CR latency.
32002 (power9-mul): Update insn latency.
32003 (power9-mul-compare): Update insn latency, bypass latency and remove
32006 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32008 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
32010 * config/aarch64/aarch64.md
32011 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
32012 aarch64_nopcrelative_literal_loads.
32013 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
32015 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
32017 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
32018 TARGET_LOONGSON_3A.
32019 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
32021 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
32024 * config.gcc (supported_defaults): Add lxc1-sxc1.
32025 (with_lxc1_sxc1): Add validation.
32026 (all_defaults): Add lxc1-sxc1.
32027 * config/mips/mips.opt (mlxc1-sxc1): New option.
32028 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
32030 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
32031 __mips_no_lxc1_sxc1.
32032 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
32033 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
32034 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
32036 2017-01-19 Richard Biener <rguenther@suse.de>
32038 PR tree-optimization/72488
32039 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
32040 sure to restore SSA info.
32041 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
32043 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
32045 PR rtl-optimization/79121
32046 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
32047 of the inner type when shifting an extended value.
32049 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
32052 * symtab.c (symtab_node::equal_address_to): Fix comparing of
32053 interposable aliases.
32055 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
32058 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
32059 Use the evmergelohi instruction.
32060 (mov_si<mode>_e500_subreg4_2_le): Likewise.
32061 (mov_sitf_e500_subreg8_2_be): Likewise.
32062 (mov_sitf_e500_subreg12_2_le): Likewise.
32063 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
32064 (mov_si<mode>_e500_subreg4_2_be): Likewise.
32065 (mov_sitf_e500_subreg8_2_le): Likewise.
32066 (mov_sitf_e500_subreg12_2_be): Likewise.
32068 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32070 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
32071 attribute from vecsimple to vecperm.
32072 (altivec_vbpermq2): Likewise.
32074 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32077 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
32079 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32080 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
32081 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
32082 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
32083 case where N arg is SIZE_MAX.
32084 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
32085 (cmpstrsi): Add pattern.
32087 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
32089 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
32090 __builtin_vec_revb builtins.
32091 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
32092 built-in functions to support generation of the ISA 3.0 XXBR<x>
32093 vector byte reverse instructions.
32094 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
32095 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
32096 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
32097 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
32098 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
32099 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
32100 (P9V_BUILTIN_VEC_REVB): Likewise.
32101 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
32102 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
32103 (p9_xxbrq_v16qi): Likewise.
32104 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
32105 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
32106 (p9_xxbrh_v8hi): Likewise.
32107 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
32108 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
32109 vec_revb built-in functions.
32111 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
32113 PR rtl-optimization/78952
32114 * config/i386/i386.md (any_extract): New code iterator.
32115 (*insvqi_2): Use any_extract for source operand.
32116 (*insvqi_3): Use any_shiftrt for source operand.
32118 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
32120 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
32122 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
32124 2017-01-18 Matthias Klose <doko@ubuntu.com>
32126 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
32128 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32130 * config/rs6000/altivec.h (vec_bperm): Change #define.
32131 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
32132 (altivec_vbpermq2): New define_insn.
32133 (altivec_vbpermd): Likewise.
32134 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
32135 function interface.
32136 (VBPERMD): Likewise.
32137 (VBPERM): New polymorphic function interface.
32138 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
32139 Add entries for P9V_BUILTIN_VEC_VBPERM.
32140 * doc/extend.texi: Add interfaces for vec_bperm.
32142 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32144 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
32145 first letter of error messages.
32146 (s390_resolve_overloaded_builtin): Likewise.
32147 * config/s390/s390.c (s390_expand_builtin): Likewise.
32148 (s390_invalid_arg_for_unprototyped_fn): Likewise.
32149 (s390_valid_target_attribute_inner_p): Likewise.
32150 * config/s390/s390.md ("tabort"): Likewise.
32152 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
32154 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
32155 (ISA_AVOID_DIV_HILO): New macro.
32156 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
32157 (ISA_HAS_DDIV): Likewise.
32159 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32161 * doc/invoke.texi (fabi-version): Correct number of occurrences.
32163 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32165 * doc/invoke.texi (fabi-version): Spelling fix.
32167 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32170 * doc/invoke.texi (fabi-version): Mention mangling fix for
32173 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32176 * doc/invoke.texi (fabi-version): Document discriminator mangling.
32178 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
32181 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
32182 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
32184 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
32186 (stack_protect_test): Ditto.
32187 * config/rs6000/rs6000.opt (mstack-protector-guard=,
32188 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
32190 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
32191 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
32192 -mstack-protector-guard-offset=.
32193 (RS/6000 and PowerPC Options): Ditto.
32195 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32197 * config/i386/i386.h (MASK_CLASS_P): New define.
32198 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
32199 there are no registers from different register sets also when
32200 mask registers are used. Update function comment.
32201 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
32202 to (*k/*r) and (*k/*km) alternatives.
32204 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
32206 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
32207 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
32208 (EH_RETURN_HANDLER_RTX): New define.
32209 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
32210 Force frame pointer in EH return functions.
32211 (aarch64_expand_epilogue): Add barrier for eh_return.
32212 (aarch64_final_eh_return_addr): Remove.
32213 (aarch64_eh_return_handler_rtx): New function.
32214 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
32216 (aarch64_eh_return_handler_rtx): New prototype.
32218 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32220 * config/rs6000/altivec.h (vec_rlmi): New #define.
32221 (vec_vrlnm): Likewise.
32222 (vec_rlnm): Likewise.
32223 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
32224 (UNSPEC_VRLNM): Likewise.
32225 (VIlong): New mode iterator.
32226 (altivec_vrl<VI_char>mi): New define_insn.
32227 (altivec_vrl<VI_char>nm): Likewise.
32228 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
32230 (VRLDNM): Likewise.
32231 (RLNM): New polymorphic function entry.
32232 (VRLWMI): New monomorphic function entry.
32233 (VRLDMI): Likewise.
32234 (RLMI): New polymorphic function entry.
32235 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
32236 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
32237 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
32240 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32243 * dwarf2out.c (field_byte_offset): Restore the
32244 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
32245 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
32246 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
32249 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32252 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
32254 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32257 * dwarf2out.c (add_data_member_location_attribute): For constant
32258 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
32259 instead of DW_AT_data_member_location, DW_AT_bit_offset and
32260 DW_AT_byte_size attributes.
32262 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32264 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
32265 after forcing to constant memory when the code model is medium.
32267 2017-01-17 Julia Koval <julia.koval@intel.com>
32270 * config/i386/avx512fintrin.h
32271 (_mm512_i32gather_ps): Change __addr type to void const*.
32272 (_mm512_mask_i32gather_ps): Ditto.
32273 (_mm512_i32gather_pd): Ditto.
32274 (_mm512_mask_i32gather_pd): Ditto.
32275 (_mm512_i64gather_ps): Ditto.
32276 (_mm512_mask_i64gather_ps): Ditto.
32277 (_mm512_i64gather_pd): Ditto.
32278 (_mm512_mask_i64gather_pd): Ditto.
32279 (_mm512_i32gather_epi32): Ditto.
32280 (_mm512_mask_i32gather_epi32): Ditto.
32281 (_mm512_i32gather_epi64): Ditto.
32282 (_mm512_mask_i32gather_epi64): Ditto.
32283 (_mm512_i64gather_epi32): Ditto.
32284 (_mm512_mask_i64gather_epi32): Ditto.
32285 (_mm512_i64gather_epi64): Ditto.
32286 (_mm512_mask_i64gather_epi64): Ditto.
32287 (_mm512_i32scatter_ps): Change __addr type to void*.
32288 (_mm512_mask_i32scatter_ps): Ditto.
32289 (_mm512_i32scatter_pd): Ditto.
32290 (_mm512_mask_i32scatter_pd): Ditto.
32291 (_mm512_i64scatter_ps): Ditto.
32292 (_mm512_mask_i64scatter_ps): Ditto.
32293 (_mm512_i64scatter_pd): Ditto.
32294 (_mm512_mask_i64scatter_pd): Ditto.
32295 (_mm512_i32scatter_epi32): Ditto.
32296 (_mm512_mask_i32scatter_epi32): Ditto.
32297 (_mm512_i32scatter_epi64): Ditto.
32298 (_mm512_mask_i32scatter_epi64): Ditto.
32299 (_mm512_i64scatter_epi32): Ditto.
32300 (_mm512_mask_i64scatter_epi32): Ditto.
32301 (_mm512_i64scatter_epi64): Ditto.
32302 (_mm512_mask_i64scatter_epi64): Ditto.
32303 * config/i386/avx512pfintrin.h
32304 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
32305 (_mm512_mask_prefetch_i32gather_ps): Ditto.
32306 (_mm512_mask_prefetch_i64gather_pd): Ditto.
32307 (_mm512_mask_prefetch_i64gather_ps): Ditto.
32308 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
32309 (_mm512_prefetch_i32scatter_ps): Ditto.
32310 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
32311 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
32312 (_mm512_prefetch_i64scatter_pd): Ditto.
32313 (_mm512_prefetch_i64scatter_ps): Ditto.
32314 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
32315 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
32316 * config/i386/avx512vlintrin.h
32317 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
32318 (_mm_mmask_i32gather_ps): Ditto.
32319 (_mm256_mmask_i32gather_pd): Ditto.
32320 (_mm_mmask_i32gather_pd): Ditto.
32321 (_mm256_mmask_i64gather_ps): Ditto.
32322 (_mm_mmask_i64gather_ps): Ditto.
32323 (_mm256_mmask_i64gather_pd): Ditto.
32324 (_mm_mmask_i64gather_pd): Ditto.
32325 (_mm256_mmask_i32gather_epi32): Ditto.
32326 (_mm_mmask_i32gather_epi32): Ditto.
32327 (_mm256_mmask_i32gather_epi64): Ditto.
32328 (_mm_mmask_i32gather_epi64): Ditto.
32329 (_mm256_mmask_i64gather_epi32): Ditto.
32330 (_mm_mmask_i64gather_epi32): Ditto.
32331 (_mm256_mmask_i64gather_epi64): Ditto.
32332 (_mm_mmask_i64gather_epi64): Ditto.
32333 (_mm256_i32scatter_ps): Change __addr type to void*.
32334 (_mm256_mask_i32scatter_ps): Ditto.
32335 (_mm_i32scatter_ps): Ditto.
32336 (_mm_mask_i32scatter_ps): Ditto.
32337 (_mm256_i32scatter_pd): Ditto.
32338 (_mm256_mask_i32scatter_pd): Ditto.
32339 (_mm_i32scatter_pd): Ditto.
32340 (_mm_mask_i32scatter_pd): Ditto.
32341 (_mm256_i64scatter_ps): Ditto.
32342 (_mm256_mask_i64scatter_ps): Ditto.
32343 (_mm_i64scatter_ps): Ditto.
32344 (_mm_mask_i64scatter_ps): Ditto.
32345 (_mm256_i64scatter_pd): Ditto.
32346 (_mm256_mask_i64scatter_pd): Ditto.
32347 (_mm_i64scatter_pd): Ditto.
32348 (_mm_mask_i64scatter_pd): Ditto.
32349 (_mm256_i32scatter_epi32): Ditto.
32350 (_mm256_mask_i32scatter_epi32): Ditto.
32351 (_mm_i32scatter_epi32): Ditto.
32352 (_mm_mask_i32scatter_epi32): Ditto.
32353 (_mm256_i32scatter_epi64): Ditto.
32354 (_mm256_mask_i32scatter_epi64): Ditto.
32355 (_mm_i32scatter_epi64): Ditto.
32356 (_mm_mask_i32scatter_epi64): Ditto.
32357 (_mm256_i64scatter_epi32): Ditto.
32358 (_mm256_mask_i64scatter_epi32): Ditto.
32359 (_mm_i64scatter_epi32): Ditto.
32360 (_mm_mask_i64scatter_epi32): Ditto.
32361 (_mm256_i64scatter_epi64): Ditto.
32362 (_mm256_mask_i64scatter_epi64): Ditto.
32363 (_mm_i64scatter_epi64): Ditto.
32364 (_mm_mask_i64scatter_epi64): Ditto.
32365 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
32366 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
32367 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
32368 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
32369 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
32370 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
32371 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
32372 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
32373 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
32374 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
32375 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
32376 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
32377 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
32378 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
32379 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
32380 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
32381 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
32382 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
32383 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
32384 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
32385 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
32386 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
32387 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
32388 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
32389 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
32390 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
32391 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
32392 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
32393 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
32394 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
32395 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
32396 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
32397 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
32398 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
32399 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
32400 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
32401 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
32402 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
32403 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
32404 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
32405 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
32406 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
32407 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
32408 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
32409 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
32410 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
32411 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
32412 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
32413 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
32414 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
32415 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
32416 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
32417 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
32418 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
32419 definitions accordingly.
32421 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
32422 Kuan-Lin Chen <kuanlinchentw@gmail.com>
32425 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
32428 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
32431 * ira-conflicts.c (ira_build_conflicts): Update total conflict
32432 hard regs for inner regno.
32434 2017-01-17 Martin Liska <mliska@suse.cz>
32437 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
32438 assumption and add comment.
32440 2017-01-17 Nathan Sidwell <nathan@acm.org>
32442 * ipa-visibility.c (localize_node): New function, broken out of ...
32443 (function_and_variable_visibility): ... here. Call it.
32445 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
32447 PR middle-end/77445
32448 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
32449 correctly set frequency of oudgoing edge.
32450 (duplicate_thread_path): Fix profile updating.
32452 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32455 * configure.ac: Add GCC_BASE_VER.
32456 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
32457 version from BASE-VER file.
32458 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
32459 (gcc.o): Depend on $(BASEVER).
32460 * common.opt (dumpfullversion): New option.
32461 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
32462 * doc/invoke.texi: Document -dumpfullversion.
32463 * doc/install.texi: Document --with-gcc-major-version-only.
32464 * configure: Regenerated.
32466 2017-01-17 Richard Biener <rguenther@suse.de>
32468 PR tree-optimization/71433
32469 * tree-vrp.c (register_new_assert_for): Merge same asserts
32470 on all incoming edges.
32471 (process_assert_insertions_for): Handle insertions at the
32474 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
32476 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
32477 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
32479 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
32482 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
32485 2017-01-17 Alan Modra <amodra@gmail.com>
32488 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
32489 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
32490 symbolic stack limit when pic.
32492 2017-01-16 Martin Sebor <msebor@redhat.com>
32494 PR tree-optimization/78608
32495 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
32497 2017-01-16 Jeff Law <law@redhat.com>
32500 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
32501 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
32502 for several include directories that may be relative to sysroot.
32503 * config/i386/x-mingw32 (gplus_includedir): Define.
32504 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
32505 (native_system_includedir): Likewise.
32506 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
32507 override if TARGET_SYSTEM_ROOT is defined.
32508 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
32510 PR tree-optimization/79090
32511 PR tree-optimization/33562
32512 PR tree-optimization/61912
32513 PR tree-optimization/77485
32514 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
32515 and computed trims into the dump file.
32517 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32519 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
32521 2017-01-16 Jakub Jelinek <jakub@redhat.com>
32524 * gimplify.c (gimplify_init_constructor): If want_value and
32525 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
32529 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
32530 sequence. Formatting fixes.
32531 (doloop_optimize): Formatting fixes.
32534 * gcc.c (debug_level_greater_than_spec_func): New function.
32535 (static_spec_functions): Add debug-level-gt spec function.
32536 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
32538 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32539 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32540 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
32541 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
32542 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
32543 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
32545 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
32547 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
32548 QImode fixups to general and mask registers only.
32550 2017-01-16 Carl Love <cel@us.ibm.com>
32552 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
32553 for built-in functions
32554 vector signed char vec_nabs (vector signed char)
32555 vector signed short vec_nabs (vector signed short)
32556 vector signed int vec_nabs (vector signed int)
32557 vector signed long long vec_nabs (vector signed long long)
32558 vector float vec_nabs (vector float)
32559 vector double vec_nabs (vector double)
32560 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
32562 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
32563 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
32564 * doc/extend.texi: Update the documentation file for the new built-in
32567 2017-01-16 Martin Sebor <msebor@redhat.com>
32569 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
32572 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32574 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
32575 UNSPEC_VSX__XXSPLTD to require special splat handling.
32577 2017-01-16 David Malcolm <dmalcolm@redhat.com>
32580 * system.h: Poison strndup.
32582 2017-01-16 Alan Modra <amodra@gmail.com>
32585 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
32588 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
32590 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
32592 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
32594 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
32595 call recog here. Assert that INSN_CODE (insn) is non-negative.
32597 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32600 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
32602 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
32603 in the currently scheduled RTL fragment.
32605 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32607 PR rtl-optimization/78751
32608 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
32611 2017-01-14 Jeff Law <law@redhat.com>
32613 PR tree-optimization/79090
32614 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
32615 variable length stores.
32616 (compute_trims): Delete dead assignment to *trim_tail.
32617 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
32620 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
32622 PR rtl-optimization/78626
32623 PR rtl-optimization/78727
32624 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
32625 of a block, and split such blocks after everything else is finished.
32627 2017-01-14 Alan Modra <amodra@gmail.com>
32630 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
32631 target legitimate_combined_insn.
32632 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
32633 (rs6000_legitimate_combined_insn): New function.
32634 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
32636 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
32637 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
32638 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
32640 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
32642 * doc/frontends.texi (G++ and GCC): Remove references to Java.
32644 2017-01-13 Jeff Law <law@redhat.com>
32646 PR tree-optimization/33562
32647 PR tree-optimization/61912
32648 PR tree-optimization/77485
32649 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
32651 (delete_dead_assignment): Likewise.
32652 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
32653 statement to delete_dead_call and delete_dead_assignment.
32655 2017-01-13 David Malcolm <dmalcolm@redhat.com>
32658 * substring-locations.c (format_warning_va): Strengthen case 1 so
32659 that both endpoints of the substring must be within the format
32660 range for just the substring to be printed.
32662 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
32664 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
32665 * config/i386/i386.c (ix86_target_string): Add missing options
32666 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
32667 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
32668 flags_other and ix86_target_other to flags2_other. Display unknown
32670 (ix86_valid_target_attribute_inner_p): Add missing options and
32671 reorder options by implied ISAs, as in ix86_target_string.
32673 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
32675 * hash-table.h (hash_table::too_empty_p): New function.
32676 (hash_table::expand): Use it.
32677 (hash_table::traverse): Likewise.
32678 (hash_table::empty_slot): Use sizeof (value_type) instead of
32679 sizeof (PTR) to convert bytes to elements. Shrink the table
32680 if the current size is excessive for the current number of
32683 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
32685 * ira-costs.c (record_reg_classes): Break from the inner loop
32686 early once alt_fail is known to be true. Update outer loop
32687 handling accordingly.
32689 2017-01-13 Jeff Law <law@redhat.com>
32691 * tree-ssa-dse.c (decrement_count): New function.
32692 (increment_start_addr, maybe_trim_memstar_call): Likewise.
32693 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
32694 when we know the partially dead statement is a mem* function.
32696 PR tree-optimization/61912
32697 PR tree-optimization/77485
32698 * tree-ssa-dse.c: Include expr.h.
32699 (maybe_trim_constructor_store): New function.
32700 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
32702 PR tree-optimization/33562
32703 PR tree-optimization/61912
32704 PR tree-optimization/77485
32705 * doc/invoke.texi: Document new dse-max-object-size param.
32706 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
32707 * tree-ssa-dse.c: Include params.h.
32708 (dse_store_status): New enum.
32709 (initialize_ao_ref_for_dse): New, partially extracted from
32711 (valid_ao_ref_for_dse, normalize_ref): New.
32712 (setup_live_bytes_from_ref, compute_trims): Likewise.
32713 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
32714 (maybe_trim_partially_dead_store): Likewise.
32715 (maybe_trim_complex_store): Likewise.
32716 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
32717 Track what bytes live from the original store. Return tri-state
32718 for dead, partially dead or live.
32719 (dse_dom_walker): Add constructor, destructor and new private members.
32720 (delete_dead_call, delete_dead_assignment): New extracted from
32722 (dse_optimize_stmt): Make a member of dse_dom_walker.
32723 Use initialize_ao_ref_for_dse.
32725 PR tree-optimization/33562
32726 PR tree-optimization/61912
32727 PR tree-optimization/77485
32728 * sbitmap.h (bitmap_count_bits): Prototype.
32729 (bitmap_clear_range, bitmap_set_range): Likewise.
32730 * sbitmap.c (bitmap_clear_range): New function.
32731 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
32733 2017-01-13 Martin Liska <mliska@suse.cz>
32736 * function.c (set_cfun): Add new argument force.
32737 * function.h (set_cfun): Likewise.
32738 * ipa-inline-transform.c (inline_call): Use the function when
32739 strict alising from is dropped for function we inline to.
32741 2017-01-13 Richard Biener <rguenther@suse.de>
32743 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
32744 for dumping GIMPLE INTEGER_CSTs.
32746 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
32748 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
32749 to 201112L since C++17.
32751 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
32754 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
32755 if -fsanitize=kernel-address is present.
32757 2017-01-13 Richard Biener <rguenther@suse.de>
32759 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
32760 as _Literal ( type ) number in case usual suffixes do not
32761 preserve all information.
32763 2017-01-13 Richard Biener <rguenther@suse.de>
32765 PR tree-optimization/77283
32766 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
32767 and ssa-iterators.h.
32768 (is_feasible_trace): Implement a cost model based on joiner
32771 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
32774 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
32775 char or short to __float128/_Float128 directly.
32777 2017-01-12 Martin Sebor <msebor@redhat.com>
32779 to -Wformat-overflow.
32780 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
32781 (min_bytes_remaining): Same.
32782 (get_string_length): Same.
32783 (format_string): Same.
32784 (format_directive): Same.
32786 (pass_sprintf_length::handle_gimple_call): Same.
32788 2017-01-12 Jakub Jelinek <jakub@redhat.com>
32790 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
32791 info.nowrite calls with no lhs that can't throw. Return bool
32792 whether gsi_remove has been called or not.
32793 (pass_sprintf_length::handle_gimple_call): Return bool whether
32794 try_substitute_return_value called gsi_remove. Formatting fix.
32795 (pass_sprintf_length::execute): Don't use gsi_remove if
32796 handle_gimple_call returned true.
32799 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
32800 be removed due to side-effects, don't remove following barrier nor
32801 turn the successor edge into fallthru edge.
32803 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32806 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
32807 element-reversing loads and stores as not swappable.
32809 2017-01-12 Nathan Sidwell <nathan@acm.org>
32810 Nicolai Stange <nicstange@gmail.com>
32812 * combine.c (try_combine): Don't ignore result of overlap checking
32813 loop. Combine overlap & asm check into single loop.
32815 2017-01-12 Richard Biener <rguenther@suse.de>
32817 * tree-pretty-print.c (dump_generic_node): Provide -gimple
32818 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
32820 2017-01-12 Richard Biener <rguenther@suse.de>
32822 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
32823 and TS_TARGET_OPTION directly derive from TS_BASE.
32824 * tree-core.h (tree_optimization_option): Derive from tree_base.
32825 (tree_target_option): Likewise.
32827 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
32829 * config/i386/i386.c (memory_address_length): Increase len
32830 only when rip_relative_addr_p returns false.
32832 2017-01-11 Julia Koval <julia.koval@intel.com>
32834 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
32835 (OPTION_MASK_ISA_SGX_SET): New.
32836 (ix86_handle_option): Handle OPT_msgx.
32837 * config.gcc: Added sgxintrin.h.
32838 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
32839 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
32840 * config/i386/i386.c (ix86_target_string): Add -msgx.
32842 (ix86_option_override_internal): Handle new options.
32843 (ix86_valid_target_attribute_inner_p): Add sgx.
32844 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
32845 * config/i386/i386.opt: Add msgx.
32846 * config/i386/sgxintrin.h: New file.
32847 * config/i386/x86intrin.h: Add sgxintrin.h.
32849 2017-01-11 Jakub Jelinek <jakub@redhat.com>
32852 * fold-const.c (maybe_nonzero_address): Return 1 for function
32854 (tree_single_nonzero_warnv_p): Don't handle function local objects
32858 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
32861 2017-01-11 David Malcolm <dmalcolm@redhat.com>
32864 * opts.c: Include "spellcheck.h"
32865 (struct string_fragment): New struct.
32866 (struct edit_distance_traits<const string_fragment &>): New
32868 (get_closest_sanitizer_option): New function.
32869 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
32871 2017-01-11 Jakub Jelinek <jakub@redhat.com>
32873 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
32875 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
32876 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
32877 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
32878 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
32879 for initial die_offset if dwarf_split_debug_info.
32880 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
32881 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
32883 (output_skeleton_debug_sections): Formatting fix. Use
32884 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
32885 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
32887 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
32889 * config/arm/cortex-a53.md: Add bypasses for
32890 cortex_a53_r2f_cvt.
32891 (cortex_a53_r2f): Only use for transfers.
32892 (cortex_a53_f2r): Likewise.
32893 (cortex_a53_r2f_cvt): Add reservation for conversions.
32894 (cortex_a53_f2r_cvt): Likewise.
32896 2017-01-11 Tamar Christina <tamar.christina@arm.com>
32898 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
32899 to all inlined functions, change static to extern.
32901 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
32904 * config/arm/arm.c (legitimize_pic_address): Handle reference to
32906 (arm_assemble_integer): Likewise.
32908 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
32910 * config.gcc: Use new awk script to check CPU, FPU and architecture
32911 parameters for --with-... options.
32912 * config/arm/parsecpu.awk: New file
32913 * config/arm/arm-cpus.in: New file.
32914 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
32916 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
32918 * config/arm/t-arm: Update dependency rules.
32919 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
32920 of processing .def files.
32921 * config/arm/genopt.sh: Deleted.
32922 * config/arm/gentune.sh: Deleted.
32923 * config/arm/arm-cores.def: Deleted.
32924 * config/arm/arm-arches.def: Deleted.
32925 * config/arm/arm-fpus.def: Deleted.
32926 * config/arm/arm-tune.md: Regenerated.
32927 * config/arm/arm-tables.opt: Regenerated.
32928 * config/arm/arm-cpu.h: New generated file.
32929 * config/arm/arm-cpu-data.h: New generated file.
32930 * config/arm/arm-cpu-cdata.h: New generated file.
32932 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
32935 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
32937 (input_varpool_node): Unpack dynamically_initialized bit.
32939 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
32941 PR rtl-optimization/79032
32942 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
32943 the alignment of the adjusted memory reference against that of MODE,
32944 instead of the alignment of the original memory reference.
32946 2017-01-11 Martin Jambor <mjambor@suse.cz>
32948 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
32950 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
32951 decorated functions.
32953 2017-01-11 Richard Biener <rguenther@suse.de>
32955 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
32956 set range/nonnull info for PHI results. Do not set it on
32957 stmts marked for removal.
32959 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
32961 * expr.c (store_field): In the bitfield case, fetch the return value
32962 from the registers before applying a single big-endian adjustment.
32963 Always do a final load for a BLKmode value not larger than a word.
32965 2017-01-10 David Malcolm <dmalcolm@redhat.com>
32968 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
32969 that we correctly handle column numbers greater than
32970 LINE_MAP_MAX_COLUMN_NUMBER.
32972 2017-01-10 Martin Sebor <msebor@redhat.com>
32974 PR middle-end/78245
32975 * gimple-ssa-sprintf.c (get_destination_size): Call
32976 {init,fini}object_sizes.
32977 * tree-object-size.c (addr_object_size): Adjust.
32978 (pass_through_call): Adjust.
32979 (pass_object_sizes::execute): Adjust.
32980 * tree-object-size.h (fini_object_sizes): Declare.
32982 2017-01-10 Martin Sebor <msebor@redhat.com>
32984 PR tree-optimization/78775
32985 * builtins.c (get_size_range): Move...
32986 * calls.c: ...to here.
32987 (alloc_max_size): Accept zero argument.
32988 (operand_signed_p): Remove.
32989 (maybe_warn_alloc_args_overflow): Call get_size_range.
32990 * calls.h (get_size_range): Declare.
32992 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
32994 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
32995 from TI's devices.csv file as of September 2016.
32996 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
32998 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
33000 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
33001 * doc/invoke.texi: Likewise.
33002 * doc/md.texi: Likewise.
33003 * doc/objc.texi: Likewise.
33005 2017-01-10 Joshua Conner <joshconner@google.com>
33007 * config/arm/fuchsia-elf.h: New file.
33008 * config/fuchsia.h: New file.
33009 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
33010 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
33012 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
33014 2016-01-10 Richard Biener <rguenther@suse.de>
33016 PR tree-optimization/79034
33017 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
33018 Propagate out degenerate PHIs in the joiner.
33020 2017-01-10 Martin Liska <mliska@suse.cz>
33022 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
33023 (sort_congruence_classes_by_decl_uid): Likewise.
33024 (sort_congruence_class_groups_by_decl_uid): Likewise.
33025 (sem_item_optimizer::merge_classes): Sort class, groups in these
33026 classes and members in the groups by DECL_UID of declarations.
33027 This would make merge operations stable.
33029 2017-01-10 Martin Liska <mliska@suse.cz>
33031 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
33032 usage of m_classes_vec.
33033 (sem_item_optimizer::~sem_item_optimizer): Likewise.
33034 (sem_item_optimizer::get_group_by_hash): Likewise.
33035 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
33036 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
33037 (sem_item_optimizer::verify_classes): Likewise.
33038 (sem_item_optimizer::process_cong_reduction): Likewise.
33039 (sem_item_optimizer::dump_cong_classes): Likewise.
33040 (sem_item_optimizer::merge_classes): Likewise.
33041 * ipa-icf.h (congruence_class_hash): Rename from
33042 congruence_class_group_hash. Remove declaration of m_classes_vec.
33044 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
33046 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
33047 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
33048 * config.gcc: Add avx512vpopcntdqintrin.h.
33049 * config/i386/avx512vpopcntdqintrin.h: New.
33050 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
33051 * config/i386/i386-builtin-types.def: Add new types.
33052 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
33053 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
33054 __builtin_ia32_vpopcountq_v8di_mask): New.
33055 * config/i386/i386-c.c (ix86_target_macros_internal): Define
33056 __AVX512VPOPCNTDQ__.
33057 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
33058 (PTA_AVX512VPOPCNTDQ): Define.
33059 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
33060 TARGET_AVX512VPOPCNTDQ_P): Define.
33061 * config/i386/i386.opt: Add mavx512vpopcntdq.
33062 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
33063 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
33065 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33067 PR middle-end/77484
33068 * predict.def (PRED_CALL): Set to 67.
33070 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
33072 * expr.c (store_field): In the bitfield case, if the value comes from
33073 a function call and is of an aggregate type returned in registers, do
33074 not modify the field mode; extract the value in all cases if the mode
33075 is BLKmode and the size is not larger than a word.
33077 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
33080 * config/i386/cpuid.h: Fix undefined behavior.
33082 2017-01-04 Jeff Law <law@redhat.com>
33084 PR tree-optimization/79007
33085 PR tree-optimization/67955
33086 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
33087 conservative for pt.null when flag_non_call_exceptions is on.
33089 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33091 PR translation/79019
33092 PR translation/79020
33093 * params.def (PARAM_INLINE_MIN_SPEEDUP,
33094 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
33095 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
33097 * config/avr/avr.opt (maccumulate-args): Likewise.
33098 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
33099 * common.opt (freport-bug): Likewise.
33100 * cif-code.def (CIF_FINAL_ERROR): Likewise.
33101 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
33102 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
33103 translatable string.
33104 * config/i386/i386.c (function_value_32): Likewise.
33105 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
33106 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
33108 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
33109 * common/config/msp430/msp430-common.c (msp430_handle_option):
33111 * symtab.c (symtab_node::verify_base): Likewise.
33112 * opts.c (set_debug_level): Likewise.
33113 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
33114 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
33115 missing whitespace to translatable strings.
33116 * config/avr/avr.md (bswapsi2): Fix typo in comment.
33117 * config/sh/superh.h: Likewise.
33118 * config/i386/xopintrin.h: Likewise.
33119 * config/i386/znver1.md: Likewise.
33120 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
33121 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
33122 * double-int.h (struct double_int): Likewise.
33123 * double-int.c (div_and_round_double): Likewise.
33124 * wide-int.cc: Likewise.
33125 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
33126 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
33127 * cfgcleanup.c (crossjumps_occured): Renamed to ...
33128 (crossjumps_occurred): ... this.
33129 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
33132 PR tree-optimization/78899
33133 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
33134 returning bool return struct loop *, NULL for failure and the new
33136 (versionable_outer_loop_p): Don't version outer loop if it has
33137 dont_vectorized bit set.
33138 (tree_if_conversion): When versioning outer loop, ensure
33139 tree_if_conversion is performed also on the inner loop of the
33140 non-vectorizable outer loop copy.
33141 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
33142 LOOP_VECTORIZED in inner loop of the scalar outer loop and
33143 prevent vectorization of it.
33144 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
33145 the outer loop vectorization of the non-scalar version is attempted
33146 before vectorization of the inner loop in scalar version. If
33147 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
33148 vectorization of its inner loop.
33149 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
33150 has 2 inner loops, rename also on edges from bb whose single pred
33151 is outer_loop->header. Fix typo in function comment.
33153 2017-01-09 Martin Sebor <msebor@redhat.com>
33156 * asan.c (asan_emit_stack_protection): Increase local buffer size
33157 to avoid snprintf truncation warning.
33159 2017-01-09 Andrew Pinski <apinski@cavium.com>
33161 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
33162 to reference thunderx2t99 for the tuning structure
33163 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
33165 (thunderx2t99_extra_costs): This.
33166 * config/aarch64/aarch64-tune.md: Regenerate.
33167 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
33168 (vulcan_addrcost_table): This.
33169 (vulcan_regmove_cost): Rename to ...
33170 (thunderx2t99_regmove_cost): This.
33171 (vulcan_vector_cost): Rename to ...
33172 (thunderx2t99_vector_cost): this.
33173 (vulcan_branch_cost): Rename to ...
33174 (thunderx2t99_branch_cost): This.
33175 (vulcan_tunings): Rename to ...
33176 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
33177 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
33179 2017-01-09 Martin Jambor <mjambor@suse.cz>
33183 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
33184 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
33185 (propagate_vr_accross_jump_function): Use the above function for all
33186 value range computations for pass-through jump functions and type
33187 converasion from explicit value range values.
33188 (ipcp_propagate_stage): Do not attempt to deduce types of formal
33189 parameters from TYPE_ARG_TYPES.
33190 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
33191 (ipa_write_node_info): Stream type of the actual argument.
33192 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
33194 2017-01-09 Martin Liska <mliska@suse.cz>
33197 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
33198 (lookup_compiler): Do not show error message with have_E.
33200 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33202 PR tree-optimization/78938
33203 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
33204 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
33205 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
33206 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
33209 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33211 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
33214 2017-01-09 Richard Biener <rguenther@suse.de>
33216 PR tree-optimization/78997
33217 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
33218 name condition properly.
33220 2017-01-09 Richard Biener <rguenther@suse.de>
33223 * dwarf2out.c (is_cxx): New overload with context.
33224 (is_naming_typedef_decl): Use it.
33226 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33228 * invoke.texi (Option Summary): Correct spacing in option lists
33229 and add line breaks to fix over-long lines.
33231 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33233 PR middle-end/17660
33235 * extend.texi (Common Variable Attributes): Add xref to GCC
33236 Internals manual to explain mode attribute keywords.
33238 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33241 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
33242 and Preprocessor Options.
33243 (Options for Linking): Document -pthread here....
33244 (RS/6000 and PowerPC Options): ...not here.
33245 (Solaris 2 Options): ...or here.
33246 * doc/cppopts.texi: Document -pthread.
33248 2017-01-08 Martin Sebor <msebor@redhat.com>
33250 PR middle-end/77708
33251 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
33252 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
33253 New member functions.
33254 (format_directive): Used them.
33256 (pass_sprintf_length::handle_gimple_call): Same.
33257 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
33258 to avoid truncation for any argument.
33259 (extract_affine_mul): Same.
33260 * tree.c (get_file_function_name): Same.
33262 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33264 PR middle-end/77484
33265 * predict.def (PRED_INDIR_CALL): Set to 86.
33267 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33269 PR preprocessor/54124
33270 * doc/cppopts.texi: Reformat -d subtable to list the full name
33271 of the options. Add cross-reference to the docs for the general
33272 compiler -d options.
33273 * doc/invoke.texi (Developer Options): Add cross-reference to the
33274 preprocessor-specific -d option documentation.
33276 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33278 PR preprocessor/13498
33279 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
33280 redudant material, and reflect new command-line options.
33281 (System Headers): Likewise.
33283 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33285 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
33286 -isystem, and -idirafter. Copy-edit.
33287 * doc/cppopts.texi: Copy-edit. Remove contradiction about
33288 default for -ftrack-macro-expansion. Delete obsolete and
33289 badly-formatted implementation details about -fdebug-cpp output.
33290 * doc/cppwarnopts.texi: Copy-edit.
33292 2017-01-07 David Malcolm <dmalcolm@redhat.com>
33295 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
33296 that the transition from a max line width >= 1<<10 to narrower
33297 lines works correctly.
33299 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
33301 * doc/options.texi (PerFunction): New.
33302 * opt-functions.awk (switch_flags): Map both Optimization and
33303 PerFunction to CL_OPTIMIZATION.
33304 * opth-gen.awk: Test for PerFunction flag along with
33306 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
33307 it only when the latter is present. Skip those that don't in
33308 the hash function generator.
33309 * common.opt (fvar-tracking): Mark as PerFunction instead of
33311 (fvar-tracking-assignments): Likewise.
33312 (fvar-tracking-assignments-toggle): Likewise.
33313 (fvar-tracking-uninit): Likewise.
33315 2017-01-07 Jakub Jelinek <jakub@redhat.com>
33317 PR translation/79018
33318 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
33321 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
33324 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
33325 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
33326 TARGET_LONG_JUMP_TABLE_OFFSETS.
33327 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
33328 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
33329 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
33330 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
33331 * config/m68k/m68k.md (tablejump expander): Likewise.
33332 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
33333 TARGET_LONG_JUMP_TABLE_OFFSETS.
33334 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
33335 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
33337 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
33338 David Holsgrove <david.holsgrove@xilinx.com>
33340 * common/config/microblaze/microblaze-common.c
33341 (TARGET_EXCEPT_UNWIND_INFO): Remove.
33342 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
33344 * config/microblaze/microblaze.c (microblaze_must_save_register)
33345 (microblaze_expand_epilogue, microblaze_return_addr): Handle
33347 (microblaze_eh_return): New function.
33348 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
33349 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
33350 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
33351 * config/microblaze/microblaze.md (eh_return): New pattern.
33353 2017-01-06 Jakub Jelinek <jakub@redhat.com>
33355 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
33356 GCC_DIAGNOSTIC_STRINGIFY): Define.
33358 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
33360 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33362 * config/arm/arm.md (<mcrr>): New.
33364 * config/arm/arm.c (arm_arch5te): New.
33365 (arm_option_override): Set arm_arch5te.
33366 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
33368 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
33369 (arm_mcrr_qualifiers): ... this. New.
33370 (MRRC_QUALIFIERS): Define to...
33371 (arm_mrrc_qualifiers): ... this. New.
33372 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
33374 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
33375 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
33376 (MRRCI, mrrc, MRRC): New.
33377 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
33378 VUNSPEC_MRRC2): New.
33380 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33382 * config/arm/arm.md (<mcr>): New.
33384 * config/arm/arm.c (arm_coproc_builtin_available): Add
33385 support for mcr, mrc, mcr2 and mrc2.
33386 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
33387 (arm_mcr_qualifiers): ... this. New.
33388 (MRC_QUALIFIERS): Define to ...
33389 (arm_mrc_qualifiers): ... this. New.
33390 (MCR_QUALIFIERS): Define to ...
33391 (arm_mcr_qualifiers): ... this. New.
33392 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
33394 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
33395 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
33396 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
33397 VUNSPEC_MRC2): New.
33399 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33401 * config/arm/arm.md (*ldc): New.
33405 * config/arm/arm.c (arm_coproc_builtin_available): Add
33406 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
33407 (arm_coproc_ldc_stc_legitimate_address): New.
33408 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
33409 'qualifier_const_pointer'.
33410 (LDC_QUALIFIERS): Define to...
33411 (arm_ldc_qualifiers): ... this. New.
33412 (STC_QUALIFIERS): Define to...
33413 (arm_stc_qualifiers): ... this. New.
33414 * config/arm/arm-protos.h
33415 (arm_coproc_ldc_stc_legitimate_address): New.
33416 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
33417 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
33418 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
33419 stc2, stcl, stc2l): New.
33420 * config/arm/constraints.md (Uz): New.
33421 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
33422 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
33423 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
33424 VUNSPEC_STC2L): New.
33426 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33428 * config/arm/arm.md (<cdp>): New.
33429 * config/arm/arm.c (neon_const_bounds): Rename this ...
33430 (arm_const_bounds): ... this.
33431 (arm_coproc_builtin_available): New.
33432 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
33433 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
33434 (CDP_QUALIFIERS): Define to...
33435 (arm_cdp_qualifiers): ... this. New.
33437 (arm_expand_builtin_args): Add case for 6 arguments.
33438 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
33439 (arm_const_bounds): ... this.
33440 (arm_coproc_builtin_available): New.
33441 * config/arm/arm_acle.h (__arm_cdp): New.
33443 * config/arm/arm_acle_builtins.def (cdp): New.
33445 * config/arm/iterators.md (CDPI,CDP,cdp): New.
33446 * config/arm/neon.md: Rename all 'neon_const_bounds' to
33447 'arm_const_bounds'.
33448 * config/arm/types.md (coproc): New.
33449 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
33450 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
33451 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
33452 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
33454 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33456 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
33457 (UBINOP_QUALIFIERS): New.
33459 (acle_builtin_data): New. Change comment.
33460 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
33461 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
33462 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
33463 arm_acle_builtins.def.
33464 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
33465 (arm_init_acle_builtins): New.
33466 (CRC32_BUILTIN): Remove.
33467 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
33468 crc32cb, crc32ch and crc32cw.
33469 (arm_init_crc32_builtins): Remove.
33470 (arm_init_builtins): Use arm_init_acle_builtins rather
33471 than arm_init_crc32_builtins.
33472 (arm_expand_acle_builtin): New.
33473 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
33474 * config/arm/arm_acle_builtins.def: New.
33476 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33478 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
33479 (arm_builtin_datum): ... this.
33480 (arm_init_neon_builtin): Rename to ...
33481 (arm_init_builtin): ... this. Add a new parameters PREFIX
33482 and USE_SIG_IN_NAME.
33483 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
33484 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
33485 'arm_builtin_datum'.
33486 (arm_init_vfp_builtins): Likewise.
33487 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
33488 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
33489 (arm_expand_neon_args): Rename to ...
33490 (arm_expand_builtin_args): ... this. Rename builtin_arg
33491 enum values and differentiate between ARG_BUILTIN_MEMORY
33492 and ARG_BUILTIN_NEON_MEMORY.
33493 (arm_expand_neon_builtin_1): Rename to ...
33494 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
33495 values, arm_expand_builtin_args and add bool parameter NEON.
33496 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
33497 (arm_expand_vfp_builtin): Likewise.
33498 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
33500 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33502 PR middle-end/77484
33503 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
33504 * predict.c (tree_estimate_probability_bb): Reverse direction of
33505 polymorphic call predictor.
33507 2017-01-06 David Malcolm <dmalcolm@redhat.com>
33509 * passes.c (execute_one_pass): Split out pass-skipping logic into...
33510 (determine_pass_name_match): ...this new function and...
33511 (should_skip_pass_p): ...this new function.
33513 2017-01-06 Nathan Sidwell <nathan@acm.org>
33515 * ipa-visibility.c (function_and_variable_visibility): Reformat
33516 comments and long lines. Remove extrneous if.
33517 * symtab.c (symtab_node::make_decl_local): Fix code format.
33518 (symtab_node::set_section_for_node): Fix comment typo.
33520 2017-01-06 Martin Liska <mliska@suse.cz>
33523 * lra-constraints.c: Rename invariant to lra_invariant.
33524 * predict.c (set_even_probabilities): Initialize e to NULL.
33526 2017-01-05 Martin Sebor <msebor@redhat.com>
33528 PR tree-optimization/78910
33529 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
33530 (format_integer): Correct off-by-one error in the handling
33531 of precision with negative numbers in signed conversions..
33533 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
33535 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
33537 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33539 PR tree-optimization/71016
33540 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
33541 factor_out_conditional_conversion. Formatting fix.
33542 (factor_out_conditional_conversion): Add cond_stmt argument.
33543 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
33544 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
33547 2017-01-05 David Malcolm <dmalcolm@redhat.com>
33549 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
33550 read-rtl-function.o, and selftest-rtl.o.
33551 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
33552 (selftest::aarch64_test_loading_full_dump): New function.
33553 (selftest::aarch64_run_selftests): New function.
33554 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
33555 selftest::aarch64_run_selftests.
33556 * config/i386/i386.c
33557 (selftest::ix86_test_loading_dump_fragment_1): New function.
33558 (selftest::ix86_test_loading_call_insn): New function.
33559 (selftest::ix86_test_loading_full_dump): New function.
33560 (selftest::ix86_test_loading_unspec): New function.
33561 (selftest::ix86_run_selftests): Call the new functions.
33562 * emit-rtl.c (maybe_set_max_label_num): New function.
33563 * emit-rtl.h (maybe_set_max_label_num): New decl.
33564 * function.c (instantiate_decls): Guard call to
33565 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
33566 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
33568 * gensupport.c (gen_reader::gen_reader): Pass "false"
33569 for new "compact" param of rtx_reader.
33570 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
33571 rather than an empty string for NULL strings.
33572 * read-md.c: Potentially include config.h rather than bconfig.h.
33573 Wrap include of errors.h with #ifdef GENERATOR_FILE.
33574 (have_error): New global, copied from errors.c.
33575 (md_reader::read_name): Rename to...
33576 (md_reader::read_name_1): ...this, adding "out_loc" param,
33577 and converting "missing name or number" to returning false, rather
33579 (md_reader::read_name): Reimplement in terms of read_name_1.
33580 (md_reader::read_name_or_nil): New function.
33581 (md_reader::read_string): Handle "(nil)" by returning NULL.
33582 (md_reader::md_reader): Add new param "compact".
33583 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
33584 (md_reader::read_file): New method.
33585 * read-md.h (md_reader::md_reader): Add new param "compact".
33586 (md_reader::read_file): New method.
33587 (md_reader::is_compact): New accessor.
33588 (md_reader::read_name): Convert return type from void to file_location.
33589 (md_reader::read_name_or_nil): New decl.
33590 (md_reader::read_name_1): New decl.
33591 (md_reader::m_compact): New field.
33592 (noop_reader::noop_reader): Pass "false" for new "compact" param
33594 (rtx_reader::rtx_reader): Add new "compact" param.
33595 (rtx_reader::read_rtx_operand): Make virtual and convert return
33596 type from void to rtx.
33597 (rtx_reader::read_until): New decl.
33598 (rtx_reader::handle_any_trailing_information): New virtual function.
33599 (rtx_reader::postprocess): New virtual function.
33600 (rtx_reader::finalize_string): New virtual function.
33601 (rtx_reader::m_in_call_function_usage): New field.
33602 (rtx_reader::m_reuse_rtx_by_id): New field.
33603 * read-rtl-function.c: New file.
33604 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
33605 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
33606 (selftest::verify_three_block_rtl_cfg): New decl.
33607 * read-rtl-function.h: New file.
33608 * read-rtl.c: Potentially include config.h rather than bconfig.h.
33609 For host, include function.h, memmodel.h, and emit-rtl.h.
33610 (one_time_initialization): New function.
33611 (struct compact_insn_name): New struct.
33612 (compact_insn_names): New array.
33613 (find_code): Handle insn codes in compact dumps.
33614 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
33615 (bind_subst_iter_and_attr): Likewise.
33616 (add_condition_to_string): Likewise.
33617 (add_condition_to_rtx): Likewise.
33618 (apply_attribute_uses): Likewise.
33619 (add_current_iterators): Likewise.
33620 (apply_iterators): Likewise.
33621 (initialize_iterators): Guard usage of apply_subst_iterator with
33622 #ifdef GENERATOR_FILE.
33623 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
33624 (md_reader::read_mapping): Likewise.
33625 (add_define_attr_for_define_subst): Likewise.
33626 (add_define_subst_attr): Likewise.
33627 (read_subst_mapping): Likewise.
33628 (check_code_iterator): Likewise.
33629 (rtx_reader::read_rtx): Likewise. Move one-time initialization
33631 (one_time_initialization): New function.
33632 (rtx_reader::read_until): New method.
33633 (read_flags): New function.
33634 (parse_reg_note_name): New function.
33635 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
33636 Handle reuse_rtx ids.
33637 Wrap iterator lookup within #ifdef GENERATOR_FILE.
33638 Add parsing support for RTL dumps, mirroring the special-cases in
33639 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
33640 values, and calling handle_any_trailing_information.
33641 (rtx_reader::read_rtx_operand): Convert return type from void
33642 to rtx, returning return_rtx. Handle case 'e'. Call
33643 finalize_string on XSTR and XTMPL fields.
33644 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
33645 "(nil)" values were omitted. Call the postprocess vfunc on the
33647 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
33648 class ctor. Initialize m_in_call_function_usage. Call
33649 one_time_initialization.
33650 * rtl-tests.c (selftest::test_uncond_jump): Call
33651 set_new_first_and_last_insn.
33652 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
33653 * selftest-rtl.c: New file.
33654 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
33655 (selftest::get_insn_by_uid): New decl.
33656 * selftest-run-tests.c (selftest::run_tests): Call
33657 read_rtl_function_c_tests.
33658 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
33659 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
33662 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
33664 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
33665 operands in a special way. Assert that pos+len <= mode precision.
33667 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33669 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
33670 3 argument Alias with unlimited for the negative form.
33671 (fno-vect-cost-model): Removed.
33673 2017-01-05 Martin Liska <mliska@suse.cz>
33675 * hsa-gen.c (gen_hsa_divmod): New function.
33676 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
33678 2017-01-05 Martin Liska <mliska@suse.cz>
33681 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
33684 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33686 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
33687 small constant length operands.
33689 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33691 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
33692 between loop iterations.
33694 2017-01-05 Martin Liska <mliska@suse.cz>
33697 * gimplify.c (gimplify_decl_expr): Compare to
33698 asan_poisoned_variables instread of checking flags.
33699 (gimplify_target_expr): Likewise.
33700 (gimplify_expr): Likewise.
33701 (gimplify_function_tree): Conditionally initialize
33702 asan_poisoned_variables.
33704 2017-01-04 Jeff Law <law@redhat.com>
33706 PR tree-optimizatin/78812
33707 * rtl.h (contains_mem_rtx_p): Prototype.
33708 * ifcvt.c (containts_mem_rtx_p): Move from here to...
33709 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
33710 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
33711 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
33712 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
33714 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33716 * input.c (assert_char_at_range): Default-initialize actual_range.
33718 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33720 * df-scan.c (df_ref_create_structure): Make regno unsigned,
33721 to match the caller.
33723 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33725 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
33726 insns after final jump in test to emit dummy move.
33728 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33730 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
33731 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
33733 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33735 * multiple_target.c (create_dispatcher_calls): Init e_next.
33736 * tree-ssa-loop-split.c (split_loop): Init border.
33737 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
33740 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
33745 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
33746 (altivec_register_operand): Do not return true if the operand
33747 contains a SUBREG mixing SImode and SFmode.
33748 (vsx_register_operand): Likewise.
33749 (vsx_reg_sfsubreg_ok): New predicate.
33750 (vfloat_operand): Do not return true if the operand contains a
33751 SUBREG mixing SImode and SFmode.
33752 (vint_operand): Likewise.
33753 (vlogical_operand): Likewise.
33754 (gpc_reg_operand): Likewise.
33755 (int_reg_operand): Likewise.
33756 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
33757 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
33758 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
33760 (rs6000_emit_move_si_sf_subreg): New helper function.
33761 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
33762 fixup SUBREGs involving SImode and SFmode.
33763 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
33764 numbers for the new peephole2 optimization.
33765 (peephole2 for SFmode unions): New peephole2 to optimize cases in
33766 the GLIBC math library that do AND/IOR/XOR operations on single
33767 precision floating point.
33768 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
33769 target macros to say whether we need to avoid SUBREGs mixing
33771 (TARGET_ALLOW_SF_SUBREG): Likewise.
33772 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
33773 (UNSPEC_SI_FROM_SF): Likewise.
33774 (iorxor): Change spacing.
33775 (and_ior_xor): New iterator for AND, IOR, and XOR.
33776 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
33777 (movdi_from_sf_zero_ext): Likewise.
33778 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
33779 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
33780 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
33781 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
33782 (fms<mode>4): Likewise.
33783 (fnma<mode>4): Likewise.
33784 (fnms<mode>4): Likewise.
33785 (nfma<mode>4): Likewise.
33786 (nfms<mode>4): Likewise.
33788 2017-01-04 Marek Polacek <polacek@redhat.com>
33791 * doc/invoke.texi: Document -Wpointer-compare.
33793 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33795 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
33798 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
33799 descriptions for -gdwarf-5 and emit them as uleb128 instead of
33802 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
33805 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
33806 documentation of the powerpc_popcntb_ok attribute.
33807 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
33808 code to issue warning messages if a requested CPU configuration is
33809 not supported by the binary (assembler and loader) toolchain.
33810 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
33811 made to define a built-in function that has been disabled.
33812 (paired_init_builtins): Add assertion to prevent ICE if attempt is
33813 made to define a built-in function that has been disabled.
33814 (altivec_init_builtins): Add comment explaining why definition
33815 of the DST built-in functions is not preceded by an assertion
33816 check. Add assertions to prevent ICE if attempts are made to
33817 define an altivec predicate or an abs* built-in function that has
33819 (htm_init_builtins): Add comment explaining why definition of the
33820 htm built-in functions is not preceded by an assertion check.
33822 2017-01-04 Jeff Law <law@redhat.com>
33824 PR tree-optimizatin/67955
33825 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
33826 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
33827 the points-to solution does not include pt_null. Use DECL_PT_UID
33830 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
33832 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
33833 Use gen_int_mode instead of gen_lopwart for const_int operands.
33835 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33837 PR tree-optimization/71563
33838 * match.pd: Simplify X << Y into X if Y is known to be 0 or
33839 out of range value - has low bits known to be zero.
33841 2017-01-04 Alan Modra <amodra@gmail.com>
33843 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
33844 * configure: Regenerate.
33845 * config.in: Regenerate.
33847 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33850 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
33851 a substring of the message, but strcmp with the whole message. Ifdef
33852 ENABLE_NLS, translate the message first using dgettext.
33854 2017-01-03 Jeff Law <law@redhat.com>
33856 PR tree-optimizatin/78856
33857 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
33858 (mark_threaded_blocks): Remove code to truncate thread paths that
33859 cross multiple loop headers. Instead invalidate the cached loop
33860 iteration information and handle case of a thread path walking
33861 into an irreducible region.
33863 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
33866 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
33867 assertions. Add support for doing the signbit if the IEEE 128-bit
33868 floating point value is in a GPR.
33869 * config/rs6000/rs6000.md (Fsignbit): Delete.
33870 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
33871 Update the length attribute if the value is in a GPR.
33872 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
33873 the sign or zero extension instruction, since the value is always 0/1.
33874 (signbit<mode>2_dm2): Delete using <Fsignbit>.
33877 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
33878 extracting SImode to a GPR register so that we can generate a
33879 store, limit the vector to be in a traditional Altivec register
33880 for the vextuwrx instruction.
33882 2017-01-03 Ian Lance Taylor <iant@google.com>
33884 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
33886 2017-01-03 Martin Sebor <msebor@redhat.com>
33888 PR tree-optimization/78696
33889 * gimple-ssa-sprintf.c (format_floating): Correct handling of
33890 precision. Use MPFR for %f for greater fidelity. Correct handling
33892 (pass_sprintf_length::compute_format_length): Set width and precision
33893 specified by asrerisk to void_node for vararg functions.
33894 (try_substitute_return_value): Adjust dump output.
33896 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
33898 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
33900 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
33902 * doc/invoke.texi (SPARC options): Document -mlra as the default.
33903 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
33904 -mlra/-mno-lra was passed to the compiler.
33906 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
33908 PR rtl-optimization/65618
33909 * emit-rtl.c (try_split): Move initialization of "before" and
33910 "after" to just before the call to emit_insn_after_setloc.
33912 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
33914 * doc/md.texi (Standard Names): Remove reference to Java frontend.
33916 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
33918 * dwarf2out.c (gen_enumeration_type_die): When
33919 -gno-strict-dwarf, add a DW_AT_encoding attribute.
33921 2017-01-03 Jakub Jelinek <jakub@redhat.com>
33923 PR tree-optimization/78965
33924 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
33925 Change first argument from const call_info & to call_info &. For %n
33926 set info.nowrite to false.
33928 PR middle-end/78901
33929 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
33930 possibly throwing calls.
33932 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
33933 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
33934 and fns handling, rather than in a separate case SSA_NAME.
33936 2017-01-02 Jeff Law <law@redhat.com>
33938 * config/darwin-driver.c (darwin_driver_init): Const-correctness
33939 fixes for first_period and second_period variables.
33941 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
33944 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
33945 (*insvqi_1): New insn pattern.
33946 (*insvqi_1_mem_rex64): Ditto.
33947 (*insvqi_2): Ditto.
33948 (*insvqi_3): Rename from *insvqi.
33950 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
33952 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
33954 * doc/cfg.texi (Edges): Remove reference to Java.
33955 (Maintaining the CFG): Ditto.
33957 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33959 PR middle-end/77674
33960 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
33961 transparent aliases.
33963 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33965 PR middle-end/77484
33966 * predict.def (PRED_CALL): Update hitrate.
33967 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
33968 * predict.c (tree_estimate_probability_bb): Split CALL predictor
33969 into direct/indirect/polymorphic variants.
33971 2017-01-01 Jakub Jelinek <jakub@redhat.com>
33973 Update copyright years.
33975 * gcc.c (process_command): Update copyright notice dates.
33976 * gcov-dump.c (print_version): Ditto.
33977 * gcov.c (print_version): Ditto.
33978 * gcov-tool.c (print_version): Ditto.
33979 * gengtype.c (create_file): Ditto.
33980 * doc/cpp.texi: Bump @copying's copyright year.
33981 * doc/cppinternals.texi: Ditto.
33982 * doc/gcc.texi: Ditto.
33983 * doc/gccint.texi: Ditto.
33984 * doc/gcov.texi: Ditto.
33985 * doc/install.texi: Ditto.
33986 * doc/invoke.texi: Ditto.
33988 Copyright (C) 2017 Free Software Foundation, Inc.
33990 Copying and distribution of this file, with or without modification,
33991 are permitted in any medium without royalty provided the copyright
33992 notice and this notice are preserved.