1 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
4 of two vec_duplicates into a vec_concat.
6 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
9 Simplify vec_merge of vec_duplicate and vec_concat.
10 * config/aarch64/constraints.md (Utq): New constraint.
11 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
14 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
17 Simplify vec_merge of vec_duplicate and const_vector.
18 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
20 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
21 mode iterator. Update predicate on operand 1 to
22 handle non-const_vec constants. Delete constraints.
23 (*aarch64_combinez_be<mode>): Likewise for operand 2.
25 2017-11-08 Jakub Jelinek <jakub@redhat.com>
27 PR tree-optimization/78821
28 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
30 (store_operand_info::store_operand_info): Initialize it to false.
31 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
32 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
33 store in the group, and if chain_info is non-NULL, to ignore altogether
35 (compatible_load_p): Fail if bit_not_p does not match.
36 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
37 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
38 (pass_store_merging::process_store): Adjust
39 terminate_all_aliasing_chains calls to pass NULL in all current spots,
40 call terminate_all_aliasing_chains newly when adding a store into
41 a chain with non-NULL chain_info.
43 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
45 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
47 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
49 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
51 (aarch64_layout_frame): Initialise emit_frame_chain.
52 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
53 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
55 2017-11-08 Martin Liska <mliska@suse.cz>
57 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
58 of gimple_call_internal_p.
60 2017-11-07 Tom de Vries <tom@codesourcery.com>
62 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
64 2017-11-07 Tom de Vries <tom@codesourcery.com>
66 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
69 2017-11-08 Martin Liska <mliska@suse.cz>
72 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
74 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
76 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
78 2017-11-07 Jakub Jelinek <jakub@redhat.com>
81 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
82 (*cmp<mode>_ccz_1): New insn with $k alternative.
85 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
86 RTX_COMM_COMPARE as commutative as well.
87 (ix86_binary_operator_ok): Formatting fix.
88 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
89 *<code><mode>3<mask_name><round_saeonly_name>,
90 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
91 *<s>mul<mode>3_highpart<mask_name>,
92 *vec_widen_umult_even_v16si<mask_name>,
93 *vec_widen_umult_even_v8si<mask_name>,
94 *vec_widen_umult_even_v4si<mask_name>,
95 *vec_widen_smult_even_v16si<mask_name>,
96 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
97 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
98 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
99 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
100 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
101 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
102 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
103 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
104 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
105 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
106 ix86_binary_operator_ok. Formatting fixes.
107 (*<plusminus_insn><mode>3<mask_name><round_name>,
108 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
111 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
113 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
114 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
115 eq and ne if TARGET_ISEL.
116 (cmp): New code_iterator.
117 (UNS, UNSU_, UNSIK): New code_attrs.
118 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
119 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
121 ("eq<mode>3"): ... this.
122 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
124 ("ne<mode>3"): ... this.
126 2017-11-07 Julia Koval <julia.koval@intel.com>
129 * common/config/i386/i386-common.c
130 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
131 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
132 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
133 * config/i386/i386.opt: Ditto.
134 * config/i386/i386.c (ix86_target_string): Ditto.
135 (ix86_option_override_internal): Ditto.
136 (ix86_init_mpx_builtins): Move MPX to args2.
137 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
138 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
139 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
140 __builtin_ia32_vgf2p8affineinvqb_v32qi,
141 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
142 __builtin_ia32_vgf2p8affineinvqb_v16qi,
143 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
145 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
148 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
149 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
150 (zero-extendsidi peephole2): Remove peephole.
152 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
155 * stor-layout.c: Include attribs.h.
156 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
157 explicit lookup of "aligned" attribute.
159 2017-11-07 Andrew Waterman <andrew@sifive.com>
161 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
162 (riscv_expand_block_move): Likewise.
163 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
165 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
166 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
167 * config/riscv/riscv.c (riscv_block_move_straight): New function.
168 (riscv_adjust_block_mem): Likewise.
169 (riscv_block_move_loop): Likewise.
170 (riscv_expand_block_move): Likewise.
171 gcc/config/riscv/riscv.md (movmemsi): New pattern.
173 2017-11-07 Michael Clark <michaeljclark@mac.com>
175 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
176 (MUSL_DYNAMIC_LINKER): Likewise.
178 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
180 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
181 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
184 2017-11-07 Richard Biener <rguenther@suse.de>
186 * match.pd: Fix build.
188 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
189 Jackson Woodruff <jackson.woodruff@arm.com>
191 PR tree-optimization/71026
192 * match.pd: Canonicalize negate in division.
194 2017-11-07 Sudakshina Das <sudi.das@arm.com>
197 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
199 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
201 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
202 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
204 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
206 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
207 non-scalar integral types.
208 * match.pd (negate_expr_p): Handle MINUS_EXPR.
209 (-(A-B), -(~A)): New transformations.
211 2017-11-07 Tom de Vries <tom@codesourcery.com>
213 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
214 semicolon after "do {} while (0)".
215 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
216 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
217 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
218 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
219 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
221 2017-11-07 Tom de Vries <tom@codesourcery.com>
223 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
224 after "do {} while (0)".
225 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
226 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
227 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
228 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
229 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
231 2017-11-07 Tom de Vries <tom@codesourcery.com>
234 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
236 (arm_rtx_costs_internal): Add missing semicolon after
237 HANDLE_NARROW_SHIFT_ARITH call.
239 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
241 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
242 disable isel if it was not set explicitly.
244 2017-11-06 James Bowman <james.bowman@ftdichip.com>
246 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
248 (add_type_attribute) likewise.
250 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
252 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
253 of struct ix86_frame.
254 (ix86_initial_elimination_offset): Likewise.
255 (ix86_expand_split_stack_prologue): Likewise.
257 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
259 * tree-vrp.h (enum value_range_type): Update stale comment.
261 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
262 Alan Hayward <alan.hayward@arm.com>
263 David Sherwood <david.sherwood@arm.com>
265 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
266 (aarch64_expand_vec_perm_const): Take the number of units too.
267 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
268 (aarch64_expand_vec_perm_const): Likewise.
269 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
270 (vec_perm<mode>): Update accordingly.
272 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
273 Alan Hayward <alan.hayward@arm.com>
274 David Sherwood <david.sherwood@arm.com>
276 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
277 Take the number of units too.
278 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
279 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
280 but check for a vector mode before rather than after the call.
281 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
282 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
283 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
284 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
285 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
286 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
287 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
288 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
289 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
290 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
291 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
292 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
293 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
294 (aarch64_sqdmull2_n<mode>): Update accordingly.
296 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
297 Alan Hayward <alan.hayward@arm.com>
298 David Sherwood <david.sherwood@arm.com>
300 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
301 the number of units too.
302 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
303 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
304 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
305 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
306 (vec_store_lanesxi<mode>): Update accordingly.
308 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
309 Alan Hayward <alan.hayward@arm.com>
310 David Sherwood <david.sherwood@arm.com>
312 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
313 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
314 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
315 of units rather than the mode.
316 * config/aarch64/iterators.md (nunits): New mode attribute.
317 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
318 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
319 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
320 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
321 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
322 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
323 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
324 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
325 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
326 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
327 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
328 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
329 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
330 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
331 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
332 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
333 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
334 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
335 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
336 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
337 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
338 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
339 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
340 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
341 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
342 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
343 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
344 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
345 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
346 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
347 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
348 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
349 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
350 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
351 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
352 (aarch64_simd_vec_setv2di): Likewise.
354 2017-11-06 Carl Love <cel@us.ibm.com>
356 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
358 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
359 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
361 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
363 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
364 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
365 Add power 8 macro expansions.
366 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
367 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
368 power 8 instructions. (VSX_XXBR): Add iterator.
370 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
372 * config/arm/arm.md (predicable_short_it): Change default to "no",
373 improve documentation, remove uses that are identical to the default.
374 (enabled_for_depr_it): Rename to enabled_for_short_it.
375 * gcc/config/arm/arm-fixed.md (predicable_short_it): Remove default uses.
376 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
377 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
378 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
379 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
381 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
384 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
385 float128 helper macros, which are no longer used after deleting
386 the old 'q' built-in functions, and moving the round to odd
387 built-in functions to being special built-in functions.
388 (BU_FLOAT128_2): Likewise.
389 (BU_FLOAT128_1_HW): Likewise.
390 (BU_FLOAT128_2_HW): Likewise.
391 (BU_FLOAT128_3_HW): Likewise.
392 (FABSQ): Delete old 'q' built-in functions.
393 (COPYSIGNQ): Likewise.
394 (SQRTF128_ODD): Move round to odd built-in functions to be
395 special built-in functions, so that we can handle
396 -mabi=ieeelongdouble.
397 (TRUNCF128_ODD): Likewise.
398 (ADDF128_ODD): Likewise.
399 (SUBF128_ODD): Likewise.
400 (MULF128_ODD): Likewise.
401 (DIVF128_ODD): Likewise.
402 (FMAF128_ODD): Likewise.
403 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
404 built-in names to 'f128'.
405 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
406 old 'q' built-in functions, as the machine independent code for
407 'f128' built-in functions handles this.
408 (rs6000_expand_builtin): Add expansion for float128 round to odd
409 functions, keying off on -mabi=ieeelongdouble of whether to use
410 the KFmode or TFmode variant.
411 (rs6000_init_builtins): Initialize the _Float128 round to odd
413 * doc/extend.texi (PowerPC Built-in Functions): Document the old
414 _Float128 'q' built-in functions are now mapped into the new
415 'f128' built-in functions.
417 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
419 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
420 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
422 2017-11-06 David Malcolm <dmalcolm@redhat.com>
425 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
426 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
427 * toplev.c: Include "ipa-fnsummary.h".
428 (toplev::finalize): Call ipa_fnsummary_c_finalize.
430 2017-11-06 Jakub Jelinek <jakub@redhat.com>
432 PR tree-optimization/82838
433 * gimple-ssa-store-merging.c
434 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
435 on a separate gimple_seq which is then appended to seq.
437 2017-11-06 Jeff Law <law@redhat.com>
440 * config/i386/i386.c (PROBE_INTERVAL): Remove.
441 (get_probe_interval): New functions.
442 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
443 (ix86_adjust_stack_and_probe): Likewise.
444 (output_adjust_stack_and_probe): Likewise.
445 (ix86_emit_probe_stack_range): Likewise.
446 (ix86_expand_prologue): Likewise.
448 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
450 PR tree-optimization/82816
451 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
452 if the modes of the two types are the same.
453 (convert_plusminus_to_widen): Likewise.
455 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
457 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
459 (usadv16qi): New define_expand.
460 (usadv8hi): New define_expand.
462 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
465 * ipa-inline-transform.c (update_noncloned_frequencies): Always
467 (inline_transform): Likewise.
468 * predict.c (counts_to_freqs): Remove useless conditional.
469 * profile-count.h (profile_count::apply_scale): Move sanity check.
470 * tree-inline.c (copy_bb): Always scale.
471 (copy_cfg_body): Likewise.
473 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
476 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
478 (*cmp_ite0): Add enabled_for_depr_it attribute.
479 (*cmp_ite1): Likewise.
481 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
483 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
486 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
488 * tree-vrp.c (vrp_int_const_binop): Return true on success and
489 return the value by pointer.
490 (extract_range_from_multiplicative_op_1): Update accordingly.
491 Return as soon as an operation fails.
493 2017-11-05 Tom de Vries <tom@codesourcery.com>
496 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
497 (DEF_SANITIZER_BUILTIN): ... here.
498 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
499 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
501 2017-11-05 Tom de Vries <tom@codesourcery.com>
504 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
506 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
507 ASM_OUTPUT_BEFORE_CASE_LABEL call.
508 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
510 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
511 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
512 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
514 2017-11-05 Tom de Vries <tom@codesourcery.com>
517 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
520 2017-11-04 Michael Clark <michaeljclark@mac.com>
522 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
523 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
525 (*addsi3_extended): Likewise.
526 (*addsi3_extended2): Likewise.
527 (<optab>si3): Likewise.
528 (<optab>di3): Likewise.
529 (<optab><mode>3): Likewise.
530 (<*optabe>si3_internal): Likewise.
531 (zero_extendqi<SUPERQI:mode>2): Likewise.
532 (*add<mode>hi3): Likewise.
533 (*xor<mode>hi3): Likewise.
534 (<optab>di3): Likewise.
535 (*<optab>si3_extend): Likewise.
536 (*sge<u>_<X:mode><GPR:mode>): Likewise.
537 (*slt<u>_<X:mode><GPR:mode>): Likewise.
538 (*sle<u>_<X:mode><GPR:mode>): Likewise.
540 2017-11-04 Andrew Waterman <andrew@sifive.com>
542 * config/riscv/riscv.c (riscv_option_override): Conditionally set
543 TARGET_STRICT_ALIGN based upon -mtune argument.
545 2017-11-04 Andrew Waterman <andrew@sifive.com>
547 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
549 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
551 config/i386/i386.c (choose_basereg): Use optional scratch
552 register and add assertion.
553 (x86_emit_outlined_ms2sysv_save): Use scratch register when
554 needed, and don't allocate stack.
555 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
556 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
557 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
559 2017-11-03 Jeff Law <law@redhat.com>
561 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
562 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
563 to probe at the start of a noreturn function.
565 2017-11-03 Jakub Jelinek <jakub@redhat.com>
567 PR tree-optimization/78821
568 * gimple-ssa-store-merging.c: Update the file comment.
569 (MAX_STORE_ALIAS_CHECKS): Define.
570 (struct store_operand_info): New type.
571 (store_operand_info::store_operand_info): New constructor.
572 (struct store_immediate_info): Add rhs_code and ops data members.
573 (store_immediate_info::store_immediate_info): Add rhscode, op0r
574 and op1r arguments to the ctor, initialize corresponding data members.
575 (struct merged_store_group): Add load_align_base and load_align
577 (merged_store_group::merged_store_group): Initialize them.
578 (merged_store_group::do_merge): Update them.
579 (merged_store_group::apply_stores): Pick the constant for
580 encode_tree_to_bitpos from one of the two operands, or skip
581 encode_tree_to_bitpos if neither operand is a constant.
582 (class pass_store_merging): Add process_store method decl. Remove
583 bool argument from terminate_all_aliasing_chains method decl.
584 (pass_store_merging::terminate_all_aliasing_chains): Remove
585 var_offset_p argument and corresponding handling.
586 (stmts_may_clobber_ref_p): New function.
587 (compatible_load_p): New function.
588 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
589 if there is overlap and rhs_code is not INTEGER_CST. For
590 non-overlapping stores terminate group if rhs is not mergeable.
591 (get_alias_type_for_stmts): Change first argument from
592 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
593 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
594 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
596 (get_location_for_stmts): Change first argument from
597 auto_vec<gimple *> & to vec<gimple *> &.
598 (struct split_store): Remove orig_stmts data member, add orig_stores.
599 (split_store::split_store): Create orig_stores rather than orig_stmts.
600 (find_constituent_stmts): Renamed to ...
601 (find_constituent_stores): ... this. Change second argument from
602 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
603 to info structures rather than the statements.
604 (split_group): Rename ALLOW_UNALIGNED argument to
605 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
606 it. Adjust find_constituent_stores caller.
607 (imm_store_chain_info::output_merged_store): Handle rhs_code other
608 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
609 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
610 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
611 (mem_valid_for_store_merging): New function.
612 (handled_load): New function.
613 (pass_store_merging::process_store): New method.
614 (pass_store_merging::execute): Use process_store method. Adjust
615 terminate_all_aliasing_chains caller.
617 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
619 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
620 Return true for more constants, symbols and label references.
621 (aarch64_valid_floating_const): Remove unused function.
623 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
626 * config/aarch64/aarch64.c (aarch64_layout_frame):
627 Undo forcing of LR at bottom of frame.
629 2017-11-03 Jeff Law <law@redhat.com>
632 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
633 for int_registers_saved.
635 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
636 extracted from tree-ssa-dom.c.
637 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
638 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
639 (record_equivalences_from_incoming_edge): Add additional argument
640 to single_pred_edge_ignoring_loop_edges call.
641 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
642 (uncprop_dom_walker::before_dom_children): Add additional argument
643 to single_pred_edge_ignoring_loop_edges call.
644 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
645 single_pred_edge_ignoring_loop_edges rather than open coding.
646 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
648 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
650 * match.pd (-(-A)): Rewrite.
652 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
654 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
655 (rs6000_emit_int_cmove): New declaration.
656 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
657 (rs6000_emit_sISEL): Delete.
658 (rs6000_emit_int_cmove): Make non-static.
659 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
660 instead of rs6000_emit_sISEL.
662 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
664 * asan.c (create_cond_insert_point): Maintain profile.
665 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
667 * basic-block.h (struct basic_block_def): Remove frequency.
668 (EDGE_FREQUENCY): Use to_frequency
669 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
671 (find_traces): Update to use to_frequency.
672 (find_traces_1_round): Likewise; use only IPA counts.
673 (bb_to_key): Likewise.
674 (connect_traces): Use IPA counts only.
675 (copy_bb_p): Update to use to_frequency.
676 (fix_up_crossing_landing_pad): Likewise.
677 (sanitize_hot_paths): Likewise.
678 * bt-load.c (basic_block_freq): Likewise.
679 * cfg.c (init_flow): Set count_max to uninitialized.
680 (check_bb_profile): Remove frequencies; check counts.
681 (dump_bb_info): Do not dump frequencies.
682 (update_bb_profile_for_threading): Update counts only.
683 (scale_bbs_frequencies_int): Likewise.
684 (MAX_SAFE_MULTIPLIER): Remove.
685 (scale_bbs_frequencies_gcov_type): Update counts only.
686 (scale_bbs_frequencies_profile_count): Update counts only.
687 (scale_bbs_frequencies): Update counts only.
688 * cfg.h (struct control_flow_graph): Add count-max.
689 (update_bb_profile_for_threading): Update prototype.
690 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
691 (find_many_sub_basic_blocks): Likewise.
692 * cfgcleanup.c (try_forward_edges): Likewise.
693 (try_crossjump_to_edge): Likewise.
694 * cfgexpand.c (expand_gimple_cond): Likewise.
695 (expand_gimple_tailcall): Likewise.
696 (construct_init_block): Likewise.
697 (construct_exit_block): Likewise.
698 * cfghooks.c (verify_flow_info): Check consistency of counts.
699 (dump_bb_for_graph): Do not dump frequencies.
700 (split_block_1): Do not update frequencies.
701 (split_edge): Do not update frequencies.
702 (make_forwarder_block): Do not update frequencies.
703 (duplicate_block): Do not update frequencies.
704 (account_profile_record): Do not update frequencies.
705 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
706 for global heuristics.
707 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
708 (expected_loop_iterations_unbounded): Use counts only.
709 * cfgloopmanip.c (scale_loop_profile): Simplify.
710 (create_empty_loop_on_edge): Simplify
712 (duplicate_loop_to_header_edge): Simplify
713 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
714 (update_br_prob_note): Take care of removing note when profile
716 (relink_block_chain): Do not dump frequency.
717 (rtl_account_profile_record): Use to_frequency.
718 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
719 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
720 (cgraph_update_edges_for_call_stmt_node): Likewise.
721 (cgraph_edge::verify_count_and_frequency): Update.
722 (cgraph_node::verify_node): Temporarily disable frequency verification.
723 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
725 (cgraph_edge::rebuild_edges): Convert to ipa counts.
726 * cgraphunit.c (init_lowered_empty_function): Do not initialize
728 (cgraph_node::expand_thunk): Update profile.
729 * except.c (dw2_build_landing_pads): Do not update frequency.
730 * final.c (compute_alignments): Use to_frequency.
731 (dump_basic_block_info): Do not dump frequency.
732 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
733 (dump_gimple_bb_header): Do not dump frequency.
734 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
736 * gimple-streamer-in.c (input_bb): Do not stream frequency.
737 * gimple-streamer-out.c (output_bb): Do not stream frequency.
738 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
739 (init_before_recovery): Do not update frequency.
740 (sched_create_recovery_edges): Do not update frequency.
741 * hsa-gen.c (convert_switch_statements): Do not update frequency.
742 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
743 (ipa_cp_c_finalize): Set max_count to uninitialized.
744 * ipa-fnsummary.c (get_minimal_bb): Use counts.
745 (param_change_prob): Use counts.
746 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
748 * ipa-split.c (consider_split): Use to_frequency.
749 (split_function): Use to_frequency.
750 * ira-build.c (loop_compare_func): Likewise.
751 (mark_loops_for_removal): Likewise.
752 (mark_all_loops_for_removal): Likewise.
753 * loop-doloop.c (doloop_modify): Do not update frequency.
754 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
756 * lto-streamer-in.c (input_function): Update count_max.
757 * omp-expand.c (expand_omp_taskreg): Update count_max.
758 * omp-simd-clone.c (simd_clone_adjust): Update profile.
759 * predict.c (maybe_hot_frequency_p): Use to_frequency.
760 (maybe_hot_count_p): Use ipa counts only.
761 (maybe_hot_bb_p): Simplify.
762 (maybe_hot_edge_p): Simplify.
763 (probably_never_executed): Do not take frequency argument.
764 (probably_never_executed_bb_p): Do not pass frequency.
765 (probably_never_executed_edge_p): Likewise.
766 (combine_predictions_for_bb): Check that profile is nonzero.
767 (propagate_freq): Do not set frequency.
768 (drop_profile): Simplify.
769 (counts_to_freqs): Simplify.
770 (expensive_function_p): Use to_frequency.
771 (propagate_unlikely_bbs_forward): Simplify.
772 (determine_unlikely_bbs): Simplify.
773 (estimate_bb_frequencies): Add hack to silence graphite issues.
774 (compute_function_frequency): Use ipa counts.
775 (pass_profile::execute): Update.
776 (rebuild_frequencies): Use counts only.
777 (force_edge_cold): Use counts only.
778 * profile-count.c (profile_count::dump): Dump new count types.
779 (profile_count::differs_from_p): Check compatiblity.
780 (profile_count::to_frequency): New function.
781 (profile_count::to_cgraph_frequency): New function.
782 * profile-count.h (struct function): Declare.
783 (enum profile_quality): Add profile_guessed_local and
784 profile_guessed_global0.
785 (class profile_proability): Decrease number of bits to 29;
786 update from_reg_br_prob_note and to_reg_br_prob_note.
787 (class profile_count: Update comment; decrease number of bits
788 to 61. Check compatibility.
789 (profile_count::compatible_p): New private member function.
790 (profile_count::ipa_p): New member function.
791 (profile_count::operator<): Handle global zero correctly.
792 (profile_count::operator>): Handle global zero correctly.
793 (profile_count::operator<=): Handle global zero correctly.
794 (profile_count::operator>=): Handle global zero correctly.
795 (profile_count::nonzero_p): New member function.
796 (profile_count::force_nonzero): New member function.
797 (profile_count::max): New member function.
798 (profile_count::apply_scale): Handle IPA scalling.
799 (profile_count::guessed_local): New member function.
800 (profile_count::global0): New member function.
801 (profile_count::ipa): New member function.
802 (profile_count::to_frequency): Declare.
803 (profile_count::to_cgraph_frequency): Declare.
804 * profile.c (OVERLAP_BASE): Delete.
805 (compute_frequency_overlap): Delete.
806 (compute_branch_probabilities): Do not use compute_frequency_overlap.
807 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
808 * sched-ebb.c (rank): Use counts only.
809 * shrink-wrap.c (handle_simple_exit): Use counts only.
810 (try_shrink_wrapping): Use counts only.
811 (place_prologue_for_one_component): Use counts only.
812 * tracer.c (find_best_predecessor): Use to_frequency.
813 (find_trace): Use to_frequency.
814 (tail_duplicate): Use to_frequency.
815 * trans-mem.c (expand_transaction): Do not update frequency.
816 * tree-call-cdce.c: Do not update frequency.
817 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
818 (gimple_merge_blocks): Likewise.
819 (gimple_split_edge): Likewise.
820 (gimple_duplicate_sese_region): Likewise.
821 (gimple_duplicate_sese_tail): Likewise.
822 (move_sese_region_to_fn): Likewise.
823 (gimple_account_profile_record): Likewise.
824 (insert_cond_bb): Likewise.
825 * tree-complex.c (expand_complex_div_wide): Likewise.
826 * tree-eh.c (lower_resx): Update profile.
827 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
829 (initialize_cfun): Do not initialize frequencies
830 (freqs_to_counts): Delete.
831 (copy_cfg_body): Ignore count parameter.
833 (expand_call_inline): Update count_max.
834 (optimize_inline_calls): Update count_max.
835 (tree_function_versioning): Update count_max.
836 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
837 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
839 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
840 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
841 (try_peel_loop): Likewise.
842 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
844 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
845 (tree_transform_and_unroll_loop): Do not use frequencies
846 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
847 Use reliable prediction only.
848 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
849 * tree-ssa-sink.c (select_best_block): Use to_frequency.
850 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
852 * tree-ssa-threadupdate.c (create_block_for_threading): Do
854 (any_remaining_duplicated_blocks): Likewise.
855 (update_profile): Likewise.
856 (estimated_freqs_path): Delete.
857 (freqs_to_counts_path): Delete.
858 (clear_counts_path): Delete.
859 (ssa_fix_duplicate_block_edges): Likewise.
860 (duplicate_thread_path): Likewise.
861 * tree-switch-conversion.c (gen_inbound_check): Use counts.
862 * tree-tailcall.c (decrease_profile): Do not update frequency.
863 (eliminate_tail_call): Likewise.
864 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
865 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
866 (optimize_mask_stores): Likewise.
867 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
868 * ubsan.c (ubsan_expand_null_ifn): Update profile.
869 (ubsan_expand_ptr_ifn): Update profile.
870 * value-prof.c (gimple_ic): Simplify.
871 * value-prof.h (gimple_ic): Update prototype.
872 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
873 * ipa-inline.c (compute_uninlined_call_time): Be sure that
875 (want_inline_self_recursive_call_p): Likewise.
876 (resolve_noninline_speculation): Only cummulate defined counts.
877 (inline_small_functions): Use nonzero_p.
878 (ipa_inline): Do not access freed node.
880 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
882 * config/aarch64/aarch64.c (aarch64_override_options_internal):
883 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
885 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
887 * config/riscv/riscv.c (riscv_legitimize_move): Handle
888 non-legitimate address.
890 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
892 * config/rs6000/rs6000.md (*lt0_disi): Delete.
893 (*lt0_<mode>di, *lt0_<mode>si): New.
895 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
897 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
900 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
901 Jim Wilson <jim.wilson@linaro.org>
903 * config/aarch64/aarch64-cores.def (saphira): New CPU.
904 * config/aarch64/aarch64-tune.md: Regenerated.
905 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
906 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
908 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
910 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
911 (arc_expand_prologue): Restore blink for millicode.
912 * config/arc/linux.h (LINK_EH_SPEC): Defined.
914 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
917 * config/i386/i386.c (ix86_vector_duplicate_value): Use
918 gen_vec_duplicate after forcing the scalar into a register.
920 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
922 * combine (try_combine): Print the insns input to try_combine to the
925 2017-11-02 Steve Ellcey <sellcey@cavium.com>
928 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
929 Remove second argument from aarch64_process_target_attr call.
930 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
932 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
934 (aarch64_handle_attr_arch): Remove second argument.
935 (aarch64_handle_attr_cpu): Ditto.
936 (aarch64_handle_attr_tune): Ditto.
937 (aarch64_handle_attr_isa_flags): Ditto.
938 (aarch64_process_one_target_attr): Ditto.
939 (aarch64_process_target_attr): Ditto.
940 (aarch64_option_valid_attribute_p): Remove second argument.
941 on aarch64_process_target_attr call.
943 2017-11-02 David Malcolm <dmalcolm@redhat.com>
945 * diagnostic.c: Include "selftest-diagnostic.h".
946 (selftest::assert_location_text): New function.
947 (selftest::test_diagnostic_get_location_text): New function.
948 (selftest::diagnostic_c_tests): Call it.
950 2017-11-02 David Malcolm <dmalcolm@redhat.com>
952 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
953 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
954 (class selftest::test_diagnostic_context): Move to...
955 * selftest-diagnostic.c: New file.
956 * selftest-diagnostic.h: New file.
958 2017-11-02 James Bowman <james.bowman@ftdichip.com>
960 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
961 offset range for FT32B.
962 * config/ft32/ft32.h: option "mcompress" enables relaxation.
963 * config/ft32/ft32.md: Add TARGET_NOPM.
964 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
965 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
967 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
969 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
971 2017-11-02 Jeff Law <law@redhat.com>
973 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
974 virtual keyword on FINAL OVERRIDE members.
976 * tree-ssa-propagate.h (ssa_propagation_engine): Group
977 virtuals together. Add virtual destructor.
978 (substitute_and_fold_engine): Similarly.
980 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
982 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
984 2017-11-02 Richard Biener <rguenther@suse.de>
986 PR tree-optimization/82795
987 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
989 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
991 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
993 (gcc_GAS_CHECK_FEATURE): Remove.
994 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
995 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
996 * configure: Regenerate.
998 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
1000 * config/arc/arc.c (hwloop_optimize): Account for empty
1003 2017-11-02 Richard Biener <rguenther@suse.de>
1006 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
1007 Truncate ARRAY_REF index and element size.
1009 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1011 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
1013 2017-11-01 Jeff Law <law@redhat.com>
1015 * tree-ssa-ccp.c (ccp_folder): New class derived from
1016 substitute_and_fold_engine.
1017 (ccp_folder::get_value): New member function.
1018 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
1019 (ccp_fold_stmt): Remove prototype.
1020 (ccp_finalize): Call substitute_and_fold from the ccp_class.
1021 * tree-ssa-copy.c (copy_folder): New class derived from
1022 substitute_and_fold_engine.
1023 (copy_folder::get_value): Renamed from get_value.
1024 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
1025 * tree-vrp.c (vrp_folder): New class derived from
1026 substitute_and_fold_engine.
1027 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
1028 (vrp_folder::get_value): New member function.
1029 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
1030 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
1031 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
1032 provide a class interface to folder/substitute routines.
1033 (ssa_prop_fold_stmt_fn): Remove typedef.
1034 (ssa_prop_get_value_fn): Likewise.
1035 (subsitute_and_fold): Remove prototype.
1036 (replace_uses_in): Likewise.
1037 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
1038 Renamed from replace_uses_in. Call the virtual member function
1039 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
1040 (substitute_and_fold_dom_walker): Remove initialization of
1041 data member entries for calbacks. Add substitute_and_fold_engine
1042 member and initialize it.
1043 (substitute_and_fold_dom_walker::before_dom_children0: Use the
1044 member functions for get_value, replace_phi_args_in c
1045 replace_uses_in, and fold_stmt calls.
1046 (substitute_and_fold_engine::substitute_and_fold): Renamed from
1047 substitute_and_fold. Remove assert. Update ctor call.
1049 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
1050 (ssa_prop_visit_phi_fn): Likewise.
1051 (class ssa_propagation_engine): New class to provide an interface
1053 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
1055 (ssa_prop_visit_phi): Likewise.
1056 (ssa_propagation_engine::simulate_stmt): Moved into class.
1057 Call visit_phi/visit_stmt from the class rather than via
1058 file scoped static variables.
1059 (ssa_propagation_engine::simulate_block): Moved into class.
1060 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
1061 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
1062 set file scoped statics for the visit_stmt/visit_phi callbacks.
1063 * tree-complex.c (complex_propagate): New class derived from
1064 ssa_propagation_engine.
1065 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
1066 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
1067 (tree_lower_complex): Call ssa_propagate via the complex_propagate
1069 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
1070 ssa_propagation_engine.
1071 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
1072 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
1073 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
1074 * tree-ssa-copy.c (copy_prop): New class derived from
1075 ssa_propagation_engine.
1076 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
1077 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
1078 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
1079 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
1080 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
1081 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
1082 (execute_vrp): Call ssa_propagate from the vrp_prop class.
1084 2017-11-01 Jakub Jelinek <jakub@redhat.com>
1086 PR rtl-optimization/82778
1087 PR rtl-optimization/82597
1088 * compare-elim.c (struct comparison): Add in_a_setter field.
1089 (find_comparison_dom_walker::before_dom_children): Remove killed
1090 bitmap and df_simulate_find_defs call, instead walk the defs.
1091 Compute last_setter and initialize in_a_setter. Merge definitions
1092 with first initialization for a few variables.
1093 (try_validate_parallel): Use insn_invalid_p instead of
1094 recog_memoized. Return insn rather than just the pattern.
1095 (try_merge_compare): Fix up comment. Don't uselessly test if
1096 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
1098 (execute_compare_elim_after_reload): Remove df_chain_add_problem
1101 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1102 Alan Hayward <alan.hayward@arm.com>
1103 David Sherwood <david.sherwood@arm.com>
1105 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
1106 aarch64_hard_regno_nregs to get the number of registers
1109 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1110 Alan Hayward <alan.hayward@arm.com>
1111 David Sherwood <david.sherwood@arm.com>
1113 * config/aarch64/constraints.md (Upl): Rename to...
1115 * config/aarch64/aarch64.md
1116 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
1119 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1120 Alan Hayward <alan.hayward@arm.com>
1121 David Sherwood <david.sherwood@arm.com>
1123 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
1124 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
1127 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1128 Alan Hayward <alan.hayward@arm.com>
1129 David Sherwood <david.sherwood@arm.com>
1131 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
1132 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
1133 (aarch64_evpc_dup): Generate rtl direcly, rather than using
1135 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
1136 of a single element.
1137 * config/aarch64/iterators.md: Add a comment above the permute
1138 unspecs to say that they are generated directly by
1139 aarch64_expand_vec_perm_const.
1140 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
1142 2017-11-01 Nathan Sidwell <nathan@acm.org>
1144 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
1146 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1148 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
1149 and medany code models, and describe what they do.
1151 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1153 Revert accidental duplicate:
1155 * combine.c (can_change_dest_mode): Reject changes in
1156 REGMODE_NATURAL_SIZE.
1158 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
1160 PR rtl-optimization/64682
1161 PR rtl-optimization/69567
1162 PR rtl-optimization/69737
1163 PR rtl-optimization/82683
1164 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
1165 register mentioned in the note, drop the note, unless it came from I3,
1166 in which case it should go to I3 again.
1168 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1170 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
1171 and return false if not.
1172 (clear_bytes_written_by, live_bytes_read): Update accordingly.
1174 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1176 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
1177 range is known to be empty.
1179 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1180 Alan Hayward <alan.hayward@arm.com>
1181 David Sherwood <david.sherwood@arm.com>
1183 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
1184 and CONST_VECTOR_NUNITS instead of computing the number of units from
1185 the byte sizes of the vector and element.
1186 (simplify_binary_operation_1): Likewise.
1187 (simplify_const_binary_operation): Likewise.
1188 (simplify_ternary_operation): Likewise.
1190 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1191 Alan Hayward <alan.hayward@arm.com>
1192 David Sherwood <david.sherwood@arm.com>
1194 * var-tracking.c (INT_MEM_OFFSET): Replace with...
1195 (int_mem_offset): ...this new function.
1196 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
1197 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
1198 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
1201 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1202 Alan Hayward <alan.hayward@arm.com>
1203 David Sherwood <david.sherwood@arm.com>
1205 * lower-subreg.c (interesting_mode_p): New function.
1206 (compute_costs, find_decomposable_subregs, decompose_register)
1207 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
1208 (resolve_clobber, dump_choices): Use it.
1210 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1211 Alan Hayward <alan.hayward@arm.com>
1212 David Sherwood <david.sherwood@arm.com>
1214 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
1216 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1217 Alan Hayward <alan.hayward@arm.com>
1218 David Sherwood <david.sherwood@arm.com>
1220 * alias.c (find_base_value, find_base_term): Only process integer
1221 truncations. Check the precision rather than the size.
1223 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1224 Alan Hayward <alan.hayward@arm.com>
1225 David Sherwood <david.sherwood@arm.com>
1227 * machmode.h (is_narrower_int_mode): New function
1228 * optabs.c (expand_float, expand_fix): Use it.
1229 * dwarf2out.c (rotate_loc_descriptor): Likewise.
1231 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1232 Alan Hayward <alan.hayward@arm.com>
1233 David Sherwood <david.sherwood@arm.com>
1235 * rtl.h (narrower_subreg_mode): New function.
1236 * ira-color.c (update_costs_from_allocno): Use it.
1238 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1239 Alan Hayward <alan.hayward@arm.com>
1240 David Sherwood <david.sherwood@arm.com>
1242 * optabs-query.h (convert_optab_p): New function, split out from...
1243 (convert_optab_handler): ...here.
1244 (widening_optab_handler): Delete.
1245 (find_widening_optab_handler): Remove permit_non_widening parameter.
1246 (find_widening_optab_handler_and_mode): Likewise. Provide an
1247 override that operates on mode class wrappers.
1248 * optabs-query.c (widening_optab_handler): Delete.
1249 (find_widening_optab_handler_and_mode): Remove permit_non_widening
1250 parameter. Assert that the two modes are the same class and that
1251 the "from" mode is narrower than the "to" mode. Use
1252 convert_optab_handler instead of widening_optab_handler.
1253 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
1254 instead of widening_optab_handler.
1255 * expr.c (expand_expr_real_2): Update calls to
1256 find_widening_optab_handler.
1257 * optabs.c (expand_widen_pattern_expr): Likewise.
1258 (expand_binop_directly): Take the insn_code as a parameter.
1259 (expand_binop): Only call find_widening_optab_handler for
1260 conversion optabs; use optab_handler otherwise. Update calls
1261 to find_widening_optab_handler and expand_binop_directly.
1262 Use convert_optab_handler instead of widening_optab_handler.
1263 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
1264 find_widening_optab_handler and use scalar_mode rather than
1266 (convert_plusminus_to_widen): Likewise.
1268 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1269 Alan Hayward <alan.hayward@arm.com>
1270 David Sherwood <david.sherwood@arm.com>
1272 * machmode.h (fixed_size_mode): New class.
1273 * rtl.h (get_pool_mode): Return fixed_size_mode.
1274 * gengtype.c (main): Add fixed_size_mode.
1275 * target.def (get_raw_result_mode): Return a fixed_size_mode.
1276 (get_raw_arg_mode): Likewise.
1277 * doc/tm.texi: Regenerate.
1278 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
1279 * targhooks.c (default_get_reg_raw_mode): Likewise.
1280 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
1281 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
1282 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
1283 (msp430_get_raw_result_mode): Likewise.
1284 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
1285 * dbxout.c (dbxout_parms): Require fixed-size modes.
1286 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
1287 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
1288 * omp-low.c (lower_oacc_reductions): Likewise.
1289 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
1290 (simplify_subreg): Update accordingly.
1291 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
1292 (force_const_mem): Update accordingly. Return NULL_RTX for modes
1293 that aren't fixed-size.
1294 (get_pool_mode): Return a fixed_size_mode.
1295 (output_constant_pool_2): Take a fixed_size_mode.
1297 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1298 Alan Hayward <alan.hayward@arm.com>
1299 David Sherwood <david.sherwood@arm.com>
1301 * doc/rtl.texi (vec_series): Document.
1302 (const): Say that the operand can be a vec_series.
1303 * rtl.def (VEC_SERIES): New rtx code.
1304 * rtl.h (const_vec_series_p_1): Declare.
1305 (const_vec_series_p): New function.
1306 * emit-rtl.h (gen_const_vec_series): Declare.
1307 (gen_vec_series): Likewise.
1308 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
1309 (gen_vec_series): Likewise.
1310 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
1311 * simplify-rtx.c (simplify_unary_operation): Handle negations
1313 (simplify_binary_operation_series): New function.
1314 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
1315 (test_vector_ops_series): New function.
1316 (test_vector_ops): Call it.
1317 * config/powerpcspe/altivec.md (altivec_lvsl): Use
1318 gen_const_vec_series.
1319 (altivec_lvsr): Likewise.
1320 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
1322 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1323 Alan Hayward <alan.hayward@arm.com>
1324 David Sherwood <david.sherwood@arm.com>
1326 * doc/rtl.texi (const): Update description of address constants.
1327 Say that vector constants are allowed too.
1328 * common.md (E, F): Use CONSTANT_P instead of checking for
1330 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
1331 checking for CONST_VECTOR.
1332 * expmed.c (make_tree): Use build_vector_from_val for a CONST
1334 * expr.c (expand_expr_real_2): Check for vector modes instead
1335 of checking for CONST_VECTOR.
1336 * rtl.h (const_vec_p): New function.
1337 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
1338 (unwrap_const_vec_duplicate): Handle them here too.
1340 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1341 David Malcolm <dmalcolm@redhat.com>
1342 Alan Hayward <alan.hayward@arm.com>
1343 David Sherwood <david.sherwood@arm.com>
1345 * rtl.h (vec_duplicate_p): New function.
1346 * selftest-rtl.c (assert_rtx_eq_at): New function.
1347 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
1348 (assert_rtx_eq_at): Declare.
1349 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
1350 * selftest-run-tests.c (selftest::run_tests): Call it.
1351 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
1352 (simplify_unary_operation_1): Recursively handle vector duplicates.
1353 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
1355 (simplify_subreg): Handle subregs of vector duplicates.
1356 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
1357 (selftest::simplify_rtx_c_tests): New functions.
1359 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1360 Alan Hayward <alan.hayward@arm.com>
1361 David Sherwood <david.sherwood@arm.com>
1363 * emit-rtl.h (gen_const_vec_duplicate): Declare.
1364 (gen_vec_duplicate): Likewise.
1365 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
1367 (gen_const_vector): ...here.
1368 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
1369 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
1370 whose elements are all equal.
1371 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
1372 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
1373 (simplify_relational_operation): Likewise.
1374 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
1376 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
1377 (aarch64_expand_vector_init): Likewise.
1378 * config/arm/arm.c (neon_vdup_constant): Likewise.
1379 (neon_expand_vector_init): Likewise.
1380 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
1381 (arm_block_set_unaligned_vect): Likewise.
1382 (arm_block_set_aligned_vect): Likewise.
1383 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
1384 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
1385 (expand_vec_perm_even_odd_pack): Likewise.
1386 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
1387 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
1388 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
1389 gen_const_vec_duplicate.
1390 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
1391 * config/mips/mips.c (mips_gen_const_int_vector): Use
1392 gen_const_vec_duplicate.
1393 (mips_expand_vector_init): Use CONST0_RTX.
1394 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
1395 (define_split): Use gen_const_vec_duplicate.
1396 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
1397 (define_split): Use gen_const_vec_duplicate.
1398 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
1399 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
1400 * config/spu/spu.c (spu_const): Likewise.
1402 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1403 Alan Hayward <alan.hayward@arm.com>
1404 David Sherwood <david.sherwood@arm.com>
1406 * combine.c (can_change_dest_mode): Reject changes in
1407 REGMODE_NATURAL_SIZE.
1409 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
1411 * configure.ac (--enable-libssp): New.
1412 (gcc_cv_libc_provides_ssp): Check for explicit setting before
1413 trying to determine target-specific default. Adjust indentation.
1414 * configure: Regenerated.
1415 * doc/install.texi (Configuration): Expand --disable-libssp
1418 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
1420 config/i386/i386.c (ix86_expand_epilogue): Correct stack
1423 2017-10-31 Martin Jambor <mjambor@suse.cz>
1426 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
1428 2017-10-31 David Malcolm <dmalcolm@redhat.com>
1430 * auto-profile.c (autofdo_source_profile::read): Use
1431 UNKNOWN_LOCATION rather than 0.
1432 * diagnostic-core.h (warning_at_rich_loc): Rename to...
1433 (warning_at): ...this overload.
1434 (warning_at_rich_loc_n): Rename to...
1435 (warning_n): ...this overload.
1436 (error_at_rich_loc): Rename to...
1437 (error_at): ...this overload.
1438 (pedwarn_at_rich_loc): Rename to...
1439 (pedwarn): ...this overload.
1440 (permerror_at_rich_loc): Rename to...
1441 (permerror): ...this overload.
1442 (inform_at_rich_loc): Rename to...
1443 (inform): ...this overload.
1444 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
1445 (diagnostic_n_impl_richloc): Rename to...
1446 (diagnostic_n_impl): ...this rich_location *-based decl.
1447 (inform_at_rich_loc): Rename to...
1448 (inform): ...this, and add an assertion.
1449 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
1450 (warning_at_rich_loc): Rename to...
1451 (warning_at): ...this, and add an assertion.
1452 (warning_at_rich_loc_n): Rename to...
1453 (warning_n): ...this, and add an assertion.
1454 (warning_n): Update location_t-based implementation for removal of
1455 location_t-based diagnostic_n_impl.
1456 (pedwarn_at_rich_loc): Rename to...
1457 (pedwarn): ...this, and add an assertion.
1458 (permerror_at_rich_loc): Rename to...
1459 (permerror): ...this, and add an assertion.
1460 (error_n): Update for removal of location_t-based diagnostic_n_impl.
1461 (error_at_rich_loc): Rename to...
1462 (error_at): ...this, and add an assertion.
1463 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
1464 (driver::do_spec_on_infiles): Likewise.
1465 * substring-locations.c (format_warning_va): Update for renaming
1466 of inform_at_rich_loc.
1468 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
1470 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
1471 _Float<N>X built-in functions so that the variant without the
1472 "__builtin_" prefix is only enabled for the GNU C and Objective C
1473 languages when they are in non-strict ANSI/ISO mode.
1474 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
1475 * target.def (floatn_builtin_p): Add a target hook to control
1476 whether _Float<N> and _Float<N>X built-in functions without the
1477 "__builtin_" prefix are enabled, and return true for C and
1478 Objective C in the default hook. Include langhooks.h in
1480 * targhooks.h (default_floatn_builtin_p): Likewise.
1481 * targhooks.c (default_floatn_builtin_p): Likewise.
1482 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
1483 floatn_builtin_p target hook.
1484 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
1486 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
1487 Eric Botcazou <ebotcazou@adacore.com>
1489 PR rtl-optimization/81803
1490 * lra-constraints.c (curr_insn_transform): Also reload the whole
1491 register for a strict subreg no wider than a word if this is for
1492 a WORD_REGISTER_OPERATIONS target.
1494 2017-10-31 Jason Merrill <jason@redhat.com>
1496 * gdbinit.in: Skip over inlines from timevar.h.
1498 2017-10-31 Martin Liska <mliska@suse.cz>
1500 * doc/gcov.texi: Document new option.
1501 * gcov.c (print_usage): Likewise print it.
1502 (process_args): Support the argument.
1503 (format_count): New function.
1504 (format_gcov): Use the function.
1506 2017-10-31 Martin Liska <mliska@suse.cz>
1508 * gcov.c (struct name_map): do not use typedef.
1509 Define operator== and operator<.
1510 (name_search): Remove.
1511 (name_sort): Remove.
1512 (main): Do not allocate names.
1513 (process_file): Add vertical space.
1514 (generate_results): Use std::find.
1515 (release_structures): Do not release memory.
1516 (find_source): Use std::find.
1518 2017-10-31 Martin Liska <mliska@suse.cz>
1520 * gcov.c (struct line_info): Remove it's typedef.
1521 (line_info::line_info): Add proper ctor.
1522 (line_info::has_block): Do not use a typedef.
1523 (struct source_info): Do not use typedef.
1524 (circuit): Likewise.
1525 (get_cycles_count): Likewise.
1526 (output_intermediate_file): Iterate via vector iterator.
1527 (add_line_counts): Use std::vector methods.
1528 (accumulate_line_counts): Likewise.
1529 (output_lines): Likewise.
1531 2017-10-31 Martin Liska <mliska@suse.cz>
1533 * gcov.c (struct source_info): Remove typedef.
1534 (source_info::source_info): Add proper ctor.
1535 (accumulate_line_counts): Use struct, not it's typedef.
1536 (output_gcov_file): Likewise.
1537 (output_lines): Likewise.
1538 (main): Do not allocate an array.
1539 (output_intermediate_file): Use size of vector container.
1540 (process_file): Resize the vector.
1541 (generate_results): Do not preallocate, use newly added vector
1543 (release_structures): Do not release sources.
1544 (find_source): Use vector methods.
1545 (add_line_counts): Do not use typedef.
1547 2017-10-31 Martin Liska <mliska@suse.cz>
1549 * doc/gcov.texi: Document that.
1550 * gcov.c (add_line_counts): Mark lines with a non-executed
1552 (output_line_beginning): Handle such lines.
1553 (output_lines): Pass new argument.
1554 (output_intermediate_file): Print it in intermediate format.
1556 2017-10-31 Martin Liska <mliska@suse.cz>
1558 * color-macros.h: New file.
1559 * diagnostic-color.c: Factor out color related to macros to
1561 * doc/gcov.texi: Document -k option.
1562 * gcov.c (INCLUDE_STRING): Include string.h.
1563 (print_usage): Add -k option.
1564 (process_args): Parse it.
1565 (pad_count_string): New function.
1566 (output_line_beginning): Likewise.
1567 (DEFAULT_LINE_START): New macro.
1568 (output_lines): Support color output.
1570 2017-10-31 Martin Liska <mliska@suse.cz>
1572 PR gcov-profile/82633
1573 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
1574 their interaction with GCOV infrastructure.
1575 * configure.ac: Add -fkeep-{inline,static}-functions to
1577 * configure: Regenerate.
1579 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
1582 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
1584 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
1587 * config/rs6000/rs6000.md (allocate_stack): Force update interval
1588 into a register if it does not fit into an immediate offset field.
1590 2017-10-31 Olivier Hainque <hainque@adacore.com>
1592 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
1594 2017-10-31 Julia Koval <julia.koval@intel.com>
1596 * config.gcc: Add gfniintrin.h.
1597 * config/i386/gfniintrin.h: New.
1598 * config/i386/i386-builtin-types.def
1599 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1600 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1601 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1602 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1603 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1604 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
1605 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
1606 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
1607 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
1608 V64QI_FTYPE_V64QI_V64QI_INT): New types.
1609 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1610 * config/i386/immintrin.h: Include gfniintrin.h.
1611 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
1613 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
1615 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
1617 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1619 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
1620 (arm_ashldi3_1bit): Remove pattern.
1621 (ashrdi3): Remove shift by 1 expansion.
1622 (arm_ashrdi3_1bit): Remove pattern.
1623 (lshrdi3): Remove shift by 1 expansion.
1624 (arm_lshrdi3_1bit): Remove pattern.
1625 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
1626 cost of ashldi3 by 1.
1627 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
1628 (<shift>di3_neon): Likewise.
1630 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
1632 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
1633 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
1634 and (*aarch64_simd_mov<VQ:mode>).
1635 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
1636 pattern alternative.
1637 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
1638 attributes to match pattern alternative.
1640 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
1642 * config.gcc (powerpc*-*-*): Add emmintrin.h.
1643 * config/rs6000/emmintrin.h: New file.
1644 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
1646 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1648 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
1649 * (movdi_vfp_cortexa8): Remove pattern.
1651 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1653 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
1655 (Specific, alpha*-dec-osf5.1): Remove.
1656 (Specific, mips-sgi-irix5): Remove.
1657 (Specific, mips-sgi-irix6): Remove.
1659 2017-10-30 Jakub Jelinek <jakub@redhat.com>
1662 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
1663 arguments to clear_bit_region_be.
1665 2017-10-30 Jim Wilson <wilson@tuliptree.org>
1667 * gimplify.c: Include memmodel.h.
1669 2017-10-30 Martin Jambor <mjambor@suse.cz>
1671 * omp-grid.c (grid_attempt_target_gridification): Also insert a
1672 condition whether loop should be executed at all.
1674 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
1676 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1677 gimple folding of vec_madd() intrinsics.
1678 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
1679 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
1680 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
1682 2017-10-30 Richard Biener <rguenther@suse.de>
1684 PR tree-optimization/82762
1686 2017-10-23 Richard Biener <rguenther@suse.de>
1688 PR tree-optimization/82129
1690 2017-08-01 Richard Biener <rguenther@suse.de>
1692 PR tree-optimization/81181
1693 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
1694 (compute_antic): ... end of iteration here.
1696 2017-10-30 Joseph Myers <joseph@codesourcery.com>
1698 * doc/invoke.texi (C Dialect Options): Document -std=c17,
1699 -std=iso9899:2017 and -std=gnu17.
1700 * doc/standards.texi (C Language): Document C17 support.
1701 * doc/cpp.texi (Overview): Mention -std=c17.
1702 (Standard Predefined Macros): Document C11 and C17 values of
1703 __STDC_VERSION__. Do not refer to C99 support as incomplete.
1704 * doc/extend.texi (Inline): Do not list individual options for
1705 standards newer than C99.
1706 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
1708 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
1711 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
1713 * asan.c (asan_finish_file): Align asan globals array by shadow
1716 2017-10-30 Jakub Jelinek <jakub@redhat.com>
1719 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
1720 (struct store_immediate_info): Add bitregion_start and bitregion_end
1722 (store_immediate_info::store_immediate_info): Add brs and bre
1723 arguments and initialize bitregion_{start,end} from those.
1724 (struct merged_store_group): Add bitregion_start, bitregion_end,
1725 align_base and mask fields. Drop unnecessary struct keyword from
1726 struct store_immediate_info. Add do_merge method.
1727 (clear_bit_region_be): Use memset instead of loop storing zeros.
1728 (merged_store_group::do_merge): New method.
1729 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
1730 stores as long as the surrounding bitregions have no gaps.
1731 (merged_store_group::merge_overlapping): Use do_merge.
1732 (merged_store_group::apply_stores): Test that bitregion_{start,end}
1733 is byte aligned, rather than requiring that start and width are
1734 byte aligned. Drop unnecessary struct keyword from
1735 struct store_immediate_info. Allocate and populate also mask array.
1736 Make start of the arrays relative to bitregion_start rather than
1737 start and size them according to bitregion_{end,start} difference.
1738 (struct imm_store_chain_info): Drop unnecessary struct keyword from
1739 struct store_immediate_info.
1740 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
1741 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
1742 struct keyword from struct store_immediate_info.
1743 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
1744 between stores as long as the surrounding bitregions have no gaps.
1746 (struct split_store): Add orig non-static data member.
1747 (split_store::split_store): Initialize orig to false.
1748 (find_constituent_stmts): Return store_immediate_info *, non-NULL
1749 if there is exactly a single original stmt. Change stmts argument
1750 to pointer from reference, if NULL, don't push anything to it. Add
1751 first argument, use it to optimize skipping over orig stmts that
1752 are known to be before bitpos already. Simplify.
1753 (split_group): Return unsigned int count how many stores are or
1754 would be needed rather than a bool. Add allow_unaligned argument.
1755 Change split_stores argument from reference to pointer, if NULL,
1756 only do a dry run computing how many stores would be produced.
1757 Rewritten algorithm to use both alignment and misalign if
1758 !allow_unaligned and handle bitfield stores with gaps.
1759 (imm_store_chain_info::output_merged_store): Set start_byte_pos
1760 from bitregion_start instead of start. Compute allow_unaligned
1761 here, if true, do 2 split_group dry runs to compute which one
1762 produces fewer stores and prefer aligned if equal. Punt if
1763 new count is bigger or equal than original before emitting any
1764 statements, rather than during that. Remove no longer needed
1765 new_ssa_names tracking. Replace num_stmts with
1766 split_stores.length (). Use 32-bit stack allocated entries
1767 in split_stores auto_vec. Try to reuse original store lhs/rhs1
1768 if possible. Handle bitfields with gaps.
1769 (pass_store_merging::execute): Ignore bitsize == 0 stores.
1770 Compute bitregion_{start,end} for the stores and construct
1771 store_immediate_info with that. Formatting fixes.
1773 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
1776 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
1777 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
1779 2017-10-29 Jim Wilson <wilson@tuliptree.org>
1781 * gimplify.c: Include tm_p.h.
1783 * common.opt (gcoff): Re-add as ignored option.
1784 (gcoff1, gcoff2, gcoff3): Likewise.
1786 * Makefile.in (OBJS): Delete sdbout.o.
1787 (GTFILES): Delete $(srcdir)/sdbout.c.
1788 * debug.h: Delete sdb_debug_hooks.
1789 * final.c: Delete sdbout.h include.
1790 (final_scan_insn): Delete SDB_DEBUG check.
1791 (rest_of_clean_state): Likewise.
1792 * output.h: Delete sdb_begin_function_line.
1795 * toplev.c: Delete sdbout.h include.
1796 (process_options): Delete SDB_DEBUG check.
1797 * tree-core.h (tree_type_common): Delete pointer field of
1799 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
1800 TYPE_SYMTAB_POINTER.
1801 * tree.h (TYPE_SYMTAB_POINTER): Delete.
1802 (TYPE_SYMTAB_IS_POINTER): Delete.
1803 (TYPE_SYMTAB_IS_DIE): Renumber.
1804 * xcoffout.c: Refer to former sdbout.c file.
1805 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
1807 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
1808 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
1809 (Debugging Options): Delete -gcoff.
1810 (-gstabs): Delete SDB reference.
1812 (-gcoff@var{level}): Delete.
1813 * doc/passes.texi (Debugging information output): Delete SDB and
1814 sdbout.c references.
1815 * doc/tm.texi: Regenerate.
1816 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
1817 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
1819 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
1820 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
1821 (SDB_DEBUGGING_INFO): Delete.
1822 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
1823 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
1824 * target.def (output_source_filename): Delete COFF reference.
1826 * common.opt (gcoff): Delete.
1827 (gxcoff+): Update Negative chain.
1828 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
1830 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
1831 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
1832 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
1834 (expand_function_start): Change sdb reference to past tense.
1835 (expand_function_end): Change sdb reference to past tense.
1836 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
1837 * opts.c (debug_type_names): Delete coff entry.
1838 (common_handle_option): Delete OPT_gcoff case.
1839 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
1841 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
1842 * config/cris/cris.h: Delete SDB reference in comment.
1843 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
1844 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
1845 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
1846 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
1848 (ix86_expand_prologue): Likewise.
1849 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
1850 * config/ia64/ia64.h: Likewise.
1851 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
1852 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
1854 * config/mmix/mmix.h: Likewise.
1855 * config/nds32/nds32.c: Likewise.
1856 * config/stormy/storym16.h: Likewise.
1857 * config/visium/visium.h: Likewise.
1858 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
1860 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
1862 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
1863 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
1864 FRAME_POINTER_REGNUM point at high end of local var area.
1866 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
1868 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
1869 Move comment around. Do not reset best_edge for a copiable
1870 destination if the copy would cause a partition change.
1871 (better_edge_p): Remove redundant check.
1873 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
1875 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
1877 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
1879 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
1880 for math functions that have _Float<N> and _Float<N>X variants.
1881 (mathfn_built_in_2): Add support for math functions that have
1882 _Float<N> and _Float<N>X variants.
1883 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
1884 (expand_builtin_mathfn_ternary): Add support for fma with
1885 _Float<N> and _Float<N>X variants.
1886 (expand_builtin): Likewise.
1887 (fold_builtin_3): Likewise.
1888 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
1889 create math function _Float<N> and _Float<N>X variants as external
1891 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
1892 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
1893 the __builtin_ prefix and if not strict ansi, without the prefix.
1894 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
1895 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
1896 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
1897 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
1898 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
1899 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
1900 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
1901 function signatures for fma _Float<N> and _Float<N>X variants.
1902 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
1903 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
1904 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
1905 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
1906 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
1907 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
1908 * gencfn-macros.c (print_case_cfn): Add support for math functions
1909 that have _Float<N> and _Float<N>X variants.
1910 (print_define_operator_list): Likewise.
1911 (fltfn_suffixes): Likewise.
1913 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
1914 for math functions that have _Float<N> and _Float<N>X variants.
1915 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
1916 and _Float<N>X variants.
1917 (COPYSIGN): Likewise.
1920 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
1921 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
1923 (integer_valued_read_call_p): Likewise.
1924 * fold-const-call.c (fold_const_call_ss): Likewise.
1925 (fold_const_call_sss): Add support for copysign, fmin, and fmax
1926 _Float<N> and _Float<N>X variants.
1927 (fold_const_call_ssss): Add support for fma _Float<N> and
1928 _Float<N>X variants.
1929 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
1930 support for copysign and fma _Float<N> and _Float<N>X variants.
1931 (backprop::process_builtin_call_use): Likewise.
1932 * tree-call-cdce.c (can_test_argument_range); Add support for
1933 sqrt _Float<N> and _Float<N>X variants.
1934 (edom_only_function): Likewise.
1935 (get_no_error_domain): Likewise.
1936 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
1937 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
1938 copysign _Float<N> and _Float<N>X variants.
1939 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
1940 handled by machine independent code.
1941 (FMAF128): Likewise.
1942 * doc/cpp.texi (Common Predefined Macros): Document defining
1943 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
1944 fma _Float<N> and _Float<N>X variants.
1946 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
1949 * config/i386/i386-modes.def (CCFPU): Remove definition.
1950 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
1951 (ix86_cc_modes_compatible): Ditto.
1952 (ix86_expand_carry_flag_compare): Ditto.
1953 (ix86_expand_int_movcc): Ditto.
1954 (ix86_expand_int_addcc): Ditto.
1955 (ix86_reverse_condition): Ditto.
1956 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
1957 Return true/false for unordered/ordered fp comparisons.
1958 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
1959 (ix86_prepare_fp_compare_args): Update for rename.
1960 (ix86_expand_fp_compare): Update for rename. Generate unordered
1961 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
1962 (ix86_expand_sse_compare_and_jump): Ditto.
1963 * config/i386/predicates.md (fcmov_comparison_operator):
1964 Remove CCFPU mode handling.
1965 (ix86_comparison_operator): Ditto.
1966 (ix86_carry_flag_operator): Ditto.
1967 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
1968 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
1969 (*cmpu<mode>_cc_i387): Ditto.
1970 (FPCMP): Remove mode iterator.
1971 (unord): Remove mode attribute.
1972 (unord_subst): New define_subst transformation
1973 (unord): New define_subst attribute.
1975 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
1976 (*cmpi<unord>xf_i387): Ditto.
1977 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
1978 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
1979 using unord_subst transformation.
1980 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
1981 (round_saeonly): Also handle CCFP mode.
1982 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
1983 Remove UNSPEC_SAHF unspec handling.
1985 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
1987 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
1989 2017-10-27 Jeff Law <law@redhat.com>
1991 * gimple-ssa-sprintf.c: Include domwalk.h.
1992 (class sprintf_dom_walker): New class, derived from dom_walker.
1993 (sprintf_dom_walker::before_dom_children): New function.
1994 (struct call_info): Moved into sprintf_dom_walker class
1995 (compute_formath_length, handle_gimple_call): Likewise.
1996 (sprintf_length::execute): Call the dominator walker rather
1997 than walking the statements.
1999 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
2000 gimple statement locations.
2001 (check_array_bounds): Corresponding changes. Get the statement's
2002 location directly from wi->stmt.
2004 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
2007 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
2009 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
2011 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
2012 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
2014 2017-10-27 Jakub Jelinek <jakub@redhat.com>
2017 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
2018 * config/i386/i386.c (maybe_get_pool_constant): Removed.
2019 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
2020 maybe_get_pool_constant.
2021 * config/i386/predicates.md (zero_extended_scalar_load_operand):
2024 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2026 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
2027 2.26 caveat. Update gas and gld versions.
2028 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
2031 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2033 * cgraph.h (set_malloc_flag): Declare.
2034 * cgraph.c (set_malloc_flag_1): New function.
2035 (set_malloc_flag): Likewise.
2036 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
2037 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
2039 (read_ipa_call_summary): Add support for reading is_return_callee.
2040 (write_ipa_call_summary): Stream is_return_callee.
2041 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
2042 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
2043 ipa-prop.h, ipa-fnsummary.h.
2044 (pure_const_names): Change to static.
2045 (malloc_state_e): Define.
2046 (malloc_state_names): Define.
2047 (funct_state_d): Add field malloc_state.
2048 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
2049 (check_retval_uses): New function.
2050 (malloc_candidate_p): Likewise.
2051 (analyze_function): Add support for malloc attribute.
2052 (pure_const_write_summary): Stream malloc_state.
2053 (pure_const_read_summary): Add support for reading malloc_state.
2054 (dump_malloc_lattice): New function.
2055 (propagate_malloc): New function.
2056 (warn_function_malloc): New function.
2057 (ipa_pure_const::execute): Call propagate_malloc and
2058 ipa_free_fn_summary.
2059 (pass_local_pure_const::execute): Add support for malloc attribute.
2060 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
2061 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
2063 2017-10-27 Martin Liska <mliska@suse.cz>
2065 PR gcov-profile/82457
2066 * doc/invoke.texi: Document that one needs a non-strict ISO mode
2067 for fork-like functions to be properly instrumented.
2069 2017-10-27 Richard Biener <rguenther@suse.de>
2072 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
2073 info when we redirected EH.
2075 2017-10-26 Michael Collison <michael.collison@arm.com>
2077 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
2079 (<optab>_trunchf<GPI:mode>2: New pattern.
2080 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
2081 * config/aarch64/iterators.md (wv): New mode attribute.
2082 (vf, VF): New mode attributes.
2083 (vgp, VGP): New mode attributes.
2084 (s): Update attribute with SImode and DImode prefixes.
2086 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2088 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
2089 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
2090 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
2091 (nios2_option_overide): Initialize it. Don't allow R0-relative
2092 addressing with PIC.
2093 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
2094 (nios2_symbolic_constant_p): Likewise.
2095 (nios2_legitimate_address_p): Likewise.
2096 (nios2_r0rel_section_name_p): New.
2097 (nios2_symbol_ref_in_r0rel_data_p): New.
2098 (nios2_emit_move_sequence): Handle r0rel_constant_p.
2099 (r0rel_constant_p): New.
2100 (nios2_print_operand_address): Handle r0rel_constant_p.
2101 (nios2_cdx_narrow_form_p): Likewise.
2102 * config/nios2/nios2.opt (mr0rel-sec=): New option.
2103 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
2104 (Nios II Options): Document -mr0rel-sec.
2106 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2108 * config/nios2/nios2.c: Include xregex.h.
2109 (nios2_gprel_sec_regex): New.
2110 (nios2_option_overide): Initialize it. Don't allow GP-relative
2111 addressing with PIC.
2112 (nios2_small_section_name_p): Check for regex match.
2113 * config/nios2/nios2.opt (mgprel-sec=): New option.
2114 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
2115 (Nios II Options): Document -mgprel-sec.
2117 2017-10-26 Jim Wilson <wilson@tuliptree.org>
2119 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
2121 2017-10-26 Tom de Vries <tom@codesourcery.com>
2123 PR tree-optimization/82707
2124 * gimple.c (gimple_copy): Fix unsharing of
2125 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
2127 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
2129 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
2130 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
2131 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
2132 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
2133 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
2134 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
2135 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
2136 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
2137 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
2138 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
2139 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
2140 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
2141 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
2142 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
2143 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
2144 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
2145 _mm512_mask_cmpunord_ps_mask): New intrinsics.
2147 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2149 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
2151 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
2152 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
2153 warning to rs6000.c. Remove the Undocumented flag, since it has
2155 (-mabi=ibmlongdouble): Likewise.
2156 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
2157 already set, set the default format for long double.
2158 (rs6000_debug_reg_global): Print whether long double is IBM or
2160 (rs6000_option_override_internal): Rework setting long double
2161 format. Only warn if the user is changing the long double default
2162 and they did not use -Wno-psabi.
2163 * doc/invoke.texi (PowerPC options): Update the documentation for
2164 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
2166 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2167 Alan Hayward <alan.hayward@arm.com>
2168 David Sherwood <david.sherwood@arm.com>
2170 * rtl.h (wider_subreg_mode): New function.
2171 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
2172 rather than an unsigned int *.
2173 * ira-color.c (regno_max_ref_width): Replace with...
2174 (regno_max_ref_mode): ...this new variable.
2175 (coalesced_pseudo_reg_slot_compare): Update accordingly.
2176 Use wider_subreg_mode.
2177 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
2178 rather than an unsigned int *.
2179 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
2180 (process_alt_operands): Likewise.
2181 (invariant_p): Likewise.
2182 * lra-spills.c (assign_mem_slot): Likewise.
2183 (add_pseudo_to_slot): Likewise.
2184 * lra.c (collect_non_operand_hard_regs): Likewise.
2185 (add_regs_to_insn_regno_info): Likewise.
2186 * reload1.c (regno_max_ref_width): Replace with...
2187 (regno_max_ref_mode): ...this new variable.
2188 (reload): Update accordingly. Update call to
2189 ira_sort_regnos_for_alter_reg.
2190 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
2191 (init_eliminable_invariants): Update to use regno_max_ref_mode.
2192 (scan_paradoxical_subregs): Likewise.
2194 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2196 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
2197 (aarch64_frame): Add emit_frame_chain boolean.
2198 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2199 Move eh_return case to aarch64_layout_frame.
2200 (aarch64_layout_frame): Initialize emit_frame_chain.
2201 (aarch64_expand_prologue): Use emit_frame_chain.
2203 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2205 * config/aarch64/aarch64.c (aarch64_layout_frame):
2206 Ensure LR is always stored at the bottom of the callee-saves.
2207 Remove rarely used frame layout which saves callee-saves at top of
2208 frame, so the store of LR can be used as a valid probe in all cases.
2210 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2212 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
2213 Improve unaligned TImode/TFmode base/offset split.
2215 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2216 Alan Hayward <alan.hayward@arm.com>
2217 David Sherwood <david.sherwood@arm.com>
2219 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
2220 * combine.c (find_single_use_1): Likewise.
2221 (expand_field_assignment): Likewise.
2222 (move_deaths): Likewise.
2223 * lra-constraints.c (simplify_operand_subreg): Likewise.
2224 (curr_insn_transform): Likewise.
2225 * lra.c (collect_non_operand_hard_regs): Likewise.
2226 (add_regs_to_insn_regno_info): Likewise.
2227 * rtlanal.c (reg_referenced_p): Likewise.
2228 (covers_regno_no_parallel_p): Likewise.
2230 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2232 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
2233 Don't print any bits outside the precision of the value.
2234 * wide-int.cc (test_printing): Add some new tests.
2236 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2238 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
2239 supports -xbrace_comment option.
2240 * configure: Regenerate.
2241 * config.in: Regenerate.
2242 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
2243 (ASM_CPU_SPEC): Use it.
2245 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2247 * target.def (static_rtx_alignment): New hook.
2248 * targhooks.h (default_static_rtx_alignment): Declare.
2249 * targhooks.c (default_static_rtx_alignment): New function.
2250 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
2251 * doc/tm.texi: Regenerate.
2252 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
2253 instead of targetm.constant_alignment. Remove call to
2255 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2256 (cris_preferred_mininum_alignment): New function, split out from...
2257 (cris_constant_alignment): ...here.
2258 (cris_static_rtx_alignment): New function.
2259 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
2261 (ix86_constant_alignment): ...here.
2262 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2263 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2264 (mmix_static_rtx_alignment): New function.
2265 * config/spu/spu.c (spu_static_rtx_alignment): New function.
2266 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2268 2017-10-26 Tamar Christina <tamar.christina@arm.com>
2271 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
2272 Add flag_trapping_math and flag_fp_int_builtin_inexact.
2274 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
2276 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
2277 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
2278 mark as a sign-extending load.
2279 (local_pic_load_u): Define.
2281 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
2284 * fold-const.c (operand_equal_for_comparison_p): Also return true
2285 if ARG0 is a simple variant of ARG1 with narrower precision.
2286 (fold_ternary_loc): Always pass unstripped operands to the predicate.
2288 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
2290 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
2292 * i386.h (processor_costs): Add gather_static, gather_per_elt,
2293 scatter_static, scatter_per_elt.
2294 * x86-tune-costs.h: Add new cost entries.
2296 2017-10-25 Richard Biener <rguenther@suse.de>
2298 * tree-ssa-sccvn.h (vn_eliminate): Declare.
2299 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
2300 class pass_fre): Move to ...
2301 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
2302 class pass_fre): ... here and adjust for statistics.
2304 2017-10-25 Jakub Jelinek <jakub@redhat.com>
2307 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
2308 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
2309 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
2310 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
2313 2017-10-25 Richard Biener <rguenther@suse.de>
2315 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
2316 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
2317 eliminate_push_avail, eliminate_insert): Move inside...
2318 (class eliminate_dom_walker): ... this class in preparation
2320 (fini_eliminate): Remove by merging with ...
2321 (eliminate): ... this function. Adjust for class changes.
2322 (pass_pre::execute): Remove fini_eliminate call.
2323 (pass_fre::execute): Likewise.
2325 2017-10-24 Jakub Jelinek <jakub@redhat.com>
2328 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
2329 (VPERMI2, VPERMI2I): New mode iterators.
2330 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
2331 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
2333 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
2334 mode iterator. Remove 3 old define_insn patterns.
2335 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
2336 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
2337 VPERMI2 mode iterator, remove the other two expanders.
2338 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
2339 to use VPERMI2 mode iterator, add another alternative for vpermi2*
2340 instructions, remove the other two patterns.
2341 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
2342 mode iterator, remove the other two patterns.
2343 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
2344 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
2345 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
2346 and adjust argument order accordingly.
2347 (ix86_expand_vec_perm): Adjust caller.
2348 (expand_vec_perm_1): Likewise.
2349 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
2350 (expand_vec_perm_vpermt2_vpshub2): ... this.
2351 (ix86_expand_vec_perm_const_1): Adjust caller.
2352 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
2355 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
2356 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
2357 (vec_shl_<mode>): Remove unused expander.
2358 (avx512bw_<shift_insn><mode>3): New define_insn.
2359 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
2360 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
2362 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
2365 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
2368 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
2370 PR rtl-optimization/82396
2371 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
2372 (autopref_multipass_init): Simplify initialization.
2373 (autopref_rank_data): Simplify sort order.
2374 * gcc/sched-int.h (autopref_multipass_data_): Remove
2375 multi_mem_insn_p, min_offset and max_offset.
2377 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
2380 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2381 Check special value of flag_omit_frame_pointer.
2382 (aarch64_can_eliminate): Likewise.
2383 (aarch64_override_options_after_change_1): Simplify handling of
2384 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
2386 2017-10-24 Richard Biener <rguenther@suse.de>
2388 PR tree-optimization/82697
2389 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
2390 zero for conditional load and unconditional store.
2392 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
2394 * doc/install.texi: Document bootstrap-cet.
2396 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
2399 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
2400 ENDBR instruction at function entrance if function is only
2403 2017-10-24 Jakub Jelinek <jakub@redhat.com>
2406 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
2407 patterns to better describe from which operation the CF is computed.
2408 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
2409 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
2410 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
2411 is 0, use _0 suffixed expanders instead of emitting a comparison
2414 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
2416 * config/i386/i386.md(*movsf_internal, *movdf_internal):
2417 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
2419 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
2422 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
2423 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
2424 * loop-iv.c (iv_get_reaching_def): Likewise.
2425 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
2426 variable is promoted and the partition contains undefined values.
2428 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2430 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
2432 (nios2_address_cost): Define.
2433 (nios2_legitimize_address): Recognize (exp + constant) directly.
2434 (TARGET_ADDRESS_COST): Define.
2436 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2438 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
2439 (nios2_symbolic_memory_operand_p): Declare.
2440 (nios2_split_large_constant): Declare.
2441 (nios2_split_symbolic_memory_operand): Declare.
2442 * config/nios2/nios2.c: Adjust includes.
2443 (nios2_symbolic_constant_allowed): New.
2444 (nios2_symbolic_constant_p): New.
2445 (nios2_plus_symbolic_constant_p): New.
2446 (nios2_valid_addr_expr_p): Recognize addresses involving
2448 (nios2_legitimate_address_p): Likewise, also LO_SUM.
2449 (nios2_symbolic_memory_operand_p): New.
2450 (nios2_large_constant_p): New.
2451 (nios2_split_large_constant): New.
2452 (nios2_split_plus_large_constant): New.
2453 (nios2_split_symbolic_memory_operand): New.
2454 (nios2_legitimize_address): Code refactoring. Handle addresses
2455 involving symbolic constants.
2456 (nios2_emit_move_sequence): Likewise.
2457 (nios2_print_operand): Improve error output.
2458 (nios2_print_operand_address): Handle LO_SUM.
2459 (nios2_cdx_narrow_form_p): Likewise.
2460 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
2461 operands involving symbolic constants.
2462 (movhi_internal, movsi_internal): Likewise.
2463 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
2464 (extendhisi2, extendqi<mode>2): Likewise.
2466 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2468 * tree-pass.h (PROP_rtl_split_insns): Define.
2469 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
2471 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2473 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
2475 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2478 * target.def (const_not_ok_for_debug_p): Default to
2479 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
2480 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
2481 * targhooks.c (default_const_not_ok_for_debug_p): New function.
2482 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
2483 which targetm.const_not_ok_for_debug_p returned true.
2484 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
2486 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
2488 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
2489 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
2490 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
2492 (ix86_const_not_ok_for_debug_p): New function.
2493 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
2494 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
2496 2017-10-23 David Malcolm <dmalcolm@redhat.com>
2499 * system.h: Conditionally include "unique-ptr.h" if
2500 INCLUDE_UNIQUE_PTR is defined.
2501 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
2502 of defining INCLUDE_UNIQUE_PTR before including "system.h".
2504 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
2506 * config/rl78/rl78.md: New define_expand "subdi3".
2508 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
2511 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
2512 DF_REF_INSN if DF_REF_INSN_INFO is false.
2514 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
2516 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
2517 xmm_move instead of sse_move.
2518 (sse_store_index): New function.
2519 (ix86_register_move_cost): Be more sensible about mismatch stall;
2520 model AVX moves correctly; make difference between sse->integer and
2522 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
2523 moves; make difference between SSE and AVX.
2524 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
2525 and zmm_move. Increase size of sse load and store tables;
2526 add unaligned load and store tables; add ssemmx_to_integer.
2527 * x86-tune-costs.h: Update all entries according to real
2528 move latencies from Agner Fog's manual and chip documentation.
2530 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2533 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
2534 * config/i386/constraints.md (Wf): New constraint.
2535 * config/i386/i386.md (UNSPEC_SBB): New unspec.
2536 (cmp<dwi>_doubleword): Removed.
2537 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
2538 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
2539 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
2540 expand with cmp<dwi>_doubleword. For LTU and GEU use
2541 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
2543 * common.opt (gcolumn-info): Enable by default.
2544 * doc/invoke.texi (gcolumn-info): Document new default.
2546 2017-10-23 Richard Biener <rguenther@suse.de>
2548 PR tree-optimization/82672
2549 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
2550 Fold the stmt if we propagated into it.
2552 2017-10-23 Richard Biener <rguenther@suse.de>
2554 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
2555 (bitmap_remove_expr_from_set): ... this. All callers call this
2556 for non-constant values.
2557 (bitmap_set_subtract): Rename to...
2558 (bitmap_set_subtract_expressions): ... this. Adjust and
2560 (bitmap_set_contains_value): Remove superfluous check.
2561 (bitmap_set_replace_value): Inline into single caller ...
2562 (bitmap_value_replace_in_set): ... here and simplify.
2563 (dependent_clean): Merge into ...
2564 (clean): ... this using an overload. Adjust.
2565 (prune_clobbered_mems): Adjust.
2566 (compute_antic_aux): Likewise.
2567 (compute_partial_antic_aux): Likewise.
2569 2017-10-23 Richard Biener <rguenther@suse.de>
2571 PR tree-optimization/82129
2573 2017-08-01 Richard Biener <rguenther@suse.de>
2575 PR tree-optimization/81181
2576 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
2577 (compute_antic): ... end of iteration here.
2579 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2581 * target.def (starting_frame_offset): New hook.
2582 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
2583 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
2584 * doc/tm.texi.in: Regenerate.
2585 * hooks.h (hook_hwi_void_0): Declare.
2586 * hooks.c (hook_hwi_void_0): New function.
2587 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
2588 STARTING_FRAME_OFFSET.
2589 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
2590 * reload1.c (reload): Likewise.
2591 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
2592 instead of STARTING_FRAME_OFFSET.
2593 * function.c (try_fit_stack_local): Likewise.
2594 (assign_stack_local_1): Likewise
2595 (instantiate_virtual_regs): Likewise.
2596 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
2597 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
2598 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
2599 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
2600 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
2601 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
2602 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
2603 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
2604 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
2605 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
2606 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
2607 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
2608 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
2609 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
2610 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
2611 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
2612 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
2613 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
2614 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
2615 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
2616 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
2617 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
2618 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
2619 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
2620 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
2621 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
2622 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
2623 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
2624 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
2625 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
2626 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
2627 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
2628 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
2629 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
2630 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
2631 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
2632 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
2633 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
2634 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
2635 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
2636 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
2637 * config/avr/avr.c (avr_starting_frame_offset): Make static and
2638 return a HOST_WIDE_INT.
2639 (avr_builtin_setjmp_frame_value): Use it instead of
2640 STARTING_FRAME_OFFSET.
2641 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2642 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
2643 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
2645 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2646 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
2647 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
2648 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2649 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
2650 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
2651 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2652 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
2653 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
2654 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2655 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
2656 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
2658 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2659 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
2660 * config/mips/mips.c (mips_compute_frame_info): Refer to
2661 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
2662 (mips_starting_frame_offset): New function.
2663 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2664 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
2665 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
2666 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
2667 and return a HOST_WIDE_INT.
2668 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2669 (mmix_initial_elimination_offset): Refer to
2670 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
2671 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
2672 * config/pa/pa.c (pa_starting_frame_offset): New function.
2673 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
2674 (pa_expand_prologue): Likewise.
2675 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2676 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
2677 !FRAME_GROWS_DOWNWARD handling to...
2678 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2679 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
2680 !FRAME_GROWS_DOWNWARD handling to...
2681 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2682 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
2683 !FRAME_GROWS_DOWNWARD handling to...
2684 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2685 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
2687 (rs6000_starting_frame_offset): New function.
2688 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
2689 !FRAME_GROWS_DOWNWARD handling to...
2690 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2691 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
2692 !FRAME_GROWS_DOWNWARD handling to...
2693 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2694 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
2695 !FRAME_GROWS_DOWNWARD handling to...
2696 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2697 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
2698 (rs6000_starting_frame_offset): New function.
2699 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
2700 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
2701 * config/vax/vax.c (vax_starting_frame_offset): New function.
2702 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
2703 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2704 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
2705 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
2706 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2707 * system.h (STARTING_FRAME_OFFSET): Poison.
2709 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2711 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
2712 SCALAR_TYPE_MODE instead of TYPE_MODE.
2714 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2715 Alan Hayward <alan.hayward@arm.com>
2716 David Sherwood <david.sherwood@arm.com>
2718 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
2720 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2721 Alan Hayward <alan.hayward@arm.com>
2722 David Sherwood <david.sherwood@arm.com>
2724 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
2726 2017-10-23 Richard Biener <rguenther@suse.de>
2728 PR tree-optimization/82129
2729 * tree-ssa-pre.c (bitmap_set_and): Remove.
2730 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
2731 canonicalizing expressions in the set to those with lowest
2732 ID rather than taking that from the first edge.
2734 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2736 * combine.c (rtx_equal_for_field_assignment_p): Use
2737 byte_lowpart_offset.
2739 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2740 Alan Hayward <alan.hayward@arm.com>
2741 David Sherwood <david.sherwood@arm.com>
2743 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
2744 to a SUBREG_PROMOTED_VAR.
2746 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2747 Alan Hayward <alan.hayward@arm.com>
2748 David Sherwood <david.sherwood@arm.com>
2750 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
2751 (expand_debug_source_expr): Likewise.
2752 * combine.c (combine_simplify_rtx): Likewise.
2753 * cse.c (fold_rtx): Likewise.
2754 * optabs.c (expand_float): Likewise.
2755 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2756 (simplify_binary_operation_1): Likewise.
2758 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2759 Alan Hayward <alan.hayward@arm.com>
2760 David Sherwood <david.sherwood@arm.com>
2762 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
2763 (record_promoted_value): Likewise.
2764 * expr.c (expand_expr_real_2): Likewise.
2765 * ree.c (update_reg_equal_equiv_notes): Likewise.
2766 (combine_set_extension): Likewise.
2767 * rtlanal.c (low_bitmask_len): Likewise.
2768 * simplify-rtx.c (neg_const_int): Likewise.
2769 (simplify_binary_operation_1): Likewise.
2771 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2772 Alan Hayward <alan.hayward@arm.com>
2773 David Sherwood <david.sherwood@arm.com>
2775 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
2776 * regcprop.c (maybe_mode_change): Likewise.
2777 * reload1.c (alter_reg): Likewise.
2779 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2781 * inchash.h (inchash::hash::add_wide_int): New function.
2782 * lto-streamer-out.c (hash_tree): Use it.
2784 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2786 * inchash.h (inchash::hash::add_wide_int): Rename to...
2787 (inchash::hash::add_hwi): ...this.
2788 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
2789 (polymorphic_call_target_hasher::hash): Likewise.
2790 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
2791 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
2792 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
2793 * lto-streamer-out.c (hash_tree): Likewise.
2794 * optc-save-gen.awk: Likewise.
2795 * tree.c (add_expr): Likewise.
2797 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
2800 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
2801 for ordered inequality comparisons even with TARGET_IEEE_FP.
2803 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
2806 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
2807 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
2808 Expand with cmp<dwi>_doubleword.
2810 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2812 * extend.texi: Add x86 specific to 'nocf_check' attribute.
2813 List CET intrinsics.
2814 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
2815 specific to -fcf-protection option.
2817 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2819 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
2820 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
2821 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
2822 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
2823 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
2824 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
2825 (extra_objs): Add cet.o for Linux/x86 targets.
2826 (tmake_file): Add i386/t-cet for Linux/x86 targets.
2827 * config/i386/cet.c: New file.
2828 * config/i386/cetintrin.h: Likewise.
2829 * config/i386/t-cet: Likewise.
2830 * config/i386/cpuid.h (bit_SHSTK): New.
2831 (bit_IBT): Likewise.
2832 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
2833 pass IBT and SHSTK bits.
2834 * config/i386/i386-builtin-types.def
2835 (VOID_FTYPE_UNSIGNED_PVOID): New.
2836 (VOID_FTYPE_UINT64_PVOID): Likewise.
2837 * config/i386/i386-builtin.def: Add CET intrinsics.
2838 * config/i386/i386-c.c (ix86_target_macros_internal): Add
2839 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
2840 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
2841 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
2843 * config/i386/i386.c (rest_of_insert_endbranch): New.
2844 (pass_data_insert_endbranch): Likewise.
2845 (pass_insert_endbranch): Likewise.
2846 (make_pass_insert_endbranch): Likewise.
2847 (ix86_notrack_prefixed_insn_p): Likewise.
2848 (ix86_target_string): Add -mibt, -mshstk flags.
2849 (ix86_option_override_internal): Add flag_cf_protection
2851 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
2852 (ix86_print_operand): Add 'notrack' prefix output.
2853 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
2854 (ix86_expand_builtin): Expand CET intrinsics.
2855 (x86_output_mi_thunk): Add 'endbranch' instruction.
2856 * config/i386/i386.h (TARGET_IBT): New.
2857 (TARGET_IBT_P): Likewise.
2858 (TARGET_SHSTK): Likewise.
2859 (TARGET_SHSTK_P): Likewise.
2860 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
2861 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
2862 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
2863 (builtin_setjmp_setup): New pattern.
2864 (builtin_longjmp): Likewise.
2865 (rdssp<mode>): Likewise.
2866 (incssp<mode>): Likewise.
2867 (saveprevssp): Likewise.
2868 (rstorssp): Likewise.
2869 (wrss<mode>): Likewise.
2870 (wruss<mode>): Likewise.
2871 (setssbsy): Likewise.
2872 (clrssbsy): Likewise.
2873 (nop_endbr): Likewise.
2874 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
2876 * config/i386/immintrin.h: Include <cetintrin.h>.
2877 * config/i386/linux-common.h
2878 (file_end_indicate_exec_stack_and_cet): New prototype.
2879 (TARGET_ASM_FILE_END): New.
2881 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
2883 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
2884 latencies instead of having separate table; make difference between
2885 integer and float costs.
2886 * i386.h (processor_costs): Remove scalar_stmt_cost,
2887 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
2888 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
2890 * x86-tune-costs.h: Remove entries which has been removed in
2891 procesor_costs from all tables; make cond_taken_branch_cost
2892 and cond_not_taken_branch_cost COST_N_INSNS based.
2894 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
2896 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
2898 2017-10-20 Jakub Jelinek <jakub@redhat.com>
2900 * config/i386/i386.md (isa): Remove fma_avx512f.
2901 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
2902 <avx512>_fmadd_<mode>_mask3<round_name>,
2903 <avx512>_fmsub_<mode>_mask<round_name>,
2904 <avx512>_fmsub_<mode>_mask3<round_name>,
2905 <avx512>_fnmadd_<mode>_mask<round_name>,
2906 <avx512>_fnmadd_<mode>_mask3<round_name>,
2907 <avx512>_fnmsub_<mode>_mask<round_name>,
2908 <avx512>_fnmsub_<mode>_mask3<round_name>,
2909 <avx512>_fmaddsub_<mode>_mask<round_name>,
2910 <avx512>_fmaddsub_<mode>_mask3<round_name>,
2911 <avx512>_fmsubadd_<mode>_mask<round_name>,
2912 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
2913 (*vec_widen_umult_even_v16si<mask_name>,
2914 *vec_widen_smult_even_v16si<mask_name>): Likewise.
2915 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
2917 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2919 * extend.texi: Add 'nocf_check' documentation.
2920 * gimple.texi: Add second parameter to
2921 gimple_build_call_from_tree.
2922 * invoke.texi: Add -fcf-protection documentation.
2923 * rtl.texi: Add REG_CALL_NOTRACK documenation.
2925 2017-10-20 Richard Biener <rguenther@suse.de>
2927 PR tree-optimization/82473
2928 * tree-vect-loop.c (vectorizable_reduction): Properly get at
2929 the largest input type.
2931 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2933 * c-attribs.c (handle_nocf_check_attribute): New function.
2934 (c_common_attribute_table): Add 'nocf_check' handling.
2935 * gimple-parser.c: Add second argument NULL to
2936 gimple_build_call_from_tree.
2937 * attrib.c (comp_type_attributes): Check nocf_check attribute.
2938 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
2940 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
2941 * common.opt: Add fcf-protection flag.
2942 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
2943 * flag-types.h: Add enum cf_protection_level.
2944 * gimple.c (gimple_build_call_from_tree): Add second parameter.
2945 Add 'nocf_check' attribute propagation to gimple call.
2946 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
2947 (gimple_build_call_from_tree): Update prototype.
2948 (gimple_call_nocf_check_p): New function.
2949 (gimple_call_set_nocf_check): Likewise.
2950 * gimplify.c: Add second argument to gimple_build_call_from_tree.
2951 * ipa-icf.c: Add nocf_check attribute in statement hash.
2952 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
2953 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
2954 * toplev.c (process_options): Add flag_cf_protection handling.
2956 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
2958 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
2960 2017-10-20 Richard Biener <rguenther@suse.de>
2962 PR tree-optimization/82603
2963 * tree-if-conv.c (predicate_mem_writes): Make sure to only
2964 remove false predicated stores.
2966 2017-10-20 Richard Biener <rguenther@suse.de>
2968 * graphite-isl-ast-to-gimple.c
2969 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
2970 Remove return value and simplify, dump copied stmt after lhs
2972 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
2973 Reduce dump verbosity.
2974 (gsi_insert_earliest): Likewise.
2975 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
2976 * graphite.c (print_global_statistics): Adjust dumping.
2977 (print_graphite_scop_statistics): Likewise.
2978 (print_graphite_statistics): Do not dump loops here.
2979 (graphite_transform_loops): But here.
2981 2017-10-20 Nicolas Roche <roche@adacore.com>
2983 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
2984 * configure: Regenerate.
2986 2017-10-20 Jakub Jelinek <jakub@redhat.com>
2989 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
2990 functions when optimizing replace GIMPLE_RETURN stmts with
2991 calls to __builtin_unreachable ().
2994 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
2995 for -fsanitize=thread link of executables.
2996 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
2997 link of executables.
3000 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
3002 (<shift_insn><mode>3<mask_name>): Change the last of the 3
3003 define_insns for logical vector shifts to use VI248_AVX512BW
3004 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
3005 condition, useless isa and prefix attributes. Change the first
3006 2 of these define_insns to ...
3007 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
3008 define_insn for avx512vl.
3009 (<shift_insn><mode>3): ... and this, new define_insn without
3010 masking for non-avx512vl.
3013 * config/i386/sse.md (*andnot<mode>3,
3014 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
3015 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
3016 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
3017 not applied use empty suffix even for TARGET_AVX512VL.
3018 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
3019 is applied, supply evex,evex or evex,evex,evex instead of just
3022 2017-10-20 Julia Koval <julia.koval@intel.com>
3024 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
3025 (OPTION_MASK_ISA_GFNI_UNSET): New.
3026 (ix86_handle_option): Handle OPT_mgfni.
3027 * config/i386/cpuid.h (bit_GFNI): New.
3028 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
3029 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
3030 * config/i386/i386.c (ix86_target_string): Add -mgfni.
3031 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
3032 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
3033 * config/i386/i386.opt: Add mgfni.
3035 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
3037 * config/msp430/msp430.c (msp430_option_override): Disable
3038 -fdelete-null-pointer-checks.
3039 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
3041 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3043 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
3044 of x87 and SSE instructions.
3046 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3048 * asan.c (create_cond_insert_point): Do not update edge count.
3049 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
3050 (afdo_propagate_circuit): Likewise.
3051 (afdo_calculate_branch_prob): Likewise.
3052 (afdo_annotate_cfg): Likewise.
3053 * basic-block.h (struct edge_def): Remove count.
3054 (edge_def::count): New accessor.
3055 * bb-reorder.c (rotate_loop): Update.
3056 (find_traces_1_round): Update.
3057 (connect_traces): Update.
3058 (sanitize_hot_paths): Update.
3059 * cfg.c (unchecked_make_edge): Update.
3060 (make_single_succ_edge): Update.
3061 (check_bb_profile): Update.
3062 (dump_edge_info): Update.
3063 (update_bb_profile_for_threading): Update.
3064 (scale_bbs_frequencies_int): Update.
3065 (scale_bbs_frequencies_gcov_type): Update.
3066 (scale_bbs_frequencies_profile_count): Update.
3067 (scale_bbs_frequencies): Update.
3068 * cfganal.c (connect_infinite_loops_to_exit): Update.
3069 * cfgbuild.c (compute_outgoing_frequencies): Update.
3070 (find_many_sub_basic_blocks): Update.
3071 * cfgcleanup.c (try_forward_edges): Update.
3072 (try_crossjump_to_edge): Update
3073 * cfgexpand.c (expand_gimple_cond): Update
3074 (expand_gimple_tailcall): Update
3075 (construct_exit_block): Update
3076 * cfghooks.c (verify_flow_info): Update
3077 (redirect_edge_succ_nodup): Update
3078 (split_edge): Update
3079 (make_forwarder_block): Update
3080 (duplicate_block): Update
3081 (account_profile_record): Update
3082 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
3083 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
3084 * cfgloopmanip.c (scale_loop_profile): Update.
3086 (lv_adjust_loop_entry_edge): Update.
3087 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
3088 (force_nonfallthru_and_redirect): Update.
3089 (purge_dead_edges): Update.
3090 (rtl_flow_call_edges_add): Update.
3091 * cgraphunit.c (init_lowered_empty_function): Update.
3092 (cgraph_node::expand_thunk): Update.
3093 * gimple-pretty-print.c (dump_probability): Update.
3094 (dump_edge_probability): Update.
3095 * gimple-ssa-isolate-paths.c (isolate_path): Update.
3096 * haifa-sched.c (sched_create_recovery_edges): Update.
3097 * hsa-gen.c (convert_switch_statements): Update.
3098 * ifcvt.c (dead_or_predicable): Update.
3099 * ipa-inline-transform.c (inline_transform): Update.
3100 * ipa-split.c (split_function): Update.
3101 * ipa-utils.c (ipa_merge_profiles): Update.
3102 * loop-doloop.c (add_test): Update.
3103 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
3104 * lto-streamer-in.c (input_cfg): Update.
3105 (input_function): Update.
3106 * lto-streamer-out.c (output_cfg): Update.
3107 * modulo-sched.c (sms_schedule): Update.
3108 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
3109 * predict.c (maybe_hot_edge_p): Update.
3110 (unlikely_executed_edge_p): Update.
3111 (probably_never_executed_edge_p): Update.
3112 (dump_prediction): Update.
3113 (drop_profile): Update.
3114 (propagate_unlikely_bbs_forward): Update.
3115 (determine_unlikely_bbs): Update.
3116 (force_edge_cold): Update.
3117 * profile.c (compute_branch_probabilities): Update.
3118 * reg-stack.c (better_edge): Update.
3119 * shrink-wrap.c (handle_simple_exit): Update.
3120 * tracer.c (better_p): Update.
3121 * trans-mem.c (expand_transaction): Update.
3122 (split_bb_make_tm_edge): Update.
3123 * tree-call-cdce.c: Update.
3124 * tree-cfg.c (gimple_find_sub_bbs): Update.
3125 (gimple_split_edge): Update.
3126 (gimple_duplicate_sese_region): Update.
3127 (gimple_duplicate_sese_tail): Update.
3128 (gimple_flow_call_edges_add): Update.
3129 (insert_cond_bb): Update.
3130 (execute_fixup_cfg): Update.
3131 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
3132 * tree-complex.c (expand_complex_div_wide): Update.
3133 * tree-eh.c (lower_resx): Update.
3134 (unsplit_eh): Update.
3135 (cleanup_empty_eh_move_lp): Update.
3136 * tree-inline.c (copy_edges_for_bb): Update.
3137 (freqs_to_counts): Update.
3138 (copy_cfg_body): Update.
3139 * tree-ssa-dce.c (remove_dead_stmt): Update.
3140 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
3141 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
3142 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
3143 (unloop_loops): Update.
3144 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3145 * tree-ssa-loop-split.c (connect_loops): Update.
3146 (split_loop): Update.
3147 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
3148 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
3149 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
3150 * tree-ssa-reassoc.c (branch_fixup): Update.
3151 * tree-ssa-tail-merge.c (replace_block_by): Update.
3152 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
3153 (compute_path_counts): Update.
3154 (update_profile): Update.
3155 (recompute_probabilities): Update.
3156 (update_joiner_offpath_counts): Update.
3157 (estimated_freqs_path): Update.
3158 (freqs_to_counts_path): Update.
3159 (clear_counts_path): Update.
3160 (ssa_fix_duplicate_block_edges): Update.
3161 (duplicate_thread_path): Update.
3162 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
3163 (case_bit_test_cmp): Update.
3164 (collect_switch_conv_info): Update.
3165 (gen_inbound_check): Update.
3166 (do_jump_if_equal): Update.
3167 (emit_cmp_and_jump_insns): Update.
3168 * tree-tailcall.c (decrease_profile): Update.
3169 (eliminate_tail_call): Update.
3170 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
3171 (vect_do_peeling): Update.
3172 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3173 * ubsan.c (ubsan_expand_null_ifn): Update.
3174 (ubsan_expand_ptr_ifn): Update.
3175 * value-prof.c (gimple_divmod_fixed_value): Update.
3176 (gimple_mod_pow2): Update.
3177 (gimple_mod_subtract): Update.
3178 (gimple_ic): Update.
3179 (gimple_stringop_fixed_value): Update.
3181 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3184 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
3186 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
3188 PR rtl-optimization/82395
3189 * ira-color.c (allocno_priority_compare_func): Fix comparison step
3190 based on non_spilled_static_chain_regno_p.
3192 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3194 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
3195 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
3196 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
3198 2017-10-19 Martin Sebor <msebor@redhat.com>
3200 PR tree-optimization/82596
3201 * tree.c (array_at_struct_end_p): Handle STRING_CST.
3203 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3205 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
3206 (get_mem_refs_of_builtin_call): Likewise.
3207 * builtins.c (expand_builtin_apply): Adjust call to
3208 allocate_dynamic_stack_space.
3209 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
3210 the third argument to allocate_dynamic_stack_space, otherwise -1.
3211 (expand_builtin): Deal with all alloca variants.
3212 (is_inexpensive_builtin): Likewise.
3213 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
3214 * calls.c (special_function_p): Deal with all alloca variants.
3215 (initialize_argument_information): Adjust call to
3216 allocate_dynamic_stack_space.
3217 (expand_call): Likewise.
3218 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
3219 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
3220 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
3221 use it for the stack usage computation.
3222 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
3223 * function.c (gimplify_parameters): Call build_alloca_call_expr.
3224 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
3225 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
3226 (in_loop_p): Remove first argument and useless check.
3227 (pass_walloca::execute): Remove useless test and adjust call to above.
3228 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
3229 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
3230 (gimplify_call_expr): Deal with all alloca variants.
3231 * hsa-gen.c (gen_hsa_alloca): Likewise.
3232 (gen_hsa_insns_for_call): Likewise.
3233 * ipa-pure-const.c (special_builtin_state): Likewise.
3234 * tree-chkp.c (chkp_build_returned_bound): Likewise.
3235 * tree-object-size.c (alloc_object_size): Likewise.
3236 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
3237 (call_may_clobber_ref_p_1): Likewise.
3238 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
3239 (ccp_fold_stmt): Likewise.
3240 (optimize_stack_restore): Likewise.
3241 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
3242 (mark_all_reaching_defs_necessary_1): Likewise.
3243 (propagate_necessity): Likewise.
3244 (eliminate_unnecessary_stmts): Likewise.
3245 * tree.c (build_common_builtin_nodes): Build
3246 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
3247 (build_alloca_call_expr): New function.
3248 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
3249 (CASE_BUILT_IN_ALLOCA): Likewise.
3250 (build_alloca_call_expr): Declare.
3251 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
3253 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3256 * dwarf2out.c (new_die_raw): New static inline function.
3257 (new_die): Use it to create the DIE.
3258 (add_AT_external_die_ref): Likewise.
3259 (clone_die): Likewise.
3260 (clone_as_declaration): Likewise.
3261 (dwarf2out_vms_debug_main_pointer): Likewise.
3262 (base_type_die): Likewise. Remove early return for corner cases.
3263 Do not call add_pubtype on the DIE here.
3264 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
3265 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
3266 typedefs for base types with DW_AT_endianity. Make sure a DIE with
3267 native order exists for base types, attach the DIE manually and call
3268 add_pubtype on it. Do not equate a reverse order DIE to the type.
3270 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
3272 * config/arm/arm.c (align_ok_ldrd_strd): New function.
3273 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
3275 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
3277 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3279 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
3280 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
3281 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
3282 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
3283 * opts.c (sanitizer_opts): Add builtin.
3284 * ubsan.c (instrument_builtin): New function.
3285 (pass_ubsan::execute): Call it.
3286 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
3287 * doc/invoke.texi: Document -fsanitize=builtin.
3289 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
3290 builtins, store max (log2 (align), 0) into uchar field instead of
3291 align into uptr field.
3292 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
3293 store uchar 0 field instead of uptr 0 field.
3294 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
3295 instead of passing one address of struct with 2 locations pass
3296 two addresses of structs with 1 location each.
3297 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
3298 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
3299 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
3300 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
3301 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
3302 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
3303 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
3304 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
3306 2017-10-19 Martin Liska <mliska@suse.cz>
3309 * file-find.c (remove_prefix): Remove.
3310 * file-find.h (remove_prefix): Likewise.
3311 * gcc-ar.c: Remove smartness of lookup.
3313 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
3315 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
3316 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
3317 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
3319 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3322 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
3323 (setcc + and to xor + setcc): New peephole2.
3325 2017-10-19 Tom de Vries <tom@codesourcery.com>
3327 * doc/sourcebuild.texi (Test Directives, Variants of
3328 dg-require-support): Add dg-require-stack-size.
3330 2017-10-19 Martin Liska <mliska@suse.cz>
3333 * gimplify.c (gimplify_decl_expr): Do not instrument variables
3334 that have a large alignment.
3335 (gimplify_target_expr): Likewise.
3337 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
3339 PR rtl-optimization/82602
3340 * ira.c (rtx_moveable_p): Return false for volatile asm.
3342 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
3345 * config/i386/i386-modes.def (CCGZ): New CC mode.
3346 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
3347 * config/i386/predicates.md (ix86_comparison_operator):
3349 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
3350 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
3351 with double-word subtraction.
3352 (put_condition_code): Handle CCGZmode.
3354 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
3356 * wide-int.cc (debug (const wide_int &)): New.
3357 (debug (const wide_int *)): New.
3358 (debug (const widest_int &)): New.
3359 (debug (const widest_int *)): New.
3361 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
3364 * lra-constraints.c (curr_insn_transform): Use non-input operand
3365 instead of output one for matched reload.
3367 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3369 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
3370 (tree-ssa-loop-ivopts.h): New header file.
3371 (struct builtin_info): New fields.
3372 (classify_builtin_1): Compute and record base and offset parts for
3373 memset builtin partition by calling strip_offset.
3374 (offset_cmp, fuse_memset_builtins): New functions.
3375 (finalize_partitions): Fuse adjacent memset partitions by calling
3377 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
3378 Expose the interface.
3379 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
3381 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3383 PR tree-optimization/82574
3384 * tree-loop-distribution.c (find_single_drs): New parameter. Check
3385 that data reference must be executed exactly once per iteration
3386 against the outermost loop in nest.
3387 (classify_partition): Update call to above function.
3389 2017-10-18 Richard Biener <rguenther@suse.de>
3391 PR tree-optimization/82591
3392 * graphite.c (graphite_transform_loops): Move code gen message
3394 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3395 Here. Handle scop_to_isl_ast failing.
3396 (scop_to_isl_ast): Limit the number of ISL operations.
3398 2017-10-18 Richard Biener <rguenther@suse.de>
3400 * graphite-isl-ast-to-gimple.c
3401 (translate_isl_ast_to_gimple::set_rename): Simplify.
3402 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
3403 (graphite_copy_stmts_from_block): ... here.
3404 (copy_bb_and_scalar_dependences): Simplify.
3405 (add_parameters_to_ivs_params): Canonicalize.
3406 (generate_entry_out_of_ssa_copies): Simplify.
3407 * graphite-sese-to-poly.c (extract_affine_name): Simplify
3408 by passing in ISL dimension.
3409 (parameter_index_in_region_1): Rename to ...
3410 (parameter_index_in_region): ... this.
3411 (extract_affine): Adjust assert, pass down parameter index.
3412 (add_param_constraints): Use range-info when available.
3413 (build_scop_context): Adjust.
3414 * sese.c (new_sese_info): Adjust.
3415 (free_sese_info): Likewise.
3416 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
3417 Remove unused typedefs.
3418 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
3420 2017-10-18 Martin Liska <mliska@suse.cz>
3422 * combine.c (simplify_compare_const): Add gcc_fallthrough.
3424 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
3426 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
3427 (s390_sched_init): Do not reset s390_sched_state if we entered the
3428 current basic block via a fallthru edge and all others are unlikely.
3430 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
3432 * config/s390/s390.c (NUM_SIDES): New variable.
3433 (LONGRUNNING_THRESHOLD): New variable.
3434 (LATENCY_FACTOR): New variable.
3435 (s390_sched_score): Decrease score for long-running instructions on
3437 (s390_sched_variable_issue): Perform bookkeeping for long-running
3440 2017-10-18 Richard Biener <rguenther@suse.de>
3442 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
3443 Simplify with removal of the parameter rename map.
3444 (set_rename): Likewise.
3445 (should_copy_to_new_region): Likewise.
3446 (graphite_copy_stmts_from_block): Likewise.
3447 (copy_bb_and_scalar_dependences): Remove initialization of
3448 unused copied_bb_map.
3450 (copy_internal_parameters): Likewise.
3451 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
3452 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
3453 Use INTEGRAL_TYPE_P.
3454 (parameter_index_in_region_1): Rename to ...
3455 (assign_parameter_index_in_region): ... this. Assert we have
3456 a parameter we handle.
3457 (scan_tree_for_params): Adjust.
3458 * sese.h (parameter_rename_map_t): Remove.
3459 (struct sese_info_t): Remove unused parameter_rename_map and
3460 copied_bb_map members.
3461 * sese.c (new_sese_info): Adjust.
3462 (free_sese_info): Likewise.
3464 2017-10-18 Martin Liska <mliska@suse.cz>
3467 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
3468 on an abnormal edge.
3470 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
3472 * doc/invoke.texi (ffunction-sections and fdata-sections):
3475 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3477 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
3478 the use statement can throw internally.
3480 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3482 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
3483 any RTX present on the RHS of a SET.
3484 * compare-elim.c (try_eliminate_compare): Restore comment.
3486 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3488 * langhooks.h (struct lang_hooks): Document that tree_size langhook
3489 may be also called on tcc_type nodes.
3490 * langhooks.c (lhd_tree_size): Likewise.
3492 2017-10-17 David Malcolm <dmalcolm@redhat.com>
3494 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
3495 format_warning_at_substring.
3496 (maybe_warn): Convert source_range * param to a location_t. Pass
3497 UNKNOWN_LOCATION rather than NULL to fmtwarn.
3498 (format_directive): Remove code to extract source_ranges and
3499 source_range * in favor of just a location_t.
3500 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
3502 * substring-locations.c (format_warning_va): Convert
3503 source_range * param to a location_t.
3504 (format_warning_at_substring): Likewise.
3505 * substring-locations.h (format_warning_va): Likewise.
3506 (format_warning_at_substring): Likewise.
3508 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
3510 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
3512 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
3513 and scatter/gather ops.
3515 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
3516 vec_gather_load and vec_scatter_store.
3517 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
3518 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
3519 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
3520 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
3521 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
3522 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
3524 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
3526 * reg-stack.c (compare_for_stack_reg): Add bool argument.
3527 Detect FTST instruction and handle its register pops. Only pop
3528 second operand if can_pop_second_op is true.
3529 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
3530 set can_pop_second_op to false in the compare_for_stack_reg call.
3532 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
3533 output_fp_compare for stack register operands.
3534 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
3535 instructions here. Do not emit stack register pops here. Assert
3536 that FCOMPP pops next to top stack register. Rewrite function.
3538 2017-10-17 Nathan Sidwell <nathan@acm.org>
3541 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
3542 use DECL_ASSEMBLER_NAME_RAW.
3545 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
3548 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
3549 Wilco Dijkstra <wilco.dijkstra@arm.com>
3551 * builtins.c (expand_builtin_update_setjmp_buf): Add a
3552 converstion to Pmode from the buf_addr.
3554 2017-10-17 Richard Biener <rguenther@suse.de>
3556 * graphite-dependences.c (scop_get_reads_and_writes): Change
3557 output parameters to references.
3559 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
3561 PR 71026/tree-optimization
3562 * fold-const.c (distribute_real_division): Removed.
3563 (fold_binary_loc): Remove calls to distribute_real_divison.
3565 2017-10-17 Richard Biener <rguenther@suse.de>
3567 * graphite-scop-detection.c
3568 (scop_detection::stmt_has_simple_data_refs_p): Always use
3569 the full nest as region.
3570 (try_generate_gimple_bb): Likewise.
3571 * sese.c (scalar_evolution_in_region): Simplify now that
3572 SCEV can handle instantiation in regions.
3573 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
3574 in the non-loop part of a function if requested.
3576 2017-10-17 Richard Biener <rguenther@suse.de>
3578 PR tree-optimization/82563
3579 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
3581 (graphite_regenerate_ast_isl): Call it.
3582 * graphite-scop-detection.c (build_scops): Remove entry edge split.
3584 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3586 PR tree-optimization/82549
3587 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
3588 Formatting fixes. Instead of calling make_bit_field_ref with negative
3591 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
3593 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
3594 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
3595 _mm_maskz_reduce_ss): New.
3596 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
3597 __builtin_ia32_reducess_mask): Ditto..
3598 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
3599 * config/i386/sse.md (reduces<mode>): Renamed to ...
3600 (reduces<mode><mask_scalar_name>): ... this.
3601 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
3603 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
3604 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
3606 2017-10-16 David Malcolm <dmalcolm@redhat.com>
3608 * Makefile.in (OBJS): Add unique-ptr-tests.o.
3609 * selftest-run-tests.c (selftest::run_tests): Call
3610 selftest::unique_ptr_tests_cc_tests.
3611 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
3612 * unique-ptr-tests.cc: New file.
3614 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
3617 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
3619 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
3620 (make_hard_regno_born, make_hard_regno_dead): Update
3621 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
3623 2017-10-16 Jeff Law <law@redhat.com>
3625 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
3627 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
3629 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
3631 2017-10-16 Olivier Hainque <hainque@adacore.com>
3633 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
3634 with_cpu if we were configured for an e500v2 target cpu name.
3636 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
3638 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
3639 * doc/invoke.texi: Document +nodsp as a valid extension for
3642 2017-10-16 Martin Liska <mliska@suse.cz>
3644 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
3645 (test_set_range): Likewise.
3646 (test_range_functions): Rename to ...
3647 (test_bit_in_range): ... this.
3648 (sbitmap_c_tests): Add new test.
3650 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3652 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
3654 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
3655 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
3657 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3659 * config/aarch64/aarch64-builtins.c
3660 (aarch64_types_quadopu_lane_qualifiers): New.
3661 (TYPES_QUADOPU_LANE): New.
3662 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
3663 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
3664 (aarch64_<sur>dot_laneq<vsi2qi>): New.
3665 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
3666 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
3667 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
3668 (Vdottype, DOTPROD): New.
3669 (sur): Add SDOT and UDOT.
3671 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3673 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
3674 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
3675 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
3677 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
3678 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
3679 Enable TARGET_DOTPROD.
3680 (cortex-a75.cortex-a55): Likewise.
3681 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
3683 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3685 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
3686 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
3688 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
3690 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
3691 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
3692 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
3693 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
3694 * config/arm/types.md (neon_dot, neon_dot_q): New.
3695 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
3697 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3699 * config/arm/arm.h (TARGET_DOTPROD): New.
3700 * config/arm/arm.c (arm_arch_dotprod): New.
3701 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
3702 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
3703 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
3704 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
3705 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
3706 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
3707 * doc/invoke.texi (armv8.2-a): Document dotprod
3709 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
3711 * i386.c (ix86_vec_cost): New function.
3712 (ix86_rtx_costs): Handle vector operations better.
3713 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
3714 * x86-tune-costs.h: Add new costs to all tables.
3716 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
3718 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
3720 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
3721 divsd, sqrtss and sqrtsd
3722 * x86-tune-costs.h: Add new entries to all costs.
3723 (znver1_cost): Fix to match real instruction latencies.
3725 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3726 Michael Collison <michael.collison@arm.com>
3728 * compare-elim.c: Include emit-rtl.h.
3729 (can_merge_compare_into_arith): New function.
3730 (try_validate_parallel): Likewise.
3731 (try_merge_compare): Likewise.
3732 (try_eliminate_compare): Call the above when no previous clobber
3734 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
3737 2017-10-14 Jakub Jelinek <jakub@redhat.com>
3741 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
3742 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
3746 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
3747 to be any operand_equal_p operands. For & (B - 1) require
3748 B to be power of 2. Recognize
3749 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
3751 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3754 * optabs.c (expand_memory_blockage): Fix call of
3755 targetm.have_memory_blockage.
3757 2017-10-14 Jakub Jelinek <jakub@redhat.com>
3760 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
3761 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
3762 objects to extra_objs instead of overwriting it.
3764 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3766 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
3767 Use any_fp_register_operand as operand[3] predicate. Simplify
3768 equality test for operands[2] and operands[4] memory location.
3769 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
3770 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
3771 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
3772 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
3773 any_fp_register_operand as operand[1] predicate. Simplify
3774 equality test for operands[0] and operands[3] memory location.
3775 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
3776 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
3777 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
3779 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3781 * target-insns.def: Add memory_blockage.
3782 * optabs.c (expand_memory_blockage): New function.
3783 (expand_asm_memory_barrier): Rename ...
3784 (expand_asm_memory_blockage): ... to this.
3785 (expand_mem_thread_fence): Call expand_memory_blockage
3786 instead of expand_asm_memory_barrier.
3787 (expand_mem_singnal_fence): Ditto.
3788 (expand_atomic_load): Ditto.
3789 (expand_atomic_store): Ditto.
3790 * doc/md.texi (Standard Pattern Names For Generation):
3791 Document memory_blockage instruction pattern.
3793 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
3795 * config/rl78/rl78.c (rl78_emit_libcall): New function.
3796 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
3797 * config/rl78/rl78.md: New define_expand "adddi3".
3799 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
3801 * cfghooks.c (verify_flow_info): Disable check that all probabilities
3804 2017-10-13 Jeff Law <law@redhat.com>
3806 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
3808 2017-10-13 Jakub Jelinek <jakub@redhat.com>
3811 * internal-fn.c (expand_mul_overflow): If both operands have
3812 the same highpart of -1 or 0 and the topmost bit of lowpart
3813 is different, overflow is if res <= 0 rather than res < 0.
3815 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
3817 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
3818 TARGET_P9_VECTOR code for unaligned_load case.
3820 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
3822 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
3824 2017-10-13 Nathan Sidwell <nathan@acm.org>
3826 * tree-core.h (tree_contains_struct): Make bool.
3827 * tree.c (tree_contains_struct): Likewise.
3828 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
3829 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
3830 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
3831 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
3832 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
3834 2017-10-13 Richard Biener <rguenther@suse.de>
3836 * graphite-isl-ast-to-gimple.c
3837 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
3838 parameters and dominance check.
3839 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
3840 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
3841 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
3842 Do not update SSA form here or do intermediate IL verification.
3843 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
3844 (graphite_initialize): Remove check on the number of loops in
3845 the function and inline into graphite_transform_loops.
3846 (graphite_finalize): Inline into graphite_transform_loops.
3847 (graphite_transform_loops): Perform SSA update and IL verification
3849 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
3851 2017-10-13 Richard Biener <rguenther@suse.de>
3853 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
3854 graphite_expression_type_precision): Avoid global constructor
3856 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
3857 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
3858 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
3859 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
3860 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
3861 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
3863 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
3866 * config/i386/i386.h (ix86_red_zone_size): New.
3867 * config/i386/i386.md (push peephole2s): Replace
3868 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
3870 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3871 Alan Hayward <alan.hayward@arm.com>
3872 David Sherwood <david.sherwood@arm.com>
3874 * combine.c (can_change_dest_mode): Reject changes in
3875 REGMODE_NATURAL_SIZE.
3877 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3878 Alan Hayward <alan.hayward@arm.com>
3879 David Sherwood <david.sherwood@arm.com>
3881 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
3882 (expand_debug_source_expr): Likewise.
3883 * combine.c (combine_simplify_rtx): Likewise.
3884 * cse.c (fold_rtx): Likewise.
3885 * fwprop.c (canonicalize_address): Likewise.
3886 * targhooks.c (default_shift_truncation_mask): Likewise.
3888 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3889 Alan Hayward <alan.hayward@arm.com>
3890 David Sherwood <david.sherwood@arm.com>
3892 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
3893 (widened_mode): Likewise.
3894 (expand_unop): Likewise.
3895 * ree.c (transform_ifelse): Likewise.
3896 (merge_def_and_ext): Likewise.
3897 (combine_reaching_defs): Likewise.
3898 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3900 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3901 Alan Hayward <alan.hayward@arm.com>
3902 David Sherwood <david.sherwood@arm.com>
3904 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
3905 * combine.c (gen_lowpart_for_combine): Likewise.
3906 * dwarf2out.c (rtl_for_decl_location): Likewise.
3907 * final.c (alter_subreg): Likewise.
3908 * rtlhooks.c (gen_lowpart_general): Likewise.
3909 (gen_lowpart_if_possible): Likewise.
3911 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3912 Alan Hayward <alan.hayward@arm.com>
3913 David Sherwood <david.sherwood@arm.com>
3915 * calls.c (expand_call): Use subreg_lowpart_offset.
3916 * cse.c (cse_insn): Likewise.
3917 * regcprop.c (copy_value): Likewise.
3918 (copyprop_hardreg_forward_1): Likewise.
3920 2017-10-13 Jakub Jelinek <jakub@redhat.com>
3923 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
3924 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
3925 =Q constraints to +Q and into insn condition add check
3926 that operands[0] and operands[1] are equal.
3927 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
3928 =Q constraints to +Q and into insn condition add check
3929 that operands[0] is equal to either operands[1] or operands[2].
3932 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
3933 instead of handling MINUS_EXPR twice (once for each argument),
3934 canonicalize operand order and handle just once, use rtype where
3935 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
3938 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
3939 any values of __C while still being pattern recognizable as a simple
3942 2017-10-13 Richard Biener <rguenther@suse.de>
3944 PR tree-optimization/82451
3946 2017-10-02 Richard Biener <rguenther@suse.de>
3948 PR tree-optimization/82355
3949 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
3950 a mapping for the enclosing loop but avoid generating one for
3952 (copy_bb_and_scalar_dependences): Remove premature codegen
3953 error on PHIs in blocks duplicated into multiple places.
3954 * graphite-scop-detection.c
3955 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
3956 in the region use it as loop and nest to analyze the DR in.
3957 (try_generate_gimple_bb): Likewise.
3958 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
3959 (add_loop_constraints): For blocks in a loop not in the region
3960 create a dimension with a single iteration.
3961 * sese.h (gbb_loop_at_index): Remove assert.
3963 * cfgloop.c (loop_preheader_edge): For the loop tree root
3964 return the single successor of the entry block.
3965 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3966 Reset the SCEV hashtable and niters.
3967 * graphite-scop-detection.c
3968 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
3969 assert that we only have POLYNOMIAL_CHREC that vary in loops
3970 contained in the region.
3971 (scop_detection::graphite_can_represent_expr): Adjust.
3972 (scop_detection::stmt_has_simple_data_refs_p): For loops
3973 not in the region set loop to NULL. The nest is now the
3974 entry edge to the region.
3975 (try_generate_gimple_bb): Likewise.
3976 * sese.c (scalar_evolution_in_region): Adjust for
3977 instantiate_scev change.
3978 * tree-data-ref.h (graphite_find_data_references_in_stmt):
3979 Make nest parameter the edge into the region.
3980 (create_data_ref): Likewise.
3981 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
3982 entry edge into a region and adjust instantiate_scev calls.
3983 (create_data_ref): Likewise.
3984 (graphite_find_data_references_in_stmt): Likewise.
3985 (find_data_references_in_stmt): Pass the loop preheader edge
3986 from the nest argument.
3987 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
3988 parameter the edge into the region.
3989 (instantiate_parameters): Use the loop preheader edge as entry.
3990 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
3992 (get_instantiated_value_entry): Make instantiate_below parameter
3993 the edge into the region.
3994 (instantiate_scev_name): Likewise. Adjust dominance checks,
3995 when we cannot use loop-based instantiation instantiate by
3996 walking use-def chains.
3997 (instantiate_scev_poly): Adjust.
3998 (instantiate_scev_binary): Likewise.
3999 (instantiate_scev_convert): Likewise.
4000 (instantiate_scev_not): Likewise.
4001 (instantiate_array_ref): Remove.
4002 (instantiate_scev_3): Likewise.
4003 (instantiate_scev_2): Likewise.
4004 (instantiate_scev_1): Likewise.
4005 (instantiate_scev_r): Do not blindly handle N-operand trees.
4006 Do not instantiate array-refs. Handle all constants and invariants.
4007 (instantiate_scev): Make instantiate_below parameter
4008 the edge into the region.
4009 (resolve_mixers): Use the loop preheader edge for the region
4010 parameter to instantiate_scev_r.
4011 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
4013 2017-10-13 Richard Biener <rguenther@suse.de>
4015 PR tree-optimization/82525
4016 * graphite-isl-ast-to-gimple.c
4017 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
4019 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
4020 Fail code generation when we cannot represent the isl integer.
4021 (binary_op_to_tree): Elide modulo operations that are no-ops
4022 in the type we code generate. Remove now superfluous code
4025 2017-10-13 Richard Biener <rguenther@suse.de>
4027 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
4028 (scop_detection::harmful_loop_in_region): Remove premature
4029 IV type restriction.
4030 (scop_detection::graphite_can_represent_scev): We can handle
4031 pointer IVs just fine.
4033 2017-10-13 Alan Modra <amodra@gmail.com>
4035 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
4036 "Clobbers and Scratch Registers". Add paragraph on
4037 alternative to clobbers for scratch registers and OpenBLAS
4040 2017-10-13 Alan Modra <amodra@gmail.com>
4042 * doc/extend.texi (Clobbers): Correct vax example. Delete old
4043 example of a memory input for a string of known length. Move
4044 commentary out of table. Add a number of new examples
4045 covering array memory inputs.
4047 2017-10-12 Martin Liska <mliska@suse.cz>
4049 PR tree-optimization/82493
4050 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
4051 (test_range_functions): New function.
4052 (sbitmap_c_tests): Likewise.
4053 * selftest-run-tests.c (selftest::run_tests): Run new tests.
4054 * selftest.h (sbitmap_c_tests): New function.
4056 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
4058 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
4060 * config/rs6000/amo.h: Fix spacing issue.
4062 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4065 * config/i386/i386.md (*ashl<mode>3_mask_1,
4066 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
4067 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
4070 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4072 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
4073 (profile_probability): Set max_probability
4074 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
4076 * profile-count.c (profile_probability::differs_from_p): Do not
4077 rely on max_probaiblity == 10000
4079 2017-10-12 Jeff Law <law@redhat.com>
4081 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
4084 2017-10-12 Martin Sebor <msebor@redhat.com>
4088 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
4089 (handle_alias_pairs): Call it.
4090 * common.opt (-Wattribute-alias): New option.
4091 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
4092 * doc/invoke.texi (-Wattribute-alias): Document.
4094 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
4097 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4099 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4101 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4102 (make_hard_regno_born, make_hard_regno_dead): Update
4103 bb_killed_pseudos and bb_gen_pseudos.
4105 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4107 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
4109 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
4111 * config/alpha/alpha.c (alpha_split_conditional_move):
4112 Use std::swap instead of manually swapping.
4113 (alpha_stdarg_optimize_hook): Ditto.
4114 (alpha_canonicalize_comparison): Ditto.
4116 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4118 * tree-loop-distribution.c (struct builtin_info): New struct.
4119 (struct partition): Refactor fields into struct builtin_info.
4120 (partition_free): Free struct builtin_info.
4121 (build_size_arg_loc, build_addr_arg_loc): Delete.
4122 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
4123 information from struct builtin_info.
4124 (find_single_drs): New function refactored from classify_partition.
4125 Also moved builtin validity checks to this function.
4126 (compute_access_range, alloc_builtin): New functions.
4127 (classify_builtin_st, classify_builtin_ldst): New functions.
4128 (classify_partition): Refactor code into functions find_single_drs,
4129 classify_builtin_st and classify_builtin_ldst.
4130 (distribute_loop): Don't do runtime alias check when distributing
4132 (find_seed_stmts_for_distribution): New function.
4133 (pass_loop_distribution::execute): Refactor code finding seed
4134 stmts into above function. Support distribution for the innermost
4135 two-level loop nest. Adjust dump information.
4137 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4139 * tree-loop-distribution.c: Adjust the general comment.
4140 (NUM_PARTITION_THRESHOLD): New macro.
4141 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
4142 (classify_partition): Skip builtin pattern of loop nest's inner loop.
4143 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
4144 in call to build_partition_graph.
4145 (finalize_partitions): New parameter. Make loop distribution more
4146 conservative by fusing more partitions.
4147 (distribute_loop): Don't do runtime alias check in case of loop nest
4149 (find_seed_stmts_for_distribution): New function.
4150 (prepare_perfect_loop_nest): New function.
4151 (pass_loop_distribution::execute): Refactor code finding seed stmts
4152 and loop nest into above functions. Support loop nest distribution.
4153 Adjust dump information accordingly.
4155 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4157 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
4158 and set PTYPE_SEQUENTIAL for merged partition.
4160 2017-10-12 Richard Biener <rguenther@suse.de>
4162 PR tree-optimization/69728
4164 2017-09-19 Richard Biener <rguenther@suse.de>
4166 PR tree-optimization/69728
4167 * graphite-sese-to-poly.c (schedule_error): New global.
4168 (add_loop_schedule): Handle empty domain by failing the
4170 (build_original_schedule): Handle schedule_error.
4172 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
4173 domain by returning an unchanged schedule.
4175 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4177 * genrecog.c (validate_pattern): For VEC_SELECT verify that
4178 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
4180 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
4182 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
4185 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4188 * expr.c (store_field): Don't optimize away bitsize == 0 store
4189 from CALL_EXPR with addressable return type.
4191 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4193 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
4194 * config/rs6000/rs6000.md (sel): Delete mode attribute.
4195 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
4196 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
4197 TARGET_ISEL instead of TARGET_ISEL<sel>.
4199 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
4201 * config/rs6000/rs6000.c
4202 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
4204 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4206 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
4208 (reg_or_cint_operand): Fix comment.
4209 (reg_or_zero_operand): New predicate.
4210 * config/rs6000/rs6000-protos.h (output_isel): Delete.
4211 * config/rs6000/rs6000.c (output_isel): Delete.
4212 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
4213 instead of reg_or_cint_operand. Output instruction directly (not via
4215 (isel_unsigned_<mode>): Ditto.
4216 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
4217 gpc_reg_operand. Add an instruction alternative for this. Output
4218 instruction directly.
4219 (*isel_reversed_unsigned_<mode>): Ditto.
4221 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4223 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
4224 (TARGET_CANONICALIZE_COMPARISON): Define.
4226 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
4229 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
4230 Check whether the dest is REG before adding REG_EQUIV note.
4232 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4235 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4237 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4238 (make_hard_regno_born, make_hard_regno_dead): Update
4239 bb_killed_pseudos and bb_gen_pseudos.
4241 2017-10-11 Nathan Sidwell <nathan@acm.org>
4243 * incpath.h (enum incpath_kind): Name enum, prefix values.
4244 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
4245 * incpath.c (heads, tails): Use INC_MAX.
4246 (add_env_var_paths, add_standard_paths): Use incpath_kind.
4247 (merge_include_chains, split_quote_chain,
4248 register_include_chains): Update incpath_kind names.
4249 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
4250 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
4252 (add_framework_path, darwin_register_objc_includes): Likewise.
4253 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
4255 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4257 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
4258 Do not use float_operator operator predicate.
4259 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
4260 * config/i386/predicates.md (float_operator): Remove predicate.
4262 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4264 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
4265 (*jccxf_i387): Ditto.
4266 (*jcc<mode>_i387): Ditto.
4267 (*jccu<mode>_i387): Ditto.
4268 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
4269 (*jcc_*_i387 splitters): Remove.
4270 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
4271 * config/i386/i386.c (ix86_split_fp_branch): Remove.
4272 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
4275 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4277 * profile-count.h (slow_safe_scale_64bit): New function.
4278 (safe_scale_64bit): New inline.
4279 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
4280 * profile-count.c: Include wide-int.h
4281 (slow_safe_scale_64bit): New.
4283 2017-10-11 Nathan Sidwell <nathan@acm.org>
4285 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
4286 HAS_DECL_ASSEMBLER_NAME_P.
4287 * gimple-expr.c (gimple_decl_printable_name: Check
4288 HAS_DECL_ASSEMBLER_NAME_P too.
4289 * ipa-utils.h (type_in_anonymous_namespace_p): Check
4290 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
4291 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
4292 * passes.c (rest_of_decl_compilation): Check
4293 HAS_DECL_ASSEMBLER_NAME_P too.
4294 * recog.c (verify_changes): Likewise.
4295 * tree-pretty-print.c (dump_decl_name): Likewise.
4296 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
4298 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
4299 (SET_DECL_ASSEMBLER_NAME): Use it.
4300 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
4301 (COPY_DECL_ASSEMBLER_NAME): Likewise.
4302 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
4304 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4306 * config.gcc (i386, x86_64): Add extra objects.
4307 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
4308 (ix86_min_insn_size): Declare.
4309 (ix86_issue_rate): Declare.
4310 (ix86_adjust_cost): Declare.
4311 (ia32_multipass_dfa_lookahead): Declare.
4312 (ix86_macro_fusion_p): Declare.
4313 (ix86_macro_fusion_pair_p): Declare.
4314 (ix86_bd_has_dispatch): Declare.
4315 (ix86_bd_do_dispatch): Declare.
4316 (ix86_core2i7_init_hooks): Declare.
4317 (ix86_atom_sched_reorder): Declare.
4318 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
4319 (COSTS_N_BYTES): Move to x86-tune-costs.h.
4320 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
4321 (rip_relative_addr_p): Rename to ...
4322 (ix86_rip_relative_addr_p): ... this one; export.
4323 (memory_address_length): Update.
4324 (ix86_issue_rate): Move to x86-tune-sched.c.
4325 (ix86_flags_dependent): Move to x86-tune-sched.c.
4326 (ix86_agi_dependent): Move to x86-tune-sched.c.
4327 (exact_dependency_1): Move to x86-tune-sched.c.
4328 (exact_store_load_dependency): Move to x86-tune-sched.c.
4329 (ix86_adjust_cost): Move to x86-tune-sched.c.
4330 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
4331 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
4332 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
4333 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
4334 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
4335 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
4336 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
4337 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
4338 (min_insn_size): Rename to ...
4339 (ix86_min_insn_size): ... this one; export.
4340 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
4341 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
4342 (core2i7_first_cycle_multipass_backtrack): Move to
4343 x86-tune-sched-core.c.
4344 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
4345 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
4346 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
4347 (ix86_avoid_jump_mispredicts): Update.
4348 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
4349 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
4350 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
4351 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
4352 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
4353 (MAX_INSN): Move to ix86-tune-sched-bd.c.
4354 (MAX_IMM): Move to ix86-tune-sched-bd.c.
4355 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
4356 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
4357 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
4358 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
4359 (MAX_STORE): Move to ix86-tune-sched-bd.c.
4360 (BIG): Move to ix86-tune-sched-bd.c.
4361 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
4362 (enum insn_path): Move to ix86-tune-sched-bd.c.
4363 (get_mem_group): Move to ix86-tune-sched-bd.c.
4364 (is_cmp): Move to ix86-tune-sched-bd.c.
4365 (dispatch_violation): Move to ix86-tune-sched-bd.c.
4366 (is_branch): Move to ix86-tune-sched-bd.c.
4367 (is_prefetch): Move to ix86-tune-sched-bd.c.
4368 (init_window): Move to ix86-tune-sched-bd.c.
4369 (allocate_window): Move to ix86-tune-sched-bd.c.
4370 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
4371 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
4372 (process_end_window): Move to ix86-tune-sched-bd.c.
4373 (allocate_next_window): Move to ix86-tune-sched-bd.c.
4374 (find_constant): Move to ix86-tune-sched-bd.c.
4375 (get_num_immediates): Move to ix86-tune-sched-bd.c.
4376 (has_immediate): Move to ix86-tune-sched-bd.c.
4377 (get_insn_path): Move to ix86-tune-sched-bd.c.
4378 (get_insn_group): Move to ix86-tune-sched-bd.c.
4379 (count_num_restricted): Move to ix86-tune-sched-bd.c.
4380 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
4381 (add_insn_window): Move to ix86-tune-sched-bd.c.
4382 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
4383 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
4384 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
4385 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
4386 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
4387 (do_dispatch): Move to ix86-tune-sched-bd.c.
4388 (has_dispatch): Move to ix86-tune-sched-bd.c.
4389 * config/i386/t-i386: Add new object files.
4390 * config/i386/x86-tune-costs.h: New file.
4391 * config/i386/x86-tune-sched-atom.c: New file.
4392 * config/i386/x86-tune-sched-bd.c: New file.
4393 * config/i386/x86-tune-sched-core.c: New file.
4394 * config/i386/x86-tune-sched.c: New file.
4396 2017-10-11 Liu Hao <lh_mouse@126.com>
4398 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
4399 the generic version below instead.
4400 (should_colorize): Recognize Windows consoles as terminals
4402 * pretty-print.c [__MINGW32__] (write_all): New function.
4403 [__MINGW32__] (find_esc_head): Likewise.
4404 [__MINGW32__] (find_esc_terminator): Likewise.
4405 [__MINGW32__] (eat_esc_sequence): Likewise.
4406 [__MINGW32__] (mingw_ansi_fputs): New function that handles
4408 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
4411 2017-10-11 Richard Biener <rguenther@suse.de>
4413 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
4414 Properly call analyze_scalar_evolution with the loop of the stmt.
4416 2017-10-11 Richard Biener <rguenther@suse.de>
4418 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
4419 * tree-core.h (tree_base): Add chrec_var union member.
4420 * tree.h (CHREC_VAR): Remove.
4421 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
4422 * tree-chrec.h (build_polynomial_chrec): Adjust.
4423 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
4424 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
4426 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
4428 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
4429 * match.pd: ... here.
4430 ((T) X == (T) Y): Relax condition.
4432 2017-10-11 Bin Cheng <bin.cheng@arm.com>
4434 PR tree-optimization/82472
4435 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
4437 (break_alias_scc_partitions): Update postorder number.
4439 2017-10-11 Martin Liska <mliska@suse.cz>
4442 * opts.c (parse_no_sanitize_attribute): Do not use error_value
4444 * opts.h (parse_no_sanitize_attribute): Remove last argument.
4446 2017-10-11 Martin Liska <mliska@suse.cz>
4448 * print-rtl.c (print_insn): Move declaration of idbuf
4449 to same scope as name.
4451 2017-10-11 Martin Liska <mliska@suse.cz>
4456 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
4459 2017-10-11 Martin Liska <mliska@suse.cz>
4462 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
4465 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
4467 * config.gcc (default_gnu_indirect_function): Default to yes for
4468 arm*-*-linux* with glibc.
4470 2017-10-11 Richard Biener <rguenther@suse.de>
4472 * tree-scalar-evolution.c (get_scalar_evolution): Handle
4473 default-defs and types we do not want to analyze.
4474 (interpret_loop_phi): Replace unreachable code with an assert.
4475 (compute_scalar_evolution_in_loop): Remove and inline ...
4476 (analyze_scalar_evolution_1): ... here, replacing condition with
4477 what makes the intent clearer. Remove handling of cases
4478 get_scalar_evolution now handles.
4480 2017-10-10 Jim Wilson <wilson@tuliptree.org>
4482 PR rtl-optimization/81434
4483 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
4484 comment for main loop. In sched_group_found if, also add checks for
4485 pass and min_cost_group.
4487 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
4489 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
4490 (rs6000_insn_cost): New function.
4491 * config/rs6000/rs6000.md (cost): New attribute.
4493 2017-10-10 Jakub Jelinek <jakub@redhat.com>
4494 H.J. Lu <hongjiu.lu@intel.com>
4498 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
4499 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
4500 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
4501 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
4502 to OPTION_MASK_ISA_AVX512VL - builtins that have both
4503 OPTION_MASK_ISA_MMX and some other bit set require both
4504 mmx and the ISAs without the mmx bit.
4505 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
4506 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
4507 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
4508 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
4509 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
4510 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
4511 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
4512 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
4513 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
4514 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
4515 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
4516 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
4517 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
4518 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
4519 Add OPTION_MASK_ISA_MMX.
4521 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
4523 * config.gcc (armv7*-*-freebsd*): New target.
4524 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
4526 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
4528 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
4529 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
4532 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
4534 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
4535 a template parameter.
4536 (WIDE_INT_REF_FOR): Update accordingly.
4537 * tree.h (wi::int_traits <const_tree>): Delete.
4538 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
4539 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
4540 (wi::tree_to_wide_ref): New typedef.
4541 (wi::to_wide): New function.
4542 * calls.c (get_size_range): Use wi::to_wide when operating on
4544 * cgraph.c (cgraph_node::create_thunk): Likewise.
4545 * config/i386/i386.c (ix86_data_alignment): Likewise.
4546 (ix86_local_alignment): Likewise.
4547 * dbxout.c (stabstr_O): Likewise.
4548 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
4549 * expr.c (const_vector_from_tree): Likewise.
4550 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
4551 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
4552 (fold_negate_expr_1, int_const_binop_1, const_binop)
4553 (fold_convert_const_int_from_real, optimize_bit_field_compare)
4554 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
4555 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
4556 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
4557 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
4558 (fold_not_const, round_up_loc): Likewise.
4559 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
4560 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
4561 (alloca_call_type): Likewise.
4562 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
4563 * godump.c (go_output_typedef): Likewise.
4564 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
4565 * internal-fn.c (get_min_precision): Likewise.
4566 * ipa-cp.c (ipcp_store_vr_results): Likewise.
4567 * ipa-polymorphic-call.c
4568 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
4569 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
4570 (ipa_modify_call_arguments): Likewise.
4571 * match.pd: Likewise.
4572 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
4573 * print-tree.c (print_node_brief, print_node): Likewise.
4574 * stmt.c (expand_case): Likewise.
4575 * stor-layout.c (layout_type): Likewise.
4576 * tree-affine.c (tree_to_aff_combination): Likewise.
4577 * tree-cfg.c (group_case_labels_stmt): Likewise.
4578 * tree-data-ref.c (dr_analyze_indices): Likewise.
4579 (prune_runtime_alias_test_list): Likewise.
4580 * tree-dump.c (dequeue_and_dump): Likewise.
4581 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
4582 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
4583 * tree-pretty-print.c (dump_generic_node): Likewise.
4584 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
4585 (simple_iv_with_niters): Likewise.
4586 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
4587 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
4588 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
4589 * tree-ssa-loop-niter.c (split_to_var_and_offset)
4590 (refine_value_range_using_guard, number_of_iterations_ne_max)
4591 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
4592 (get_cst_init_from_scev, record_nonwrapping_iv)
4593 (scev_var_range_cant_overflow): Likewise.
4594 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
4595 * tree-ssa-pre.c (compute_avail): Likewise.
4596 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
4597 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
4598 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
4599 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
4600 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
4601 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
4602 (dump_case_nodes, try_switch_expansion): Likewise.
4603 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
4604 (vect_do_peeling): Likewise.
4605 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
4606 * tree-vect-stmts.c (vectorizable_load): Likewise.
4607 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
4608 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
4609 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
4610 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
4611 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
4612 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
4613 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
4614 (evrp_dom_walker::before_dom_children): Likewise.
4615 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
4616 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
4617 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
4618 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
4619 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
4620 (get_range_pos_neg): Likewise.
4621 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
4622 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
4623 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
4624 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
4625 * config/avr/avr.c (avr_fold_builtin): Likewise.
4626 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
4627 * config/msp430/msp430.c (msp430_attr): Likewise.
4628 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
4629 * config/powerpcspe/powerpcspe-c.c
4630 (altivec_resolve_overloaded_builtin): Likewise.
4631 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
4632 (rs6000_expand_ternop_builtin): Likewise.
4633 * config/rs6000/rs6000-c.c
4634 (altivec_resolve_overloaded_builtin): Likewise.
4635 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
4636 (rs6000_expand_ternop_builtin): Likewise.
4637 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
4639 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4641 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
4642 when copying loop nest with only one inner loop.
4644 2017-10-10 Richard Biener <rguenther@suse.de>
4646 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
4647 blocks if SCEV is active.
4648 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
4650 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
4651 (scev_initialize): Assert we are not yet initialized.
4653 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4655 * tree-loop-distribution.c (generate_loops_for_partition): Remove
4656 inner loop's exit stmt by making it always exit the loop, otherwise
4657 we would generate an infinite empty loop.
4659 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4661 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
4662 renaming variables in new preheader if it's deleted.
4664 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4666 * tree-loop-distribution.c (struct partition): Remove unused field
4667 loops of the structure.
4668 (partition_alloc, partition_free): Ditto.
4669 (build_rdg_partition_for_vertex): Ditto.
4671 2017-10-09 Jeff Law <law@redhat.com>
4673 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
4674 return type to match prototype and documentation.
4676 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4678 * config/rs6000/rs6000.c (processor_costs): Move to ...
4679 * config/rs6000/rs6000.h: ... here.
4680 (rs6000_cost): Declare.
4682 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
4684 * except.c (setjmp_fn): New global variable.
4685 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
4686 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
4687 if DONT_USE_BUILTIN_SETJMP is defined.
4689 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4691 * target.def (insn_cost): New hook.
4692 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
4693 * doc/tm.texi: Regenerate.
4694 * rtlanal.c (insn_cost): Use the new hook.
4696 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4698 * combine.c (combine_validate_cost): Compute the new insn_cost,
4699 not just pattern_cost.
4700 (try_combine): Adjust comment.
4702 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4704 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
4706 * combine.c (uid_insn_cost): Adjust comment.
4707 (combine_validate_cost): Adjust comment. Use pattern_cost instead
4709 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
4710 * dse.c (find_shift_sequence): Ditto.
4711 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
4712 (bb_valid_for_noce_process_p): Use pattern_cost.
4713 * rtl.h (insn_rtx_cost): Delete.
4714 (pattern_cost): New prototype.
4715 (insn_cost): New prototype.
4716 * rtlanal.c (insn_rtx_cost): Rename to...
4717 (pattern_cost): ... this.
4720 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
4722 * config/i386/i386.md (*jcc_2): Remove insn pattern.
4723 (*jcc<mode>_0_r_i387): Ditto.
4724 (*jccxf_r_i387): Ditto.
4725 (*jcc<mode>_r_i387): Ditto.
4726 (*jccu<mode>_r_i387): Ditto.
4727 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
4728 (*jcc): Rename from *jcc_1.
4730 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4732 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
4733 deferred rescans after the lvx/stvx recombination pre-pass.
4735 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
4737 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
4738 memory operation instruction support.
4739 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
4740 (rs6000-ibm-aix[789]*): Likewise.
4741 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
4742 Document new functions.
4744 2017-10-09 Richard Biener <rguenther@suse.de>
4746 PR tree-optimization/82397
4747 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
4748 equality only for semantically equal trees.
4750 2017-10-09 Richard Biener <rguenther@suse.de>
4752 PR tree-optimization/82449
4753 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
4754 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
4755 allow constant addresses.
4756 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
4759 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4761 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
4764 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4767 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
4770 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4773 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
4775 2017-10-09 Jakub Jelinek <jakub@redhat.com>
4778 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
4779 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
4781 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
4783 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
4784 (WI_BINARY_PREDICATE_RESULT): Likewise.
4785 (wi::binary_traits::operator_result): New type.
4786 (wi::binary_traits::predicate_result): Likewise.
4787 (generic_wide_int::operator~, unary generic_wide_int::operator-)
4788 (generic_wide_int::operator==, generic_wide_int::operator!=)
4789 (generic_wide_int::operator&, generic_wide_int::and_not)
4790 (generic_wide_int::operator|, generic_wide_int::or_not)
4791 (generic_wide_int::operator^, generic_wide_int::operator+
4792 (binary generic_wide_int::operator-, generic_wide_int::operator*):
4794 (operator~, unary operator-, operator==, operator!=, operator&)
4795 (operator|, operator^, operator+, binary operator-, operator*): New
4797 * expr.c (get_inner_reference): Use wi::bit_and_not.
4798 * fold-const.c (fold_binary_loc): Likewise.
4799 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
4800 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
4801 (bit_value_binop): Likewise.
4802 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
4803 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
4804 (extract_range_from_binary_expr_1): Likewise.
4805 (masked_increment): Likewise.
4806 (simplify_bit_ops_using_ranges): Likewise.
4808 2017-10-09 Martin Jambor <mjambor@suse.cz>
4811 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
4812 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
4813 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
4815 (hsa_fixup_mov_insn_type): New function.
4816 (hsa_op_with_type::get_in_type): Use it.
4817 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
4818 immediates in an assert.
4819 (hsa_op_with_type::extend_int_to_32bit): New method.
4820 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
4821 types. Convert to dest type if necessary.
4822 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
4823 (reg_for_gimple_ssa): Pass false as min32int to
4824 hsa_type_for_scalar_tree_type.
4825 (gen_hsa_addr): Fixup type when creating addresable temporary.
4826 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
4827 (gen_hsa_unary_operation): Extend operands and convert to dest type if
4828 necessary. Call hsa_fixup_mov_insn_type.
4829 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
4830 extend operands and convert to dest type if necessary.
4831 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
4832 to dest type if necessary.
4833 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
4834 if conversion nt necessary and size matches.
4835 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
4836 to dest type if necessary.
4837 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
4838 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
4840 (gen_hsa_clrsb): Likewise.
4841 (gen_hsa_ffs): Likewise.
4842 (gen_hsa_divmod): Extend operands and convert to dest type if
4844 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
4846 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
4848 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
4849 Remove empty default arguments. Use a brace block as output
4851 (conditional return): Ditto.
4853 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
4854 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
4856 (group_ending_nop): Ditto.
4857 (doloop_end): Ditto.
4858 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
4859 (splitters for those): Ditto.
4861 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
4863 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
4864 a conditional jump (and the compare for it) so that pc_rtx is the
4866 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
4867 for the deleted and renamed ctr<mode>_internal[234] patterns.
4868 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
4869 Delete second conditional return pattern.
4870 (ctr<mode>_internal2): Delete this second bdnz pattern.
4871 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
4872 (ctr<mode>_internal4): Delete this second bdz pattern.
4874 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
4876 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
4877 (always_initialized_rtx_for_ssa_name_p): New predicate.
4878 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
4879 (finish_out_of_ssa): Free new field of SA.
4880 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
4881 * tree-ssa-coalesce.c: Include tree-ssa.h.
4882 (get_parm_default_def_partitions): Remove extern keyword.
4883 (get_undefined_value_partitions): New function.
4884 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
4885 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
4887 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
4889 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
4891 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
4893 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
4895 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
4896 for targets that preffer 128bit.
4898 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
4900 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
4902 2017-10-08 Olivier Hainque <hainque@adacore.com>
4904 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
4905 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
4907 (thumb_set_return_address): Likewise.
4909 2017-10-08 Olivier Hainque <hainque@adacore.com>
4911 * common/config/arm/arm-common.c (arm_except_unwind_info):
4912 Handle DWARF2_UNWIND_INFO.
4914 2017-10-07 Michael Collison <michael.collison@arm.com>
4916 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
4919 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
4921 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
4922 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
4923 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
4924 defined, force the creation of a new block for a dispatch label.
4926 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
4928 * invoke.texi (Wsuggest-attribute=cold): Document.
4929 * common.opt (Wsuggest-attribute=cold): New
4930 * ipa-pure-const.c (warn_function_cold): New function.
4931 * predict.c (compute_function_frequency): Use it.
4932 * predict.h (warn_function_cold): Declare.
4934 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
4936 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
4939 2017-10-06 Martin Liska <mliska@suse.cz>
4941 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
4942 keyword for member functions.
4943 (struct sanopt_tree_couple): New struct.
4944 (struct sanopt_tree_couple_hash): New function.
4945 (struct sanopt_ctx): Add new hash_map.
4946 (has_dominating_ubsan_ptr_check): New function.
4947 (record_ubsan_ptr_check_stmt): Likewise.
4948 (maybe_optimize_ubsan_ptr_ifn): Likewise.
4949 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
4950 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
4952 2017-10-06 Sudakshina Das <sudi.das@arm.com>
4955 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
4956 aarch64_simd_valid_immediate on CONST_VECTORs.
4957 (aarch64_reg_or_bic_imm): Likewise.
4959 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
4961 PR rtl-optimization/82396
4962 * haifa-sched.c (ready_sort_real): Disable qsort checking.
4964 2017-10-06 Sebastian Pop <sebpop@gmail.com>
4966 * graphite-dependences.c (scop_get_reads): Move code to...
4967 (scop_get_must_writes): Move code to...
4968 (scop_get_may_writes): Move code to...
4969 (scop_get_reads_and_writes): ... here.
4970 (scop_get_dependences): Call scop_get_reads_and_writes.
4972 2017-10-06 Jakub Jelinek <jakub@redhat.com>
4974 PR tree-optimization/82434
4975 * fold-const.h (can_native_encode_type_p,
4976 can_native_encode_string_p): Remove.
4977 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
4978 don't encode anything, just return what would be otherwise returned.
4979 (native_encode_fixed, native_encode_complex, native_encode_vector):
4981 (native_encode_string): Likewise. Inline by hand
4982 can_native_encode_string_p.
4983 (can_native_encode_type_p): Remove.
4984 (can_native_encode_string_p): Remove.
4985 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
4986 STRING_CSTs using can_native_encode_string_p, test all
4987 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
4988 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
4989 argument from native_encode_expr.
4990 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
4991 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
4994 2017-10-06 Richard Biener <rguenther@suse.de>
4996 PR tree-optimization/82397
4997 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
4998 operand_equal_p but rely on data_ref_compare_tree for detecting
5000 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
5001 to match up with dr_group_sort_cmp.
5003 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5006 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
5008 * config/s390/s390-builtin-types.def: Regenerate.
5010 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5013 * config/s390/s390-builtin-types.def: Regenerate.
5014 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
5015 Change flag from B_VXE to B_VX.
5016 (s390_vec_min_dbl): Remove B_VXE flag.
5018 2017-10-06 Richard Biener <rguenther@suse.de>
5020 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
5021 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
5022 translate_isl_ast_to_gimple::is_valid_rename,
5023 translate_isl_ast_to_gimple::get_rename,
5024 translate_isl_ast_to_gimple::get_def_bb_for_const,
5025 translate_isl_ast_to_gimple::get_new_name,
5026 translate_isl_ast_to_gimple::collect_all_ssa_names,
5027 translate_isl_ast_to_gimple::copy_loop_phi_args,
5028 translate_isl_ast_to_gimple::collect_all_ssa_names,
5029 translate_isl_ast_to_gimple::copy_loop_phi_args,
5030 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
5031 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
5032 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
5033 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
5034 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
5035 translate_isl_ast_to_gimple::copy_cond_phi_args,
5036 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
5037 translate_isl_ast_to_gimple::edge_for_new_close_phis,
5038 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
5039 translate_isl_ast_to_gimple::rename_uses,
5040 translate_isl_ast_to_gimple::rename_all_uses): Remove.
5041 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
5042 (set_rename_for_each_def): Likewise.
5043 (graphite_copy_stmts_from_block): Handle debug stmt resetting
5044 here. Handle rewriting SCEV analyzable uses here.
5045 (copy_bb_and_scalar_dependences): Generate code for PHI
5047 (graphite_regenerate_ast_isl): Adjust.
5048 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
5049 (add_write, add_read): New functions.
5050 (build_cross_bb_scalars_def): Use it and simplify.
5051 (build_cross_bb_scalars_use): Likewise.
5052 (graphite_find_cross_bb_scalar_vars): Inline into...
5053 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
5054 simulating out-of-SSA. Compute liveout and add dependencies.
5055 (build_scops): Force an empty entry block.
5056 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
5058 (sese_build_liveouts): Declare.
5059 (sese_trivially_empty_bb_p): Likewise.
5060 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
5061 compute liveout and debug_liveout.
5062 (sese_bad_liveouts_use): Remove.
5063 (sese_reset_debug_liveouts_bb): Likewise.
5064 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
5065 (sese_build_liveouts): Build liveout and debug_liveout and store
5067 (new_sese_info): Adjust.
5068 (free_sese_info): Likewise.
5069 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
5070 do not build liveout here.
5071 (move_sese_in_condition): Adjust region entry.
5072 (scev_analyzable_p): Match up with chrec_apply requirements.
5073 (sese_trivially_empty_bb_p): New.
5074 * tree-into-ssa.c (get_reaching_def): Properly support generating
5075 default-defs for incremental rewrite of anonymous names.
5077 2017-10-06 Richard Biener <rguenther@suse.de>
5079 * graphite-sese-to-poly.c (extract_affine): For casts increasing
5080 precision do not perform modulo reduction.
5082 2017-10-06 Richard Biener <rguenther@suse.de>
5084 PR tree-optimization/82436
5085 * tree-vect-slp.c (vect_supported_load_permutation_p): More
5086 conservatively choose the vectorization factor when checking
5087 whether we can perform the required load permutation.
5088 (vect_transform_slp_perm_load): Assert when we may not fail.
5090 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
5092 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
5093 message for incompatible -msdata=* and -mcall-* options.
5095 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5097 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
5098 rate for post-reload scheduling.
5100 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5102 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
5104 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5106 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
5107 to improve monte carlo in scimark.
5109 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5111 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
5112 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
5113 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
5114 pentium4_cost, nocona_cost): Set reassociation width to 1.
5115 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
5116 width to 2 for fp operations and 1 otherwise.
5117 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
5119 (atom_cost): Set reassociation width to 2.
5120 (slm_cost, generic_cost): Set fp reassociation width
5121 to 2 and 1 otherwise.
5122 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
5123 (core_cost): Set fp reassociation width to 4 and vector to 2.
5124 (ix86_reassociation_width): Rewrite using cost table; special case
5125 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
5126 and TARGET_AVX128_OPTIMAL.
5127 * config/i386/i386.h (processor_costs): Add
5128 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
5129 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
5130 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
5131 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
5132 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
5133 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
5135 2017-10-05 Nathan Sidwell <nathan@acm.org>
5137 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
5139 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5141 * config/arm/arm.c (arm_test_fpu_data): New.
5142 (arm_run_selftests): Call arm_test_fpu_data.
5144 2017-10-04 Nathan Sidwell <nathan@acm.org>
5146 * toplev.c (toplev::main): Remove excess parens on pretty_printer
5148 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
5150 2017-10-04 Sudakshina Das <sudi.das@arm.com>
5152 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
5153 check type for aarch64_simd_valid_immediate.
5154 (aarch64_output_simd_mov_immediate): Update prototype.
5155 (aarch64_simd_valid_immediate): Update prototype.
5156 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
5157 support for ORR-immediate.
5158 (and<mode>3): modified pattern to add support for BIC-immediate.
5159 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
5160 now checks for valid immediate for BIC and ORR based on new enum
5162 (aarch64_output_simd_mov_immediate): Function now used to output
5163 BIC/ORR imm as well based on new enum argument.
5164 * config/aarch64/constraints.md (Do): New vector immediate constraint.
5166 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
5167 (aarch64_reg_or_bic_imm): Likewise.
5169 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5171 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
5172 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
5174 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5178 PR rtl-optimization/82396
5179 * haifa-sched.c (autopref_multipass_init): Simplify
5181 (autopref_rank_data): Simplify sort order.
5182 * sched-int.h (autopref_multipass_data_): Remove
5183 multi_mem_insn_p, min_offset and max_offset.
5185 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5187 * doc/sourcebuild.texi: Document vect_peeling_profitable.
5189 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5191 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
5194 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5196 * doc/sourcebuild.texi: Document vect_long_mult.
5198 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
5200 PR tree-optimization/82413
5201 * fold-const.c (build_range_check): Use widest_int when comparing
5202 the maximum ETYPE value with HIGH.
5204 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5206 PR rtl-optimization/82396
5207 * haifa-sched.c (autopref_multipass_init): Simplify
5209 (autopref_rank_data): Simplify sort order.
5210 * sched-int.h (autopref_multipass_data_): Remove
5211 multi_mem_insn_p, min_offset and max_offset.
5213 2017-10-04 Jakub Jelinek <jakub@redhat.com>
5215 PR tree-optimization/82381
5216 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
5217 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
5220 PR tree-optimization/82374
5221 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
5222 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
5223 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
5224 current_function_decl to the new decl.
5226 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
5228 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
5229 helper macro for IEEE float128 hardware built-in functions.
5230 (SQRTF128_ODD): Add built-in functions with the round-to-odd
5232 (TRUNCF128_ODD): Likewise.
5233 (ADDF128_ODD): Likewise.
5234 (SUBF128_ODD): Likewise.
5235 (MULF128_ODD): Likewise.
5236 (DIVF128_ODD): Likewise.
5237 (FMAF128_ODD): Likewise.
5238 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
5239 UNSPEC_TRUNC_ROUND_TO_ODD.
5240 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
5241 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
5242 floating point round to odd instructions.
5243 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
5244 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
5245 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
5246 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
5247 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
5248 (trunc<mode>sf2_hw): Change the truncate with round to odd
5249 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
5250 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
5251 to odd hardware instructions.
5252 (sub<mode>3_odd): Likewise.
5253 (mul<mode>3_odd): Likewise.
5254 (div<mode>3_odd): Likewise.
5255 (sqrt<mode>2_odd): Likewise.
5256 (fma<mode>4_odd): Likewise.
5257 (fms<mode>4_odd): Likewise.
5258 (nfma<mode>4_odd): Likewise.
5259 (nfms<mode>4_odd): Likewise.
5260 (trunc<mode>df2_odd): Change the truncate with round to odd
5261 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
5263 * doc/extend.texi (PowerPC built-in functions): Update documentation
5264 for existing IEEE float128-bit built-in functions. Add built-in
5265 functions that generate the IEEE 128-bit floating point round to
5268 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
5270 PR rtl-optimization/77729
5271 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
5272 to (X&(C1&~C2))|C2 transformations.
5274 2017-10-03 Martin Jambor <mjambor@suse.cz>
5276 PR tree-optimization/82363
5277 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
5278 mismatch, mark lacc written regardless of racc.
5280 2017-10-03 Jakub Jelinek <jakub@redhat.com>
5282 PR tree-optimization/82381
5283 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
5284 stmt_to_insert nor wheather SSA_NAMEs are default defs.
5285 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
5286 fallthrough into reassoc_stmt_dominates_stmt_p.
5289 * combine.c (combine_instructions): Don't combine in unreachable
5292 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
5295 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
5296 function to not use the have_cpu variable. Do not set cpu_index,
5297 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
5299 (rs6000_valid_attribute_p): Remove duplicate initializations of
5300 old_optimize and func_optimize.
5301 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
5302 (rs6000_activate_target_options): Make global.
5303 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
5306 2017-10-02 Jakub Jelinek <jakub@redhat.com>
5308 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
5309 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
5310 Set *poffset to 0 and *psize and *pmax_size to -1 if
5311 *poffset + *psize overflows in HOST_WIDE_INT.
5313 PR tree-optimization/82387
5314 PR tree-optimization/82388
5315 PR tree-optimization/82389
5316 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
5317 instead of live_bytes non-NULL.
5319 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
5322 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
5325 2017-10-02 Richard Biener <rguenther@suse.de>
5327 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
5328 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
5329 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
5331 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
5333 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
5334 requested precision matches the type's.
5335 * calls.c (alloc_max_size): Calculate the new candidate size as
5336 a widest_int and use wi::to_widest when comparing it with the
5337 current candidate size.
5338 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
5339 zero rather than integer_zero_node.
5340 * match.pd: Check for a no-op conversion before using wi::add
5341 rather than after. Use tree_to_uhwi when summing small shift
5342 counts into an unsigned int.
5344 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
5345 Alan Hayward <alan.hayward@arm.com>
5346 David Sherwood <david.sherwood@arm.com>
5349 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
5350 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
5351 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
5352 POINTER_AND_FP_REGS.
5354 2017-10-02 Richard Biener <rguenther@suse.de>
5356 PR tree-optimization/82355
5357 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5358 a mapping for the enclosing loop but avoid generating one for
5360 (copy_bb_and_scalar_dependences): Remove premature codegen
5361 error on PHIs in blocks duplicated into multiple places.
5362 * graphite-scop-detection.c
5363 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5364 in the region use it as loop and nest to analyze the DR in.
5365 (try_generate_gimple_bb): Likewise.
5366 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5367 (add_loop_constraints): For blocks in a loop not in the region
5368 create a dimension with a single iteration.
5369 * sese.h (gbb_loop_at_index): Remove assert.
5371 2017-10-01 Kevin Buettner <kevinb@redhat.com>
5373 * omp-expand.c (adjust_context_scope): New function.
5374 (expand_parallel_call): Call adjust_context_scope.
5376 2017-10-01 Jeff Law <law@redhat.com>
5378 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
5379 dom_opt_dom_walker class with direct access to private members.
5380 Add comments. Call test_for_singularity.
5381 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
5382 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
5383 m_dummy_cond anymore.
5384 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
5386 (pass_dominator:execute): Build the dummy_cond here and pass it
5387 to the dom_opt_dom_walker ctor.
5388 (test_for_singularity): New function.
5390 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
5391 Maya Rashish <coypu@sdf.org>
5393 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
5394 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
5395 (alpha*-*-netbsd*) Use nbsd_tm_file.
5396 (arm*-*-netbsdelf*) Likewise.
5397 (i[34567]86-*-netbsdelf*) Likewise.
5398 (x86_64-*-netbsd*) Likewise.
5399 (mips*-*-netbsd*) Likewise.
5400 (powerpc-*-netbsd*) Likewise.
5401 (sh*-*-netbsd*) Likewise.
5402 (sparc-*-netbsdelf*) Likewise.
5403 (sparc64-*-netbsd*) Likewise.
5404 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
5406 (vax-*-netbsdelf*) Likewise.
5407 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
5408 (UINT_FAST8_TYPE) Likewise.
5409 (INT_FAST16_TYPE) Check CHAR_FAST16.
5410 (UINT_FAST16_TYPE) Likewise.
5412 2017-09-30 Jakub Jelinek <jakub@redhat.com>
5415 * config/i386/i386.md
5416 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
5417 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
5418 *divmodsi4_zext_2): New define_insn_and_split.
5419 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
5420 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
5421 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
5422 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
5423 New define_insn_and_split.
5424 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
5425 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
5426 operands[1] having DImode when mode is SImode.
5428 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
5429 always SImode for DIV and MOD in REG_EQUAL notes.
5431 2017-09-29 Yury Gribov <tetra2005@gmail.com>
5434 * match.pd: Fix handling of NaNs in pattern.
5436 2017-09-29 Jeff Law <law@redhat.com>
5438 * sbitmap.c (bitmap_bit_in_range_p): New function.
5439 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
5440 * tree-ssa-dse.c (live_bytes_read): New function.
5441 (dse_classify_store): Ignore reads of dead bytes.
5443 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
5444 typos and whitespace errors.
5445 * config/i386/predicates.md (address_no_seg_operand): Likewise.
5446 * config/s390/s390.c (s390_emit_prologue): Likewise.
5448 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
5451 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
5452 with a symbol for LRA.
5454 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
5456 PR rtl-optimization/82338
5457 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
5459 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
5461 * genmodes.c (calc_wider_mode): Suppress qsort macro.
5462 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
5463 (qsort_chk): Declare.
5464 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
5465 (qsort_chk): New function.
5467 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5469 PR tree-optimization/82337
5470 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
5471 phi definition if the PHI result appears in an abnormal PHI.
5472 (find_basis_for_base_expr): Don't record a basis if the LHS of the
5473 basis appears in an abnormal PHI.
5475 2017-09-29 Richard Biener <rguenther@suse.de>
5477 * graphite-isl-ast-to-gimple.c
5478 (translate_isl_ast_to_gimple::set_codegen_error): New function.
5479 (binary_op_to_tree): Use it.
5480 (get_rename_from_scev): Likewise.
5481 (copy_loop_phi_nodes): Likewise.
5482 (copy_bb_and_scalar_dependences): Likewise.
5483 (translate_pending_phi_nodes): Likewise.
5485 2017-09-29 Jakub Jelinek <jakub@redhat.com>
5488 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
5489 for movabsq $(i32 << shift), r64.
5491 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
5493 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
5494 index when encoding %esp as %rsp to avoid 0x67 prefix.
5496 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
5498 * config/i386/i386.md (*movsf_internal, *movdf_internal):
5499 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
5501 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
5503 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
5504 Extensions with more than 16 double VFP registers.
5505 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
5506 to_clear_mask and all code related to it. Replace the remaining
5507 entry by a sbitmap and adapt code accordingly.
5509 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
5511 * brig-builtins.def: Change pure attributes to const.
5513 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5515 * config.gcc (default_gnu_indirect_function): Default to yes for
5516 sparc*-*-linux* with glibc.
5518 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5520 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
5521 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
5522 when creating .init_array and .fini_array sections with priority
5525 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
5528 * config/aarch64/aarch64.c
5529 (aarch64_builtin_support_vector_misalignment): Always return false
5530 when misalignment is unknown.
5532 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
5534 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
5535 this function to return false if the definition used by the swap
5536 instruction is artificial, or if the memory address from which the
5537 constant value is loaded is not represented by a base address held
5538 in a register or if the base address register is a frame or stack
5539 pointer. Additionally, return false if the base address of the
5540 loaded constant is a SYMBOL_REF but is not considered to be a
5542 (replace_swapped_load_constant): New function.
5543 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
5544 loaded constant vector with a load of a swapped constant vector.
5546 2017-09-27 Carl Love <cel@us.ibm.com>
5548 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
5549 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
5550 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
5553 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
5555 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
5556 first, always call autopref_rank_data otherwise.
5558 2017-09-27 Richard Biener <rguenther@suse.de>
5560 * graphite-scop-detection.c (find_scop_parameters): Move
5561 loop bound handling ...
5562 (gather_bbs::before_dom_children): ... here, avoiding the need
5563 to build scop_info->loop_nest.
5564 (record_loop_in_sese): Remove.
5565 * sese.h (sese_info_t::loop_nest): Remove.
5566 * sese.c (new_sese_info): Do not allocate loop_nest.
5567 (free_sese_info): Do not free loop_nest.
5569 2017-09-27 Jakub Jelinek <jakub@redhat.com>
5572 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
5573 lhs from calls if the lhs has addressable type.
5575 2017-09-27 Richard Biener <rguenther@suse.de>
5577 * graphite.h (scop::max_alias_set): New member.
5578 * graphite-scop-detection.c: Remove references to non-existing
5579 --param in comments.
5580 (build_alias_sets): Record the maximum alias set used for drs.
5581 (build_scops): Support zero as unlimited for
5582 --param graphite-max-arrays-per-scop.
5583 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
5585 (build_poly_sr_1): ... here. Compute alias set based on the
5586 maximum alias set used for drs rather than
5587 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
5589 2017-09-27 Richard Biener <rguenther@suse.de>
5591 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
5592 --param loop-block-tile-size=0 to disable tiling.
5594 2017-09-27 Richard Biener <rguenther@suse.de>
5596 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
5597 (graphite-max-nb-scop-params): Document special value zero.
5598 * domwalk.h (dom_walker::STOP): New symbolical constant.
5599 (dom_walker::dom_walker): Add optional parameter for bb to
5601 (dom_walker::~dom_walker): Declare.
5602 (dom_walker::before_dom_children): Document STOP return value.
5603 (dom_walker::m_user_bb_to_rpo): New member.
5604 (dom_walker::m_bb_to_rpo): Likewise.
5605 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
5606 mapping here if not provided by the user.
5607 (dom_walker::~dom_walker): Free bb to RPO mapping if not
5608 provided by the user.
5609 (dom_walker::STOP): Define.
5610 (dom_walker::walk): Do not compute bb to RPO mapping here.
5611 Support STOP return value from before_dom_children to stop
5613 * graphite-optimize-isl.c (optimize_isl): If the schedule
5614 is the same still generate code if -fgraphite-identity
5615 or -floop-parallelize-all are given.
5616 * graphite-scop-detection.c: Include cfganal.h.
5617 (gather_bbs::gather_bbs): Get and pass through bb to RPO
5619 (gather_bbs::before_dom_children): Return STOP for BBs
5621 (build_scops): Compute bb to RPO mapping and pass it to
5622 the domwalk. Treat --param graphite-max-nb-scop-params=0
5623 as not limiting the number of params.
5624 * graphite.c (graphite_initialize): Remove limit on the
5625 number of basic-blocks in a function.
5626 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
5627 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
5628 default value of 10.
5630 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5632 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
5633 Adjust code to eliminate needing to do the shift right 32-bits
5634 operation after XSCVDPSPN.
5636 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5638 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
5639 ((X / Y) != 0 -> X >= Y): Likewise.
5641 2017-09-26 Carl Love <cel@us.ibm.com>
5643 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
5644 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
5645 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
5646 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
5647 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
5648 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
5649 definitions and overloading.
5650 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
5651 statement for P9V_BUILTIN_XST_LEN_R.
5652 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
5653 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
5654 define_expand and define_insn for the instructions and builtins.
5655 * doc/extend.texi: Update the built-in documentation file for the new
5657 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
5658 define_insn for the instructions
5660 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
5663 * gcc/config/netbsd-protos.h: New file.
5664 * gcc/config/netbsd.c: New file.
5665 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
5666 * gcc/config/t-netbsd: New file.
5667 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
5668 (tmake_file) Add t-netbsd.
5669 (extra_objs) Add netbsd.o.
5671 2017-09-26 Janus Weil <janus@gcc.gnu.org>
5675 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
5677 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5679 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
5680 sign extension from a vector register to a GPR by doing a 32-bit
5681 direct move and then an EXTSW.
5682 (extendsi<mode>2 splitter): Likewise.
5683 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
5684 right or vector extract after doing XSCVDPSPN. Use
5685 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
5687 (movdi_from_sf_zero_ext): Likewise.
5688 (reload_gpr_from_vsxsf): Likewise.
5689 (p8_mfvsrd_4_disf): Delete, no longer used.
5690 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
5691 then needing to move the SFmode to a GPR to use the XSCVDPSP
5692 instruction instead of FRSP and XSCVDPSPN.
5693 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
5694 it is adjacent to the other XSCVSPDP insns.
5695 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
5696 SFmode to be in traditional Altivec registers.
5697 (vsx_xscvdpspn): Eliminate useless alternative constraint.
5698 (vsx_xscvspdpn): Likewise.
5699 (vsx_xscvspdpn_scalar): Likewise.
5701 2017-09-26 Martin Jambor <mjambor@suse.cz>
5703 * tree-sra.c (compare_access_positions): Put integral types first,
5704 stabilize sorting of integral types, remove conditions putting
5705 non-full-precision integers last.
5706 (sort_and_splice_var_accesses): Disable scalarization if a
5707 non-integert would be represented by a non-full-precision integer.
5709 2017-09-26 Joseph Myers <joseph@codesourcery.com>
5711 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
5712 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
5713 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
5714 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
5715 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
5716 conditionals inside the function instead of around it. Call
5717 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
5718 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
5720 2017-09-26 Richard Biener <rguenther@suse.de>
5722 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
5724 (scop_detection::build_scop_breadth): ... this. Removed.
5725 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
5726 (scop_detection::harmful_stmt_in_bb): Likewise.
5727 (scop_detection::graphite_can_represent_stmt): Likewise.
5728 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
5729 (scop_detection::can_represent_loop): Remove recursion, fold in ...
5730 (scop_detection::can_represent_loop_1): ... this. Removed.
5731 (scop_detection::harmful_loop_in_region): Simplify after inlining
5732 the above and remove more quadraticness.
5733 (build_scops): Adjust.
5734 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
5737 2017-09-26 Jakub Jelinek <jakub@redhat.com>
5740 * config/i386/i386.c (ix86_print_operand_address_as): Only test
5741 REGNO (base) == SP_REG if base is a REG.
5744 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
5745 if it is different SSA_NAME.
5746 (optimize_range_tests_cmp_bitwise): New function.
5747 (optimize_range_tests): Call it.
5749 2017-09-26 Richard Biener <rguenther@suse.de>
5751 PR tree-optimization/82321
5752 * graphite.c (canonicalize_loop_closed_ssa): Properly check
5753 for the def being inside the loop.
5755 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5757 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
5759 * config/s390/s390-builtins.def: Fix constraint on op4.
5761 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5763 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
5764 independent expanders.
5765 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
5766 ("vec_ordered", "vec_unordered"): New expanders.
5768 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5770 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
5773 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5775 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
5776 vec_unpacks_lo_v16qi.
5777 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
5779 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5781 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
5782 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
5783 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
5785 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5787 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
5789 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
5791 ("*vec_slb<mode>"): New insn pattern.
5792 ("vec_shr_<mode>"): New expander.
5793 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
5794 and force the shift count operand to V16QImode.
5795 ("vec_srb<mode>"): Set shift count mode to V16QI.
5797 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5799 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
5800 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
5801 ("vec_widen_smult_hi_<mode>"): New expander definitions.
5803 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
5806 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
5808 2017-09-26 Richard Biener <rguenther@suse.de>
5810 PR tree-optimization/82320
5811 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
5814 2017-09-25 Jeff Law <law@redhat.com>
5816 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
5817 prototype for new argument.
5818 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
5819 mostly extracted from rs6000_emit_allocate_stack.
5820 (rs6000_emit_probe_stack_range_stack_clash): New function.
5821 (rs6000_emit_allocate_stack): Call
5822 rs6000_emit_probe_stack_range_stack_clash as needed.
5823 (rs6000_emit_probe_stack_range): Add additional argument
5824 to call to gen_probe_stack_range{si,di}.
5825 (output_probe_stack_range): New.
5826 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
5827 (output_probe_stack_range_stack_clash): New.
5828 (rs6000_emit_prologue): Emit notes into dump file as requested.
5829 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
5830 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
5831 Add additional operand and pass it to output_probe_stack_range.
5833 2017-09-25 Bin Cheng <bin.cheng@arm.com>
5835 PR tree-optimization/82163
5836 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
5837 (checking_verify_loop_closed_ssa): New parameter.
5838 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
5839 (check_loop_closed_ssa_stmt): Delete.
5840 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
5841 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
5842 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
5845 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
5847 * brig-builtins.def: Treat HSAIL barrier builtins as
5848 setjmp/longjump style functions.
5850 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
5852 * target.def (constant_alignment): New hook.
5853 * defaults.h (CONSTANT_ALIGNMENT): Delete.
5854 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
5855 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
5856 * doc/tm.texi: Regenerate.
5857 * targhooks.h (default_constant_alignment): Declare.
5858 (constant_alignment_word_strings): Likewise.
5859 * targhooks.c (default_constant_alignment): New function.
5860 (constant_alignment_word_strings): Likewise.
5861 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
5862 instead of CONSTANT_ALIGNMENT.
5863 * varasm.c (align_variable, get_variable_align, build_constant_desc)
5864 (force_const_mem): Likewise.
5865 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
5866 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
5867 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
5868 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5869 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
5871 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
5872 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5873 constant_alignment_word_strings.
5874 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
5875 (CONSTANT_ALIGNMENT): Likewise.
5876 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5877 (arm_constant_alignment): New function.
5878 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
5879 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5880 constant_alignment_word_strings.
5881 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
5882 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5883 constant_alignment_word_strings.
5884 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
5885 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5886 (cris_constant_alignment): New function.
5887 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
5888 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5889 (epiphany_constant_alignment): New function.
5890 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
5891 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5892 constant_alignment_word_strings.
5893 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
5894 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5895 constant_alignment_word_strings.
5896 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
5897 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5898 constant_alignment_word_strings.
5899 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
5900 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
5901 * config/i386/i386.c (ix86_constant_alignment): Make static.
5902 Use the same interface as the target hook.
5903 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5904 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
5905 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5906 constant_alignment_word_strings.
5907 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
5908 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
5909 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5910 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
5911 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5912 constant_alignment_word_strings.
5913 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
5914 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5915 constant_alignment_word_strings.
5916 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
5917 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5918 constant_alignment_word_strings.
5919 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
5920 * config/microblaze/microblaze.c (microblaze_constant_alignment):
5922 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5923 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
5924 * config/mips/mips.c (mips_constant_alignment): New function.
5925 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5926 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
5927 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
5928 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5929 (mmix_constant_alignment): Make static. Use the same interface
5931 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
5932 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5933 constant_alignment_word_strings.
5934 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
5935 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5936 constant_alignment_word_strings.
5937 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
5938 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5939 constant_alignment_word_strings.
5940 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
5941 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5942 (rs6000_constant_alignment): New function.
5943 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
5944 * config/riscv/riscv.c (riscv_constant_alignment): New function.
5945 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5946 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
5947 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5948 (rs6000_constant_alignment): New function.
5949 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
5950 * config/s390/s390.c (s390_constant_alignment): New function.
5951 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5952 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
5953 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5954 constant_alignment_word_strings.
5955 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
5956 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5957 (sparc_constant_alignment): New function.
5958 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
5959 * config/spu/spu.c (spu_constant_alignment): New function.
5960 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5961 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
5962 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5963 constant_alignment_word_strings.
5964 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
5965 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5966 constant_alignment_word_strings.
5967 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
5968 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5969 constant_alignment_word_strings.
5970 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
5971 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5972 (visium_constant_alignment): New function.
5973 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
5974 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5975 (xtensa_constant_alignment): New function.
5976 * system.h (CONSTANT_ALIGNMENT): Poison.
5978 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
5980 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
5981 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
5982 (rs6000_builtin_valid_without_lhs): New helper function.
5983 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5984 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
5986 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
5988 * target.h (vec_perm_indices): Use unsigned short rather than
5990 (auto_vec_perm_indices): Likewise.
5991 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
5992 Use unsigned int rather than unsigned char.
5993 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
5995 2017-09-25 Richard Biener <rguenther@suse.de>
5997 * cfgloop.h (sort_sibling_loops): Declare.
5998 * cfgloop.c (sort_sibling_loops_cmp): New helper.
5999 (sort_sibling_loops): New function sorting the sibling loop list
6001 * graphite.c (graphite_transform_loops): Sort sibling loops.
6003 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
6005 * target.def (vec_perm_const_ok): Change sel parameter to
6007 * optabs-query.c (can_vec_perm_p): Update accordingly.
6008 * doc/tm.texi: Regenerate.
6009 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
6010 auto_vec_perm_indices and remove separate nelt field.
6011 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
6012 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
6013 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
6014 (aarch64_expand_vec_perm_const): Update accordingly.
6015 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
6016 to vec_perm_indices.
6017 * config/arm/arm.c (expand_vec_perm_d): Change perm to
6018 auto_vec_perm_indices and remove separate nelt field.
6019 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
6020 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
6021 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
6023 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
6024 to vec_perm_indices.
6025 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
6026 sel to vec_perm_indices.
6027 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
6028 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
6029 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
6031 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
6034 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
6037 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
6038 on the FUNCTION_DECL function context if it has a DIE that is a
6041 2017-09-25 Richard Biener <rguenther@suse.de>
6043 PR tree-optimization/82285
6044 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
6047 2017-09-25 Tom de Vries <tom@codesourcery.com>
6051 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
6054 2017-09-25 Richard Biener <rguenther@suse.de>
6056 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
6057 ISL errors other than isl_error_quota happen. Dump if the
6058 schedule is the same.
6059 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
6060 errors instead of aborting inside ISL.
6062 2017-09-25 Iain Sandoe <iain@codesourcery.com>
6065 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
6066 of libgcc_eh for m64.
6067 * config/i386/darwin64.h: Likewise.
6069 2017-09-25 Richard Biener <rguenther@suse.de>
6072 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
6073 attribute for incomplete types nor twice for complete ones.
6075 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
6078 * config/i386/i386.c (ix86_print_operand_address_as): Encode
6079 %esp as %rsp to avoid 0x67 prefix if there is no index or base
6082 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
6085 * config/i386/i386.opt (mprefer-avx256): Use
6086 ix86_target_flags variable.
6087 * config/i386/i386.c (ix86_target_string): Move
6088 -mprefer-avx256 to flag2_opts.
6090 2017-09-22 Jakub Jelinek <jakub@redhat.com>
6093 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
6094 and x != -1 | y != -1 into (x & y) != -1.
6096 2017-09-22 Steve Ellcey <sellcey@cavium.com>
6098 * config.gcc: Add new case statement to set
6099 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
6100 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
6101 s390x-*-linux* case statements. Added aarch64 to the list of
6102 supported architectures.
6104 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6106 PR tree-optimization/82289
6107 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
6108 STMT_VINFO_RELEVANT_P.
6110 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6111 Alan Hayward <alan.hayward@arm.com>
6112 David Sherwood <david.sherwood@arm.com>
6114 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
6115 for VR_RANGE only; don't allow VR_ANTI_RANGE.
6116 (extract_range_from_binary_expr_1): Don't call
6117 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
6119 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6120 Alan Hayward <alan.hayward@arm.com>
6121 David Sherwood <david.sherwood@arm.com>
6123 * target.def (preferred_vector_alignment): New hook.
6124 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
6126 * doc/tm.texi: Regenerate.
6127 * targhooks.h (default_preferred_vector_alignment): Declare.
6128 * targhooks.c (default_preferred_vector_alignment): New function.
6129 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
6131 (DR_TARGET_ALIGNMENT): New macro.
6132 (aligned_access_p): Update commentary.
6133 (vect_known_alignment_in_bytes): New function.
6134 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
6136 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
6137 Calculate the misalignment based on the target alignment rather than
6139 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
6140 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
6141 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
6142 the target alignment, rather than masking the element misalignment
6143 with the number of elements in a vector. Also use the target
6144 alignment when calculating the maximum number of peels.
6145 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
6146 instead of TYPE_ALIGN_UNIT.
6147 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
6148 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
6149 (vect_create_addr_base_for_vector_ref): Update call accordingly.
6150 (vect_create_data_ref_ptr): Likewise.
6151 (vect_setup_realignment): Realign by ANDing with
6152 -DR_TARGET_MISALIGNMENT.
6153 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
6154 the number of peels based on DR_TARGET_ALIGNMENT.
6155 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
6156 with the guaranteed alignment boundary when deciding whether
6158 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
6159 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
6160 (ensure_base_align): Remove stmt_info parameter. Get the
6161 target base alignment from DR_TARGET_ALIGNMENT.
6162 (vectorizable_store): Update call accordingly. Interpret
6163 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
6165 (vectorizable_load): Likewise.
6167 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6168 Alan Hayward <alan.hayward@arm.com>
6169 David Sherwood <david.sherwood@arm.com>
6171 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
6172 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
6173 (vect_enhance_data_refs_alignment): Likewise.
6175 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6177 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
6178 error. Only quit immediately if parsing is complete.
6179 (BEGIN): Initialize fatal_err and parse_done.
6180 (begin fpu, end fpu): Check number of arguments.
6181 (begin arch, end arch): Likewise.
6182 (begin cpu, end cpu): Likewise.
6183 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
6184 (optalias): Likewise.
6186 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6188 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
6189 * config/arm/arm-isa.h: Delete. Move definitions to ...
6190 * arm-cpus.in: ... here. Use new feature and fgroup values.
6191 * config/arm/arm.c (arm_option_override): Use lower case for feature
6193 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
6194 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
6195 * config/arm/parsecpu.awk (END): Add new command 'isa'.
6197 (print_isa_bits_for): New function.
6198 (gen_isa): New function.
6199 (gen_comm_data): Use print_isa_bits_for.
6200 (define feature): New keyword.
6201 (define fgroup): New keyword.
6202 * config/arm/t-arm (TM_H): Remove.
6203 (GTM_H): Add arm-isa.h.
6204 (arm-isa.h): Add rule to generate file.
6205 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
6206 case for feature bit names.
6208 2017-09-22 Richard Biener <rguenther@suse.de>
6210 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
6212 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
6213 print of no dependency loops ...
6214 * graphite.c (graphite_transform_loops): ... here.
6215 (canonicalize_loop_closed_ssa_form): Work from inner to outer
6217 (same_close_phi_node, remove_duplicate_close_phi,
6218 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
6219 (canonicalize_loop_closed_ssa): ... here and simplify.
6220 * graphite-optimize-isl.c: Include tree-vectorizer.h.
6221 (optimize_isl): Use dump_printf_loc to tell when we stopped
6222 optimizing because of an ISL timeout.
6224 2017-09-22 Richard Biener <rguenther@suse.de>
6226 PR tree-optimization/82291
6227 * tree-if-conv.c (predicate_mem_writes): Make sure to
6228 remove writes in blocks predicated with false.
6230 2017-09-22 Richard Biener <rguenther@suse.de>
6232 * sese.c: Include cfganal.h.
6233 (if_region_set_false_region): Remove.
6234 (create_if_region_on_edge): Likewise.
6235 (move_sese_in_condition): Re-implement without destroying
6238 2017-09-22 Richard Biener <rguenther@suse.de>
6240 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
6241 Verify both BBs contain loop PHI nodes before dispatching to
6243 (graphite_regenerate_ast_isl): Do not recompute dominators,
6244 do not verify three times. Restructure for clarity.
6245 * graphite-scop-detection.c (same_close_phi_node,
6246 remove_duplicate_close_phi, make_close_phi_nodes_unique,
6247 defined_in_loop_p, canonicalize_loop_closed_ssa,
6248 canonicalize_loop_closed_ssa_form): Simplify, remove excess
6249 checking and SSA rewrite, move to ...
6250 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
6251 (graphite_initialize): Do not pass in ctx, do not reset the
6252 SCEV cache, compute only dominators.
6253 (graphite_transform_loops): Allocate ISL ctx after
6254 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
6255 Maintain post-dominators only around build_scops.
6256 * sese.c (if_region_set_false_region): Make static. Free
6257 and recompute dominators.
6258 (move_sese_in_condition): Assert we don't get called with
6259 post-dominators computed.
6260 * sese.h (if_region_set_false_region): Remove.
6262 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
6264 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
6265 mode attribute for TARGET_AVX512VL.
6267 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
6269 * config/i386/i386.opt (mprefer-avx256): New option.
6270 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
6272 (ix86_preferred_simd_mode): Return 256-bit AVX modes
6273 for TARGET_PREFER_AVX256.
6274 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
6276 2017-09-21 Jeff Law <law@redhat.com>
6278 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
6279 Fix dump output if the only stack space is for pushed registers.
6281 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6283 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
6286 2017-09-21 Martin Sebor <msebor@redhat.com>
6289 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
6290 code (in C++) or code that triggers warnings.
6292 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
6294 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
6296 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
6298 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
6299 * sched-rgn.c: Ditto.
6300 * sel-sched-ir.c: Ditto.
6302 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
6304 * toplev.h (set_random_seed): Adjust return type.
6305 * toplev.c (init_local_tick): Move eager initialization of random_seed
6306 to get_random_seed. Adjust comment.
6307 (init_random_seed): Inline to get_random_seed, delete.
6308 (get_random_seed): Initialize random_seed lazily.
6309 (set_random_seed): Do not return previous value.
6310 (print_switch_value): Do not call get_random_seed.
6312 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
6314 * cgraph.c (delete_function_version): New, broken out from...
6315 (cgraph_node::delete_function_version): ...here. Rename to
6316 cgraph_node::delete_function_version_by_decl. Update all uses.
6317 (cgraph_node::remove): Call delete_function_version.
6319 2017-09-21 Jakub Jelinek <jakub@redhat.com>
6322 * tree-inline.c (expand_call_inline): Emit clobber stmts for
6323 VAR_DECLs to which addressable non-volatile parameters are mapped
6324 and for id->retvar after the return value assignment. Clear
6325 id->retval and id->retbnd after inlining.
6327 2017-09-21 Richard Biener <rguenther@suse.de>
6329 PR tree-optimization/82276
6330 PR tree-optimization/82244
6331 * tree-vrp.c (build_assert_expr_for): Set
6332 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
6334 (remove_range_assertions): Revert earlier change.
6336 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
6339 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
6341 2017-09-21 Richard Biener <rguenther@suse.de>
6343 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
6344 Restore valid IL after code generation errors.
6345 * graphite.c (graphite_transform_loops): Diagnose code
6346 generation issues as MSG_MISSED_OPTIMIZATION and continue
6347 with processing SCOPs.
6349 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6350 Alan Hayward <alan.hayward@arm.com>
6351 David Sherwood <david.sherwood@arm.com>
6353 * calls.c (compute_argument_addresses): Use simplify_gen_binary
6354 rather than choosing between plus_constant and gen_rtx_<CODE>.
6355 * expr.c (emit_push_insn): Likewise.
6356 (expand_expr_real_2): Likewise.
6358 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6359 Alan Hayward <alan.hayward@arm.com>
6360 David Sherwood <david.sherwood@arm.com>
6362 * loop-unroll.c (split_iv): Call copy_rtx on the step.
6364 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6365 Alan Hayward <alan.hayward@arm.com>
6366 David Sherwood <david.sherwood@arm.com>
6368 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
6369 calling tree_to_uhwi.
6371 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6372 Alan Hayward <alan.hayward@arm.com>
6373 David Sherwood <david.sherwood@arm.com>
6375 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
6376 INTEGER_CST rather than a negative test for ADDR_EXPR.
6378 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6379 Alan Hayward <alan.hayward@arm.com>
6380 David Sherwood <david.sherwood@arm.com>
6382 * tree-vrp.c (extract_range_from_binary_expr_1): Check
6383 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
6385 2017-09-21 Richard Biener <rguenther@suse.de>
6387 PR tree-optimization/71351
6388 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
6389 graphite_create_new_loop_guard): Remove, fold remaining parts
6391 (translate_isl_ast_node_for): ... here and simplify.
6393 2017-09-21 Jakub Jelinek <jakub@redhat.com>
6396 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
6397 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
6398 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
6399 alternative always use QI mode, for -Os imov (=R,R) alternative
6400 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
6403 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6404 Jeff Law <law@redhat.com>
6406 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
6407 (allocate_stack_space): New function, partially extracted from
6409 (s390_emit_prologue): Track offset to most recent stack probe.
6410 Code to allocate space moved into allocate_stack_space.
6411 Dump actions when no stack is allocated.
6412 (s390_prologue_plus_offset): New function.
6413 (s390_emit_stack_probe): Likewise.
6415 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
6417 * common.opt (Wa, Wl, Wp, g, gz=): Add
6419 (gno-column-info): Remove.
6420 (gcolumn-info): Drop RejectNegative.
6422 (gno-record-gcc-switches): Remove.
6423 (grecord-gcc-switches): Drop RejectNegative.
6424 (gno-split-dwarf): Remove.
6425 (gsplit-dwarf): Drop RejectNegative.
6426 (gno-strict-dwarf): Remove.
6427 (gstrict-dwarf): Drop RejectNegative.
6428 * config/darwin.opt (gfull, gused): Add RejectNegative.
6429 * dwarf2out.c (gen_producer_string): Drop
6430 gno-record-gcc-switches handler.
6431 * optc-gen.awk: Add g to prefixes with negative forms.
6432 * opts-common.c (remapping_prefix_p): New.
6433 (find_opt): Check it.
6434 (generate_canonical_option): Test g prefix.
6435 (option_map): Add -gno- mapping.
6436 (add_misspelling_candidates): Check remapping_prefix_p.
6438 2017-09-20 Jeff Law <law@redhat.com>
6440 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
6441 thinko in stack clash protection support.
6443 * explow.c (compute_stack_clash_protection_loop_data): Use
6444 CONST_INT_P instead of explicit test. Verify object is a
6445 CONST_INT_P before looking at INTVAL.
6446 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
6447 instead of explicit test.
6449 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
6452 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
6453 address instead of to r1 and r11.
6455 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
6457 * config.gcc: Support "knm".
6458 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
6459 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
6461 * config/i386/i386.c (m_KNM): Define.
6462 (processor_target_table): Add "knm".
6464 (ix86_option_override_internal): Add "knm".
6465 (ix86_issue_rate): Add PROCESSOR_KNM.
6466 (ix86_adjust_cost): Ditto.
6467 (ia32_multipass_dfa_lookahead): Ditto.
6468 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
6469 (fold_builtin_cpu): Add M_INTEL_KNM.
6470 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
6471 (processor_type): Add PROCESSOR_KNM.
6472 * config/i386/x86-tune.def: Add m_KNM.
6473 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
6475 2017-09-20 Richard Biener <rguenther@suse.de>
6477 PR tree-optimization/80213
6478 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
6479 are allowed in empty BBs as well.
6480 (canonicalize_loop_closed_ssa): Also look for other complex
6482 (scop_detection::get_sese): Include the loop-closed PHI block
6484 (scop_detection::merge_sese): Remove code adding extra blocks.
6485 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
6486 (build_scops): Assert the final returned scop is invalid.
6488 2017-09-20 Richard Biener <rguenther@suse.de>
6490 PR tree-optimization/82264
6491 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
6493 (vn_phi_lookup): Likewise.
6494 (vn_phi_insert): Likewise.
6496 2017-09-20 Jakub Jelinek <jakub@redhat.com>
6498 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
6499 that fits into uhwi or shwi, add DW_AT_const_value regardless
6500 of early_dwarf without going through RTL, using add_AT_unsigned
6503 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
6504 (DEBUG_LTO_ABBREV_SECTION): Likewise.
6505 (DEBUG_LTO_MACINFO_SECTION): Likewise.
6506 (DEBUG_MACRO_SECTION): Likewise.
6507 (DEBUG_LTO_MACRO_SECTION): Likewise.
6508 (DEBUG_STR_DWO_SECTION): Likewise.
6509 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
6510 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
6511 (DEBUG_LTO_DWO_LINE_SECTION): Define.
6512 (DEBUG_LTO_LINE_STR_SECTION): Define.
6513 (init_sections_and_labels): Initialize debug_line_str_section
6514 variable. Initialize debug_loc_section for -gdwarf-5 to
6515 DEBUG_LOCLISTS_SECTION. Formatting fixes.
6517 2017-09-20 Richard Biener <rguenther@suse.de>
6519 * graphite-sese-to-poly.c (extract_affine): Properly handle
6520 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
6522 2017-09-20 Richard Biener <rguenther@suse.de>
6524 PR tree-optimization/81373
6525 * graphite-scop-detection.c (build_cross_bb_scalars_def):
6526 Force SESE live-out defs to be handled even if they are
6529 2017-09-19 Jeff Law <law@redhat.com>
6531 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
6532 nothing for stack adjustments with REG_STACK_CHECK.
6533 * sched-deps.c (parse_add_or_inc): Reject insns with
6534 REG_STACK_CHECK from dependency breaking.
6535 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
6536 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
6537 * reg-notes.def (STACK_CHECK): New note.
6539 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
6540 (ix86_expand_prologue): Dump stack clash info as needed.
6541 Call ix86_adjust_stack_and_probe_stack_clash as needed.
6543 * function.c (dump_stack_clash_frame_info): New function.
6544 * function.h (dump_stack_clash_frame_info): Prototype.
6545 (enum stack_clash_probes): New enum.
6547 * config/alpha/alpha.c (alpha_expand_prologue): Also check
6548 flag_stack_clash_protection.
6549 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
6550 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
6551 (arm_frame_pointer_required): Likewise.
6552 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
6553 (ia64_expand_prologue): Likewise.
6554 * config/mips/mips.c (mips_expand_prologue): Likewise.
6555 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
6556 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6557 (sparc_flat_expand_prologue): Likewise.
6558 * config/spu/spu.c (spu_expand_prologue): Likewise.
6560 * explow.c: Include "params.h".
6561 (anti_adjust_stack_and_probe_stack_clash): New function.
6562 (get_stack_check_protect): Likewise.
6563 (compute_stack_clash_protection_loop_data): Likewise.
6564 (emit_stack_clash_protection_loop_start): Likewise.
6565 (emit_stack_clash_protection_loop_end): Likewise.
6566 (allocate_dynamic_stack_space): Use get_stack_check_protect.
6567 Use anti_adjust_stack_and_probe_stack_clash.
6568 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
6569 (emit_stack_clash_protection_loop_start): Likewise.
6570 (emit_stack_clash_protection_loop_end): Likewise.
6571 * rtl.h (get_stack_check_protect): Prototype.
6572 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
6573 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
6574 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
6576 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
6578 * doc/tm.texi: Rebuilt.
6579 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
6580 get_stack_check_protect.
6581 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
6582 * config/arm/arm.c (arm_expand_prologue): Likewise.
6583 (arm_frame_pointer_required): Likewise.
6584 * config/i386/i386.c (ix86_expand_prologue): Likewise.
6585 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
6586 * config/mips/mips.c (mips_expand_prologue): Likewise.
6587 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
6588 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
6589 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6590 (sparc_flat_expand_prologue): Likewise.
6592 * common.opt (-fstack-clash-protection): New option.
6593 * flag-types.h (enum stack_check_type): Note difference between
6594 -fstack-check= and -fstack-clash-protection.
6595 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
6596 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
6597 * toplev.c (process_options): Issue warnings/errors for cases
6598 not handled with -fstack-clash-protection.
6599 * doc/invoke.texi (-fstack-clash-protection): Document new option.
6600 (-fstack-check): Note additional problem with -fstack-check=generic.
6601 Note that -fstack-check is primarily for Ada and refer users
6602 to -fstack-clash-protection for stack-clash-protection.
6603 Document new params for stack clash protection.
6605 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
6607 * config/i386/i386.c (ix86_split_long_move): Do not handle
6608 address used for LEA in a special way.
6610 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
6612 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
6614 2017-09-19 Martin Sebor <msebor@redhat.com>
6617 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
6618 of incompatible types.
6620 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
6622 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6623 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
6624 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6625 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
6627 2017-09-19 Richard Biener <rguenther@suse.de>
6629 PR tree-optimization/82244
6630 * tree-vrp.c (remove_range_assertions): Do not propagate
6631 a constant to abnormals but replace the assert with a copy.
6633 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
6635 PR rtl-optimization/57878
6636 PR rtl-optimization/68988
6637 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
6638 avoidance test involving non_reload_pseudos. Move frequency test
6639 below the general fragmentation avoidance test.
6641 2017-09-19 Richard Biener <rguenther@suse.de>
6643 PR tree-optimization/69728
6644 * graphite-sese-to-poly.c (schedule_error): New global.
6645 (add_loop_schedule): Handle empty domain by failing the
6647 (build_original_schedule): Handle schedule_error.
6649 2017-09-19 Richard Biener <rguenther@suse.de>
6651 * graphite-scop-detection.c (scop_detection::can_represent_loop):
6652 Do not iterate to sibling loops but only to siblings of inner
6655 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
6658 * config/m68k/m68k.md (moveq feeding equality comparison): Check
6659 that the registers are different.
6661 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
6663 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
6664 to processor_model and "amdfam17h" to arch_names_table.
6665 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
6667 2017-09-18 Jakub Jelinek <jakub@redhat.com>
6670 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
6672 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6673 Alan Hayward <alan.hayward@arm.com>
6674 David Sherwood <david.sherwood@arm.com>
6676 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
6678 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
6680 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
6681 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
6682 vect_schedule_slp_instance.
6683 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
6684 Update call to vect_slp_analyze_node_operations. Simplify return
6686 (vect_slp_analyze_bb_1): Update call accordingly.
6687 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
6688 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
6689 (vect_schedule_slp): Update call accordingly.
6691 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6692 Alan Hayward <alan.hayward@arm.com>
6693 David Sherwood <david.sherwood@arm.com>
6695 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
6696 with types that aren't in fact scalar.
6698 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6700 * tree-vect-slp.c (vect_record_max_nunits): New function,
6702 (vect_build_slp_tree_1): ...here.
6703 (vect_build_slp_tree_2): Call it for phis too.
6705 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6707 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
6708 to vect_get_vec_def_for_operand when getting the mask operand.
6710 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6711 Alan Hayward <alan.hayward@arm.com>
6712 David Sherwood <david.sherwood@arm.com>
6714 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
6717 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6718 Alan Hayward <alan.hayward@arm.com>
6719 David Sherwood <david.sherwood@arm.com>
6721 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
6722 calculation for vector booleans.
6724 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6725 Alan Hayward <alan.hayward@arm.com>
6726 David Sherwood <david.sherwood@arm.com>
6728 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
6730 (vect_transform_stmt): ...here.
6731 (vect_analyze_stmt): Use it instead of calling
6732 vectorizable_live_operation directly.
6734 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
6736 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
6737 non-SIMT targets in acc vector loops.
6739 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
6741 * configure.ac: Add arc and check if assembler supports gdwarf2.
6742 * configure: Regenerate.
6744 2017-09-18 Richard Biener <rguenther@suse.de>
6746 PR tree-optimization/82220
6747 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
6748 epilogue niters from the min_profitable_iters compute.
6750 2017-09-18 Jakub Jelinek <jakub@redhat.com>
6753 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
6754 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
6755 (ix86_init_pic_reg): Revert 2017-09-01 changes.
6757 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
6760 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
6761 switching to a new text section.
6763 2017-09-18 Richard Biener <rguenther@suse.de>
6765 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
6767 (build_alias_set): Reject aliases with no access function.
6769 2017-09-18 Richard Biener <rguenther@suse.de>
6771 PR tree-optimization/79622
6772 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
6774 (build_cross_bb_scalars_use): Likewise.
6776 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
6778 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
6780 2017-09-18 Alan Modra <amodra@gmail.com>
6783 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
6784 stack_pointer_rtx for count 0. Update comments. Break up
6785 large rtl expression.
6787 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
6789 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
6790 Increase to 20 bytes.
6791 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
6792 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
6793 or avx version of the stub.
6795 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
6798 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
6799 compute the minimum stack alignment. Also update preferred stack
6800 boundary for leaf functions.
6802 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
6804 PR tree-optimization/82228
6805 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
6808 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
6810 * common/config/nds32/nds32-common.c
6811 (nds32_option_optimization_table): Refine formatting.
6812 (nds32_option_optimization_table): Use -fsched-pressure and
6813 -fomit-frame-pointer for specific optimization level.
6815 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
6817 * config/nds32/nds32.c: Refine formatting and comments.
6818 * config/nds32/nds32.h: Likewise.
6819 * config/nds32/nds32.md: Likewise.
6820 * config/nds32/nds32-cost.c: Likewise.
6821 * config/nds32/nds32-isr.c: Likewise.
6822 * config/nds32/nds32-md-auxiliary.c: Likewise.
6823 * config/nds32/nds32-multiple.md: Likewise.
6824 * config/nds32/nds32-predicates.c: Likewise.
6826 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
6827 Jakub Jelinek <jakub@redhat.com>
6829 Add support for -std=c++2a.
6830 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
6832 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
6834 2017-09-15 Steve Ellcey <sellcey@cavium.com>
6837 * doc/extend.texi (Common Function Attributes): Add
6838 references to ARM, AArch64, and S/390 specific attributes.
6839 (Function Specific Option Pragmas): Add AArch64 and S/390
6840 to list of back ends that support the target pragma.
6842 2017-09-15 Nathan Sidwell <nathan@acm.org>
6844 * doc/standards.texi: Fix C++17 description. Update URLs for
6847 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
6849 * common.opt (Wcast-align=strict): New warning option.
6850 * doc/invoke.texi: Document -Wcast-align=strict.
6852 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
6854 * cgraph.h (cgraph_thunk_info): Add comments.
6855 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
6856 assert for VIRTUAL_* arguments stricter.
6858 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
6860 PR tree-optimization/71026
6861 * match.pd: Move RDIV patterns from fold-const.c
6862 * fold-const.c (distribute_real_division): Removed.
6863 (fold_binary_loc): Remove calls to distribute_real_divison.
6865 2017-09-15 Jakub Jelinek <jakub@redhat.com>
6867 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
6868 c++1z and gnu++1z as deprecated. Change other references to
6869 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
6870 Change -Wc++1z-compat to -Wc++17-compat.
6871 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
6872 * dwarf2out.c (highest_c_language): Handle C++17.
6873 (gen_compile_unit_die): Likewise.
6875 2017-09-15 Jakub Jelinek <jakub@redhat.com>
6877 PR rtl-optimization/82192
6878 * combine.c (make_extraction): Don't look through non-paradoxical
6879 SUBREGs or TRUNCATE if pos + len is or might be bigger than
6882 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
6883 Alan Hayward <alan.hayward@arm.com>
6884 David Sherwood <david.sherwood@arm.com>
6886 * target.def (function_arg_offset): New hook.
6887 * targhooks.h (default_function_arg_offset): Declare.
6888 * targhooks.c (default_function_arg_offset): New function.
6889 * function.c (locate_and_pad_parm): Use
6890 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
6891 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
6892 (TARGET_FUNCTION_ARG_OFFSET): ...this.
6893 * doc/tm.texi: Regenerate.
6894 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
6895 * config/spu/spu.c (spu_function_arg_offset): New function.
6896 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
6897 * system.h (FUNCTION_ARG_OFFSET): Poison.
6899 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
6900 Alan Hayard <alan.hayward@arm.com>
6901 David Sherwood <david.sherwood@arm.com>
6903 * target.def (truly_noop_truncation): New hook.
6904 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
6905 than TRULY_NOOP_TRUNCATION.
6906 * hooks.h (hook_bool_uint_uint_true): Declare.
6907 * hooks.c (hook_bool_uint_uint_true): New function.
6908 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
6909 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
6910 * doc/tm.texi: Regenerate.
6911 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
6912 rather than TRULY_NOOP_TRUNCATION in comments.
6913 (simplify_comparison): Likewise.
6914 (record_truncated_value): Likewise.
6915 * expmed.c (extract_bit_field_1): Likewise.
6916 (extract_split_bit_field): Likewise.
6917 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
6918 instead of TRULY_NOOP_TRUNCATION.
6919 * function.c (assign_parm_setup_block): Likewise.
6920 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
6921 * rtlhooks.c: Include target.h.
6922 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
6923 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
6924 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
6925 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
6926 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
6927 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
6928 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
6929 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
6930 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
6931 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
6932 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
6933 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
6934 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
6935 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
6936 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
6937 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
6938 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
6939 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
6940 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
6941 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
6942 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
6943 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
6944 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
6945 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
6946 * config/mips/mips.c (mips_truly_noop_truncation): New function.
6947 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
6948 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
6949 rather than TRULY_NOOP_TRUNCATION in comments.
6950 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
6951 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
6952 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
6953 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
6954 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
6955 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
6956 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
6957 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
6958 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
6959 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
6960 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
6961 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
6962 rather than TRULY_NOOP_TRUNCATION in comments.
6963 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
6964 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
6965 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
6966 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
6967 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
6968 TRULY_NOOP_TRUNCATION condition.
6969 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
6970 (TRULY_NOOP_TRUNCATION): Delete.
6971 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
6972 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
6973 * config/spu/spu.c (spu_truly_noop_truncation): New function.
6974 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
6975 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
6976 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
6977 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
6978 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
6979 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
6980 rather than TRULY_NOOP_TRUNCATION in comments.
6981 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
6982 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
6983 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
6984 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
6985 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
6986 * system.h (TRULY_NOOP_TRUNCATION): Poison.
6988 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
6991 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
6992 (*cmp_ior): Likewise.
6993 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
6994 (*ior_scc_scc_cmp): Likewise.
6995 (*and_scc_scc): Likewise.
6996 (*and_scc_scc_cmp): Likewise.
6998 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
6999 Alan Hayard <alan.hayward@arm.com>
7000 David Sherwood <david.sherwood@arm.com>
7002 * target.def (can_change_mode_class): New hook.
7003 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
7004 (hard_regno_nregs): Likewise.
7005 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
7006 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
7007 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
7008 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
7009 (LOAD_EXTEND_OP): Update accordingly.
7010 * doc/tm.texi: Regenerate.
7011 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
7012 CANNOT_CHANGE_MODE_CLASS.
7013 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
7014 (REG_CAN_CHANGE_MODE_P): ...this new macro.
7015 * combine.c (simplify_set): Update accordingly.
7016 * emit-rtl.c (validate_subreg): Likewise.
7017 * recog.c (general_operand): Likewise.
7018 * regcprop.c (mode_change_ok): Likewise.
7019 * reload1.c (choose_reload_regs): Likewise.
7020 (inherit_piecemeal_p): Likewise.
7021 * rtlanal.c (simplify_subreg_regno): Likewise.
7022 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
7023 instead of CANNOT_CHANGE_MODE_CLASS.
7024 (reload_cse_simplify_operands): Likewise.
7025 * reload.c (push_reload): Use targetm.can_change_mode_class
7026 instead of CANNOT_CHANGE_MODE_CLASS.
7027 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
7028 REG_CANNOT_CHANGE_MODE_P.
7029 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7030 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
7031 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7032 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7033 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7034 (arm_can_change_mode_class): New function.
7035 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7036 than CANNOT_CHANGE_MODE_CLASS in comments.
7037 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7038 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
7039 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
7040 (ix86_can_change_mode_class): ...this new function, inverting the
7041 sense of the return value.
7042 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7043 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7044 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7045 (ia64_can_change_mode_class): New function.
7046 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7047 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
7048 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
7049 (m32c_can_change_mode_class): ...this new function, inverting the
7050 sense of the return value.
7051 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7052 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7053 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
7054 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
7055 (mips_can_change_mode_class): ...this new function, inverting the
7056 sense of the return value.
7057 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7058 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7059 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7060 (msp430_can_change_mode_class): New function.
7061 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7062 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
7063 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7064 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7065 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7066 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
7067 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7068 (pa_cannot_change_mode_class): Replace with...
7069 (pa_can_change_mode_class): ...this new function, inverting the
7070 sense of the return value.
7071 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7072 than CANNOT_CHANGE_MODE_CLASS in comments.
7073 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7074 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
7075 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7076 (pdp11_cannot_change_mode_class): Replace with...
7077 (pdp11_can_change_mode_class): ...this new function, inverting the
7078 sense of the return value.
7079 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7080 * config/powerpcspe/powerpcspe-protos.h
7081 (rs6000_cannot_change_mode_class_ptr): Delete.
7082 * config/powerpcspe/powerpcspe.c
7083 (rs6000_cannot_change_mode_class_ptr): Delete.
7084 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7085 (rs6000_option_override_internal): Assign to
7086 targetm.can_change_mode_class instead of
7087 rs6000_cannot_change_mode_class_ptr.
7088 (rs6000_cannot_change_mode_class): Replace with...
7089 (rs6000_can_change_mode_class): ...this new function, inverting the
7090 sense of the return value.
7091 (rs6000_debug_cannot_change_mode_class): Replace with...
7092 (rs6000_debug_can_change_mode_class): ...this new function.
7093 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7094 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
7095 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7096 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7097 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
7099 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
7100 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7101 (rs6000_option_override_internal): Assign to
7102 targetm.can_change_mode_class instead of
7103 rs6000_cannot_change_mode_class_ptr.
7104 (rs6000_cannot_change_mode_class): Replace with...
7105 (rs6000_can_change_mode_class): ...this new function, inverting the
7106 sense of the return value.
7107 (rs6000_debug_cannot_change_mode_class): Replace with...
7108 (rs6000_debug_can_change_mode_class): ...this new function.
7109 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7110 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
7111 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
7112 (s390_can_change_mode_class): ...this new function, inverting the
7113 sense of the return value.
7114 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7115 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7116 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
7117 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7118 (sh_cannot_change_mode_class): Replace with...
7119 (sh_can_change_mode_class): ...this new function, inverting the
7120 sense of the return value.
7121 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7122 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7123 (sparc_can_change_mode_class): New function.
7124 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7125 * config/spu/spu.c (spu_can_change_mode_class): New function.
7126 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7127 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7128 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7129 (visium_can_change_mode_class): New function.
7130 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
7132 2017-09-15 Richard Biener <rguenther@suse.de>
7134 PR tree-optimization/82217
7135 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
7136 but not undefined case.
7138 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7141 * postreload.c (reload_cse_simplify_operands): Skip
7142 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
7144 2017-09-15 Richard Biener <rguenther@suse.de>
7146 PR tree-optimization/68823
7147 * graphite-scop-detection.c (build_alias_set): If we have a
7148 possible dependence check whether we can handle them by just
7149 looking at the DRs DR_ACCESS_FNs.
7150 (build_scops): If build_alias_set fails, fail the SCOP.
7152 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
7154 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
7155 to support float128 built-in functions that require the ISA 3.0
7157 (BU_FLOAT128_3_HW): Likewise.
7158 (SQRTF128): Add support for the IEEE 128-bit square root and fma
7160 (FMAF128): Likewise.
7162 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
7163 support for built-in functions that need the ISA 3.0 IEEE 128-bit
7164 floating point instructions.
7165 (rs6000_invalid_builtin): Likewise.
7166 (rs6000_builtin_mask_names): Likewise.
7167 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
7168 (RS6000_BTM_FLOAT128_HW): Likewise.
7169 (RS6000_BTM_COMMON): Likewise.
7170 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
7172 * doc/extend.texi (RS/6000 built-in functions): Document the
7173 IEEE 128-bit floating point square root and fused multiply-add
7176 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
7178 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
7179 reg (r2) isn't in the set of registers defined in the prologue.
7181 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7182 Alan Hayward <alan.hayward@arm.com>
7183 David Sherwood <david.sherwood@arm.com>
7185 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
7186 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
7187 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
7188 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
7189 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
7191 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
7192 max_vectorization_factor.
7193 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
7195 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7196 Alan Hayward <alan.hayward@arm.com>
7197 David Sherwood <david.sherwood@arm.com>
7199 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
7200 (vect_worthwhile_without_simd_p): Declare.
7201 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
7202 (vectorizable_reduction): Use it.
7203 * tree-vect-stmts.c (vectorizable_shift): Likewise.
7204 (vectorizable_operation): Likewise.
7206 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7207 Alan Hayward <alan.hayward@arm.com>
7208 David Sherwood <david.sherwood@arm.com>
7210 * tree-vectorizer.h (vect_get_num_copies): New function.
7211 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
7212 * tree-vect-loop.c (vectorizable_reduction): Likewise.
7213 (vectorizable_induction): Likewise.
7214 (vectorizable_live_operation): Likewise.
7215 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7216 (vectorizable_bswap): Likewise.
7217 (vectorizable_call): Likewise.
7218 (vectorizable_conversion): Likewise.
7219 (vectorizable_assignment): Likewise.
7220 (vectorizable_shift): Likewise.
7221 (vectorizable_operation): Likewise.
7222 (vectorizable_store): Likewise.
7223 (vectorizable_load): Likewise.
7224 (vectorizable_condition): Likewise.
7225 (vectorizable_comparison): Likewise.
7226 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
7228 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7229 Alan Hayward <alan.hayward@arm.com>
7230 David Sherwood <david.sherwood@arm.com>
7232 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
7233 of vect_init_vector.
7235 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7236 Alan Hayward <alan.hayward@arm.com>
7237 David Sherwood <david.sherwood@arm.com>
7239 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
7240 an inline wrapper that provides a location.
7241 (gimple_build_vector): Likewise.
7242 * gimple-fold.c (gimple_build_vector_from_val): New function.
7243 (gimple_build_vector): Likewise.
7244 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
7245 functions to build the initial value. Always return a gimple value.
7246 (get_initial_defs_for_reduction): Likewise. Only compute
7248 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
7249 vect_init_vector on the results from get_initial_def(s)_for_reduction.
7250 (vectorizable_induction): Use gimple_build_vector rather than
7253 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7254 Alan Hayward <alan.hayward@arm.com>
7255 David Sherwood <david.sherwood@arm.com>
7257 * target.h (vec_perm_indices): New typedef.
7258 (auto_vec_perm_indices): Likewise.
7259 * optabs-query.h: Include target.h
7260 (can_vec_perm_p): Take a vec_perm_indices *.
7261 * optabs-query.c (can_vec_perm_p): Likewise.
7262 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
7263 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7264 * tree-vect-generic.c (lower_vec_perm): Likewise.
7265 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
7266 (vect_grouped_load_supported): Likewise.
7267 (vect_shift_permute_load_chain): Likewise.
7268 (vect_permute_store_chain): Use auto_vec_perm_indices.
7269 (vect_permute_load_chain): Likewise.
7270 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
7271 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
7272 Update uses of can_vec_perm_p.
7273 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
7274 mode with a number of elements. Take a vec_perm_indices *.
7275 (vect_create_epilog_for_reduction): Update accordingly.
7276 Use auto_vec_perm_indices.
7277 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
7278 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
7279 (vect_transform_slp_perm_load): Likewise.
7280 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
7281 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
7282 (vect_gen_perm_mask_checked): Likewise.
7283 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
7284 (vect_gen_perm_mask_checked): Likewise.
7285 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
7286 (vectorizable_store): Likewise.
7287 (vectorizable_load): Likewise.
7288 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
7289 (vectorizable_bswap): Likewise.
7291 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7292 Alan Hayward <alan.hayward@arm.com>
7293 David Sherwood <david.sherwood@arm.com>
7295 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
7296 * tree.c (build_vector): Likewise.
7297 (build_vector_from_ctor): Update accordingly.
7298 (build_vector_from_val): Likewise.
7299 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
7300 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7301 * tree-vect-generic.c (add_rshift): Likewise.
7302 (expand_vector_divmod): Likewise.
7303 (optimize_vector_constructor): Likewise.
7304 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
7305 (vect_transform_slp_perm_load): Likewise.
7306 (vect_schedule_slp_instance): Likewise.
7307 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
7308 (vectorizable_call): Likewise.
7309 (vect_gen_perm_mask_any): Likewise. Add elements in order.
7310 * expmed.c (make_tree): Likewise.
7311 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
7312 a vector passed to build_vector.
7313 (fold_convert_const): Likewise.
7314 (exact_inverse): Likewise.
7315 (fold_ternary_loc): Likewise.
7316 (fold_relational_const): Likewise.
7317 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
7318 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
7319 (const_unop): Likewise. Store the reduction accumulator in a
7320 variable rather than an array.
7321 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
7322 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
7323 the new vector, rather than constructing it after the input arrays.
7324 (native_interpret_vector): Use auto_vec<tree> when building
7325 a vector passed to build_vector. Add elements in order.
7326 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
7327 auto_vec<tree> when building a vector passed to build_vector.
7328 (vect_create_epilog_for_reduction): Likewise.
7329 (vectorizable_induction): Likewise.
7330 (get_initial_def_for_reduction): Likewise. Fix indentation of
7332 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
7334 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
7335 passed to build_vector.
7337 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7338 Alan Hayward <alan.hayward@arm.com>
7339 David Sherwood <david.sherwood@arm.com>
7341 * tree-core.h (tree_base::u): Add an "nelts" field.
7342 (tree_vector): Use VECTOR_CST_NELTS as the length.
7343 * tree.c (tree_size): Likewise.
7344 (make_vector): Initialize VECTOR_CST_NELTS.
7345 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
7346 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
7347 TYPE_VECTOR_SUBPARTS.
7348 * expr.c (const_vector_mask_from_tree): Consistently use "units"
7349 as the number of units, setting it from VECTOR_CST_NELTS.
7350 (const_vector_from_tree): Likewise.
7351 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
7352 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
7353 (fold_negate_expr_1): Likewise.
7354 (fold_convert_const): Likewise.
7355 (const_binop): Likewise. Differentiate the number of output and
7357 (const_unop): Likewise.
7358 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
7359 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
7360 in cases that did the opposite.
7362 2017-09-14 Richard Biener <rguenther@suse.de>
7364 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
7367 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
7369 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
7371 2017-09-14 Jakub Jelinek <jakub@redhat.com>
7374 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
7375 if and where to split a bb, except for splitting before debug insn
7376 sequences followed by non-label real insn. Delete debug insns
7377 in between basic blocks.
7379 * combine.c (make_compound_operation_int): Formatting fixes.
7381 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
7382 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7383 * config/netbsd.h (LINK_EH_SPEC): Likewise.
7384 * config/sol2.h (LINK_EH_SPEC): Likewise.
7385 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7386 * config/s390/linux.h (LINK_SPEC): Likewise.
7387 * config/freebsd.h (LINK_EH_SPEC): Likewise.
7388 * config/openbsd.h (LINK_EH_SPEC): Likewise.
7389 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7390 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
7391 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
7392 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7393 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
7394 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
7396 2017-09-13 Jakub Jelinek <jakub@redhat.com>
7398 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
7400 (ENDFILE_LINUX_SPEC): Likewise.
7401 (LINK_EH_SPEC): Likewise.
7402 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
7403 (LINK_OS_LINUX_SPEC32): Likewise.
7404 (LINK_OS_LINUX_SPEC64): Likewise.
7405 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
7406 (LINK_OS_LINUX_SPEC): Likewise.
7408 2017-09-13 Martin Liska <mliska@suse.cz>
7411 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
7412 CASE_HIGH is NULL_TREE.
7414 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
7415 Alan Hayward <alan.hayward@arm.com>
7416 David Sherwood <david.sherwood@arm.com>
7418 * target.def (secondary_memory_needed): New hook.
7419 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
7420 instead of SECONDARY_MEMORY_NEEDED.
7421 (secondary_memory_needed_mode): Likewise.
7422 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
7423 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
7424 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
7425 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
7426 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
7427 * doc/tm.texi: Regenerate.
7428 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
7429 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
7430 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7431 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
7432 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
7433 * config/i386/i386.c (inline_secondary_memory_needed): Put the
7434 mode argument first and change the reg_class arguments to reg_class_t.
7435 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
7436 Make static. Update the call to inline_secondary_memory_needed.
7437 (ix86_register_move_cost): Update the call to
7438 inline_secondary_memory_needed.
7439 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7440 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
7442 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
7443 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7445 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
7446 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
7447 * config/mips/mips.c (mips_secondary_memory_needed): Make static
7448 and match hook interface. Add comment from mips.h.
7449 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7450 * config/mmix/mmix.md (truncdfsf2): Refer to
7451 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7453 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
7454 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
7455 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7456 (pa_secondary_memory_needed): New function.
7457 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
7458 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
7459 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7460 (pdp11_secondary_memory_needed): Make static and match hook interface.
7461 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
7462 * config/powerpcspe/powerpcspe-protos.h
7463 (rs6000_secondary_memory_needed_ptr): Delete.
7464 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
7466 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7467 (rs6000_option_override_internal): Assign to
7468 targetm.secondary_memory_needed rather than
7469 rs6000_secondary_memory_needed_ptr.
7470 (rs6000_secondary_memory_needed): Match hook interface.
7471 (rs6000_debug_secondary_memory_needed): Likewise.
7472 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
7473 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
7474 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
7475 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7476 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
7477 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
7479 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
7480 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7481 (rs6000_option_override_internal): Assign to
7482 targetm.secondary_memory_needed rather than
7483 rs6000_secondary_memory_needed_ptr.
7484 (rs6000_secondary_memory_needed): Match hook interface.
7485 (rs6000_debug_secondary_memory_needed): Likewise.
7486 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
7487 * config/s390/s390.c (s390_secondary_memory_needed): New function.
7488 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7489 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
7490 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7491 (sparc_secondary_memory_needed): New function.
7492 * lra-constraints.c (check_and_process_move): Refer to
7493 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7495 (curr_insn_transform): Likewise.
7496 (process_alt_operands): Use targetm.secondary_memory_needed
7497 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7498 (check_secondary_memory_needed_p): Likewise.
7499 (choose_split_class): Likewise.
7500 * reload.c: Unconditionally include code that was previously
7501 conditional on SECONDARY_MEMORY_NEEDED.
7502 (push_secondary_reload): Use targetm.secondary_memory_needed
7503 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7504 (push_reload): Likewise.
7505 * reload1.c: Unconditionally include code that was previously
7506 conditional on SECONDARY_MEMORY_NEEDED.
7507 (choose_reload_regs): Use targetm.secondary_memory_needed
7508 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7509 (gen_reload): Likewise.
7510 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
7512 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
7513 Alan Hayward <alan.hayward@arm.com>
7514 David Sherwood <david.sherwood@arm.com>
7516 * target.def (secondary_memory_needed_mode): New hook:
7517 * targhooks.c (default_secondary_memory_needed_mode): Declare.
7518 * targhooks.h (default_secondary_memory_needed_mode): New function.
7519 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
7520 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
7521 * doc/tm.texi: Regenerate.
7522 * lra-constraints.c (check_and_process_move): Use
7523 targetm.secondary_memory_needed_mode instead of
7524 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
7525 (curr_insn_transform): Likewise.
7526 * reload.c (get_secondary_mem): Likewise.
7527 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7528 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
7530 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7531 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7532 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
7533 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7534 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
7536 * config/powerpcspe/powerpcspe-protos.h
7537 (rs6000_secondary_memory_needed_mode): Delete.
7538 * config/powerpcspe/powerpcspe.c
7539 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7540 (rs6000_secondary_memory_needed_mode): Make static.
7541 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7542 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
7544 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7546 (rs6000_secondary_memory_needed_mode): Make static.
7547 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7548 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
7549 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7550 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7551 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7553 (sparc_secondary_memory_needed_mode): New function.
7554 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
7556 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
7558 * config/aarch64/constraints.md (Umq): New constraint.
7559 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
7561 (mov<mode>): Update condition.
7563 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7565 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
7566 when bitposition is the same.
7568 2017-09-13 Richard Biener <rguenther@suse.de>
7570 * dwarf2out.c (output_die_symbol): Remove.
7571 (output_die): Do not output a DIEs symbol.
7573 2017-09-13 Richard Biener <rguenther@suse.de>
7576 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
7577 default-def to avoid breaking iterator update with the weird
7578 interaction with cgraph_update_edges_for_call_stmt_node.
7580 2017-09-13 Richard Biener <rguenther@suse.de>
7582 * tree-cfg.c (verify_gimple_assign_binary): Add verification
7583 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
7584 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
7585 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
7587 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
7589 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
7590 Disable pc relative literal load irrespective of
7591 TARGET_FIX_ERR_A53_84341 for default.
7593 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
7595 * config/sparc/sparc.c (output_return): Output the source location of
7596 the insn in the delay slot, if any.
7597 (output_sibcall): Likewise.
7599 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
7602 * common.opt (-static-pie): New alias.
7603 (shared): Negate static-pie.
7604 (-no-pie): Update help text.
7606 (static-pie): New option.
7607 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
7608 -static-pie support.
7609 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
7610 (LINK_EH_SPEC): Likewise.
7611 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7612 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7613 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7614 * gcc.c (LINK_COMMAND_SPEC): Likewise.
7615 (init_gcc_specs): Likewise.
7616 (init_spec): Likewise.
7617 (display_help): Update help message for -pie.
7618 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
7621 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
7623 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
7624 (movdi_aarch64): Likewise.
7625 (movti_aarch64): Likewise.
7627 2017-09-12 Simon Wright <simon@pushface.org>
7630 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
7631 calculation of the minor version, always output as 0.
7633 2017-09-12 Jakub Jelinek <jakub@redhat.com>
7636 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
7637 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
7638 on it early, rather than manual conversion late. For
7639 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
7640 instead of performing manual conversion.
7642 2017-09-12 Carl Love <cel@us.ibm.com>
7644 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
7645 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
7647 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7648 VMULOSW): Add definitions.
7649 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7650 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7651 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7652 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7653 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7655 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
7657 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
7659 (movti_aarch64): Likewise.
7660 (movdf_aarch64): Likewise.
7661 (movtf_aarch64): Likewise.
7662 (load_pairdi): Likewise.
7663 (store_pairdi): Likewise.
7664 (load_pairdf): Likewise.
7665 (store_pairdf): Likewise.
7666 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
7667 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
7668 (ldr_got_small_<mode>): Likewise.
7669 (ldr_got_small_28k_<mode>): Likewise.
7670 (ldr_got_tiny): Likewise.
7671 * config/aarch64/iterators.md (ldst_sz): New.
7672 (ldpstp_sz): Likewise.
7673 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
7675 (thunderx_load): Split load_8 to load_16.
7676 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
7678 (thunderx2t99_storepair_basic): Split store_8 to store_16.
7679 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
7680 (xgene1_store_pair): Split store_8 to store_16.
7681 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
7682 (falkor_st_0_st_sd): Split store_8 to store_16.
7684 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
7686 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
7687 and store1/2/3/4 to store_4/8/12/16.
7688 * config/aarch64/aarch64.md: Update for rename.
7689 * config/arm/arm.md: Likewise.: Likewise.
7690 * config/arm/arm.c: Likewise.
7691 * config/arm/thumb1.md: Likewise.
7692 * config/arm/thumb2.md: Likewise.
7693 * config/arm/vfp.md: Likewise.
7694 * config/arm/arm-generic.md: Likewise.
7695 * config/arm/arm1020e.md: Likewise.
7696 * config/arm/arm1026ejs.md: Likewise.
7697 * config/arm/arm1136jfs.md: Likewise.
7698 * config/arm/arm926ejs.md: Likewise.
7699 * config/arm/cortex-a15.md: Likewise.
7700 * config/arm/cortex-a17.md: Likewise.
7701 * config/arm/cortex-a5.md: Likewise.
7702 * config/arm/cortex-a53.md: Likewise.
7703 * config/arm/cortex-a57.md: Likewise.
7704 * config/arm/cortex-a7.md: Likewise.
7705 * config/arm/cortex-a8.md: Likewise.
7706 * config/arm/cortex-a9.md: Likewise.
7707 * config/arm/cortex-m4.md: Likewise.
7708 * config/arm/cortex-m7.md: Likewise.
7709 * config/arm/cortex-r4.md: Likewise.
7710 * config/arm/exynos-m1.md: Likewise.
7711 * config/arm/fa526.md: Likewise.
7712 * config/arm/fa606te.md: Likewise.
7713 * config/arm/fa626te.md: Likewise.
7714 * config/arm/fa726te.md: Likewise.
7715 * config/arm/fmp626.md: Likewise.
7716 * config/arm/iwmmxt.md: Likewise.
7717 * config/arm/ldmstm.md: Likewise.
7718 * config/arm/marvell-pj4.md: Likewise.
7719 * config/arm/xgene1.md: Likewise.
7720 * config/aarch64/thunderx.md: Likewise.
7721 * config/aarch64/thunderx2t99.md: Likewise.
7722 * config/aarch64/falkor.md: Likewise.
7724 2017-09-12 Martin Liska <mliska@suse.cz>
7726 * attribs.c (private_lookup_attribute): New function.
7727 * attribs.h (private_lookup_attribute): Declared here.
7728 (lookup_attribute): Called from this place.
7730 2017-09-12 Richard Biener <rguenther@suse.de>
7732 PR tree-optimization/82157
7733 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
7734 stmts with side-effects.
7736 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7737 Alan Hayward <alan.hayward@arm.com>
7738 David Sherwood <david.sherwood@arm.com>
7740 * target.def (hard_regno_nregs): New hook.
7741 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
7742 * targhooks.h (default_hard_regno_nregs): Declare.
7743 * targhooks.c (default_hard_regno_nregs): New function.
7744 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
7745 (TARGET_HARD_REGNO_NREGS): ...this hook.
7746 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
7747 (CLASS_MAX_NREGS): Likewise.
7748 * doc/tm.texi: Regenerate.
7749 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
7750 instead of HARD_REGNO_NREGS.
7751 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
7752 HARD_REGNO_NREGS in the comment.
7753 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
7754 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
7755 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
7756 Return an unsigned int.
7757 (TARGET_HARD_REGNO_NREGS): Redefine.
7758 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
7759 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
7760 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
7761 (arc_hard_regno_nregs): New function.
7762 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
7763 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
7764 (arm_hard_regno_nregs): New function.
7765 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
7766 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
7767 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
7768 (TARGET_HARD_REGNO_NREGS): Redefine.
7769 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
7770 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
7771 (HARD_REGNO_NREGS): Delete.
7772 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
7773 (cr16_hard_regno_nregs): New function.
7774 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
7775 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
7776 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
7777 (cris_hard_regno_nregs): New function.
7778 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
7779 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
7780 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
7781 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
7782 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
7783 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
7784 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
7785 (frv_hard_regno_nregs): Make static. Take and return an
7787 (frv_class_max_nregs): Remove outdated copy of documentation.
7788 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
7789 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
7790 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
7791 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
7792 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
7793 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
7794 (TARGET_HARD_REGNO_NREGS): Redefine.
7795 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
7796 (CLASS_MAX_NREGS): Update comment.
7797 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
7798 (ia64_hard_regno_nregs): New function.
7799 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
7800 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
7801 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
7802 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
7803 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
7805 (m32c_hard_regno_nregs): Likewise. Make static.
7806 (TARGET_HARD_REGNO_NREGS): Redefine.
7807 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
7808 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
7809 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
7810 (m68k_hard_regno_nregs): New function.
7811 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
7812 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
7813 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
7814 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
7815 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
7816 Take and return an unsigned int.
7817 (TARGET_HARD_REGNO_NREGS): Redefine.
7818 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
7819 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
7820 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
7821 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
7822 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
7823 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
7824 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
7825 (msp430_hard_regno_nregs): Make static. Take and return an
7827 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
7828 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
7829 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
7830 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
7831 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
7832 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
7833 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
7834 (TARGET_HARD_REGNO_NREGS): Redefine.
7835 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
7836 (PA_HARD_REGNO_NREGS): ...this.
7837 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
7838 (PA_HARD_REGNO_NREGS): ...this.
7839 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
7840 (pa_hard_regno_nregs): New function.
7841 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
7842 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
7843 (pdp11_hard_regno_nregs): New function.
7844 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
7845 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
7846 (rs6000_hard_regno_nregs_hook): New function.
7847 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
7848 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
7849 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
7850 Take and return an unsigned int. Move earlier in file.
7851 (TARGET_HARD_REGNO_NREGS): Redefine.
7852 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
7853 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
7854 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
7855 (rl78_hard_regno_nregs): Make static. Take and return an
7857 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
7858 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
7859 (rs6000_hard_regno_nregs_hook): New function.
7860 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
7861 * config/rx/rx.c (rx_hard_regno_nregs): New function.
7862 (TARGET_HARD_REGNO_NREGS): Redefine.
7863 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
7864 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
7865 instead of HARD_REGNO_NREGS.
7866 (s390_hard_regno_nregs): New function.
7867 (s390_hard_regno_mode_ok): Add comment from s390.h.
7868 (TARGET_HARD_REGNO_NREGS): Redefine.
7869 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
7870 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
7871 (sh_hard_regno_nregs): New function.
7872 (sh_pass_in_reg_p): Use it.
7873 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
7874 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
7875 (sparc_hard_regno_nregs): New function.
7876 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
7877 * config/spu/spu.c (spu_hard_regno_nregs): New function.
7878 (spu_function_arg_advance): Use it, supplying a valid register number.
7879 (TARGET_HARD_REGNO_NREGS): Redefine.
7880 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
7881 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
7882 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
7883 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
7884 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
7885 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
7886 (CLASS_MAX_NREGS): Remove copy of old documentation.
7887 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
7888 (visium_hard_regno_nregs): New function.
7889 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
7890 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
7891 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
7892 xtensa_hard_regno_nregs): New function.
7893 * system.h (HARD_REGNO_NREGS): Poison.
7895 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7897 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
7898 hard_regno_nregs instead of HARD_REGNO_NREGS.
7899 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
7900 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
7901 (c6x_expand_epilogue): Likewise.
7902 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
7903 (frv_read_iacc_argument): Likewise.
7904 * config/sh/sh.c: Include regs.h.
7905 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
7906 (regs_used): Likewise.
7907 (output_stack_adjust): Likewise.
7908 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
7909 * expmed.c: Include regs.h.
7910 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
7911 * ree.c: Include regs.h.
7912 (combine_reaching_defs): Use hard_regno_nregs instead of
7914 (add_removable_extension): Likewise.
7916 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7918 * regs.h (hard_regno_nregs): Turn into a function.
7919 (end_hard_regno): Update accordingly.
7920 * caller-save.c (setup_save_areas): Likewise.
7921 (save_call_clobbered_regs): Likewise.
7922 (replace_reg_with_saved_mem): Likewise.
7923 (insert_restore): Likewise.
7924 (insert_save): Likewise.
7925 * combine.c (can_change_dest_mode): Likewise.
7926 (move_deaths): Likewise.
7927 (distribute_notes): Likewise.
7928 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
7929 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
7930 (rs6000_split_multireg_move): Likewise.
7931 (rs6000_register_move_cost): Likewise.
7932 (rs6000_memory_move_cost): Likewise.
7933 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
7934 (rs6000_split_multireg_move): Likewise.
7935 (rs6000_register_move_cost): Likewise.
7936 (rs6000_memory_move_cost): Likewise.
7937 * cselib.c (cselib_reset_table): Likewise.
7938 (cselib_lookup_1): Likewise.
7939 * emit-rtl.c (set_mode_and_regno): Likewise.
7940 * function.c (aggregate_value_p): Likewise.
7941 * ira-color.c (setup_profitable_hard_regs): Likewise.
7942 (check_hard_reg_p): Likewise.
7943 (calculate_saved_nregs): Likewise.
7944 (assign_hard_reg): Likewise.
7945 (improve_allocation): Likewise.
7946 (calculate_spill_cost): Likewise.
7947 * ira-emit.c (modify_move_list): Likewise.
7948 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
7949 (ira_hard_reg_in_set_p): Likewise.
7950 * ira.c (setup_reg_mode_hard_regset): Likewise.
7951 (clarify_prohibited_class_mode_regs): Likewise.
7952 (check_allocation): Likewise.
7953 * lra-assigns.c (find_hard_regno_for_1): Likewise.
7954 (lra_setup_reg_renumber): Likewise.
7955 (setup_try_hard_regno_pseudos): Likewise.
7956 (spill_for): Likewise.
7957 (assign_hard_regno): Likewise.
7958 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
7959 * lra-constraints.c (in_class_p): Likewise.
7960 (lra_constraint_offset): Likewise.
7961 (simplify_operand_subreg): Likewise.
7962 (lra_constraints): Likewise.
7963 (split_reg): Likewise.
7964 (split_if_necessary): Likewise.
7965 (invariant_p): Likewise.
7966 (inherit_in_ebb): Likewise.
7967 * lra-lives.c (process_bb_lives): Likewise.
7968 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
7969 (get_hard_regs): Likewise.
7970 (do_remat): Likewise.
7971 * lra-spills.c (assign_spill_hard_regs): Likewise.
7972 * mode-switching.c (create_pre_exit): Likewise.
7973 * postreload.c (reload_combine_recognize_pattern): Likewise.
7974 * recog.c (peep2_find_free_register): Likewise.
7975 * regcprop.c (kill_value_regno): Likewise.
7976 (set_value_regno): Likewise.
7977 (copy_value): Likewise.
7978 (maybe_mode_change): Likewise.
7979 (find_oldest_value_reg): Likewise.
7980 (copyprop_hardreg_forward_1): Likewise.
7981 * regrename.c (check_new_reg_p): Likewise.
7982 (regrename_do_replace): Likewise.
7983 * reload.c (push_reload): Likewise.
7984 (combine_reloads): Likewise.
7985 (find_dummy_reload): Likewise.
7986 (operands_match_p): Likewise.
7987 (find_reloads): Likewise.
7988 (find_equiv_reg): Likewise.
7989 (reload_adjust_reg_for_mode): Likewise.
7990 * reload1.c (count_pseudo): Likewise.
7991 (count_spilled_pseudo): Likewise.
7992 (find_reg): Likewise.
7993 (clear_reload_reg_in_use): Likewise.
7994 (free_for_value_p): Likewise.
7995 (allocate_reload_reg): Likewise.
7996 (choose_reload_regs): Likewise.
7997 (reload_adjust_reg_for_temp): Likewise.
7998 (emit_reload_insns): Likewise.
7999 (delete_output_reload): Likewise.
8000 * rtlanal.c (subreg_get_info): Likewise.
8001 * sched-deps.c (sched_analyze_reg): Likewise.
8002 * sel-sched.c (init_regs_for_mode): Likewise.
8003 (mark_unavailable_hard_regs): Likewise.
8004 (choose_best_reg_1): Likewise.
8005 (verify_target_availability): Likewise.
8006 * valtrack.c (dead_debug_insert_temp): Likewise.
8007 * var-tracking.c (track_loc_p): Likewise.
8008 (emit_note_insn_var_location): Likewise.
8009 * varasm.c (make_decl_rtl): Likewise.
8010 * reginfo.c (choose_hard_reg_mode): Likewise.
8011 (init_reg_modes_target): Refer directly to
8012 this_target_regs->x_hard_regno_nregs.
8014 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8016 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
8017 instead of hard_regno_nregs.
8019 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8021 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
8022 end_hard_regno instead of hard_regno_nregs.
8023 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
8024 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
8025 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
8026 * ira-color.c (improve_allocation): Likewise.
8027 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8028 * lra-lives.c (mark_regno_live): Likewise.
8029 (mark_regno_dead): Likewise.
8030 * lra-remat.c (operand_to_remat): Likewise.
8031 * lra.c (collect_non_operand_hard_regs): Likewise.
8032 * postreload.c (reload_combine_note_store): Likewise.
8033 (move2add_valid_value_p): Likewise.
8034 * reload.c (regno_clobbered_p): Likewise.
8036 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8038 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
8040 * config/v850/v850.c (v850_reorg): Likewise.
8041 * reload.c (refers_to_regno_for_reload_p): Likewise.
8042 (find_equiv_reg): Likewise.
8043 * reload1.c (reload_reg_reaches_end_p): Likewise.
8045 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8047 * caller-save.c (add_used_regs): Use REG_NREGS instead of
8049 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
8050 * config/arm/arm.c (output_move_neon): Likewise.
8051 (arm_attr_length_move_neon): Likewise.
8052 (neon_split_vcombine): Likewise.
8053 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
8054 (c6x_mark_reg_written): Likewise.
8055 (c6x_dwarf_register_span): Likewise.
8056 * config/i386/i386.c (ix86_save_reg): Likewise.
8057 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
8058 (rws_access_reg): Likewise.
8059 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
8060 * mode-switching.c (create_pre_exit): Likewise.
8061 * ree.c (combine_reaching_defs): Likewise.
8062 (add_removable_extension): Likewise.
8063 * regcprop.c (find_oldest_value_reg): Likewise.
8064 (copyprop_hardreg_forward_1): Likewise.
8065 * reload.c (reload_inner_reg_of_subreg): Likewise.
8066 (push_reload): Likewise.
8067 (combine_reloads): Likewise.
8068 (find_dummy_reload): Likewise.
8069 (reload_adjust_reg_for_mode): Likewise.
8070 * reload1.c (find_reload_regs): Likewise.
8071 (forget_old_reloads_1): Likewise.
8072 (reload_reg_free_for_value_p): Likewise.
8073 (reload_adjust_reg_for_temp): Likewise.
8074 (emit_reload_insns): Likewise.
8075 (delete_output_reload): Likewise.
8076 * sel-sched.c (choose_best_reg_1): Likewise.
8077 (choose_best_pseudo_reg): Likewise.
8079 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8080 Alan Hayward <alan.hayward@arm.com>
8081 David Sherwood <david.sherwood@arm.com>
8083 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
8084 * target.def (slow_unaligned_access): New hook.
8085 * targhooks.h (default_slow_unaligned_access): Declare.
8086 * targhooks.c (default_slow_unaligned_access): New function.
8087 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
8088 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
8089 * doc/tm.texi: Regenerate.
8090 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
8091 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
8092 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
8094 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
8095 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
8097 (rs6000_slow_unaligned_access): New function.
8098 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8099 (expand_block_compare): Likewise.
8100 (expand_strn_compare): Likewise.
8101 (rs6000_rtx_costs): Likewise.
8102 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
8103 (riscv_slow_unaligned_access): Likewise.
8104 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
8105 (riscv_slow_unaligned_access_p): ...this and make static.
8106 (riscv_option_override): Update accordingly.
8107 (riscv_slow_unaligned_access): New function.
8108 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8109 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
8110 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8111 (rs6000_slow_unaligned_access): New function.
8112 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8113 (rs6000_rtx_costs): Likewise.
8114 * config/rs6000/rs6000-string.c (expand_block_compare)
8115 (expand_strn_compare): Use targetm.slow_unaligned_access instead
8116 of SLOW_UNALIGNED_ACCESS.
8117 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
8118 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
8119 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
8120 of SLOW_UNALIGNED_ACCESS.
8121 * expmed.c (simple_mem_bitfield_p): Likewise.
8122 * expr.c (alignment_for_piecewise_move): Likewise.
8123 (emit_group_load_1): Likewise.
8124 (emit_group_store): Likewise.
8125 (copy_blkmode_from_reg): Likewise.
8126 (emit_push_insn): Likewise.
8127 (expand_assignment): Likewise.
8128 (store_field): Likewise.
8129 (expand_expr_real_1): Likewise.
8130 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
8131 * lra-constraints.c (simplify_operand_subreg): Likewise.
8132 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
8133 * gimple-ssa-store-merging.c: Likewise in block comment at start
8135 * tree-ssa-strlen.c: Include target.h.
8136 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
8137 of SLOW_UNALIGNED_ACCESS.
8138 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
8140 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8142 PR rtl-optimization/82185
8143 * expmed.c (emit_store_flag_int): Only test tem if it has been
8146 2017-09-12 Richard Biener <rguenther@suse.de>
8149 * match.pd ((FTYPE) N CMP CST): Fix typo.
8151 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
8153 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
8155 (mips_near_type_p): Add 'short_call' attribute as a synonym
8157 * doc/extend.texi (short_call): Document new function attribute.
8159 2017-09-12 Jakub Jelinek <jakub@redhat.com>
8162 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
8163 assertion check that in the condition.
8164 (get_atomic_generic_size): Likewise. Before testing if parameter
8165 has pointer type, if it has array type, call for C++
8166 default_conversion to perform array-to-pointer conversion.
8168 2017-09-12 Richard Biener <rguenther@suse.de>
8170 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
8171 for operations we cannot scalarize.
8173 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
8175 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
8176 vectors heap vectors. Clean up comments.
8177 Make visited_bbs a reference.
8178 (profitable_jump_thread_path): Make GC
8179 vectors heap vectors. Clean up comments.
8181 (convert_and_register_jump_thread_path): Make GC vectors heap
8183 (check_subpath_and_update_thread_path): Same. Clean up comments.
8184 Make visited_bbs a reference.
8185 (handle_phi): Abstract common code to to
8186 register_jump_thread_path_if_profitable.
8187 Rename VAR_BB to DEF_BB.
8189 Make GC vectors heap vectors.
8190 Make visited_bbs a reference.
8191 (handle_assignment): Same.
8192 (register_jump_thread_path_if_profitable): New.
8193 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
8195 Make GC vectors heap vectors. Clean up comments.
8196 Make visited_bbs a reference.
8197 (find_jump_threads_backwards): Make visited_bbs live in the stack.
8198 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
8201 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
8204 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
8205 words of E_DImode object are reachable by xtensa_uimm8x4 access.
8207 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
8209 Revert r251800 and r251799.
8211 2017-09-11 Martin Jambor <mjambor@suse.cz>
8214 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
8215 arguments in advance.
8216 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
8217 use it to find predecessor edges.
8218 (naive_outof_ssa): Collect vector of predecessors.
8220 2017-09-08 Jason Merrill <jason@redhat.com>
8222 PR c++/70029 - ICE with ref-qualifier and -flto
8223 * langhooks.h (struct lang_hooks_for_types): Add
8224 copy_lang_qualifiers.
8225 * attribs.c (build_type_attribute_qual_variant): Use it.
8226 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
8228 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
8229 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
8231 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
8234 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
8235 (*mulsi3_sp64): New instruction.
8236 (mulsi3): New expander.
8238 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
8240 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
8242 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8244 * sancov.c: Include memmodel.h.
8246 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
8249 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
8252 2017-09-07 Carl Love <cel@us.ibm.com>
8254 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
8255 the sldi instruction.
8257 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
8259 * sancov.c: Include tm_p.h.
8261 2017-09-07 Jakub Jelinek <jakub@redhat.com>
8264 * output.h (switch_to_other_text_partition): New declaration.
8265 * varasm.c (switch_to_other_text_partition): New function.
8266 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
8267 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
8268 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
8269 to the other text partition before emitting LCL label and switch back
8270 after emitting the word after it.
8272 2017-09-07 Richard Biener <rguenther@suse.de>
8274 * passes.def (pass_split_crit_edges): Remove instance before PRE.
8275 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
8276 critical edges here, after loop init.
8277 (pass_data_pre): Remove PROP_no_crit_edges flags.
8278 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
8279 for valueization of call args to avoid leaking VN_TOP.
8280 (visit_use): Assert we do not visit default defs.
8281 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
8282 Use error_mark_node to more easily detect leaking VN_TOP.
8283 All default-defs are varying, not VN_TOP. Mark them visited.
8284 (run_scc_vn): Make code match comment.
8286 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8288 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
8289 OPTION_MASK_FLOAT128_KEYWORD.
8290 (POWERPC_MASKS): Likewise.
8291 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
8292 support for the -mfloat128-type option, and make -mfloat128
8293 default on PowerPC Linux systems. Define or undefine
8294 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
8295 Define __float128 to be __ieee128 if IEEE 128-bit support is
8296 enabled, or undefine it.
8297 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
8298 Delete defining __FLOAT128_TYPE__.
8299 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
8300 -mfloat128-type option and make -mfloat128 default on PowerPC
8302 (TARGET_FLOAT128_TYPE): Likewise.
8303 (-mfloat128-type): Likewise.
8304 * config/rs6000/rs6000.c (rs6000_option_override_internal):
8305 Delete the -mfloat128-type option and make -mfloat128 default on
8306 PowerPC Linux systems. Always use __ieee128 to be the keyword for
8307 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
8308 128-bit floating point is enabled. Change tests from using
8309 -mfloat128-type to -mfloat128.
8310 (rs6000_mangle_type): Use the correct mangling for the __float128
8311 type even if normal long double is restricted to 64-bits.
8312 (floatn_mode): Enable the _Float128 type by default on VSX Linux
8314 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
8315 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
8316 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
8318 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
8319 documentation for -mfloat128.
8321 2017-09-06 Olivier Hainque <hainque@adacore.com>
8323 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
8325 2017-09-06 Wish Wu <wishwu007@gmail.com>
8326 Jakub Jelinek <jakub@redhat.com>
8328 * asan.c (initialize_sanitizer_builtins): Add
8329 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
8330 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
8331 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
8332 BT_FN_VOID_UINT64_PTR variables.
8333 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
8334 (BT_FN_VOID_UINT16_UINT16): Likewise.
8335 (BT_FN_VOID_UINT32_UINT32): Likewise.
8336 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
8337 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
8338 (BT_FN_VOID_UINT64_PTR): Likewise.
8339 * common.opt (flag_sanitize_coverage): New variable.
8340 (fsanitize-coverage=trace-pc): Remove.
8341 (fsanitize-coverage=): Add.
8342 * flag-types.h (enum sanitize_coverage_code): New enum.
8343 * fold-const.c (fold_range_test): Disable non-short-circuit
8344 optimization if flag_sanitize_coverage.
8345 (fold_truth_andor): Likewise.
8346 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
8347 * opts.c (COVERAGE_SANITIZER_OPT): Define.
8348 (coverage_sanitizer_opts): New array.
8349 (get_closest_sanitizer_option): Add OPTS argument, handle also
8350 OPT_fsanitize_coverage_.
8351 (parse_sanitizer_options): Adjusted to also handle
8352 OPT_fsanitize_coverage_.
8353 (common_handle_option): Add OPT_fsanitize_coverage_.
8354 * sancov.c (instrument_comparison, instrument_switch): New function.
8355 (sancov_pass): Add trace-cmp support.
8356 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
8357 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
8358 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
8359 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
8360 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
8361 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
8362 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
8363 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
8364 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
8365 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
8367 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
8369 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
8370 error. Only quit immediately if parsing is complete.
8371 (BEGIN): Initialize fatal_err and parse_done.
8372 (begin fpu, end fpu): Check number of arguments.
8373 (begin arch, end arch): Likewise.
8374 (begin cpu, end cpu): Likewise.
8375 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
8376 (optalias): Likewise.
8378 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
8380 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
8381 * config/arm/arm-isa.h: Delete. Move definitions to ...
8382 * arm-cpus.in: ... here. Use new feature and fgroup values.
8383 * config/arm/arm.c (arm_option_override): Use lower case for feature
8385 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
8386 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
8387 * config/arm/parsecpu.awk (END): Add new command 'isa'.
8389 (print_isa_bits_for): New function.
8390 (gen_isa): New function.
8391 (gen_comm_data): Use print_isa_bits_for.
8392 (define feature): New keyword.
8393 (define fgroup): New keyword.
8394 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
8395 (arm-isa.h): Add rule to generate file.
8396 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
8397 case for feature bit names.
8399 2017-09-06 Richard Biener <rguenther@suse.de>
8401 * tree-ssa-pre.c (NECESSARY): Remove.
8402 (create_expression_by_pieces): Do not touch pass-local flags.
8403 (insert_into_preds_of_block): Likewise.
8404 (do_pre_regular_insertion): Likewise.
8405 (eliminate_insert): Likewise.
8406 (eliminate_dom_walker::before_dom_children): Likewise.
8407 (fini_eliminate): Do not look at inserted_exprs.
8408 (mark_operand_necessary): Remove.
8409 (remove_dead_inserted_code): Replace with simple work-list
8410 algorithm based on inserted_exprs and SSA uses.
8411 (pass_pre::execute): Re-order fini_eliminate and
8412 remove_dead_inserted_code.
8414 2017-09-06 Olivier Hainque <hainque@adacore.com>
8416 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
8417 for VxWorks 7. Adjust surrounding comments.
8419 2017-09-06 Richard Biener <rguenther@suse.de>
8421 * gimple-ssa-strength-reduction.c
8422 (find_candidates_dom_walker::before_dom_children): Also allow
8425 2017-09-06 Richard Biener <rguenther@suse.de>
8427 PR tree-optimization/82108
8428 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
8429 for gap in the non-permutation SLP case.
8431 2017-09-06 Martin Jambor <mjambor@suse.cz>
8433 PR tree-optimization/82078
8434 * tree-sra.c (sort_and_splice_var_accesses): Move call to
8435 add_access_to_work_queue...
8436 (build_accesses_from_assign): ...here.
8437 (propagate_all_subaccesses): Make sure racc is the group
8438 representative, if there is one.
8440 2017-09-06 Jakub Jelinek <jakub@redhat.com>
8443 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
8446 2017-09-06 Richard Biener <rguenther@suse.de>
8448 * gimple-ssa-strength-reduction.c
8449 (find_candidates_dom_walker::before_doom_children): Use a
8450 type and not a mode check.
8452 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
8455 * config/arm/predicates.md (arm_general_adddi_operand): Create new
8457 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
8459 2017-09-05 Jeff Law <law@redhat.com>
8461 PR tree-optimization/64910
8462 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
8463 cases where we have 3 or more operands.
8465 2017-09-05 Jakub Jelinek <jakub@redhat.com>
8468 * omp-low.c (lower_omp_for): Recompute tree invariant if
8469 gimple_omp_for_initial/final is ADDR_EXPR.
8472 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
8473 into gimple val before gimplification fo the COND_EXPR.
8475 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
8477 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
8478 REGION_COPY argument.
8479 (thread_through_all_blocks): Remove unused argument to
8480 duplicate_thread_path.
8482 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8483 Alan Hayward <alan.hayward@arm.com>
8484 David Sherwood <david.sherwood@arm.com>
8486 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
8487 Take a scalar_mode rather than a machine_mode.
8488 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8489 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
8490 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8491 (aarch64_gen_adjusted_ldpstp): Likewise.
8492 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
8494 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8495 Alan Hayward <alan.hayward@arm.com>
8496 David Sherwood <david.sherwood@arm.com>
8498 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
8499 Take a scalar_int_mode instead of a machine_mode.
8500 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8501 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8502 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8503 (aarch64_simd_attr_length_rglist): Delete.
8504 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
8505 a scalar_int_mode instead of a machine_mode.
8506 (aarch64_add_offset): Likewise.
8507 (aarch64_internal_mov_immediate): Likewise
8508 (aarch64_add_constant_internal): Likewise.
8509 (aarch64_add_constant): Likewise.
8510 (aarch64_movw_imm): Likewise.
8511 (aarch64_rtx_arith_op_extract_p): Likewise.
8512 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8513 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8514 Remove assert that the mode isn't a vector.
8515 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8516 (aarch64_expand_mov_immediate): Update calls after above changes.
8517 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
8518 (aarch64_and_bitmask_imm): Check for scalar integer modes.
8519 (aarch64_move_imm): Likewise.
8520 (aarch64_can_const_movi_rtx_p): Likewise.
8521 (aarch64_strip_extend): Likewise.
8522 (aarch64_extr_rtx_p): Likewise.
8523 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
8524 a CONST_INT when the mode parameter is VOIDmode.
8525 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
8527 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8529 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
8530 * stor-layout.c (bitwise_mode_for_mode): Likewise.
8531 (bitwise_type_for_mode): Update accordingly.
8533 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8535 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
8536 * stor-layout.c (mode_for_size_tree): Likewise.
8537 (mode_for_array): Update accordingly.
8538 (layout_decl): Likewise.
8539 (compute_record_mode): Likewise. Only set the mode once.
8541 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8543 * target.def (get_mask_mode): Change return type to opt_mode.
8545 * doc/tm.texi: Regenerate.
8546 * targhooks.h (default_get_mask_mode): Return an opt_mode.
8547 * targhooks.c (default_get_mask_mode): Likewise.
8548 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
8549 * optabs-query.c (can_vec_mask_load_store_p): Update use of
8550 targetm.get_mask_mode.
8551 * tree.c (build_truth_vector_type): Likewise.
8553 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8555 * machmode.h (mode_for_vector): Return an opt_mode.
8556 * stor-layout.c (mode_for_vector): Likewise.
8557 (mode_for_int_vector): Update accordingly.
8558 (layout_type): Likewise.
8559 * config/i386/i386.c (emit_memmov): Likewise.
8560 (ix86_expand_set_or_movmem): Likewise.
8561 (ix86_expand_vector_init): Likewise.
8562 (ix86_get_mask_mode): Likewise.
8563 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
8565 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
8566 * expmed.c (extract_bit_field_1): Likewise.
8567 * expr.c (expand_expr_real_2): Likewise.
8568 * optabs-query.c (can_vec_perm_p): Likewise.
8569 (can_vec_mask_load_store_p): Likewise.
8570 * optabs.c (expand_vec_perm): Likewise.
8571 * targhooks.c (default_get_mask_mode): Likewise.
8572 * tree-vect-stmts.c (vectorizable_store): Likewise.
8573 (vectorizable_load): Likewise.
8574 (get_vectype_for_scalar_type_and_size): Likewise.
8576 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8578 * machmode.h (mode_for_int_vector): New function.
8579 * stor-layout.c (mode_for_int_vector): Likewise.
8580 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
8581 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
8582 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
8583 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
8584 (s390_expand_vcond): Likewise.
8586 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8588 * machmode.h (opt_machine_mode): New type.
8589 (opt_mode<T>): Allow construction from anything that can be
8591 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
8592 (mode_for_size): Return an opt_machine_mode.
8593 * stor-layout.c (mode_for_size): Likewise.
8594 (mode_for_size_tree): Update call accordingly.
8595 (bitwise_mode_for_mode): Likewise.
8596 (make_fract_type): Likewise.
8597 (make_accum_type): Likewise.
8598 * caller-save.c (replace_reg_with_saved_mem): Update call
8600 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8601 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8602 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8603 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8604 * expmed.c (extract_bit_field_1): Likewise.
8605 * reload.c (get_secondary_mem): Likewise.
8606 * varasm.c (assemble_integer): Likewise.
8607 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
8610 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8612 * machmode.h (decimal_float_mode_for_size): New function.
8613 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
8614 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
8615 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
8616 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
8617 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
8618 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
8620 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8622 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
8623 (get_builtin_sync_mode): Likewise.
8624 (expand_ifn_atomic_compare_exchange): Likewise.
8625 (expand_builtin_atomic_clear): Likewise.
8626 (expand_builtin_atomic_test_and_set): Likewise.
8627 (fold_builtin_atomic_always_lock_free): Likewise.
8628 * calls.c (compute_argument_addresses): Likewise.
8629 (emit_library_call_value_1): Likewise.
8630 (store_one_arg): Likewise.
8631 * combine.c (combine_instructions): Likewise.
8632 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
8633 * config/arm/arm.c (arm_function_value): Likewise.
8634 (aapcs_allocate_return_reg): Likewise.
8635 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
8636 * config/i386/i386.c (construct_container): Likewise.
8637 (ix86_gimplify_va_arg): Likewise.
8638 (ix86_expand_sse_cmp): Likewise.
8639 (emit_memmov): Likewise.
8640 (emit_memset): Likewise.
8641 (expand_small_movmem_or_setmem): Likewise.
8642 (ix86_expand_pextr): Likewise.
8643 (ix86_expand_pinsr): Likewise.
8644 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
8645 * config/microblaze/microblaze.c (microblaze_block_move_straight):
8647 * config/mips/mips.c (mips_function_value_1) Likewise.
8648 (mips_block_move_straight): Likewise.
8649 (mips_expand_ins_as_unaligned_store): Likewise.
8650 * config/powerpcspe/powerpcspe.c
8651 (rs6000_darwin64_record_arg_advance_flush): Likewise.
8652 (rs6000_darwin64_record_arg_flush): Likewise.
8653 * config/rs6000/rs6000.c
8654 (rs6000_darwin64_record_arg_advance_flush): Likewise.
8655 (rs6000_darwin64_record_arg_flush): Likewise.
8656 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
8657 (sparc_function_value_1): Likewise.
8658 * config/spu/spu.c (adjust_operand): Likewise.
8659 (spu_emit_branch_or_set): Likewise.
8660 (arith_immediate_p): Likewise.
8661 * emit-rtl.c (gen_lowpart_common): Likewise.
8662 * expr.c (expand_expr_real_1): Likewise.
8663 * function.c (assign_parm_setup_block): Likewise.
8664 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
8665 * reload1.c (alter_reg): Likewise.
8666 * stor-layout.c (mode_for_vector): Likewise.
8667 (layout_type): Likewise.
8669 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8671 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
8672 (spu_convert_move): Likewise.
8673 * lower-subreg.c (resolve_simple_move): Likewise.
8675 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8678 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
8679 define_insn to a define_expand.
8680 (altivec_vsum2sws_direct): New define_insn.
8681 (altivec_vsumsws): Convert from a define_insn to a define_expand.
8683 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
8685 * config/arm/arm.c (arm_option_params_internal): Improve setting of
8688 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
8693 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
8694 realign stack if stack alignment needed is less than incoming
8697 2017-09-05 Marek Polacek <polacek@redhat.com>
8700 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
8703 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
8705 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
8707 2017-09-05 Richard Biener <rguenther@suse.de>
8709 PR tree-optimization/82084
8710 * fold-const.c (can_native_encode_string_p): Handle wide characters.
8712 2017-09-05 Richard Biener <rguenther@suse.de>
8714 PR tree-optimization/82102
8715 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
8717 2017-09-05 Martin Liska <mliska@suse.cz>
8719 PR tree-optimization/82032
8720 * tree-cfg.c (generate_range_test): New function.
8721 * tree-cfg.h (generate_range_test): Declared here.
8722 * tree-cfgcleanup.c (convert_single_case_switch): New function.
8723 (cleanup_control_expr_graph): Use it.
8724 * tree-switch-conversion.c (try_switch_expansion): Remove
8726 (emit_case_nodes): Use generate_range_test.
8728 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
8731 * config/i386/i386.md (*<btsc><mode>_mask): Add
8732 TARGET_USE_BT to insn constraint.
8733 (*btr<mode>_mask): Ditto.
8735 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
8737 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
8738 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
8740 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
8743 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
8744 TARGET_NEON and TARGET_IWMMXT.
8745 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
8746 TARGET_NEON and TARGET_IWMMXT.
8747 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
8749 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
8751 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
8752 (ix86_rewrite_tls_address): Ditto.
8753 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
8754 (ix86_rewrite_tls_address_1): Ditto.
8755 (ix86_rewrite_tls_address): Ditto.
8756 * config/i386/predicates.md (tls_address_pattern): New predicate.
8757 * config/i386/i386.md (TLS address splitter): New splitter.
8759 2017-09-04 Richard Biener <rguenther@suse.de>
8761 PR tree-optimization/82084
8762 * fold-const.h (can_native_encode_string_p): Declare.
8763 * fold-const.c (can_native_encode_string_p): Factor out from ...
8764 (native_encode_string): ... here.
8765 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
8766 vectorizing stores from constants we later cannot handle.
8768 2017-09-04 Marek Polacek <polacek@redhat.com>
8771 * doc/invoke.texi: Update -Wtautological-compare documentation.
8773 2017-09-04 Jeff Law <law@redhat.com>
8775 PR tree-optimization/64910
8776 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
8777 swap the first and last operand if the last is a constant.
8779 2017-09-04 Marek Polacek <polacek@redhat.com>
8782 * convert.c (do_narrow): When sanitizing signed integer overflows,
8783 bail out for signed types.
8784 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
8786 2017-09-04 Richard Biener <rguenther@suse.de>
8788 PR tree-optimization/82060
8789 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
8790 Move devirtualization after stmt folding and before EH/AB/noreturn
8791 cleanup to get the stmt refs canonicalized. Use a bool instead
8792 of gimple_modified_p since that doesn't work for NOPs. Schedule
8793 NOPs generated by folding for removal.
8795 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8796 Alan Hayward <alan.hayward@arm.com>
8797 David Sherwood <david.sherwood@arm.com>
8799 * coretypes.h (pad_direction): New enum.
8800 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
8801 (FUNCTION_ARG_PADDING): Likewise.
8802 * target.def (function_arg_padding): New hook.
8803 * targhooks.h (default_function_arg_padding): Declare.
8804 * targhooks.c (default_function_arg_padding): New function.
8805 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
8806 (TARGET_FUNCTION_ARG_PADDING): ...this.
8807 * doc/tm.texi: Regenerate.
8808 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
8809 instead of direction.
8810 (compute_argument_addresses): Likewise.
8811 (load_register_parameters): Likewise.
8812 (emit_library_call_value_1): Likewise.
8813 (store_one_arg): Use targetm.calls.function_arg_padding instead
8814 of FUNCTION_ARG_PADDING.
8815 (must_pass_in_stack_var_size_or_pad): Likewise.
8816 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
8817 (emit_group_store): Likewise.
8818 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
8819 instead of FUNCTION_ARG_PADDING.
8820 (emit_push_insn): Likewise, and propagate enum change throughout
8822 * function.h (direction): Delete.
8823 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
8825 * function.c (assign_parm_find_stack_rtl): Likewise.
8826 (assign_parm_setup_block_p): Likewise.
8827 (assign_parm_setup_block): Likewise.
8828 (gimplify_parameters): Likewise.
8829 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
8830 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
8832 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
8833 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8834 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
8835 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
8836 (aarch64_function_arg_padding): ...this new function.
8837 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
8838 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8839 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
8840 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8841 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
8842 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8843 (arm_pad_arg_upward): Replace with...
8844 (arm_function_arg_padding): ...this new function.
8845 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
8847 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
8848 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
8849 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8850 (ia64_hpux_function_arg_padding): Replace with...
8851 (ia64_function_arg_padding): ...this new function. Use pad_direction
8852 instead of direction. Check for TARGET_HPUX.
8853 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
8854 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8855 (iq2000_function_arg_padding): New function.
8856 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
8857 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
8858 (mips_function_arg_padding): ...this new function.
8859 (mips_pad_reg_upward): Update accordingly.
8860 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8861 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
8862 targetm.calls.function_arg_padding.
8863 (FUNCTION_ARG_PADDING): Delete.
8864 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8865 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
8866 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8867 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
8868 (nios2_block_reg_padding): Return pad_direction instead of direction.
8869 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
8870 instead of direction.
8871 (nios2_function_arg_padding): Likewise. Make static.
8872 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8873 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
8874 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
8875 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
8876 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8877 (pa_function_arg_padding): Make static. Return pad_direction instead
8879 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
8880 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8881 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
8882 instead of direction. Use targetm.calls.function_arg_padding.
8883 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
8884 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
8885 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
8886 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
8887 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
8889 (function_arg_padding): Rename to...
8890 (rs6000_function_arg_padding): ...this. Make static. Return
8891 pad_direction instead of direction.
8892 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
8893 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
8894 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8895 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
8896 instead of direction. Use targetm.calls.function_arg_padding.
8897 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
8898 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
8899 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
8900 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
8901 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8902 (function_arg_padding): Rename to...
8903 (rs6000_function_arg_padding): ...this. Make static. Return
8904 pad_direction instead of direction.
8905 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
8906 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
8907 * config/s390/s390.c (s390_function_arg_padding): New function.
8908 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8909 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
8910 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
8911 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8912 (function_arg_padding): Rename to...
8913 (sparc_function_arg_padding): ...this. Make static. Return
8914 pad_direction instead of direction.
8915 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
8916 * config/spu/spu.c (spu_function_arg_padding): New function.
8917 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8918 * system.h (FUNCTION_ARG_PADDING): Poison.
8920 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8921 Alan Hayward <alan.hayward@arm.com>
8922 David Sherwood <david.sherwood@arm.com>
8924 * target.def (modes_tieable_p): New hook.
8925 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
8926 (TARGET_MODES_TIEABLE_P): ...this.
8927 * doc/tm.texi.in: Regenerate.
8928 * hooks.h (hook_bool_mode_mode_true): Declare.
8929 * hooks.c (hook_bool_mode_mode_true): New function.
8930 * combine.c (subst): Use targetm.modes_tieable_p instead of
8932 * dse.c (find_shift_sequence): Likewise.
8933 * expmed.c (extract_low_bits): Likewise.
8934 * lower-subreg.c: Include target.h.
8935 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
8937 * rtlanal.c (rtx_cost): Likewise.
8938 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
8939 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
8940 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
8941 (TARGET_MODES_TIEABLE_P): Redefine.
8942 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
8943 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
8944 (TARGET_MODES_TIEABLE_P): Redefine.
8945 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
8946 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
8947 (arc_modes_tieable_p): New function.
8948 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
8949 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
8950 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
8951 (arm_modes_tieable_p): Make static.
8952 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
8953 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
8954 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
8955 (TARGET_MODES_TIEABLE_P): Redefine.
8956 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
8957 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
8958 (TARGET_MODES_TIEABLE_P): Redefine.
8959 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
8960 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
8961 (cr16_modes_tieable_p): New function.
8962 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
8963 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
8964 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
8965 (TRULY_NOOP_TRUNCATION): Update comment.
8966 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
8967 (TRULY_NOOP_TRUNCATION): Update comment.
8968 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
8969 (frv_modes_tieable_p): New function.
8970 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
8971 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
8972 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
8973 (TARGET_MODES_TIEABLE_P): Redefine.
8974 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
8975 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
8976 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
8977 (TARGET_MODES_TIEABLE_P): Redefine.
8978 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
8979 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
8980 (ia64_modes_tieable_p): New function.
8981 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
8982 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
8983 (iq2000_modes_tieable_p): New function.
8984 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
8985 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
8986 (lm32_modes_tieable_p): New function.
8987 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
8988 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
8989 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
8990 (TARGET_MODES_TIEABLE_P): Redefine.
8991 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
8992 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
8993 (m32r_modes_tieable_p): New function.
8994 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
8995 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
8996 (m68k_modes_tieable_p): New function.
8997 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
8998 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
8999 (mcore_modes_tieable_p): New function.
9000 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
9001 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
9003 (TARGET_MODES_TIEABLE_P): Redefine.
9004 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
9005 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
9006 * config/mips/mips.c (mips_modes_tieable_p): Make static.
9007 (TARGET_MODES_TIEABLE_P): Redefine.
9008 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
9009 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
9010 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
9011 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
9012 (mn10300_modes_tieable_p): ...this and make static.
9013 (TARGET_MODES_TIEABLE_P): Redefine.
9014 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
9015 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
9016 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
9017 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
9018 (msp430_modes_tieable_p): Make static.
9019 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
9020 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
9021 (TARGET_MODES_TIEABLE_P): Redefine.
9022 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
9023 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
9024 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
9025 (TARGET_MODES_TIEABLE_P): Redefine.
9026 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
9027 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
9028 * config/pa/pa.c (pa_modes_tieable_p): Make static.
9029 (TARGET_MODES_TIEABLE_P): Redefine.
9030 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
9031 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
9032 (pdp11_modes_tieable_p): New function.
9033 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
9034 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
9035 (rs6000_modes_tieable_p): New function.
9036 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9037 * config/powerpcspe/powerpcspe.md: Update comment.
9038 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
9039 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
9040 (TARGET_MODES_TIEABLE_P): Redefine.
9041 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
9042 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
9043 (rl78_modes_tieable_p): New function.
9044 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
9045 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
9046 (rs6000_modes_tieable_p): New function.
9047 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9048 * config/rs6000/rs6000.md: Update comment.
9049 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
9050 * config/rx/rx.c (rx_modes_tieable_p): New function.
9051 (TARGET_MODES_TIEABLE_P): Redefine.
9052 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
9053 * config/s390/s390.c (s390_modes_tieable_p): New function.
9054 (TARGET_MODES_TIEABLE_P): Redefine.
9055 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
9056 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
9057 (sh_modes_tieable_p): New function.
9058 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
9059 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
9060 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
9061 (sparc_modes_tieable_p): Make static.
9062 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
9063 * config/spu/spu.c (spu_modes_tieable_p): New function.
9064 (TARGET_MODES_TIEABLE_P): Redefine.
9065 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
9066 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
9067 (TARGET_MODES_TIEABLE_P): Redefine.
9068 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
9069 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
9070 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
9071 * config/v850/v850.c (v850_modes_tieable_p): New function.
9072 (TARGET_MODES_TIEABLE_P): Redefine.
9073 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
9074 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
9075 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
9076 (visium_modes_tieable_p): New function.
9077 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
9078 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
9079 (xtensa_modes_tieable_p): New function.
9080 * system.h (MODES_TIEABLE_P): Poison.
9082 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9083 Alan Hayward <alan.hayward@arm.com>
9084 David Sherwood <david.sherwood@arm.com>
9086 * target.def (hard_regno_mode_ok): New hook.
9087 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
9088 (TARGET_HARD_REGNO_MODE_OK): ...this.
9089 * doc/tm.texi.in: Regenerate.
9090 * hooks.h (hook_bool_uint_mode_true): Declare.
9091 * hooks.c (hook_bool_uint_mode_true): New function.
9092 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
9094 * genpreds.c (write_insn_preds_c): Add an include of target.h.
9095 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
9096 instead of HARD_REGNO_MODE_OK.
9097 * caller-save.c: Include target.h.
9098 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
9100 * combine.c (can_combine_p): Likewise.
9101 (combinable_i3pat): Likewise.
9102 (can_change_dest_mode): Likewise.
9103 * expr.c (init_expr_target): Likewise.
9104 (convert_move): Likewise.
9105 (convert_modes): Likewise.
9106 * ira.c (setup_prohibited_class_mode_regs): Likewise.
9107 (setup_prohibited_mode_move_regs): Likewise.
9108 * ira.h (target_ira): Likewise.
9109 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9110 * lra-constraints.c (process_alt_operands): Likewise.
9111 (split_reg): Likewise.
9112 * recog.c (peep2_find_free_register): Likewise.
9113 * ree.c (combine_reaching_defs): Likewise.
9114 * regcprop.c (maybe_mode_change): Likewise.
9115 * reginfo.c (init_reg_sets_1): Likewise.
9116 (choose_hard_reg_mode): Likewise.
9117 (simplifiable_subregs): Likewise.
9118 * regrename.c (check_new_reg_p): Likewise.
9119 * reload.c (find_valid_class): Likewise.
9120 (find_valid_class_1): Likewise.
9121 (reload_inner_reg_of_subreg): Likewise.
9122 (push_reload): Likewise.
9123 (combine_reloads): Likewise.
9124 (find_dummy_reload): Likewise.
9125 (find_reloads): Likewise.
9126 * reload1.c (find_reg): Likewise.
9127 (set_reload_reg): Likewise.
9128 (allocate_reload_reg): Likewise.
9129 (choose_reload_regs): Likewise.
9130 (reload_adjust_reg_for_temp): Likewise.
9131 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
9132 (simplify_subreg_regno): Likewise.
9133 * sel-sched.c (init_regs_for_mode): Likewise.
9134 * varasm.c (make_decl_rtl): Likewise.
9135 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
9136 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
9138 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
9139 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
9140 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9141 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
9142 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
9143 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9144 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
9145 (arc_mode_class): Delete.
9146 (HARD_REGNO_MODE_OK): Delete.
9147 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9148 (arc_hard_regno_mode_ok): Rename old array to...
9149 (arc_hard_regno_mode_ok_modes): ...this.
9150 (arc_conditional_register_usage): Update accordingly.
9151 (arc_mode_class): Make static.
9152 (arc_hard_regno_mode_ok): New function.
9153 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
9154 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
9155 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9156 (arm_hard_regno_mode_ok): Make static.
9157 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
9159 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
9160 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
9161 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
9163 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9164 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
9165 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
9166 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
9167 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
9168 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9169 * config/bfin/predicates.md (valid_reg_operand): Use
9170 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
9171 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
9172 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
9173 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9174 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
9175 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
9176 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9177 (cr16_hard_regno_mode_ok): Make static and return a bool.
9178 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
9179 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9180 (cris_hard_regno_mode_ok): New function.
9181 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
9182 (epiphany_mode_class): Delete.
9183 (HARD_REGNO_MODE_OK): Delete.
9184 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
9185 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9186 (hard_regno_mode_ok): Rename to...
9187 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
9188 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
9189 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
9191 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
9192 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
9193 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9194 (frv_hard_regno_mode_ok): Make static and return a bool.
9195 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
9197 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
9198 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
9199 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
9200 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
9202 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9203 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
9204 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
9205 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
9207 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9208 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
9209 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9210 (ia64_hard_regno_mode_ok): New function.
9211 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
9212 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9213 (iq2000_hard_regno_mode_ok): New function.
9214 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
9215 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9216 (lm32_hard_regno_mode_ok): New function.
9217 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
9218 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
9219 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
9220 instead of HARD_REGNO_MODE_OK.
9221 (m32c_hard_regno_ok): Rename to...
9222 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
9223 (m32c_cannot_change_mode_class): Update accordingly.
9224 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9225 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
9226 (m32r_mode_class): Delete.
9227 (HARD_REGNO_MODE_OK): Delete.
9228 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9229 (m32r_hard_regno_mode_ok): Rename to...
9230 (m32r_hard_regno_modes): ...this.
9231 (m32r_mode_class): Make static.
9232 (m32r_hard_regno_mode_ok): New function.
9233 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
9234 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
9235 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9236 (m68k_hard_regno_mode_ok): Make static.
9237 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
9238 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9239 (mcore_hard_regno_mode_ok): New function.
9240 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
9241 (HARD_REGNO_MODE_OK): Delete.
9242 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
9244 (microblaze_hard_regno_mode_ok_p): ...this and make static.
9245 (microblaze_hard_regno_mode_ok): New function.
9246 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9247 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
9248 (mips_hard_regno_mode_ok): Delete.
9249 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
9250 (mips_hard_regno_mode_ok_p): ...this and make static.
9251 (mips_hard_regno_mode_ok_p): Rename to...
9252 (mips_hard_regno_mode_ok_uncached): ...this.
9253 (mips_hard_regno_mode_ok): New function.
9254 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
9255 of HARD_REGNO_MODE_OK.
9256 (mips_option_override): Update after above name changes.
9257 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9258 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
9259 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
9260 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
9261 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
9262 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9263 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
9264 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
9265 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
9266 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9267 (msp430_hard_regno_mode_ok): Make static and return a bool.
9268 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
9269 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
9270 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
9272 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9273 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
9274 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
9275 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
9276 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
9277 (PA_HARD_REGNO_MODE_OK): ...this
9278 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
9279 (PA_HARD_REGNO_MODE_OK): ...this.
9280 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9281 (pa_hard_regno_mode_ok): New function.
9282 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
9283 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9284 (pdp11_hard_regno_mode_ok): New function.
9285 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
9286 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
9288 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
9290 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9291 (rs6000_hard_regno_mode_ok): Rename to...
9292 (rs6000_hard_regno_mode_ok_uncached): ...this.
9293 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9294 (rs6000_hard_regno_mode_ok): New function.
9295 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
9296 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
9297 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
9298 (riscv_hard_regno_mode_ok): ...this and make static.
9299 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9300 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
9301 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
9302 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9303 (rl78_hard_regno_mode_ok): Make static and return bool.
9304 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
9305 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
9307 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
9308 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9309 (rs6000_hard_regno_mode_ok): Rename to...
9310 (rs6000_hard_regno_mode_ok_uncached): ...this.
9311 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9312 (rs6000_hard_regno_mode_ok): New function.
9313 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
9314 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
9315 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9316 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
9317 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
9318 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
9319 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9320 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
9321 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
9322 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9323 (sh_hard_regno_mode_ok): Make static.
9324 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
9325 instead of HARD_REGNO_MODE_OK.
9326 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
9327 (sparc_mode_class): Delete.
9328 (HARD_REGNO_MODE_OK): Delete.
9329 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9330 (hard_regno_mode_classes): Make static.
9331 (sparc_mode_class): Likewise.
9332 (sparc_hard_regno_mode_ok): New function.
9333 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
9334 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
9335 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
9337 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9338 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
9339 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
9340 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
9341 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
9342 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9343 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
9344 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
9345 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9346 (visium_hard_regno_mode_ok): New function.
9347 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
9348 instead of HARD_REGNO_MODE_OK.
9349 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
9350 (HARD_REGNO_MODE_OK): Delete.
9351 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
9352 (xtensa_hard_regno_mode_ok_p): ...this and make static.
9353 (xtensa_option_override): Update accordingly.
9354 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9355 (xtensa_hard_regno_mode_ok): New function.
9356 * system.h (HARD_REGNO_MODE_OK): Poison.
9358 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9359 Alan Hayward <alan.hayward@arm.com>
9360 David Sherwood <david.sherwood@arm.com>
9362 * target.def (hard_regno_call_part_clobbered): New hook.
9363 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
9364 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
9365 * doc/tm.texi: Regenerate.
9366 * hooks.h (hook_bool_uint_mode_false): Declare.
9367 * hooks.c (hook_bool_uint_mode_false): New function.
9368 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9369 * cselib.c (cselib_process_insn): Use
9370 targetm.hard_regno_call_part_clobbered instead of
9371 HARD_REGNO_CALL_PART_CLOBBERED.
9372 * ira-conflicts.c (ira_build_conflicts): Likewise.
9373 * ira-costs.c (ira_tune_allocno_costs): Likewise.
9374 * lra-constraints.c (need_for_call_save_p): Likewise.
9375 * lra-lives.c: Include target.h.
9376 (check_pseudos_live_through_calls): Use
9377 targetm.hard_regno_call_part_clobbered instead of
9378 HARD_REGNO_CALL_PART_CLOBBERED.
9379 * regcprop.c: Include target.h.
9380 (copyprop_hardreg_forward_1): Use
9381 targetm.hard_regno_call_part_clobbered instead of
9382 HARD_REGNO_CALL_PART_CLOBBERED.
9383 * reginfo.c (choose_hard_reg_mode): Likewise.
9384 * regrename.c (check_new_reg_p): Likewise.
9385 * reload.c (find_equiv_reg): Likewise.
9386 * reload1.c (emit_reload_insns): Likewise.
9387 * sched-deps.c (deps_analyze_insn): Likewise.
9388 * sel-sched.c (init_regs_for_mode): Likewise.
9389 (mark_unavailable_hard_regs): Likewise.
9390 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
9391 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9392 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
9394 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9395 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9396 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
9398 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
9400 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9401 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9402 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
9404 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9405 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9406 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
9408 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9409 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
9411 * config/powerpcspe/powerpcspe.c
9412 (rs6000_hard_regno_call_part_clobbered): New function.
9413 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9414 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9415 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
9417 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9418 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9419 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
9421 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9422 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9423 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
9425 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9426 Alan Hayward <alan.hayward@arm.com>
9427 David Sherwood <david.sherwood@arm.com>
9429 * rtl.h (subreg_memory_offset): Declare.
9430 * emit-rtl.c (subreg_memory_offset): New function.
9431 * expmed.c (store_bit_field_1): Use it.
9432 * expr.c (undefined_operand_subword_p): Likewise.
9433 * simplify-rtx.c (simplify_subreg): Likewise.
9435 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
9437 PR rtl-optimization/57448
9440 * optabs.c (expand_atomic_load): Place compiler memory barriers if
9441 using atomic_load pattern.
9442 (expand_atomic_store): Likewise.
9444 2017-09-04 Jakub Jelinek <jakub@redhat.com>
9447 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
9448 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
9449 handling. Use replace_call_with_value with NULL instead of
9450 gsi_replace, unlink_stmt_vdef and release_defs.
9452 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
9455 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
9457 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9460 * rtl.h (emit_library_call_value_1): Declare.
9461 (emit_library_call): Replace declaration with a series of overloads.
9462 Remove the parameter count argument.
9463 (emit_library_call_value): Likewise.
9464 * calls.c (emit_library_call_value_1): Make global. Replace varargs
9465 with an "rtx_mode_t *".
9466 (emit_library_call_value): Delete.
9467 (emit_library_call): Likewise.
9468 * asan.c (asan_emit_stack_protection): Update calls accordingly.
9469 (asan_emit_allocas_unpoison): Likewise.
9470 * builtins.c (expand_builtin_powi): Likewise.
9471 (expand_asan_emit_allocas_unpoison): Likewise.
9472 * cfgexpand.c (expand_main_function): Likewise.
9473 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
9474 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
9475 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
9476 * config/arm/arm.c (arm_trampoline_init): Likewise.
9477 (arm_call_tls_get_addr): Likewise.
9478 (arm_expand_divmod_libfunc): Likewise.
9479 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
9480 (smulsi3_highpart): Likewise.
9481 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
9482 (c6x_expand_compare): Likewise.
9483 (c6x_expand_movmem): Likewise.
9484 * config/frv/frv.c (frv_trampoline_init): Likewise.
9485 * config/i386/i386.c (ix86_trampoline_init): Likewise.
9486 (ix86_expand_divmod_libfunc): Likewise.
9487 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
9488 (ia64_expand_compare): Likewise.
9489 (ia64_profile_hook): Likewise.
9490 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
9491 (nonlocal_goto): Likewise.
9492 (restore_stack_nonlocal): Likewise.
9493 * config/m32r/m32r.c (block_move_call): Likewise.
9494 (m32r_trampoline_init): Likewise.
9495 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
9496 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
9497 (m68k_call_m68k_read_tp): Likewise.
9498 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
9499 (microblaze_expand_divide): Likewise.
9500 * config/mips/mips.h (mips_args): Likewise.
9501 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
9502 (MIPS_ICACHE_SYNC): Likewise.
9503 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
9504 (nios2_trampoline_init): Likewise.
9505 * config/pa/pa.c (hppa_tls_call): Likewise.
9506 (pa_trampoline_init): Likewise.
9507 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
9508 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
9509 (expand_strn_compare): Likewise.
9510 (rs6000_generate_compare): Likewise.
9511 (rs6000_expand_float128_convert): Likewise.
9512 (output_profile_hook): Likewise.
9513 (rs6000_trampoline_init): Likewise.
9514 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
9515 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
9516 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
9517 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
9518 (rs6000_generate_compare): Likewise.
9519 (rs6000_expand_float128_convert): Likewise.
9520 (output_profile_hook): Likewise.
9521 (rs6000_trampoline_init): Likewise.
9522 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
9523 * config/sh/sh.c (sh_trampoline_init): Likewise.
9524 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
9525 (sparc_emit_float_lib_cmp): Likewise.
9526 (sparc32_initialize_trampoline): Likewise.
9527 (sparc64_initialize_trampoline): Likewise.
9528 (sparc_profile_hook): Likewise.
9529 * config/spu/spu.c (ea_load_store): Likewise.
9530 * config/spu/spu.md (floatunssidf2): Likewise.
9531 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
9532 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
9533 * config/visium/visium.c (expand_block_move_4): Likewise.
9534 (expand_block_move_2): Likewise.
9535 (expand_block_move_1): Likewise.
9536 (expand_block_set_4): Likewise.
9537 (expand_block_set_2): Likewise.
9538 (expand_block_set_1): Likewise.
9539 (visium_trampoline_init): Likewise.
9540 (visium_profile_hook): Likewise.
9541 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
9542 (xtensa_setup_frame_addresses): Likewise.
9543 (xtensa_trampoline_init): Likewise.
9544 * except.c (sjlj_emit_function_enter): Likewise.
9545 (sjlj_emit_function_exit): Likewise.
9546 * explow.c (allocate_dynamic_stack_space): Likewise.
9547 (probe_stack_range): Likewise.
9548 * expr.c (convert_mode_scalar): Likewise.
9549 * optabs.c (expand_binop): Likewise.
9550 (expand_twoval_binop_libfunc): Likewise.
9551 (expand_unop): Likewise.
9552 (prepare_cmp_insn): Likewise.
9553 (prepare_float_lib_cmp): Likewise.
9554 (expand_float): Likewise.
9555 (expand_fix): Likewise.
9556 (expand_fixed_convert): Likewise.
9557 (maybe_emit_sync_lock_test_and_set): Likewise.
9558 (expand_atomic_compare_and_swap): Likewise.
9559 (expand_mem_thread_fence): Likewise.
9560 (expand_atomic_fetch_op): Likewise.
9562 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
9564 * doc/generic.texi (OpenACC): Adjust URL.
9565 * doc/invoke.texi (C Dialect Options): Ditto.
9567 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
9569 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
9570 predicate for operand 1. Add (m,<S>) constraint.
9571 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
9572 Prevent memory operand 1 with register operand 2.
9574 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
9576 PR rtl-optimization/82024
9577 * combine.c (try_combine): If the combination result is a PARALLEL,
9578 and we only need to retain the SET in there that would be placed
9579 at I2, check that we can place that at I3 instead, before doing so.
9581 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9584 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
9586 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
9587 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
9590 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
9591 Jeff Law <law@redhat.com>
9593 * varasm.c (bss_initializer_p): Do not put constants into .bss
9594 (categorize_decl_for_section): Handle bss_initializer_p returning
9595 false when DECL_INITIAL is NULL.
9597 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9600 * config/s390/s390.c (s390_can_inline_p): New function.
9602 2017-09-01 Jeff Law <law@redhat.com>
9604 PR tree-optimization/82052
9605 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
9606 Always initialize the returned slot after a hash table miss
9607 when INSERT is true.
9609 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
9611 * config/s390/s390.md (mem_signal_fence): Remove.
9612 * doc/md.texi (mem_signal_fence): Remove.
9613 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
9615 * target-insns.def (mem_signal_fence): Remove.
9617 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9620 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
9623 * asan.c (create_odr_indicator): Strip name encoding from assembler
9624 name before appending it after __odr_asan_.
9626 2017-09-01 Martin Liska <mliska@suse.cz>
9628 PR tree-optimization/82059
9629 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
9630 frequency only when an edge is redirected.
9632 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9634 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
9635 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
9636 (arc_conditional_register_usage): Remove ARC600 lp_count
9638 (arc_file_start): Emit Tag_ARC_CPU_variation.
9639 (arc_can_use_doloop_p): New conditions to use ZOLs.
9640 (hwloop_fail): New function.
9641 (hwloop_optimize): Likewise.
9642 (hwloop_pattern_reg): Likewise.
9643 (arc_doloop_hooks): New struct, to be used with reorg_loops.
9644 (arc_reorg_loops): New function, calls reorg_loops.
9645 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
9646 (arc600_corereg_hazard): Remove ZOL checking, case handled by
9648 (arc_loop_hazard): Remove function, functionality moved into
9650 (arc_hazard): Remove arc_loop_hazard call.
9651 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
9652 into hwloop_optimize.
9653 (arc_label_align): Remove ZOL handling.
9654 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
9655 * config/arc/arc.md (doloop_begin): Remove pattern.
9656 (doloop_begin_i): Likewise.
9657 (doloop_end_i): Likewise.
9658 (doloop_fallback): Likewise.
9659 (doloop_fallback_m): Likewise.
9660 (doloop_end): Reimplement expand.
9661 (arc_lp): New pattern for LP instruction.
9662 (loop_end): New pattern.
9663 (loop_fail): Likewise.
9664 (decrement_and_branch_until_zero): Likewise.
9665 * config/arc/arc.opt (mlpc-width): New option.
9666 * doc/invoke.texi (mlpc-width): Document option.
9668 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9670 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
9671 (arc_ccfsm_advance): Fix checking for delay slots.
9672 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
9674 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9676 * config/arc/arc.md (movqi_insn): Add stores to save constant long
9678 (movhi_insn): Update store instruction constraint which are saving
9679 6-bit short immediates.
9680 (movsi_insn): Consider also short scaled load operations.
9681 (zero_extendhisi2_i): Use Usd constraint instead of T.
9682 (extendhisi2_i): Add q constraint.
9683 (arc_clzsi2): Add type and length attributes.
9684 (arc_ctzsi2): Likewise.
9685 * config/arc/constraints.md (Usc): Update constraint, the
9686 assembler can parse two relocations for a single instruction.
9688 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9690 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
9691 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
9693 2017-08-31 Olivier Hainque <hainque@adacore.com>
9695 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
9696 match as powerpc-wrs-vxworks*.
9698 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
9700 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
9701 register constraint for by-element operand.
9702 (aarch64_mls_elt_merge<mode>): Likewise.
9704 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9706 * config/arc/arc.c (arc_can_follow_jump): Check for short
9709 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9711 * config.gcc: Use g.opt for arc.
9712 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
9713 functionality moved to ...
9714 (legitimate_scaled_address_p): New function, ...here.
9715 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
9716 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
9717 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
9719 (arc_override_options): Handle G option.
9720 (arc_output_pic_addr_const): Correct function definition.
9721 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
9722 (arc_decl_anon_ns_mem_p): Delete.
9723 (arc_in_small_data_p): Overhaul this function to take into
9724 consideration the value given via G option.
9725 (arc_rewrite_small_data_1): Renamed and corrected old
9726 arc_rewrite_small_data function.
9727 (arc_rewrite_small_data): New function.
9728 (small_data_pattern): Don't use pic_offset_table_rtx.
9729 * config/arc/arc.h (CC1_SPEC): Recognize G option.
9730 * config/arc/simdext.md (movmisalignv2hi): Use
9731 prepare_move_operands function.
9733 (movmisalign*): Likewise.
9734 * doc/invoke.texi (ARC options): Document -G option.
9736 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9738 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
9740 * config/arc/arc.c (arc_print_operand): Output scalled address for
9741 sdata whenever is possible.
9742 (arc_in_small_data_p): Allow sdata for 64bit datum when double
9743 load/stores are available.
9744 (compact_sda_memory_operand): Check for the alignment required by
9745 code density instructions.
9746 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
9748 * config/arc/constraints.md (Usd): Update constraint.
9749 (Us0): New constraint.
9750 (Usc): Update constraint.
9752 2017-08-31 Richard Biener <rguenther@suse.de>
9755 * dwarf2out.c (dwarf2out_early_global_decl): Process each
9758 2017-08-31 Tamar Christina <tamar.christina@arm.com>
9760 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
9761 Resize type_signature.
9763 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9764 Alan Hayward <alan.hayward@arm.com>
9765 David Sherwood <david.sherwood@arm.com>
9767 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
9768 subregs whose inner modes can be stored in GPRs.
9769 (aarch64_classify_index): Likewise.
9771 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9772 Alan Hayward <alan.hayward@arm.com>
9773 David Sherwood <david.sherwood@arm.com>
9775 * config/aarch64/iterators.md (V_cmp_result): Rename to...
9776 (V_INT_EQUIV): ...this.
9777 (v_cmp_result): Rename to...
9778 (v_int_equiv): ...this.
9779 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
9780 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
9781 (copysign<mode>3): Likewise.
9782 (aarch64_simd_bsl<mode>_internal): Likewise.
9783 (aarch64_simd_bsl<mode>): Likewise.
9784 (vec_cmp<mode><mode>): Likewise.
9785 (vcond<mode><mode>): Likewise.
9786 (vcond<v_cmp_mixed><mode>): Likewise.
9787 (vcondu<mode><v_cmp_mixed>): Likewise.
9788 (aarch64_cm<optab><mode>): Likewise.
9789 (aarch64_cmtst<mode>): Likewise.
9790 (aarch64_fac<optab><mode>): Likewise.
9791 (vec_perm_const<mode>): Likewise.
9792 (vcond_mask_<mode><v_cmp_result>): Rename to...
9793 (vcond_mask_<mode><v_int_equiv>): ...this.
9794 (vec_cmp<mode><v_cmp_result>): Rename to...
9795 (vec_cmp<mode><v_int_equiv>): ...this.
9797 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9798 Alan Hayward <alan.hayward@arm.com>
9799 David Sherwood <david.sherwood@arm.com>
9801 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
9803 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
9804 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
9805 (UNSPEC_LD4_DREG): New unspecs.
9806 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
9807 (aarch64_ld2<mode>_dreg_be): Replace with...
9808 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
9810 (aarch64_ld3<mode>_dreg_le)
9811 (aarch64_ld3<mode>_dreg_be): Replace with...
9812 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
9814 (aarch64_ld4<mode>_dreg_le)
9815 (aarch64_ld4<mode>_dreg_be): Replace with...
9816 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
9819 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9821 PR tree-optimization/81987
9822 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
9823 insert an initializer in a location not dominated by the stride
9826 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
9828 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
9829 on the entire header of the finally block in the fallthru case.
9831 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
9833 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
9835 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
9837 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
9838 rs6000_emit_move_from_cr and call renamed function.
9839 (rs6000_emit_prologue): Call renamed functions.
9840 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
9841 movesi_from_cr, remove volatile CRs.
9843 2017-08-30 Jon Beniston <jon@beniston.com>
9844 Richard Biener <rguenther@suse.de>
9846 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
9847 of VECTOR_MODE_P check.
9848 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
9849 element vector types.
9851 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9853 * df.h (df_read_modify_subreg_p): Remove in favor of...
9854 * rtl.h (read_modify_subreg_p): ...this new function. Take a
9855 const_rtx instead of an rtx.
9856 * cprop.c (local_cprop_find_used_regs): Update accordingly.
9857 * df-problems.c (df_word_lr_mark_ref): Likewise.
9858 * ira-lives.c (mark_pseudo_reg_live): Likewise.
9859 (mark_pseudo_reg_dead): Likewise.
9860 (mark_ref_dead): Likewise.
9861 * reginfo.c (init_subregs_of_mode): Likewise.
9862 * sched-deps.c (sched_analyze_1): Likewise.
9863 * df-scan.c (df_def_record_1): Likewise.
9864 (df_uses_record): Likewise.
9865 (df_read_modify_subreg_p): Remove in favor of...
9866 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
9867 const_rtx instead of an rtx.
9869 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9870 Alan Hayward <alan.hayward@arm.com>
9871 David Sherwood <david.sherwood@arm.com>
9873 * rtl.h (partial_subreg_p): New function.
9874 * caller-save.c (save_call_clobbered_regs): Use it.
9875 * calls.c (expand_call): Likewise.
9876 * combine.c (combinable_i3pat): Likewise.
9877 (simplify_set): Likewise.
9878 (make_extraction): Likewise.
9879 (make_compound_operation_int): Likewise.
9880 (gen_lowpart_or_truncate): Likewise.
9881 (force_to_mode): Likewise.
9882 (make_field_assignment): Likewise.
9883 (reg_truncated_to_mode): Likewise.
9884 (record_truncated_value): Likewise.
9885 (move_deaths): Likewise.
9886 * cse.c (record_jump_cond): Likewise.
9887 (cse_insn): Likewise.
9888 * cselib.c (cselib_lookup_1): Likewise.
9889 * expmed.c (extract_bit_field_using_extv): Likewise.
9890 * function.c (assign_parm_setup_reg): Likewise.
9891 * ifcvt.c (noce_convert_multiple_sets): Likewise.
9892 * ira-build.c (create_insn_allocnos): Likewise.
9893 * lra-coalesce.c (merge_pseudos): Likewise.
9894 * lra-constraints.c (match_reload): Likewise.
9895 (simplify_operand_subreg): Likewise.
9896 (curr_insn_transform): Likewise.
9897 * lra-lives.c (process_bb_lives): Likewise.
9898 * lra.c (new_insn_reg): Likewise.
9899 (lra_substitute_pseudo): Likewise.
9900 * regcprop.c (mode_change_ok): Likewise.
9901 (maybe_mode_change): Likewise.
9902 (copyprop_hardreg_forward_1): Likewise.
9903 * reload.c (push_reload): Likewise.
9904 (find_reloads): Likewise.
9905 (find_reloads_subreg_address): Likewise.
9906 * reload1.c (alter_reg): Likewise.
9907 (eliminate_regs_1): Likewise.
9908 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9910 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
9912 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
9913 back to if statements, including unpack.
9915 2017-08-30 Martin Liska <mliska@suse.cz>
9918 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
9920 (func_checker::compare_asm_inputs_outputs): ... this function.
9921 (func_checker::compare_gimple_asm): Use the function to compare
9922 also ASM constrains.
9923 * ipa-icf-gimple.h: Rename the function.
9925 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9926 Alan Hayward <alan.hayward@arm.com>
9927 David Sherwood <david.sherwood@arm.com>
9929 * coretypes.h (complex_mode): New type.
9930 * gdbhooks.py (build_pretty_printer): Handle it.
9931 * machmode.h (complex_mode): New class.
9932 (complex_mode::includes_p): New function.
9933 (is_complex_int_mode): Likewise.
9934 (is_complex_float_mode): Likewise.
9935 * genmodes.c (get_mode_class): Handle complex mode classes.
9936 * function.c (expand_function_end): Use is_complex_int_mode.
9938 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9939 Alan Hayward <alan.hayward@arm.com>
9940 David Sherwood <david.sherwood@arm.com>
9942 * coretypes.h (scalar_mode_pod): New typedef.
9943 * gdbhooks.py (build_pretty_printer): Handle it.
9944 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
9945 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
9946 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
9947 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
9950 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9951 Alan Hayward <alan.hayward@arm.com>
9952 David Sherwood <david.sherwood@arm.com>
9954 * machmode.h (mode_for_vector): Take a scalar_mode instead
9956 * stor-layout.c (mode_for_vector): Likewise.
9957 * explow.c (promote_mode): Use as_a <scalar_mode>.
9958 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
9960 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9961 Alan Hayward <alan.hayward@arm.com>
9962 David Sherwood <david.sherwood@arm.com>
9964 * target.def (preferred_simd_mode): Take a scalar_mode
9965 instead of a machine_mode.
9966 * targhooks.h (default_preferred_simd_mode): Likewise.
9967 * targhooks.c (default_preferred_simd_mode): Likewise.
9968 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
9969 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
9970 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
9971 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
9972 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
9973 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
9974 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
9975 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
9977 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
9978 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
9979 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
9980 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
9981 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
9982 * doc/tm.texi: Regenerate.
9983 * optabs-query.c (can_vec_mask_load_store_p): Return false for
9986 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9987 Alan Hayward <alan.hayward@arm.com>
9988 David Sherwood <david.sherwood@arm.com>
9990 * target.def (scalar_mode_supported_p): Take a scalar_mode
9991 instead of a machine_mode.
9992 * targhooks.h (default_scalar_mode_supported_p): Likewise.
9993 * targhooks.c (default_scalar_mode_supported_p): Likewise.
9994 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
9995 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
9996 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
9997 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
9998 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
9999 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
10000 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
10001 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
10002 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
10003 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
10004 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
10005 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
10007 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
10008 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
10009 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
10010 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
10011 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
10013 * doc/tm.texi: Regenerate.
10015 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10016 Alan Hayward <alan.hayward@arm.com>
10017 David Sherwood <david.sherwood@arm.com>
10019 * coretypes.h (opt_scalar_mode): New typedef.
10020 * gdbhooks.py (build_pretty_printers): Handle it.
10021 * machmode.h (mode_iterator::get_2xwider): Add overload for
10023 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
10025 * expr.c (convert_mode_scalar): Likewise.
10026 * omp-low.c (omp_clause_aligned_alignment): Likewise.
10027 * optabs.c (expand_float): Likewise.
10028 (expand_fix): Likewise.
10029 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10031 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10032 Alan Hayward <alan.hayward@arm.com>
10033 David Sherwood <david.sherwood@arm.com>
10035 * optabs.c (expand_float): Explicitly check for scalars before
10036 using a branching expansion.
10037 (expand_fix): Likewise.
10039 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10040 Alan Hayward <alan.hayward@arm.com>
10041 David Sherwood <david.sherwood@arm.com>
10043 * expr.c (convert_mode): Split scalar handling out into...
10044 (convert_mode_scalar): ...this new function. Treat the modes
10047 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10048 Alan Hayward <alan.hayward@arm.com>
10049 David Sherwood <david.sherwood@arm.com>
10051 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
10053 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
10055 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10056 Alan Hayward <alan.hayward@arm.com>
10057 David Sherwood <david.sherwood@arm.com>
10059 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
10060 rather than a machine_mode.
10061 (fixed_from_string): Likewise.
10062 (fixed_convert): Likewise.
10063 (fixed_convert_from_int): Likewise.
10064 (fixed_convert_from_real): Likewise.
10065 (real_convert_from_fixed): Likewise.
10066 * fixed-value.c (fixed_from_double_int): Likewise.
10067 (fixed_from_string): Likewise.
10068 (fixed_convert): Likewise.
10069 (fixed_convert_from_int): Likewise.
10070 (fixed_convert_from_real): Likewise.
10071 (real_convert_from_fixed): Likewise.
10072 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
10074 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10075 Alan Hayward <alan.hayward@arm.com>
10076 David Sherwood <david.sherwood@arm.com>
10078 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
10079 of separate mode class checks. Do not allow vector modes here.
10080 (immed_wide_int_const): Use as_a <scalar_mode>.
10081 * explow.c (trunc_int_for_mode): Likewise.
10082 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
10083 (wi::shwi): Likewise.
10084 (wi::min_value): Likewise.
10085 (wi::max_value): Likewise.
10086 * dwarf2out.c (loc_descriptor): Likewise.
10087 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
10088 for CONST_WIDE_INT.
10090 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10091 Alan Hayward <alan.hayward@arm.com>
10092 David Sherwood <david.sherwood@arm.com>
10094 * tree.h (SCALAR_TYPE_MODE): New macro.
10095 * expr.c (expand_expr_addr_expr_1): Use it.
10096 (expand_expr_real_2): Likewise.
10097 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
10098 (fold_convert_const_fixed_from_int): Likewise.
10099 (fold_convert_const_fixed_from_real): Likewise.
10100 (native_encode_fixed): Likewise
10101 (native_encode_complex): Likewise
10102 (native_encode_vector): Likewise.
10103 (native_interpret_fixed): Likewise.
10104 (native_interpret_real): Likewise.
10105 (native_interpret_complex): Likewise.
10106 (native_interpret_vector): Likewise.
10107 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
10108 (simd_clone_adjust_argument_types): Likewise.
10109 (simd_clone_init_simd_arrays): Likewise.
10110 (simd_clone_adjust): Likewise.
10111 * stor-layout.c (layout_type): Likewise.
10112 * tree.c (build_minus_one_cst): Likewise.
10113 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
10114 * tree-inline.c (estimate_move_cost): Likewise.
10115 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
10116 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
10117 (vectorizable_reduction): Likewise.
10118 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
10119 (vect_recog_mixed_size_cond_pattern): Likewise.
10120 (check_bool_pattern): Likewise.
10121 (adjust_bool_pattern): Likewise.
10122 (search_type_for_mask_1): Likewise.
10123 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
10124 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10125 (vectorizable_load): Likewise.
10126 (vectorizable_store): Likewise.
10127 * ubsan.c (ubsan_encode_value): Likewise.
10128 * varasm.c (output_constant): Likewise.
10130 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10131 Alan Hayward <alan.hayward@arm.com>
10132 David Sherwood <david.sherwood@arm.com>
10134 * coretypes.h (scalar_mode): New class.
10135 * machmode.h (scalar_mode): Likewise.
10136 (scalar_mode::includes_p): New function.
10137 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
10138 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
10139 * genmodes.c (get_mode_class): Handle remaining scalar modes.
10140 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
10141 * expmed.c (store_bit_field_1): Likewise.
10142 (extract_bit_field_1): Likewise.
10143 * expr.c (write_complex_part): Likewise.
10144 (read_complex_part): Likewise.
10145 (emit_move_complex_push): Likewise.
10146 (expand_expr_real_2): Likewise.
10147 * function.c (assign_parm_setup_reg): Likewise.
10148 (assign_parms_unsplit_complex): Likewise.
10149 * optabs.c (expand_binop): Likewise.
10150 * rtlanal.c (subreg_get_info): Likewise.
10151 * simplify-rtx.c (simplify_immed_subreg): Likewise.
10152 * varasm.c (output_constant_pool_2): Likewise.
10154 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10155 Alan Hayward <alan.hayward@arm.com>
10156 David Sherwood <david.sherwood@arm.com>
10158 * expmed.c (extract_high_half): Use scalar_int_mode and remove
10160 (expmed_mult_highpart_optab): Likewise.
10161 (expmed_mult_highpart): Likewise.
10163 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10164 Alan Hayward <alan.hayward@arm.com>
10165 David Sherwood <david.sherwood@arm.com>
10167 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
10168 instead of a machine_mode.
10169 (builtin_memset_read_str): Likewise.
10170 * builtins.c (c_readstr): Likewise.
10171 (builtin_memcpy_read_str): Likewise.
10172 (builtin_strncpy_read_str): Likewise.
10173 (builtin_memset_read_str): Likewise.
10174 (builtin_memset_gen_str): Likewise.
10175 (expand_builtin_signbit): Use scalar_int_mode for local variables.
10176 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
10177 instead of a machine_mode.
10178 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
10180 (make_extraction): Likewise.
10181 (try_widen_shift_mode): Take and return scalar_int_modes instead
10183 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
10184 a scalar_int_mode instead of a machine_mode.
10185 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
10186 (avr_addr_space_pointer_mode): Likewise.
10187 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
10188 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
10189 (msp430_unwind_word_mode): Likewise.
10190 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
10191 (spu_addr_space_pointer_mode): Likewise.
10192 (spu_addr_space_address_mode): Likewise.
10193 (spu_libgcc_cmp_return_mode): Likewise.
10194 (spu_libgcc_shift_count_mode): Likewise.
10195 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
10196 (rl78_addr_space_pointer_mode): Likewise.
10197 (fl78_unwind_word_mode): Likewise.
10198 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
10200 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
10201 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
10202 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
10203 (mips_valid_pointer_mode): Likewise.
10204 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
10205 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
10206 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
10208 (ft32_addr_space_address_mode): Likewise.
10209 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
10210 scalar_int_mode instead of a machine_mode.
10211 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
10213 (m32c_addr_space_address_mode): Likewise.
10214 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
10215 (rs6000_eh_return_filter_mode): Likewise.
10216 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
10217 (rs6000_eh_return_filter_mode): Likewise.
10218 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
10219 (s390_libgcc_shift_count_mode): Likewise.
10220 (s390_unwind_word_mode): Likewise.
10221 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
10223 * target.def (mode_rep_extended): Likewise.
10224 (valid_pointer_mode): Likewise.
10225 (addr_space.valid_pointer_mode): Likewise.
10226 (eh_return_filter_mode): Return a scalar_int_mode rather than
10228 (libgcc_cmp_return_mode): Likewise.
10229 (libgcc_shift_count_mode): Likewise.
10230 (unwind_word_mode): Likewise.
10231 (addr_space.pointer_mode): Likewise.
10232 (addr_space.address_mode): Likewise.
10233 * doc/tm.texi: Regenerate.
10234 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
10236 (do_jump): Use scalar_int_mode for local variables.
10237 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
10238 rather than a machine_mode.
10239 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
10240 (scompare_loc_descriptor_wide): Likewise.
10241 (scompare_loc_descriptor_narrow): Likewise.
10242 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
10244 * except.c (sjlj_emit_dispatch_table): Likewise.
10245 (expand_builtin_eh_copy_values): Likewise.
10246 * explow.c (convert_memory_address_addr_space_1): Likewise.
10247 Take a scalar_int_mode rather than a machine_mode.
10248 (convert_memory_address_addr_space): Take a scalar_int_mode rather
10249 than a machine_mode.
10250 (memory_address_addr_space): Use scalar_int_mode for local variables.
10251 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
10252 rather than a machine_mode.
10253 * expmed.c (mask_rtx): Likewise.
10254 (init_expmed_one_conv): Likewise.
10255 (expand_mult_highpart_adjust): Likewise.
10256 (extract_high_half): Likewise.
10257 (expmed_mult_highpart_optab): Likewise.
10258 (expmed_mult_highpart): Likewise.
10259 (expand_smod_pow2): Likewise.
10260 (expand_sdiv_pow2): Likewise.
10261 (emit_store_flag_int): Likewise.
10262 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
10264 (extract_low_bits): Likewise.
10265 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
10267 * expr.c (pieces_addr::adjust): Likewise.
10268 (can_store_by_pieces): Likewise.
10269 (store_by_pieces): Likewise.
10270 (clear_by_pieces_1): Likewise.
10271 (expand_expr_addr_expr_1): Likewise.
10272 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
10273 (expand_expr_real_1): Likewise.
10274 (try_casesi): Likewise.
10275 * final.c (shorten_branches): Likewise.
10276 * fold-const.c (fold_convert_const_int_from_fixed): Change the
10277 type of "mode" to machine_mode.
10278 * internal-fn.c (expand_arith_overflow_result_store): Take a
10279 scalar_int_mode rather than a machine_mode.
10280 (expand_mul_overflow): Use scalar_int_mode for local variables.
10281 * loop-doloop.c (doloop_modify): Likewise.
10282 (doloop_optimize): Likewise.
10283 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
10284 than a machine_mode.
10285 (expand_doubleword_shift_condmove): Likewise.
10286 (expand_doubleword_shift): Likewise.
10287 (expand_doubleword_clz): Likewise.
10288 (expand_doubleword_popcount): Likewise.
10289 (expand_doubleword_parity): Likewise.
10290 (expand_absneg_bit): Use scalar_int_mode for local variables.
10291 (prepare_float_lib_cmp): Likewise.
10292 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
10293 rather than a machine_mode.
10294 (convert_memory_address_addr_space): Likewise.
10295 (get_mode_bounds): Likewise.
10296 (get_address_mode): Return a scalar_int_mode rather than a
10298 * rtlanal.c (get_address_mode): Likewise.
10299 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
10300 than a machine_mode.
10301 * targhooks.c (default_mode_rep_extended): Likewise.
10302 (default_valid_pointer_mode): Likewise.
10303 (default_addr_space_valid_pointer_mode): Likewise.
10304 (default_eh_return_filter_mode): Return a scalar_int_mode rather
10305 than a machine_mode.
10306 (default_libgcc_cmp_return_mode): Likewise.
10307 (default_libgcc_shift_count_mode): Likewise.
10308 (default_unwind_word_mode): Likewise.
10309 (default_addr_space_pointer_mode): Likewise.
10310 (default_addr_space_address_mode): Likewise.
10311 * targhooks.h (default_eh_return_filter_mode): Likewise.
10312 (default_libgcc_cmp_return_mode): Likewise.
10313 (default_libgcc_shift_count_mode): Likewise.
10314 (default_unwind_word_mode): Likewise.
10315 (default_addr_space_pointer_mode): Likewise.
10316 (default_addr_space_address_mode): Likewise.
10317 (default_mode_rep_extended): Take a scalar_int_mode rather than
10319 (default_valid_pointer_mode): Likewise.
10320 (default_addr_space_valid_pointer_mode): Likewise.
10321 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
10323 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
10324 rather than a machine_mode.
10325 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
10326 for local variables.
10327 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10328 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
10329 than a machine_mode.
10331 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10332 Alan Hayward <alan.hayward@arm.com>
10333 David Sherwood <david.sherwood@arm.com>
10335 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
10336 the mode argument to scalar_int_mode.
10337 (do_jump_by_parts_zero_rtx): Likewise.
10338 (do_jump_by_parts_equality_rtx): Likewise.
10339 (do_jump_by_parts_greater): Take a mode argument.
10340 (do_jump_by_parts_equality): Likewise.
10341 (do_jump_1): Update calls accordingly.
10343 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10344 Alan Hayward <alan.hayward@arm.com>
10345 David Sherwood <david.sherwood@arm.com>
10347 * is-a.h (safe_dyn_cast): New function.
10348 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
10349 (jump_table_for_label): Likewise.
10350 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
10351 instead of an rtx_insn *.
10352 (shorten_branches): Use dyn_cast instead of LABEL_P and
10353 JUMP_TABLE_DATA_P. Use jump_table_for_label and
10354 rtx_jump_table_data::get_data_mode.
10355 (final_scan_insn): Likewise.
10357 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10358 Alan Hayward <alan.hayward@arm.com>
10359 David Sherwood <david.sherwood@arm.com>
10361 * combine.c (try_combine): Use is_a <scalar_int_mode> when
10362 trying to combine a full-register integer set with a subreg
10365 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10366 Alan Hayward <alan.hayward@arm.com>
10367 David Sherwood <david.sherwood@arm.com>
10369 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
10370 that is always either address_mode or pointer_mode.
10372 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10373 Alan Hayward <alan.hayward@arm.com>
10374 David Sherwood <david.sherwood@arm.com>
10376 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
10377 when the two are known to be equal.
10379 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10380 Alan Hayward <alan.hayward@arm.com>
10381 David Sherwood <david.sherwood@arm.com>
10383 * simplify-rtx.c (simplify_const_unary_operation): Use
10384 is_a <scalar_int_mode> instead of checking for a nonzero
10385 precision. Forcibly convert op_mode to a scalar_int_mode
10386 in that case. More clearly differentiate the operand and
10387 result modes and use the former when deciding what the value
10388 of a count-bits operation should be. Use is_int_mode instead
10389 of checking for a MODE_INT. Remove redundant check for whether
10390 this mode has a zero precision.
10392 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10393 Alan Hayward <alan.hayward@arm.com>
10394 David Sherwood <david.sherwood@arm.com>
10396 * optabs.c (widen_leading): Change the type of the mode argument
10397 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
10398 (widen_bswap): Likewise.
10399 (expand_parity): Likewise.
10400 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
10401 (expand_ffs): Likewise.
10402 (epand_unop): Check for scalar integer modes before calling the
10405 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10406 Alan Hayward <alan.hayward@arm.com>
10407 David Sherwood <david.sherwood@arm.com>
10409 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
10411 (expand_expr_real_1): Update call accordingly.
10413 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10414 Alan Hayward <alan.hayward@arm.com>
10415 David Sherwood <david.sherwood@arm.com>
10417 * expmed.c (store_bit_field_using_insv): Add op0_mode and
10418 value_mode arguments. Use scalar_int_mode internally.
10419 (store_bit_field_1): Rename the new integer mode from imode
10420 to op0_mode and use it instead of GET_MODE (op0). Update calls
10421 to store_split_bit_field, store_bit_field_using_insv and
10422 store_fixed_bit_field.
10423 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
10424 Use scalar_int_mode internally. Use a bit count rather than a mode
10425 when calculating the largest bit size for get_best_mode.
10426 Update calls to store_split_bit_field and store_fixed_bit_field_1.
10427 (store_fixed_bit_field_1): Add mode and value_mode arguments.
10428 Remove assertion that OP0 has a scalar integer mode.
10429 (store_split_bit_field): Add op0_mode and value_mode arguments.
10430 Update calls to extract_fixed_bit_field.
10431 (extract_bit_field_using_extv): Add an op0_mode argument.
10432 Use scalar_int_mode internally.
10433 (extract_bit_field_1): Rename the new integer mode from imode to
10434 op0_mode and use it instead of GET_MODE (op0). Update calls to
10435 extract_split_bit_field, extract_bit_field_using_extv and
10436 extract_fixed_bit_field.
10437 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
10438 to extract_split_bit_field and extract_fixed_bit_field_1.
10439 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
10440 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
10441 on the target mode.
10442 (extract_split_bit_field): Add an op0_mode argument. Update call
10443 to extract_fixed_bit_field.
10445 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10446 Alan Hayward <alan.hayward@arm.com>
10447 David Sherwood <david.sherwood@arm.com>
10449 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
10450 * explow.c (hard_function_value): Likewise.
10451 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
10452 convert_to_mode call outside the loop.
10453 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
10454 for the mode iterator. Require the mode specified by max_pieces
10456 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
10458 (copy_blkmode_to_reg): Likewise.
10459 (set_storage_via_setmem): Likewise.
10460 * optabs.c (prepare_cmp_insn): Likewise.
10461 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
10462 * stor-layout.c (finish_bitfield_representative): Likewise.
10464 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10465 Alan Hayward <alan.hayward@arm.com>
10466 David Sherwood <david.sherwood@arm.com>
10468 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
10469 * expr.c (convert_move): Use them.
10470 (convert_modes): Likewise.
10471 (store_expr_with_bounds): Likewise.
10473 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10474 Alan Hayward <alan.hayward@arm.com>
10475 David Sherwood <david.sherwood@arm.com>
10477 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
10478 parameter for the mode of "x". Remove the "known_x", "known_mode"
10479 and "known_ret" arguments. Change the type of the mode argument
10480 to scalar_int_mode.
10481 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
10482 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
10483 (reg_num_sign_bit_copies_for_combine): Likewise.
10484 * rtlanal.c (nonzero_bits1): Likewise.
10485 (num_sign_bit_copies1): Likewise.
10486 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
10487 (reg_num_sign_bit_copies_general): Likewise.
10488 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
10489 (reg_nonzero_bits_general): Likewise.
10491 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10492 Alan Hayward <alan.hayward@arm.com>
10493 David Sherwood <david.sherwood@arm.com>
10495 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
10496 than in subroutines. Return 1 for non-integer modes.
10497 (cached_num_sign_bit_copies): Change the type of the mode parameter
10498 to scalar_int_mode.
10499 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
10500 classes. Handle CONST_INT_P first and then check whether X also
10501 has a scalar integer mode. Check the same thing for inner registers
10502 of a SUBREG and for values that are being extended or truncated.
10504 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10505 Alan Hayward <alan.hayward@arm.com>
10506 David Sherwood <david.sherwood@arm.com>
10508 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
10509 in subroutines. Return the mode mask for non-integer modes.
10510 (cached_nonzero_bits): Change the type of the mode parameter
10511 to scalar_int_mode.
10512 (nonzero_bits1): Likewise. Remove early exit for other mode
10513 classes. Handle CONST_INT_P first and then check whether X
10514 also has a scalar integer mode.
10516 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10517 Alan Hayward <alan.hayward@arm.com>
10518 David Sherwood <david.sherwood@arm.com>
10520 * expr.c (widest_int_mode_for_size): Make the comment match the code.
10521 Return a scalar_int_mode and assert that the size is greater than
10523 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
10524 (op_by_pieces_d::op_by_pieces_d): Likewise.
10525 (op_by_pieces_d::run): Likewise.
10526 (can_store_by_pieces): Likewise.
10528 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10529 Alan Hayward <alan.hayward@arm.com>
10530 David Sherwood <david.sherwood@arm.com>
10532 * combine.c (extract_left_shift): Add a mode argument and update
10534 (make_compound_operation_int): Change the type of the mode parameter
10535 to scalar_int_mode and update the call to extract_left_shift.
10537 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10538 Alan Hayward <alan.hayward@arm.com>
10539 David Sherwood <david.sherwood@arm.com>
10541 * combine.c (simplify_and_const_int): Change the type of the mode
10542 parameter to scalar_int_mode.
10543 (simplify_and_const_int_1): Likewise. Update recursive call.
10545 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10546 Alan Hayward <alan.hayward@arm.com>
10547 David Sherwood <david.sherwood@arm.com>
10549 * combine.c (simplify_compare_const): Check that the mode is a
10550 scalar_int_mode (rather than VOIDmode) before testing its
10552 (simplify_comparison): Move COMPARISON_P handling out of the
10553 loop and restrict the latter part of the loop to scalar_int_modes.
10554 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
10555 and when considering SUBREG_REGs. Use is_int_mode instead of
10556 checking GET_MODE_CLASS against MODE_INT.
10558 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10559 Alan Hayward <alan.hayward@arm.com>
10560 David Sherwood <david.sherwood@arm.com>
10562 * combine.c (try_widen_shift_mode): Move check for equal modes to...
10563 (simplify_shift_const_1): ...here. Use scalar_int_mode for
10564 shift_unit_mode and for modes involved in scalar shifts.
10566 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10567 Alan Hayward <alan.hayward@arm.com>
10568 David Sherwood <david.sherwood@arm.com>
10570 * combine.c (force_int_to_mode): New function, split out from...
10571 (force_to_mode): ...here. Keep xmode up-to-date and use it
10572 instead of GET_MODE (x).
10574 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10575 Alan Hayward <alan.hayward@arm.com>
10576 David Sherwood <david.sherwood@arm.com>
10578 * optabs-query.h (extraction_insn::struct_mode): Change type to
10579 opt_scalar_int_mode and update comment.
10580 (extraction_insn::field_mode): Change type to scalar_int_mode.
10581 (extraction_insn::pos_mode): Likewise.
10582 * combine.c (make_extraction): Update accordingly.
10583 * optabs-query.c (get_traditional_extraction_insn): Likewise.
10584 (get_optab_extraction_insn): Likewise.
10585 * recog.c (simplify_while_replacing): Likewise.
10586 * expmed.c (narrow_bit_field_mem): Change the type of the mode
10587 parameter to opt_scalar_int_mode.
10589 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10590 Alan Hayward <alan.hayward@arm.com>
10591 David Sherwood <david.sherwood@arm.com>
10593 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
10594 to a scalar_int_mode instead of a machine_mode.
10595 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
10596 (get_best_mode): Return a boolean and use a pointer argument to store
10597 the selected mode. Replace the limit mode parameter with a bit limit.
10598 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
10599 for the values returned by bit_field_mode_iterator::next_mode.
10600 (store_bit_field): Update call to get_best_mode.
10601 (store_fixed_bit_field): Likewise.
10602 (extract_fixed_bit_field): Likewise.
10603 * expr.c (optimize_bitfield_assignment_op): Likewise.
10604 * fold-const.c (optimize_bit_field_compare): Likewise.
10605 (fold_truth_andor_1): Likewise.
10606 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
10607 Update for new type of m_mode.
10608 (get_best_mode): As above.
10610 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10611 Alan Hayward <alan.hayward@arm.com>
10612 David Sherwood <david.sherwood@arm.com>
10614 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
10615 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
10616 (store_bit_field): Check is_a <scalar_int_mode> before calling
10617 strict_volatile_bitfield_p.
10618 (extract_bit_field): Likewise.
10620 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10621 Alan Hayward <alan.hayward@arm.com>
10622 David Sherwood <david.sherwood@arm.com>
10624 * target.def (cstore_mode): Return a scalar_int_mode.
10625 * doc/tm.texi: Regenerate.
10626 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
10627 * targhooks.h (default_cstore_mode): Likewise.
10628 * targhooks.c (default_cstore_mode): Likewise, using a forced
10630 * expmed.c (emit_cstore): Expect the target of the cstore to be
10633 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10634 Alan Hayward <alan.hayward@arm.com>
10635 David Sherwood <david.sherwood@arm.com>
10637 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
10639 (niter_desc): Likewise mode.
10640 (iv_analyze): Add a mode parameter.
10642 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
10643 and change its type to scalar_int_mode.
10644 * loop-iv.c: Update commentary at head of file.
10645 (iv_constant): Pass the mode paraeter before the rtx it describes
10646 and change its type to scalar_int_mode. Remove VOIDmode handling.
10647 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
10648 (iv_extend): Likewise.
10649 (shorten_into_mode): Likewise.
10650 (iv_add): Use scalar_int_mode.
10651 (iv_mult): Likewise.
10652 (iv_shift): Likewise.
10653 (canonicalize_iv_subregs): Likewise.
10654 (get_biv_step_1): Pass the outer_mode parameter before the rtx
10655 it describes and change its mode to scalar_int_mode. Also change
10656 the type of the returned inner_mode to scalar_int_mode.
10657 (get_biv_step): Likewise, turning outer_mode from a pointer
10658 into a direct parameter. Update call to get_biv_step_1.
10659 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
10660 iv_constant and get_biv_step.
10661 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
10662 and change its type to scalar_int_mode. Don't initialise iv->mode
10663 to VOIDmode and remove later checks for its still being VOIDmode.
10664 Update calls to iv_analyze_op and iv_analyze_expr. Check
10665 is_a <scalar_int_mode> when changing the mode under consideration.
10666 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
10667 Update call to iv_analyze_expr.
10668 (iv_analyze_op): Add a mode parameter. Reject subregs whose
10669 inner register is not also a scalar_int_mode. Update call to
10671 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
10672 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
10673 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
10674 separate mode class checks. Update calls to iv_analyze. Remove
10675 fix-up of VOIDmodes after iv_analyze_biv.
10676 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
10677 don't have a scalar_int_mode. Update call to biv_p.
10679 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10680 Alan Hayward <alan.hayward@arm.com>
10681 David Sherwood <david.sherwood@arm.com>
10683 * cfgexpand.c (convert_debug_memory_address): Use
10684 as_a <scalar_int_mode>.
10685 * combine.c (expand_compound_operation): Likewise.
10686 (make_extraction): Likewise.
10687 (change_zero_ext): Likewise.
10688 (simplify_comparison): Likewise.
10689 * cse.c (cse_insn): Likewise.
10690 * dwarf2out.c (minmax_loc_descriptor): Likewise.
10691 (mem_loc_descriptor): Likewise.
10692 (loc_descriptor): Likewise.
10693 * expmed.c (init_expmed_one_mode): Likewise.
10694 (synth_mult): Likewise.
10695 (emit_store_flag_1): Likewise.
10696 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
10697 of a comparison with size.
10698 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
10699 (reduce_to_bit_field_precision): Likewise.
10700 * function.c (expand_function_end): Likewise.
10701 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
10702 * loop-doloop.c (doloop_modify): Likewise.
10703 * optabs.c (expand_binop): Likewise.
10704 (expand_unop): Likewise.
10705 (expand_copysign_absneg): Likewise.
10706 (prepare_cmp_insn): Likewise.
10707 (maybe_legitimize_operand): Likewise.
10708 * recog.c (const_scalar_int_operand): Likewise.
10709 * rtlanal.c (get_address_mode): Likewise.
10710 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10711 (simplify_cond_clz_ctz): Likewise.
10712 * tree-nested.c (get_nl_goto_field): Likewise.
10713 * tree.c (build_vector_type_for_mode): Likewise.
10714 * var-tracking.c (use_narrower_mode): Likewise.
10716 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10717 Alan Hayward <alan.hayward@arm.com>
10718 David Sherwood <david.sherwood@arm.com>
10720 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
10721 * builtins.c (expand_builtin_signbit): Use it.
10722 * cfgexpand.c (expand_debug_expr): Likewise.
10723 * dojump.c (do_jump): Likewise.
10724 (do_compare_and_jump): Likewise.
10725 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
10726 * expmed.c (make_tree): Likewise.
10727 * expr.c (expand_expr_real_2): Likewise.
10728 (expand_expr_real_1): Likewise.
10729 (try_casesi): Likewise.
10730 * fold-const-call.c (fold_const_call_ss): Likewise.
10731 * fold-const.c (unextend): Likewise.
10732 (extract_muldiv_1): Likewise.
10733 (fold_single_bit_test): Likewise.
10734 (native_encode_int): Likewise.
10735 (native_encode_string): Likewise.
10736 (native_interpret_int): Likewise.
10737 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
10738 * internal-fn.c (expand_addsub_overflow): Likewise.
10739 (expand_neg_overflow): Likewise.
10740 (expand_mul_overflow): Likewise.
10741 (expand_arith_overflow): Likewise.
10742 * match.pd: Likewise.
10743 * stor-layout.c (layout_type): Likewise.
10744 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
10745 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
10746 * tree-ssanames.c (get_range_info): Likewise.
10747 * tree-switch-conversion.c (array_value_type) Likewise.
10748 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
10749 (vect_recog_divmod_pattern): Likewise.
10750 (vect_recog_mixed_size_cond_pattern): Likewise.
10751 * tree-vrp.c (extract_range_basic): Likewise.
10752 (simplify_float_conversion_using_ranges): Likewise.
10753 * tree.c (int_fits_type_p): Likewise.
10754 * ubsan.c (instrument_bool_enum_load): Likewise.
10755 * varasm.c (mergeable_string_section): Likewise.
10756 (narrowing_initializer_constant_valid_p): Likewise.
10757 (output_constant): Likewise.
10759 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10760 Alan Hayward <alan.hayward@arm.com>
10761 David Sherwood <david.sherwood@arm.com>
10763 * machmode.h (NARROWEST_INT_MODE): New macro.
10764 * expr.c (alignment_for_piecewise_move): Use it instead of
10765 GET_CLASS_NARROWEST_MODE (MODE_INT).
10766 (push_block): Likewise.
10767 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
10769 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10771 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10772 Alan Hayward <alan.hayward@arm.com>
10773 David Sherwood <david.sherwood@arm.com>
10775 * postreload.c (move2add_valid_value_p): Change the type of the
10776 mode parameter to scalar_int_mode.
10777 (move2add_use_add2_insn): Add a mode parameter and use it instead
10779 (move2add_use_add3_insn): Likewise.
10780 (reload_cse_move2add): Update accordingly.
10782 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10783 Alan Hayward <alan.hayward@arm.com>
10784 David Sherwood <david.sherwood@arm.com>
10786 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
10788 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
10789 * optabs.c (expand_unop): Likewise.
10791 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10792 Alan Hayward <alan.hayward@arm.com>
10793 David Sherwood <david.sherwood@arm.com>
10795 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
10796 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
10797 (popcount_loc_descriptor): Likewise.
10798 (bswap_loc_descriptor): Likewise.
10799 (rotate_loc_descriptor): Likewise.
10800 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
10801 calling the functions above.
10803 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10804 Alan Hayward <alan.hayward@arm.com>
10805 David Sherwood <david.sherwood@arm.com>
10807 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
10809 (try_combine): Likewise.
10810 (simplify_if_then_else): Likewise.
10811 * cse.c (cse_insn): Likewise.
10812 * dwarf2out.c (mem_loc_descriptor): Likewise.
10813 * emit-rtl.c (gen_lowpart_common): Likewise.
10814 * simplify-rtx.c (simplify_truncation): Likewise.
10815 (simplify_binary_operation_1): Likewise.
10816 (simplify_const_relational_operation): Likewise.
10817 (simplify_ternary_operation): Likewise.
10818 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
10820 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10821 Alan Hayward <alan.hayward@arm.com>
10822 David Sherwood <david.sherwood@arm.com>
10824 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
10825 * reload.c (push_reload): Likewise.
10826 (find_reloads): Likewise.
10828 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10829 Alan Hayward <alan.hayward@arm.com>
10830 David Sherwood <david.sherwood@arm.com>
10832 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
10833 (make_compound_operation_int): Likewise.
10834 (change_zero_ext): Likewise.
10835 * expr.c (convert_move): Likewise.
10836 (convert_modes): Likewise.
10837 * fwprop.c (forward_propagate_subreg): Likewise.
10838 * loop-iv.c (get_biv_step_1): Likewise.
10839 * optabs.c (widen_operand): Likewise.
10840 * postreload.c (move2add_valid_value_p): Likewise.
10841 * recog.c (simplify_while_replacing): Likewise.
10842 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10843 (simplify_binary_operation_1): Likewise. Remove redundant
10844 mode equality check.
10846 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10847 Alan Hayward <alan.hayward@arm.com>
10848 David Sherwood <david.sherwood@arm.com>
10850 * combine.c (combine_simplify_rtx): Add checks for
10851 is_a <scalar_int_mode>.
10852 (simplify_if_then_else): Likewise.
10853 (make_field_assignment): Likewise.
10854 (simplify_comparison): Likewise.
10855 * ifcvt.c (noce_try_bitop): Likewise.
10856 * loop-invariant.c (canonicalize_address_mult): Likewise.
10857 * simplify-rtx.c (simplify_unary_operation_1): 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 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
10864 is_a <scalar_int_mode> instead of != BLKmode.
10866 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10867 Alan Hayward <alan.hayward@arm.com>
10868 David Sherwood <david.sherwood@arm.com>
10870 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
10871 instead of != VOIDmode.
10872 * combine.c (if_then_else_cond): Likewise.
10873 (change_zero_ext): Likewise.
10874 * dwarf2out.c (mem_loc_descriptor): Likewise.
10875 (loc_descriptor): Likewise.
10876 * rtlanal.c (canonicalize_condition): Likewise.
10877 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
10879 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10880 Alan Hayward <alan.hayward@arm.com>
10881 David Sherwood <david.sherwood@arm.com>
10883 * simplify-rtx.c (simplify_binary_operation_1): Use
10884 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
10886 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10887 Alan Hayward <alan.hayward@arm.com>
10888 David Sherwood <david.sherwood@arm.com>
10890 * wide-int.h (int_traits<unsigned char>) New class.
10891 (int_traits<unsigned short>) Likewise.
10892 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
10893 Use GET_MODE_UNIT_PRECISION and remove redundant test for
10895 * combine.c (set_nonzero_bits_and_sign_copies): Use
10896 is_a <scalar_int_mode>.
10897 (find_split_point): Likewise.
10898 (combine_simplify_rtx): Likewise.
10899 (simplify_logical): Likewise.
10900 (expand_compound_operation): Likewise.
10901 (expand_field_assignment): Likewise.
10902 (make_compound_operation): Likewise.
10903 (extended_count): Likewise.
10904 (change_zero_ext): Likewise.
10905 (simplify_comparison): Likewise.
10906 * dwarf2out.c (scompare_loc_descriptor): Likewise.
10907 (ucompare_loc_descriptor): Likewise.
10908 (minmax_loc_descriptor): Likewise.
10909 (mem_loc_descriptor): Likewise.
10910 (loc_descriptor): Likewise.
10911 * expmed.c (init_expmed_one_mode): Likewise.
10912 * lra-constraints.c (lra_constraint_offset): Likewise.
10913 * optabs.c (prepare_libcall_arg): Likewise.
10914 * postreload.c (move2add_note_store): Likewise.
10915 * reload.c (operands_match_p): Likewise.
10916 * rtl.h (load_extend_op): Likewise.
10917 * rtlhooks.c (gen_lowpart_general): Likewise.
10918 * simplify-rtx.c (simplify_truncation): Likewise.
10919 (simplify_unary_operation_1): Likewise.
10920 (simplify_binary_operation_1): Likewise.
10921 (simplify_const_binary_operation): Likewise.
10922 (simplify_const_relational_operation): Likewise.
10923 (simplify_subreg): Likewise.
10924 * stor-layout.c (bitwise_mode_for_mode): Likewise.
10925 * var-tracking.c (adjust_mems): Likewise.
10926 (prepare_call_arguments): Likewise.
10928 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10929 Alan Hayward <alan.hayward@arm.com>
10930 David Sherwood <david.sherwood@arm.com>
10932 * machmode.h (is_int_mode): New fuction.
10933 * combine.c (find_split_point): Use it.
10934 (combine_simplify_rtx): Likewise.
10935 (simplify_if_then_else): Likewise.
10936 (simplify_set): Likewise.
10937 (simplify_shift_const_1): Likewise.
10938 (simplify_comparison): Likewise.
10939 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
10940 * cse.c (notreg_cost): Likewise.
10941 (cse_insn): Likewise.
10942 * cselib.c (cselib_lookup_1): Likewise.
10943 * dojump.c (do_jump_1): Likewise.
10944 (do_compare_rtx_and_jump): Likewise.
10945 * dse.c (get_call_args): Likewise.
10946 * dwarf2out.c (rtl_for_decl_init): Likewise.
10947 (native_encode_initializer): Likewise.
10948 * expmed.c (emit_store_flag_1): Likewise.
10949 (emit_store_flag): Likewise.
10950 * expr.c (convert_modes): Likewise.
10951 (store_field): Likewise.
10952 (expand_expr_real_1): Likewise.
10953 * fold-const.c (fold_read_from_constant_string): Likewise.
10954 * gimple-ssa-sprintf.c (get_format_string): Likewise.
10955 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
10956 * optabs.c (expand_binop): Likewise.
10957 (expand_unop): Likewise.
10958 (expand_abs_nojump): Likewise.
10959 (expand_one_cmpl_abs_nojump): Likewise.
10960 * simplify-rtx.c (mode_signbit_p): Likewise.
10961 (val_signbit_p): Likewise.
10962 (val_signbit_known_set_p): Likewise.
10963 (val_signbit_known_clear_p): Likewise.
10964 (simplify_relational_operation_1): Likewise.
10965 * tree.c (vector_type_mode): 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 (smallest_mode_for_size): Fix formatting.
10972 (smallest_int_mode_for_size): New function.
10973 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
10974 instead of smallest_mode_for_size.
10975 * combine.c (make_extraction): Likewise.
10976 * config/arc/arc.c (arc_expand_movmem): Likewise.
10977 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
10978 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
10979 * config/s390/s390.c (s390_expand_insv): Likewise.
10980 * config/sparc/sparc.c (assign_int_registers): Likewise.
10981 * config/spu/spu.c (spu_function_value): Likewise.
10982 (spu_function_arg): Likewise.
10983 * coverage.c (get_gcov_type): Likewise.
10984 (get_gcov_unsigned_t): Likewise.
10985 * dse.c (find_shift_sequence): Likewise.
10986 * expmed.c (store_bit_field_1): Likewise.
10987 * expr.c (convert_move): Likewise.
10988 (store_field): Likewise.
10989 * internal-fn.c (expand_arith_overflow): Likewise.
10990 * optabs-query.c (get_best_extraction_insn): Likewise.
10991 * optabs.c (expand_twoval_binop_libfunc): Likewise.
10992 * stor-layout.c (layout_type): Likewise.
10993 (initialize_sizetypes): Likewise.
10994 * targhooks.c (default_get_mask_mode): Likewise.
10995 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
10997 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10998 Alan Hayward <alan.hayward@arm.com>
10999 David Sherwood <david.sherwood@arm.com>
11001 * machmode.h (opt_mode::else_blk): New function.
11002 (int_mode_for_mode): Declare.
11003 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
11004 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
11006 * cfgexpand.c (expand_debug_expr): Likewise.
11007 * combine.c (gen_lowpart_or_truncate): Likewise.
11008 (gen_lowpart_for_combine): Likewise.
11009 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
11010 * config/avr/avr.c (avr_to_int_mode): Likewise.
11011 (avr_out_plus_1): Likewise.
11012 (avr_out_plus): Likewise.
11013 (avr_out_round): Likewise.
11014 * config/i386/i386.c (ix86_split_to_parts): Likewise.
11015 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
11016 (s390_expand_vcond): Likewise.
11017 * config/spu/spu.c (spu_split_immediate): Likewise.
11018 (spu_expand_mov): Likewise.
11019 * dse.c (get_stored_val): Likewise.
11020 * expmed.c (store_bit_field_1): Likewise.
11021 (convert_extracted_bit_field): Use int_mode_for_mode instead of
11023 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
11024 (extract_low_bits): Likewise.
11025 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
11026 handling rather than repeating the check.
11027 (emit_group_store): Likewise.
11028 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
11029 * optabs.c (expand_absneg_bit): Likewise.
11030 (expand_copysign_absneg): Likewise.
11031 (expand_copysign_bit): Likewise.
11032 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
11033 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
11034 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
11035 * var-tracking.c (prepare_call_arguments): Likewise.
11036 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
11037 int_mode_for_mode instead of mode_for_size.
11038 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
11040 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11041 Alan Hayward <alan.hayward@arm.com>
11042 David Sherwood <david.sherwood@arm.com>
11044 * machmode.h (int_mode_for_size): New function.
11045 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
11046 instead of mode_for_size.
11047 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
11049 * combine.c (expand_field_assignment): Use int_mode_for_size
11050 instead of mode_for_size.
11051 (make_extraction): Likewise.
11052 (simplify_shift_const_1): Likewise.
11053 (simplify_comparison): Likewise.
11054 * dojump.c (do_jump): Likewise.
11055 * dwarf2out.c (mem_loc_descriptor): Likewise.
11056 * emit-rtl.c (init_derived_machine_modes): Likewise.
11057 * expmed.c (flip_storage_order): Likewise.
11058 (convert_extracted_bit_field): Likewise.
11059 * expr.c (copy_blkmode_from_reg): Likewise.
11060 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
11061 * internal-fn.c (expand_mul_overflow): Likewise.
11062 * lower-subreg.c (simple_move): Likewise.
11063 * optabs-libfuncs.c (init_optabs): Likewise.
11064 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11065 * tree.c (vector_type_mode): Likewise.
11066 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
11067 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
11068 * tree-vect-generic.c (expand_vector_parallel): Likewise.
11069 * tree-vect-stmts.c (vectorizable_load): Likewise.
11070 (vectorizable_store): Likewise.
11072 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11073 Alan Hayward <alan.hayward@arm.com>
11074 David Sherwood <david.sherwood@arm.com>
11076 * coretypes.h (pod_mode): New type.
11077 (scalar_int_mode_pod): New typedef.
11078 * machmode.h (pod_mode): New class.
11079 (int_n_data_t::m): Change type to scalar_int_mode_pod.
11080 * genmodes.c (emit_mode_int_n): Update accordingly.
11081 * lower-subreg.h (target_lower_subreg): Change type to
11082 scalar_int_mode_pod.
11083 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
11084 scalar_int_mode_pod.
11086 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11087 Alan Hayward <alan.hayward@arm.com>
11088 David Sherwood <david.sherwood@arm.com>
11090 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
11091 machine_mode to scalar_int_mode.
11092 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
11093 (rs6000_option_override_internal): Remove cast to int.
11094 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
11095 machine_mode to scalar_int_mode.
11096 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
11097 (rs6000_option_override_internal): Remove cast to int.
11098 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
11099 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
11101 * config/s390/s390.c (s390_expand_builtin): Likewise.
11102 * coretypes.h (scalar_int_mode): New type.
11103 (opt_scalar_int_mode): New typedef.
11104 * machmode.h (scalar_int_mode): New class.
11105 (scalar_int_mode::includes_p): New function.
11106 (byte_mode): Change type to scalar_int_mode.
11107 (word_mode): Likewise.
11108 (ptr_mode): Likewise.
11109 * emit-rtl.c (byte_mode): Likewise.
11110 (word_mode): Likewise.
11111 (ptr_mode): Likewise.
11112 (init_derived_machine_modes): Update accordingly.
11113 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
11114 and MODE_PARTIAL_INT.
11115 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
11116 opt_scalar_int_mode.
11118 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11119 Alan Hayward <alan.hayward@arm.com>
11120 David Sherwood <david.sherwood@arm.com>
11122 * target.def (libgcc_floating_mode_supported_p): Take a
11124 * doc/tm.texi: Regenerate.
11125 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
11127 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
11128 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
11131 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11132 Alan Hayward <alan.hayward@arm.com>
11133 David Sherwood <david.sherwood@arm.com>
11135 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
11136 * doc/tm.texi: Regenerate.
11137 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
11138 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
11139 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
11140 * targhooks.h (default_floatn_mode): Likewise.
11141 * targhooks.c (default_floatn_mode): Likewise.
11142 * tree.c (build_common_tree_nodes): Update accordingly.
11144 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11145 Alan Hayward <alan.hayward@arm.com>
11146 David Sherwood <david.sherwood@arm.com>
11148 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
11149 (mode_iterator::iterate_p): Likewise.
11150 (mode_iterator::get_wider): Likewise.
11151 * expr.c (init_expr_target): Use opt_scalar_float_mode.
11153 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11154 Alan Hayward <alan.hayward@arm.com>
11155 David Sherwood <david.sherwood@arm.com>
11157 * coretypes.h (opt_scalar_float_mode): New typedef.
11158 * machmode.h (float_mode_for_size): New function.
11159 * emit-rtl.c (double_mode): Delete.
11160 (init_emit_once): Use float_mode_for_size.
11161 * stor-layout.c (layout_type): Likewise.
11162 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
11164 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11165 Alan Hayward <alan.hayward@arm.com>
11166 David Sherwood <david.sherwood@arm.com>
11168 * output.h (assemble_real): Take a scalar_float_mode.
11169 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
11170 * config/arm/arm.md (consttable_4): Likewise.
11171 (consttable_8): Likewise.
11172 (consttable_16): Likewise.
11173 * config/mips/mips.md (consttable_float): Likewise.
11174 * config/s390/s390.c (s390_output_pool_entry): Likewise.
11175 * varasm.c (assemble_real): Take a scalar_float_mode.
11176 (output_constant_pool_2): Update accordingly.
11177 (output_constant): Likewise.
11179 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11180 Alan Hayward <alan.hayward@arm.com>
11181 David Sherwood <david.sherwood@arm.com>
11183 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
11184 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
11185 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
11186 (native_encode_real): Likewise.
11187 (native_interpret_real): Likewise.
11188 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
11189 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11191 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11192 Alan Hayward <alan.hayward@arm.com>
11193 David Sherwood <david.sherwood@arm.com>
11195 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
11196 <scalar_float_mode>. Simplify.
11197 (gen_extend_conv_libfunc): Likewise.
11199 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11200 Alan Hayward <alan.hayward@arm.com>
11201 David Sherwood <david.sherwood@arm.com>
11203 * coretypes.h (scalar_float_mode): New type.
11204 * machmode.h (mode_traits::from_int): Use machine_mode if
11205 USE_ENUM_MODES is defined.
11206 (is_a): New function.
11208 (dyn_cast): Likewise.
11209 (scalar_float_mode): New class.
11210 (scalar_float_mode::includes_p): New function.
11211 (is_float_mode): Likewise.
11212 * gdbhooks.py (MachineModePrinter): New class.
11213 (build_pretty_printer): Use it for scalar_float_mode.
11214 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
11215 (format_helper::format_helper): Turn into a template.
11216 * genmodes.c (get_mode_class): New function.
11217 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
11218 or machine_mode if none.
11219 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
11220 as_a <scalar_float_mode>.
11221 * dwarf2out.c (mem_loc_descriptor): Likewise.
11222 (insert_float): Likewise.
11223 (add_const_value_attribute): Likewise.
11224 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11225 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
11226 (expand_unop): Update accordingly.
11227 (expand_abs_nojump): Likewise.
11228 (expand_copysign_absneg): Take a scalar_float_mode.
11229 (expand_copysign_bit): Likewise.
11230 (expand_copysign): Update accordingly.
11232 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11233 Alan Hayward <alan.hayward@arm.com>
11234 David Sherwood <david.sherwood@arm.com>
11236 * coretypes.h (opt_mode): New class.
11237 * machmode.h (opt_mode): Likewise.
11238 (opt_mode::else_void): New function.
11239 (opt_mode::require): Likewise.
11240 (opt_mode::exists): Likewise.
11241 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
11242 (GET_MODE_2XWIDER_MODE): Likewise.
11243 (mode_iterator::get_wider): Update accordingly.
11244 (mode_iterator::get_2xwider): Likewise.
11245 (mode_iterator::get_known_wider): Likewise, turning into a template.
11246 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
11247 forcing a wider mode to exist.
11248 * config/cr16/cr16.h (LONG_REG_P): Likewise.
11249 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11250 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
11251 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
11252 * lower-subreg.c (init_lower_subreg): Likewise.
11253 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
11254 on the final iteration.
11255 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
11256 a wider mode exists before asking for a move pattern.
11257 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
11258 forcing a wider mode to exist.
11259 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
11260 returning false if no such mode exists.
11261 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
11262 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
11263 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
11264 Avoid checking for a MODE_INT if we already know the mode is not a
11266 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
11267 forcing a wider mode to exist.
11268 (expmed_mult_highpart_optab): Likewise.
11269 (expmed_mult_highpart): Likewise.
11270 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
11272 * lto-streamer-in.c (lto_input_mode_table): Likewise.
11273 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
11274 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
11275 * internal-fn.c (expand_mul_overflow): Update use of
11276 GET_MODE_2XWIDER_MODE.
11277 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11278 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
11279 GET_MODE_WIDER_MODE.
11280 (convert_plusminus_to_widen): Likewise.
11281 * tree-switch-conversion.c (array_value_type): Likewise.
11282 * var-tracking.c (emit_note_insn_var_location): Likewise.
11283 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11284 Return false inside rather than outside the loop if no wider mode
11286 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
11287 and GET_MODE_2XWIDER_MODE
11288 (can_compare_p): Use else_void.
11289 * gdbhooks.py (OptMachineModePrinter): New class.
11290 (build_pretty_printer): Use it for opt_mode.
11292 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11293 Alan Hayward <alan.hayward@arm.com>
11294 David Sherwood <david.sherwood@arm.com>
11296 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
11297 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
11299 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11300 Alan Hayward <alan.hayward@arm.com>
11301 David Sherwood <david.sherwood@arm.com>
11303 * machmode.h (mode_traits): New structure.
11304 (get_narrowest_mode): New function.
11305 (mode_iterator::start): Likewise.
11306 (mode_iterator::iterate_p): Likewise.
11307 (mode_iterator::get_wider): Likewise.
11308 (mode_iterator::get_known_wider): Likewise.
11309 (mode_iterator::get_2xwider): Likewise.
11310 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
11311 (FOR_EACH_MODE): Likewise.
11312 (FOR_EACH_MODE_FROM): Likewise.
11313 (FOR_EACH_MODE_UNTIL): Likewise.
11314 (FOR_EACH_WIDER_MODE): Likewise.
11315 (FOR_EACH_2XWIDER_MODE): Likewise.
11316 * builtins.c (expand_builtin_strlen): Use new mode iterators.
11317 * combine.c (simplify_comparison): Likewise
11318 * config/i386/i386.c (type_natural_mode): Likewise.
11319 * cse.c (cse_insn): Likewise.
11320 * dse.c (find_shift_sequence): Likewise.
11321 * emit-rtl.c (init_derived_machine_modes): Likewise.
11322 (init_emit_once): Likewise.
11323 * explow.c (hard_function_value): Likewise.
11324 * expmed.c (extract_fixed_bit_field_1): Likewise.
11325 (extract_bit_field_1): Likewise.
11326 (expand_divmod): Likewise.
11327 (emit_store_flag_1): Likewise.
11328 * expr.c (init_expr_target): Likewise.
11329 (convert_move): Likewise.
11330 (alignment_for_piecewise_move): Likewise.
11331 (widest_int_mode_for_size): Likewise.
11332 (emit_block_move_via_movmem): Likewise.
11333 (copy_blkmode_to_reg): Likewise.
11334 (set_storage_via_setmem): Likewise.
11335 (compress_float_constant): Likewise.
11336 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11337 * optabs-query.c (get_best_extraction_insn): Likewise.
11338 * optabs.c (expand_binop): Likewise.
11339 (expand_twoval_unop): Likewise.
11340 (expand_twoval_binop): Likewise.
11341 (widen_leading): Likewise.
11342 (widen_bswap): Likewise.
11343 (expand_parity): Likewise.
11344 (expand_unop): Likewise.
11345 (prepare_cmp_insn): Likewise.
11346 (prepare_float_lib_cmp): Likewise.
11347 (expand_float): Likewise.
11348 (expand_fix): Likewise.
11349 (expand_sfix_optab): Likewise.
11350 * postreload.c (move2add_use_add2_insn): Likewise.
11351 * reg-stack.c (reg_to_stack): Likewise.
11352 * reginfo.c (choose_hard_reg_mode): Likewise.
11353 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11354 * stor-layout.c (mode_for_size): Likewise.
11355 (smallest_mode_for_size): Likewise.
11356 (mode_for_vector): Likewise.
11357 (finish_bitfield_representative): Likewise.
11358 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
11359 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
11360 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11361 * var-tracking.c (prepare_call_arguments): Likewise.
11363 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11364 Alan Hayward <alan.hayward@arm.com>
11365 David Sherwood <david.sherwood@arm.com>
11367 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
11368 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
11369 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
11370 * machmode.h (mode_size): Move earlier in file.
11371 (mode_precision): Likewise.
11372 (mode_inner): Likewise.
11373 (mode_nunits): Likewise.
11374 (mode_unit_size): Likewise.
11375 (unit_unit_precision): Likewise.
11376 (mode_wider): Likewise.
11377 (mode_2xwider): Likewise.
11378 (machine_mode): New class.
11379 (mode_to_bytes): New function.
11380 (mode_to_bits): Likewise.
11381 (mode_to_precision): Likewise.
11382 (mode_to_inner): Likewise.
11383 (mode_to_unit_size): Likewise.
11384 (mode_to_unit_precision): Likewise.
11385 (mode_to_nunits): Likewise.
11386 (GET_MODE_SIZE): Use mode_to_bytes.
11387 (GET_MODE_BITSIZE): Use mode_to_bits.
11388 (GET_MODE_PRECISION): Use mode_to_precision.
11389 (GET_MODE_INNER): Use mode_to_inner.
11390 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
11391 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
11392 (GET_MODE_NUNITS): Use mode_to_nunits.
11393 * system.h (ALWAYS_INLINE): New macro.
11394 * config/powerpcspe/powerpcspe-c.c
11395 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
11396 int for arg1_mode and arg2_mode.
11398 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11399 Alan Hayward <alan.hayward@arm.com>
11400 David Sherwood <david.sherwood@arm.com>
11402 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
11403 Prefix mode names with E_ in case statements.
11404 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11405 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
11406 (aarch64_split_simd_move): Likewise.
11407 (aarch64_gen_storewb_pair): Likewise.
11408 (aarch64_gen_loadwb_pair): Likewise.
11409 (aarch64_gen_store_pair): Likewise.
11410 (aarch64_gen_load_pair): Likewise.
11411 (aarch64_get_condition_code_1): Likewise.
11412 (aarch64_constant_pool_reload_icode): Likewise.
11413 (get_rsqrte_type): Likewise.
11414 (get_rsqrts_type): Likewise.
11415 (get_recpe_type): Likewise.
11416 (get_recps_type): Likewise.
11417 (aarch64_gimplify_va_arg_expr): Likewise.
11418 (aarch64_simd_container_mode): Likewise.
11419 (aarch64_emit_load_exclusive): Likewise.
11420 (aarch64_emit_store_exclusive): Likewise.
11421 (aarch64_expand_compare_and_swap): Likewise.
11422 (aarch64_gen_atomic_cas): Likewise.
11423 (aarch64_emit_bic): Likewise.
11424 (aarch64_emit_atomic_swap): Likewise.
11425 (aarch64_emit_atomic_load_op): Likewise.
11426 (aarch64_evpc_trn): Likewise.
11427 (aarch64_evpc_uzp): Likewise.
11428 (aarch64_evpc_zip): Likewise.
11429 (aarch64_evpc_ext): Likewise.
11430 (aarch64_evpc_rev): Likewise.
11431 (aarch64_evpc_dup): Likewise.
11432 (aarch64_gen_ccmp_first): Likewise.
11433 (aarch64_gen_ccmp_next): Likewise.
11434 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11435 (alpha_emit_xfloating_libcall): Likewise.
11436 (emit_insxl): Likewise.
11437 (alpha_arg_type): Likewise.
11438 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
11439 (arc_preferred_simd_mode): Likewise.
11440 (arc_secondary_reload): Likewise.
11441 (get_arc_condition_code): Likewise.
11442 (arc_print_operand): Likewise.
11443 (arc_legitimate_constant_p): Likewise.
11444 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11445 * config/arc/arc.md (casesi_load): Likewise.
11446 (casesi_compact_jump): Likewise.
11447 * config/arc/predicates.md (proper_comparison_operator): Likewise.
11448 (cc_use_register): Likewise.
11449 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11450 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
11451 (arm_init_iwmmxt_builtins): Likewise.
11452 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
11453 (neon_expand_vector_init): Likewise.
11454 (arm_attr_length_move_neon): Likewise.
11455 (maybe_get_arm_condition_code): Likewise.
11456 (arm_emit_vector_const): Likewise.
11457 (arm_preferred_simd_mode): Likewise.
11458 (arm_output_iwmmxt_tinsr): Likewise.
11459 (thumb1_output_casesi): Likewise.
11460 (thumb2_output_casesi): Likewise.
11461 (arm_emit_load_exclusive): Likewise.
11462 (arm_emit_store_exclusive): Likewise.
11463 (arm_expand_compare_and_swap): Likewise.
11464 (arm_evpc_neon_vuzp): Likewise.
11465 (arm_evpc_neon_vzip): Likewise.
11466 (arm_evpc_neon_vrev): Likewise.
11467 (arm_evpc_neon_vtrn): Likewise.
11468 (arm_evpc_neon_vext): Likewise.
11469 (arm_validize_comparison): Likewise.
11470 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
11471 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
11472 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
11473 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
11474 (c6x_preferred_simd_mode): Likewise.
11475 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
11476 (epiphany_rtx_costs): Likewise.
11477 * config/epiphany/predicates.md (proper_comparison_operator):
11479 * config/frv/frv.c (condexec_memory_operand): Likewise.
11480 (frv_emit_move): Likewise.
11481 (output_move_single): Likewise.
11482 (output_condmove_single): Likewise.
11483 (frv_hard_regno_mode_ok): Likewise.
11484 (frv_matching_accg_mode): Likewise.
11485 * config/h8300/h8300.c (split_adds_subs): Likewise.
11486 (h8300_rtx_costs): Likewise.
11487 (h8300_print_operand): Likewise.
11488 (compute_mov_length): Likewise.
11489 (output_logical_op): Likewise.
11490 (compute_logical_op_length): Likewise.
11491 (compute_logical_op_cc): Likewise.
11492 (h8300_shift_needs_scratch_p): Likewise.
11493 (output_a_shift): Likewise.
11494 (compute_a_shift_length): Likewise.
11495 (compute_a_shift_cc): Likewise.
11496 (expand_a_rotate): Likewise.
11497 (output_a_rotate): Likewise.
11498 * config/i386/i386.c (classify_argument): Likewise.
11499 (function_arg_advance_32): Likewise.
11500 (function_arg_32): Likewise.
11501 (function_arg_64): Likewise.
11502 (function_value_64): Likewise.
11503 (ix86_gimplify_va_arg): Likewise.
11504 (ix86_legitimate_constant_p): Likewise.
11505 (put_condition_code): Likewise.
11506 (split_double_mode): Likewise.
11507 (ix86_avx256_split_vector_move_misalign): Likewise.
11508 (ix86_expand_vector_logical_operator): Likewise.
11509 (ix86_split_idivmod): Likewise.
11510 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
11511 (ix86_build_const_vector): Likewise.
11512 (ix86_build_signbit_mask): Likewise.
11513 (ix86_match_ccmode): Likewise.
11514 (ix86_cc_modes_compatible): Likewise.
11515 (ix86_expand_branch): Likewise.
11516 (ix86_expand_sse_cmp): Likewise.
11517 (ix86_expand_sse_movcc): Likewise.
11518 (ix86_expand_int_sse_cmp): Likewise.
11519 (ix86_expand_vec_perm_vpermi2): Likewise.
11520 (ix86_expand_vec_perm): Likewise.
11521 (ix86_expand_sse_unpack): Likewise.
11522 (ix86_expand_int_addcc): Likewise.
11523 (ix86_split_to_parts): Likewise.
11524 (ix86_vectorize_builtin_gather): Likewise.
11525 (ix86_vectorize_builtin_scatter): Likewise.
11526 (avx_vpermilp_parallel): Likewise.
11527 (inline_memory_move_cost): Likewise.
11528 (ix86_tieable_integer_mode_p): Likewise.
11529 (x86_maybe_negate_const_int): Likewise.
11530 (ix86_expand_vector_init_duplicate): Likewise.
11531 (ix86_expand_vector_init_one_nonzero): Likewise.
11532 (ix86_expand_vector_init_one_var): Likewise.
11533 (ix86_expand_vector_init_concat): Likewise.
11534 (ix86_expand_vector_init_interleave): Likewise.
11535 (ix86_expand_vector_init_general): Likewise.
11536 (ix86_expand_vector_set): Likewise.
11537 (ix86_expand_vector_extract): Likewise.
11538 (emit_reduc_half): Likewise.
11539 (ix86_emit_i387_round): Likewise.
11540 (ix86_mangle_type): Likewise.
11541 (ix86_expand_round_sse4): Likewise.
11542 (expand_vec_perm_blend): Likewise.
11543 (canonicalize_vector_int_perm): Likewise.
11544 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
11545 (expand_vec_perm_1): Likewise.
11546 (expand_vec_perm_interleave3): Likewise.
11547 (expand_vec_perm_even_odd_pack): Likewise.
11548 (expand_vec_perm_even_odd_1): Likewise.
11549 (expand_vec_perm_broadcast_1): Likewise.
11550 (ix86_vectorize_vec_perm_const_ok): Likewise.
11551 (ix86_expand_vecop_qihi): Likewise.
11552 (ix86_expand_mul_widen_hilo): Likewise.
11553 (ix86_expand_sse2_abs): Likewise.
11554 (ix86_expand_pextr): Likewise.
11555 (ix86_expand_pinsr): Likewise.
11556 (ix86_preferred_simd_mode): Likewise.
11557 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
11558 * config/i386/sse.md (*andnot<mode>3): Likewise.
11559 (<mask_codefor><code><mode>3<mask_name>): Likewise.
11560 (*<code><mode>3): Likewise.
11561 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
11562 (ia64_expand_atomic_op): Likewise.
11563 (ia64_arg_type): Likewise.
11564 (ia64_mode_to_int): Likewise.
11565 (ia64_scalar_mode_supported_p): Likewise.
11566 (ia64_vector_mode_supported_p): Likewise.
11567 (expand_vec_perm_broadcast): Likewise.
11568 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
11569 (iq2000_function_arg_advance): Likewise.
11570 (iq2000_function_arg): Likewise.
11571 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
11572 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
11573 (m68k_libcall_value): Likewise.
11574 (m68k_function_value): Likewise.
11575 (sched_attr_op_type): Likewise.
11576 * config/mcore/mcore.c (mcore_output_move): Likewise.
11577 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
11579 (microblaze_function_arg): Likewise.
11580 * config/mips/mips.c (mips16_build_call_stub): Likewise.
11581 (mips_print_operand): Likewise.
11582 (mips_mode_ok_for_mov_fmt_p): Likewise.
11583 (mips_vector_mode_supported_p): Likewise.
11584 (mips_preferred_simd_mode): Likewise.
11585 (mips_expand_vpc_loongson_even_odd): Likewise.
11586 (mips_expand_vec_unpack): Likewise.
11587 (mips_expand_vi_broadcast): Likewise.
11588 (mips_expand_vector_init): Likewise.
11589 (mips_expand_vec_reduc): Likewise.
11590 (mips_expand_msa_cmp): Likewise.
11591 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
11592 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
11593 (cc_flags_for_mode): Likewise.
11594 * config/msp430/msp430.c (msp430_print_operand): Likewise.
11595 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
11596 (nds32_output_casesi_pc_relative): Likewise.
11597 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11598 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
11599 (nvptx_gen_unpack): Likewise.
11600 (nvptx_gen_pack): Likewise.
11601 (nvptx_gen_shuffle): Likewise.
11602 (nvptx_gen_wcast): Likewise.
11603 (nvptx_preferred_simd_mode): Likewise.
11604 * config/pa/pa.c (pa_secondary_reload): Likewise.
11605 * config/pa/predicates.md (base14_operand): Likewise.
11606 * config/powerpcspe/powerpcspe-c.c
11607 (altivec_resolve_overloaded_builtin): Likewise.
11608 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
11610 (rs6000_preferred_simd_mode): Likewise.
11611 (output_vec_const_move): Likewise.
11612 (rs6000_expand_vector_extract): Likewise.
11613 (rs6000_split_vec_extract_var): Likewise.
11614 (reg_offset_addressing_ok_p): Likewise.
11615 (rs6000_legitimate_offset_address_p): Likewise.
11616 (rs6000_legitimize_address): Likewise.
11617 (rs6000_emit_set_const): Likewise.
11618 (rs6000_const_vec): Likewise.
11619 (rs6000_emit_move): Likewise.
11620 (spe_build_register_parallel): Likewise.
11621 (rs6000_darwin64_record_arg_recurse): Likewise.
11622 (swap_selector_for_mode): Likewise.
11623 (spe_init_builtins): Likewise.
11624 (paired_init_builtins): Likewise.
11625 (altivec_init_builtins): Likewise.
11626 (do_load_for_compare): Likewise.
11627 (rs6000_generate_compare): Likewise.
11628 (rs6000_expand_float128_convert): Likewise.
11629 (emit_load_locked): Likewise.
11630 (emit_store_conditional): Likewise.
11631 (rs6000_output_function_epilogue): Likewise.
11632 (rs6000_handle_altivec_attribute): Likewise.
11633 (rs6000_function_value): Likewise.
11634 (emit_fusion_gpr_load): Likewise.
11635 (emit_fusion_p9_load): Likewise.
11636 (emit_fusion_p9_store): Likewise.
11637 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
11638 (fusion_gpr_mem_load): Likewise.
11639 (fusion_addis_mem_combo_load): Likewise.
11640 (fusion_addis_mem_combo_store): Likewise.
11641 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
11642 (fusion_gpr_mem_load): Likewise.
11643 (fusion_addis_mem_combo_load): Likewise.
11644 (fusion_addis_mem_combo_store): Likewise.
11645 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11647 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
11648 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
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 (rs6000_darwin64_record_arg_recurse): Likewise.
11660 (swap_selector_for_mode): Likewise.
11661 (paired_init_builtins): Likewise.
11662 (altivec_init_builtins): Likewise.
11663 (rs6000_expand_float128_convert): Likewise.
11664 (emit_load_locked): Likewise.
11665 (emit_store_conditional): Likewise.
11666 (rs6000_output_function_epilogue): Likewise.
11667 (rs6000_handle_altivec_attribute): Likewise.
11668 (rs6000_function_value): Likewise.
11669 (emit_fusion_gpr_load): Likewise.
11670 (emit_fusion_p9_load): Likewise.
11671 (emit_fusion_p9_store): Likewise.
11672 * config/rx/rx.c (rx_gen_move_template): Likewise.
11673 (flags_from_mode): Likewise.
11674 * config/s390/predicates.md (s390_alc_comparison): Likewise.
11675 (s390_slb_comparison): Likewise.
11676 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
11677 (s390_vector_mode_supported_p): Likewise.
11678 (s390_cc_modes_compatible): Likewise.
11679 (s390_match_ccmode_set): Likewise.
11680 (s390_canonicalize_comparison): Likewise.
11681 (s390_emit_compare_and_swap): Likewise.
11682 (s390_branch_condition_mask): Likewise.
11683 (s390_rtx_costs): Likewise.
11684 (s390_secondary_reload): Likewise.
11685 (__SECONDARY_RELOAD_CASE): Likewise.
11686 (s390_expand_cs): Likewise.
11687 (s390_preferred_simd_mode): Likewise.
11688 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
11689 * config/sh/sh.c (sh_print_operand): Likewise.
11690 (dump_table): Likewise.
11691 (sh_secondary_reload): Likewise.
11692 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11693 * config/sh/sh.md (casesi_worker_1): Likewise.
11694 (casesi_worker_2): Likewise.
11695 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
11696 (fcc_comparison_operator): Likewise.
11697 * config/sparc/sparc.c (sparc_expand_move): Likewise.
11698 (emit_soft_tfmode_cvt): Likewise.
11699 (sparc_preferred_simd_mode): Likewise.
11700 (output_cbranch): Likewise.
11701 (sparc_print_operand): Likewise.
11702 (sparc_expand_vec_perm_bmask): Likewise.
11703 (vector_init_bshuffle): Likewise.
11704 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11705 (spu_vector_mode_supported_p): Likewise.
11706 (spu_expand_insv): Likewise.
11707 (spu_emit_branch_or_set): Likewise.
11708 (spu_handle_vector_attribute): Likewise.
11709 (spu_builtin_splats): Likewise.
11710 (spu_builtin_extract): Likewise.
11711 (spu_builtin_promote): Likewise.
11712 (spu_expand_sign_extend): Likewise.
11713 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11714 (tilegx_simd_int): Likewise.
11715 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
11716 (tilepro_simd_int): Likewise.
11717 * config/v850/v850.c (const_double_split): Likewise.
11718 (v850_print_operand): Likewise.
11719 (ep_memory_offset): Likewise.
11720 * config/vax/vax.c (vax_rtx_costs): Likewise.
11721 (vax_output_int_move): Likewise.
11722 (vax_output_int_add): Likewise.
11723 (vax_output_int_subtract): Likewise.
11724 * config/visium/predicates.md (visium_branch_operator): Likewise.
11725 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
11726 (visium_print_operand_address): Likewise.
11727 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11728 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
11729 (xtensa_expand_conditional_branch): Likewise.
11730 (xtensa_copy_incoming_a7): Likewise.
11731 (xtensa_output_literal): Likewise.
11732 * dfp.c (decimal_real_maxval): Likewise.
11733 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
11735 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11736 Alan Hayward <alan.hayward@arm.com>
11737 David Sherwood <david.sherwood@arm.com>
11739 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
11740 (mode_nunits_inline): Likewise.
11741 (mode_inner_inline): Likewise.
11742 (mode_unit_size_inline): Likewise.
11743 (mode_unit_precision_inline): Likewise.
11744 (emit_insn_modes_h): Likewise. Also emit a #define of the
11746 (emit_mode_wider): Add an E_ prefix to mode names.
11747 (emit_mode_complex): Likewise.
11748 (emit_mode_inner): Likewise.
11749 (emit_mode_adjustments): Likewise.
11750 (emit_mode_int_n): Likewise.
11751 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
11752 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
11753 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
11754 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
11755 (CRC32_BUILTIN, ENTRY): Likewise.
11756 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
11757 (aarch64_pop_regs): Likewise.
11758 (aarch64_process_components): Likewise.
11759 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
11760 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
11761 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
11762 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
11763 * config/arm/arm.c (arm_init_libfuncs): Likewise.
11764 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
11766 * config/i386/i386-builtin.def (pcmpestr): Likewise.
11767 (pcmpistr): Likewise.
11768 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
11769 * config/mmix/mmix.c (mmix_output_condition): Likewise.
11770 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
11772 * config/rl78/rl78.c (mduc_regs): Likewise.
11773 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
11774 (htm_expand_builtin): Likewise.
11775 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
11776 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
11777 * config/xtensa/xtensa.c (print_operand): Likewise.
11778 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
11779 (NUM_MODE_VECTOR_INT): Likewise.
11780 * genoutput.c (null_operand): Likewise.
11781 (output_operand_data): Likewise.
11782 * genrecog.c (print_parameter_value): Likewise.
11783 * lra.c (debug_operand_data): Likewise.
11785 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11787 * dwarf2out.c (scompare_loc_descriptor_wide)
11788 (scompare_loc_descriptor_narrow): New functions, split out from...
11789 (scompare_loc_descriptor): ...here.
11790 * expmed.c (emit_store_flag_int): New function, split out from...
11791 (emit_store_flag): ...here.
11793 2017-08-30 Richard Biener <rguenther@suse.de>
11795 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
11796 (dwarf2out_early_finish): Move setting of AT_pubnames from
11797 early debug output to early finish.
11799 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11801 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
11802 and -mdata-region to the assembler.
11804 2017-08-30 Richard Biener <rguenther@suse.de>
11806 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
11808 (gen_subprogram_die): Add DW_AT_object_pointer only early.
11809 (dwarf2out_early_global_decl): Only generate a DIE for the
11810 abstract origin if it doesn't already exist or is a declaration DIE.
11811 (resolve_addr): Do not add the linkage name twice when
11812 generating a stub DIE for the DW_TAG_GNU_call_site target.
11814 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11816 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11817 Use machine_mode rather than int for arg1_mode.
11819 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
11822 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
11823 that the second argument of the built-in functions to unpack
11824 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
11825 switch statement instead a lot of if statements.
11826 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
11827 Allow 64-bit values to be in Altivec registers as well as
11828 traditional floating point registers.
11829 (pack<mode>, FMOVE128_VSX iterator): Likewise.
11831 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
11833 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
11834 MAX_REGS_PER_ADDRESS == 1.
11836 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
11838 * config/i386/i386.opt (flag_fentry): Do not init to -1.
11839 * config/i386/i386.c (ix86_option_override_internal): Simplify
11840 setting of opts->x_flag_entry.
11842 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11843 Jakub Jelinek <jakub@redhat.com>
11844 Richard Biener <rguenther@suse.de>
11846 PR tree-optimization/81503
11847 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
11848 folded constant fits in the target type; reorder tests for clarity.
11850 2017-08-29 Martin Liska <mliska@suse.cz>
11852 * passes.def: Include pass_lower_switch.
11853 * stmt.c (dump_case_nodes): Remove and move to
11854 tree-switch-conversion.
11855 (case_values_threshold): Likewise.
11856 (expand_switch_as_decision_tree_p): Likewise.
11857 (emit_case_decision_tree): Likewise.
11858 (expand_case): Likewise.
11859 (balance_case_nodes): Likewise.
11860 (node_has_low_bound): Likewise.
11861 (node_has_high_bound): Likewise.
11862 (node_is_bounded): Likewise.
11863 (emit_case_nodes): Likewise.
11864 (struct simple_case_node): New struct.
11865 (add_case_node): Remove.
11866 (emit_case_dispatch_table): Use vector instead of case_list.
11867 (reset_out_edges_aux): Remove.
11868 (compute_cases_per_edge): Likewise.
11869 (expand_case): Build list of simple_case_node.
11870 (expand_sjlj_dispatch_table): Use it.
11871 * tree-switch-conversion.c (struct case_node): Moved from
11872 stmt.c and adjusted.
11873 (emit_case_nodes): Likewise.
11874 (node_has_low_bound): Likewise.
11875 (node_has_high_bound): Likewise.
11876 (node_is_bounded): Likewise.
11877 (case_values_threshold): Likewise.
11878 (reset_out_edges_aux): Likewise.
11879 (compute_cases_per_edge): Likewise.
11880 (add_case_node): Likewise.
11881 (dump_case_nodes): Likewise.
11882 (balance_case_nodes): Likewise.
11883 (expand_switch_as_decision_tree_p): Likewise.
11884 (emit_jump): Likewise.
11885 (emit_case_decision_tree): Likewise.
11886 (try_switch_expansion): Likewise.
11887 (do_jump_if_equal): Likewise.
11888 (emit_cmp_and_jump_insns): Likewise.
11889 (fix_phi_operands_for_edge): New function.
11890 (record_phi_operand_mapping): Likewise.
11891 (class pass_lower_switch): New pass.
11892 (pass_lower_switch::execute): New function.
11893 (make_pass_lower_switch): Likewise.
11894 (conditional_probability):
11895 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
11896 * tree-pass.h: Add make_pass_lower_switch.
11898 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11901 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
11904 2017-08-29 Richard Biener <rguenther@suse.de>
11906 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
11907 we do not add a DW_AT_inline attribute twice.
11908 (gen_subprogram_die): Remove code setting DW_AT_inline on
11909 DECL_ABSTRACT_P nodes.
11911 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
11913 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
11914 calls to internal functions.
11915 (gimplify_modify_expr): Likewise.
11916 * tree-call-cdce.c (use_internal_fn): Likewise.
11917 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
11918 (convert_to_divmod): Set the nothrow flag.
11919 * tree-if-conv.c (predicate_mem_writes): Likewise.
11920 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
11921 (vectorizable_call): Likewise.
11922 (vectorizable_store): Likewise.
11923 (vectorizable_load): Likewise.
11924 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
11925 (vect_recog_mask_conversion_pattern): Likewise.
11927 2017-08-29 Martin Liska <mliska@suse.cz>
11930 * gcc.c (driver_handle_option): Add new argument.
11931 * opts-common.c (handle_option): Pass
11932 target_option_override_hook.
11933 * opts-global.c (lang_handle_option): Add new option.
11934 (set_default_handlers): Add new argument.
11935 (decode_options): Likewise.
11936 * opts.c (target_handle_option): Likewise.
11937 (common_handle_option): Call target_option_override_hook.
11938 * opts.h (struct cl_option_handler_func): Add hook for
11939 target option override.
11940 (struct cl_option_handlers): Likewise.
11941 (set_default_handlers): Add new argument.
11942 (decode_options): Likewise.
11943 (common_handle_option): Likewise.
11944 (target_handle_option): Likewise.
11945 * toplev.c (toplev::main): Pass targetm.target_option.override
11948 2017-08-29 Richard Biener <rguenther@suse.de>
11949 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
11951 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
11952 life to the active subtree.
11954 2017-08-28 Jeff Law <law@redhat.com>
11956 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
11957 derive_equivalences.
11958 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
11959 Code moved into....
11960 (edge_info::derive_equivalences): New private member function
11962 * tree-ssa-dom.c (class edge_info): Changed from a struct
11963 to a class. Add ctor/dtor, methods and data members.
11964 (edge_info::edge_info): Renamed from allocate_edge_info.
11965 Initialize additional members.
11966 (edge_info::~edge_info): New.
11967 (free_dom_edge_info): Delete the edge info.
11968 (record_edge_info): Use new class & associated member functions.
11969 Tighten forms for testing for edge equivalences.
11970 (record_temporary_equivalences): Iterate over the simple
11971 equivalences rather than assuming there's only one per edge.
11972 (cprop_into_successor_phis): Iterate over the simple
11973 equivalences rather than assuming there's only one per edge.
11974 (optimize_stmt): Use operand_equal_p rather than pointer
11975 equality for mini-DSE code.
11977 2017-08-28 Nathan Sidwell <nathan@acm.org>
11979 * gcc.c (execute): Fold SIGPIPE handling into switch
11980 statement. Adjust internal error message.
11982 2017-08-28 Richard Biener <rguenther@suse.de>
11985 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
11986 Do nothing for removed DIEs.
11988 2017-08-28 Richard Biener <rguenther@suse.de>
11990 PR tree-optimization/81977
11991 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
11994 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
11997 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
11998 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
11999 using targetm.gen_mem_thread_fence.
12001 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
12004 * config/i386/i386.md (*<btsc><mode>): Change operand 2
12005 predicate to register_operand. Reorder operands.
12006 (*btr<mode>): Ditto.
12007 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
12008 (*btr<mode>_mask): Ditto.
12010 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
12012 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
12013 * config/rs6000/xmmintrin.h: New file.
12014 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
12016 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12019 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
12020 parameter and_insn and return it.
12021 (recombine_lvx_pattern): Insert a copy to ensure availability of
12022 the base register of the copied masking operation at the point of
12023 the instruction replacement.
12024 (recombine_stvx_pattern): Likewise.
12026 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
12028 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
12029 undocumented switches.
12030 (-mpower9-dform-vector): Likewise.
12031 (-mpower9-dform): Likewise.
12032 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
12033 comments to delete references to -mpower9-dform* switches.
12034 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
12035 Delete reference to -mpower9-dform* switches, test for
12036 -mpower9-vector instead.
12037 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
12038 (OTHER_P9_VECTOR_MASKS): Likewise.
12039 (POWERPC_MASKS): Likewise.
12040 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
12041 tests against -mpower9-dform* to -mpower9-vector. Delete code
12042 that checked for -mpower9-dform* consistancy with other options.
12043 Add test for -mpower9-misc to enable other power9 switches.
12044 (rs6000_init_hard_regno_mode_ok): Likewise.
12045 (rs6000_option_override_internal): Likewise.
12046 (rs6000_emit_prologue): Likewise.
12047 (rs6000_emit_epilogue): Likewise.
12048 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
12049 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
12050 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
12052 (emit_fusion_p9_store): Likewise.
12053 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
12054 resetting these macros if the assembler does not support ISA 3.0
12056 (TARGET_P9_DFORM_VECTOR): Likewise.
12057 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
12058 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
12060 2017-08-25 Alan Modra <amodra@gmail.com>
12062 PR rtl-optimization/81747
12063 * cse.c (cse_extended_basic_block): Don't attempt to record
12064 equivalences for degenerate conditional branches that branch
12065 to their fall-through.
12067 2017-08-24 Martin Sebor <msebor@redhat.com>
12069 PR middle-end/81908
12070 * gimple-fold.c (size_must_be_zero_p): New function.
12071 (gimple_fold_builtin_memory_op): Call it.
12073 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
12075 * config/rs6000/mm_malloc.h: New file.
12077 2017-08-24 Bin Cheng <bin.cheng@arm.com>
12079 PR tree-optimization/81913
12080 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
12081 analysis when either IVs in condition can wrap.
12083 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
12085 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
12086 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
12088 2017-08-24 Richard Biener <rguenther@suse.de>
12091 * targhooks.c (default_target_can_inline_p): Properly
12092 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
12093 is present and always compare.
12094 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
12095 infer -mfpmath=sse from TARGET_SSE_P.
12096 (ix86_can_inline_p): Properly use target_option_default_node when
12097 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
12099 2017-08-24 Richard Biener <rguenther@suse.de>
12102 * dwarf2out.c (output_die): Handle flag_generate_offload like
12104 (output_comp_unit): Likewise.
12105 (gen_array_type_die): Likewise.
12106 (dwarf2out_early_finish): Likewise.
12107 (note_variable_value_in_expr): Likewise.
12108 (dwarf2out_finish): Likewise. Adjust assert.
12109 * cgraphunit.c (symbol_table::compile): Move setting of
12110 flag_generate_offload earlier ...
12111 (symbol_table::finalize_compilation_unit): ... here, before
12112 early debug finalization.
12114 2017-08-24 Richard Biener <rguenther@suse.de>
12116 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
12117 and ipa-fnsummary.h.
12118 (ix86_can_inline_p): When ix86_fpmath flags do not match
12119 check whether the callee uses FP math at all.
12121 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
12123 PR middle-end/81931
12124 * tree-ssanames.c (get_nonzero_bits): Use element_precision
12125 instead of TYPE_PRECISION.
12127 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
12128 Alan Hayward <alan.hayward@arm.com>
12129 David Sherwood <david.sherwood@arm.com>
12131 * combine.c (make_extraction): Use subreg_offset_from_lsb.
12133 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
12135 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
12137 (ix86_frame::stack_realign_allocate): New field.
12138 (struct machine_frame_state): Modify comments.
12139 (machine_frame_state::sp_realigned_fp_end): New field.
12140 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
12141 layout calculation.
12142 (sp_valid_at): Add assertion to assure no attempt to access invalid
12143 offset of a realigned stack.
12144 (fp_valid_at): Likewise.
12145 (choose_baseaddr): Modify comments.
12146 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
12147 ix86_expand_prologue.
12148 (ix86_expand_prologue): Modify stack realignment and allocation.
12149 (ix86_expand_epilogue): Modify comments.
12150 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
12151 avx2_runtime, avx512f, and avx512f_runtime.
12153 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
12155 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
12156 (mstackrealign): Do not init to -1.
12157 * config/i386/i386.c (ix86_option_override_internal):
12158 Check opts_set, not opts when setting default value of
12159 opts->x_ix86_force_align_arg_pointer.
12161 2017-08-23 Richard Biener <rguenther@suse.de>
12163 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
12164 lang_hooks.decl_printable_name.
12165 * print-rtl-function.c (print_rtx_function): Likewise.
12166 * tree-pretty-print.c (dump_function_header): Likewise.
12168 2017-08-23 Richard Biener <rguenther@suse.de>
12171 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
12172 -g0 at compile-time.
12174 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12176 PR middle-end/19706
12177 * doc/sourcebuild.texi (Other hardware attributes):
12180 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12182 PR middle-end/19706
12183 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
12184 Fix single-use check.
12186 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
12188 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
12190 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12192 * doc/install.texi: Modify to add more details on running selected
12195 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12197 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
12198 is combined with -mabi=ms.
12199 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
12202 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12204 PR tree-optimization/81488
12205 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
12206 and cached_basis fields.
12207 (MAX_SPREAD): New constant.
12208 (alloc_cand_and_find_basis): Initialize new fields.
12209 (clear_visited): New function.
12210 (create_phi_basis_1): Rename from create_phi_basis, set visited
12211 and cached_basis fields.
12212 (create_phi_basis): New wrapper function.
12213 (phi_add_costs_1): Rename from phi_add_costs, add spread
12214 parameter, set visited field, short-circuit when limits reached.
12215 (phi_add_costs): New wrapper function.
12216 (record_phi_increments_1): Rename from record_phi_increments, set
12218 (record_phi_increments): New wrapper function.
12219 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
12220 (phi_incr_cost): New wrapper function.
12221 (all_phi_incrs_profitable_1): Rename from
12222 all_phi_incrs_profitable, set visited field.
12223 (all_phi_incrs_profitable): New wrapper function.
12225 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12226 Alan Hayward <alan.hayward@arm.com>
12227 David Sherwood <david.sherwood@arm.com>
12229 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
12230 that takes the outer and inner modes.
12231 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
12232 comparison as the canonical test for a paradoxical subreg.
12233 * combine.c (simplify_set): Use paradoxical_subreg_p.
12234 (make_extraction): Likewise.
12235 (force_to_mode): Likewise.
12236 (rtx_equal_for_field_assignment_p): Likewise.
12237 (gen_lowpart_for_combine): Likewise.
12238 (simplify_comparison): Likewise.
12239 * cse.c (equiv_constant): Likewise.
12240 * expmed.c (store_bit_field_1): Likewise.
12241 * final.c (alter_subreg): Likewise.
12242 * fwprop.c (propagate_rtx): Likewise.
12243 (forward_propagate_subreg): Likewise.
12244 * ira-conflicts.c (ira_build_conflicts): Likewise.
12245 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
12246 * lra-constraints.c (curr_insn_transform): Likewise.
12247 (split_reg): Likewise.
12248 * lra-eliminations.c (move_plus_up): Likewise.
12249 (lra_eliminate_regs_1): Likewise.
12250 * recog.c (general_operand): Likewise.
12251 * ree.c (combine_reaching_defs): Likewise.
12252 * reload.c (push_reload): Likewise.
12253 (find_reloads): Likewise.
12254 * reload1.c (elimination_effects): Likewise.
12255 (compute_reload_subreg_offset): Likewise.
12256 (choose_reload_regs): Likewise.
12257 * rtlanal.c (subreg_lsb_1): Likewise.
12258 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12259 (simplify_subreg): Likewise.
12260 * var-tracking.c (track_loc_p): Likewise.
12261 * emit-rtl.c (byte_lowpart_offset): Likewise.
12262 (paradoxical_subreg_p): Delete out-of-line definition.
12264 2017-08-22 Jeff Law <law@redhat.com>
12266 PR tree-optimization/81741
12267 PR tree-optimization/71947
12268 * tree-ssa-dom.c: Include tree-inline.h.
12269 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
12270 equivalences if one is more expensive to compute than the other.
12271 * tree-ssa-scopedtables.h (class const_or_copies): Make
12272 record_const_or_copy_raw method private.
12273 (class avail_exprs_stack): New method simplify_binary_operation.
12274 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
12275 avail_exprs_stack::simplify_binary_operation as needed.
12276 (avail_exprs_stack::simplify_binary_operation): New function.
12278 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
12280 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
12281 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
12282 (DOT_SYMBOLS): Likewise.
12283 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
12284 (RELOCATABLE_NEEDS_FIXUP): Likewise.
12285 (RS6000_ABI_NAME): Likewise.
12286 (TARGET_CMODEL): Likewise.
12287 (TOC_SECTION_ASM_OP): Likewise.
12288 (SET_CMODEL): New macro.
12289 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
12291 2017-08-22 Richard Biener <rguenther@suse.de>
12293 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
12294 to allow for free-lang-data replacements similar to verify_type_variant.
12296 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12297 Alan Hayward <alan.hayward@arm.com>
12298 David Sherwood <david.sherwood@arm.com>
12300 * config/aarch64/aarch64.md (casesi): Use DImode rather than
12301 VOIDmode for the LABEL_REF.
12303 2017-08-22 Richard Biener <rguenther@suse.de>
12305 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
12307 2017-08-22 Richard Biener <rguenther@suse.de>
12309 * common.opt (feliminate-dwarf2-dups): Ignore.
12310 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
12311 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
12312 same_die_p_wrap, compute_section_prefix,
12313 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
12314 (comdat_symbol_id, comdat_symbol_number): Likewise.
12315 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
12317 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
12318 (output_die): Mark unreachable path unreachable.
12319 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
12320 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
12321 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
12322 (dwarf2out_early_finish): Likewise.
12324 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
12326 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
12328 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
12331 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
12332 not VAR_P. Filter attribute warnings with OPT_Wattributes.
12333 (avr_attribute_table) <io, io_low, address>: Initialize
12334 .decl_required with true.
12336 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
12338 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
12339 undocumented debugging options.
12340 (-mvsx-scalar-double): Likewise.
12341 (-mallow-df-permute): Likewise.
12342 (-mvectorize-builtins): Likewise.
12343 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12344 (rs6000_builtin_vectorized_function): Likewise.
12345 (rs6000_builtin_md_vectorized_function): Likewise.
12346 (rs6000_opt_vars): Likewise.
12348 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
12351 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
12352 (*btrq_imm): Rename from *btrq.
12353 (*btcq_imm): Rename from *btcq.
12354 (btsc): New code attribute.
12355 (*<btsc><mode>): New insn pattern.
12356 (*btr<mode>): Ditto.
12357 (*<btsc><mode>_mask): New insn_and_split pattern.
12358 (*btr<mode>_mask): Ditto.
12360 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12361 Alan Hayward <alan.hayward@arm.com>
12362 David Sherwood <david.sherwood@arm.com>
12364 * function.c (pad_below): Simplify padding calculation.
12366 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12367 Alan Hayward <alan.hayward@arm.com>
12368 David Sherwood <david.sherwood@arm.com>
12370 * target.def (function_prologue): Remove frame size argument.
12371 (function_epilogue): Likewise.
12372 * doc/tm.texi: Regenerate.
12373 * final.c (final_start_function): Update call to function_prologue.
12374 (final_end_function): Update call to function_epilogue.
12375 (default_function_pro_epilogue): Remove frame size argument.
12376 * output.h (default_function_pro_epilogue): Likewise.
12377 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
12378 (arm_output_function_prologue): Likewise.
12379 * config/frv/frv.c (frv_function_prologue): Likewise.
12380 (frv_function_epilogue): Likewise.
12381 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
12382 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
12383 (ia64_output_function_epilogue): Likewise.
12384 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
12385 (m32r_output_function_epilogue): Likewise.
12386 * config/microblaze/microblaze.c (microblaze_function_prologue)
12387 (microblaze_function_epilogue): Likewise.
12388 * config/mips/mips.c (mips_output_function_prologue): Likewise.
12389 (mips_output_function_epilogue): Likewise.
12390 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
12391 (mmix_target_asm_function_epilogue): Likewise.
12392 * config/msp430/msp430.c (msp430_start_function): Likewise.
12393 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
12394 (nds32_asm_function_epilogue): Likewise.
12395 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
12396 * config/pa/pa.c (pa_output_function_prologue): Likewise.
12397 (pa_output_function_epilogue): Likewise.
12398 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
12399 (rs6000_output_function_epilogue): Likewise.
12400 * config/rl78/rl78.c (rl78_start_function): Likewise.
12401 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
12402 (rs6000_output_function_epilogue): Likewise.
12403 * config/rx/rx.c (rx_output_function_prologue): Likewise.
12404 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
12405 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
12406 (sparc_asm_function_epilogue): Likewise.
12408 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12410 * tree.h (type_has_mode_precision_p): New function.
12411 * convert.c (convert_to_integer_1): Use it.
12412 * expr.c (expand_expr_real_2): Likewise.
12413 (expand_expr_real_1): Likewise.
12414 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
12415 * match.pd: Likewise.
12416 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
12417 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
12418 * tree-tailcall.c (process_assignment): Likewise.
12419 * tree-vect-loop.c (vectorizable_reduction): Likewise.
12420 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
12421 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
12422 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12423 (vectorizable_assignment): Likewise.
12424 (vectorizable_shift): Likewise.
12425 (vectorizable_operation): Likewise.
12426 * tree-vrp.c (register_edge_assert_for_2): Likewise.
12428 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
12430 * match.pd: Add pow (C, x) simplification.
12432 2017-08-21 Richard Biener <rguenther@suse.de>
12434 PR tree-optimization/81900
12435 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
12436 for blocks with abnormal predecessors.
12437 (compute_antic): Do not set visited flag prematurely.
12439 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
12442 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
12444 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12446 * stor-layout.h (vector_type_mode): Move to...
12447 * tree.h (vector_type_mode): ...here.
12448 * stor-layout.c (vector_type_mode): Move to...
12449 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
12451 2017-08-21 Richard Biener <rguenther@suse.de>
12453 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
12454 register_external_die hooks.
12455 (debug_false_tree_charstarstar_uhwistar): Declare.
12456 (debug_nothing_tree_charstar_uhwi): Likewise.
12457 * debug.c (do_nothing_debug_hooks): Adjust.
12458 (debug_false_tree_charstarstar_uhwistar): New do nothing.
12459 (debug_nothing_tree_charstar_uhwi): Likewise.
12460 * dbxout.c (dbx_debug_hooks): Adjust.
12461 (xcoff_debug_hooks): Likewise.
12462 * sdbout.c (sdb_debug_hooks): Likewise.
12463 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
12464 * dwarf2out.c (macinfo_label_base): New global.
12465 (dwarf2out_register_external_die): New function for the
12466 register_external_die hook.
12467 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
12468 (dwarf2_debug_hooks): Use them.
12469 (dwarf2_lineno_debug_hooks): Adjust.
12470 (struct die_struct): Add with_offset flag.
12471 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
12472 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
12473 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
12474 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
12475 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
12476 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
12477 defining section names for the early LTO debug variants.
12478 (reset_indirect_string): New helper.
12479 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
12480 (print_dw_val): Add support for offsetted symbol references.
12481 (get_ultimate_context): Split out from is_cxx.
12482 (is_cxx): Use get_ultimate_context.
12483 (is_fortran): Add decl overload.
12484 (compute_comp_unit_symbol): Split out worker from
12485 compute_section_prefix.
12486 (compute_section_prefix): Call compute_comp_unit_symbol and
12487 set comdat_type_p here.
12488 (output_die): Skip DIE symbol output for the LTO added one.
12489 Handle DIE symbol references with offset.
12490 (output_comp_unit): Guard section name mangling properly.
12491 For LTO debug sections emit a symbol at the section beginning
12492 which we use to refer to its DIEs.
12493 (add_abstract_origin_attribute): For DIEs registered via
12494 dwarf2out_register_external_die directly refer to the early
12495 DIE rather than indirectly through the shadow one we created.
12496 Remove obsolete call to dwarf2out_abstract_function for
12497 non-function/block origins.
12498 (gen_array_type_die): When generating early LTO debug do
12499 not emit DW_AT_string_length.
12500 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
12501 late when in LTO. As suggested place a gcc_unreachable for
12502 the DECL_ABSTRACT_P case.
12503 (gen_subprogram_die): Avoid another specification DIE
12504 for early built declarations/definitions for the late LTO case.
12505 (gen_variable_die): Add type references for late duplicated VLA dies
12507 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
12508 we have the abstract instance already.
12509 (process_scope_var): Adjust decl DIE contexts in LTO which
12510 first puts them in limbo.
12511 (gen_decl_die): Do not generate type DIEs late apart from
12512 types for VLAs or for decls we do not yet have a DIE. Do not
12513 call dwarf2out_abstract_function late.
12514 (dwarf2out_early_global_decl): Make sure to create DIEs
12515 for abstract instances of a decl first.
12516 (dwarf2out_late_global_decl): Adjust comment.
12517 (output_macinfo_op): With multiple macro sections use
12518 macinfo_label_base to distinguish labels.
12519 (output_macinfo): Likewise. Update macinfo_label_base.
12520 Pass in the line info label.
12521 (note_variable_value_in_expr): When generating LTO resolve
12522 all variable values here by generating DIEs as needed.
12523 (init_sections_and_labels): Add early LTO debug flag parameter
12524 and generate different sections and names if set. Add generation
12525 counter for the labels so we can have multiple of them.
12526 (reset_dies): Helper to allow DIEs to be output multiple times.
12527 (dwarf2out_finish): When outputting DIEs to the fat part of an
12528 LTO object first reset DIEs.
12529 (dwarf2out_early_finish): Output early DIEs when generating LTO.
12530 (modified_type_die): Check for decl_ultimate_origin being self
12532 (gen_type_die_with_usage): Likewise.
12533 (gen_typedef_die): Allow decl_ultimate_origin being self.
12534 (set_decl_abstract_flags): Remove.
12535 (set_block_abstract_flags): Likewise.
12536 (dwarf2out_abstract_function): Treat the early generated DIEs
12537 as the abstract copy and only add DW_AT_inline and
12538 DW_AT_artificial here and call set_decl_origin_self.
12539 If the DIE has an abstract origin don't do anything.
12540 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
12541 if we have none yet (Go fails to build one, PR78628).
12542 (variably_modified_type_p): Prevent endless recursion for Ada
12543 cyclic pointer types.
12544 * lto-streamer-in.c: Include debug.h.
12545 (dref_queue): New global.
12546 (lto_read_tree_1): Stream in DIE references.
12547 (lto_input_tree): Register DIE references.
12548 (input_function): Stream DECL_DEBUG_ARGS.
12549 * lto-streamer-out.c: Include debug.h.
12550 (lto_write_tree_1): Output DIE references.
12551 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
12552 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
12553 (output_function): Stream DECL_DEBUG_ARGS.
12554 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
12555 Stream DECL_ABSTRACT_ORIGIN.
12556 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
12557 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
12558 DECL_CONTEXT for file-scope decls.
12559 * lto-streamer.h (struct dref_entry): Declare.
12560 (dref_queue): Likewise.
12561 * cfgexpand.c (pass_expand::execute): Do not call the
12562 outlining_inline_function hook here.
12563 * lto-wrapper.c (debug_obj): New global.
12564 (tool_cleanup): Unlink it if required.
12565 (debug_objcopy): New function.
12566 (run_gcc): Handle early debug sections in the IL files by
12567 extracting them to separate files, partially linkin them and
12568 feeding the result back as result to the linker.
12569 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
12570 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
12571 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
12572 sections into a separate segment.
12573 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
12575 (darwin_asm_dwarf_section): Likewise.
12576 (darwin_asm_output_dwarf_offset): Likewise.
12577 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
12579 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12580 Alan Hayward <alan.hayward@arm.com>
12581 David Sherwood <david.sherwood@arm.com>
12583 * read-md.h (md_reader::record_potential_iterator_use): Replace
12584 pointer argument with an rtx and an index.
12585 * read-rtl.c (iterator_group::apply_iterator): Likewise.
12586 (apply_mode_iterator): Likewise.
12587 (apply_code_iterator): Likewise.
12588 (apply_int_iterator): Likewise.
12589 (apply_subst_iterator): Likewise.
12590 (record_iterator_use): Likewise.
12591 (record_attribute_use): Likewise.
12592 (md_reader::record_potential_iterator_use): Likewise. Update calls
12593 to record_iterator_use and apply_iterator.
12594 (iterator_use): Replace ptr with x and index.
12595 (attribute_use): Likewise.
12596 (apply_attribute_uses): Update calls to apply_iterator.
12597 (apply_iterators): Likewise. Update initialization of iterator_use.
12598 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
12599 and record_potential_iterator_use.
12600 (rtx_reader::read_rtx_operand): Likewise.
12602 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12603 Alan Hayward <alan.hayward@arm.com>
12604 David Sherwood <david.sherwood@arm.com>
12606 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
12609 2017-08-21 Richard Biener <rguenther@suse.de>
12611 PR middle-end/81884
12612 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
12613 at struct end conservatively when comparing common bases.
12615 2017-08-21 Richard Biener <rguenther@suse.de>
12617 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
12618 (mem_ref_in_stmt): Remove.
12619 (determine_max_movement): Use ref index to get at the reference.
12620 (invariantness_dom_walker::before_dom_children): Deal with
12621 lim data already initialized.
12622 (gather_mem_refs_stmt): Initialize lim data and record ref index.
12624 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12626 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
12627 (TARGET_ISA_ROUND): Ditto.
12628 (TARGET_ROUND): Ditto.
12629 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
12630 * config/i386/i386.md: Ditto.
12631 * config/i386/sse.md: Ditto.
12632 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
12633 with OPTION_MASK_ISA_SSE4_1.
12635 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12638 * doc/extend.texi (x86 Built-in Functions): Correct the name of
12639 __builtin_ia32_lzcnt_u16.
12641 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
12644 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
12645 (rs6000_set_current_function): Rewrite function to use it.
12647 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
12650 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
12651 and TYPE_WARN_IF_NOT_ALIGN.
12652 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
12653 (handle_warn_if_not_align): New.
12654 (place_union_field): Call handle_warn_if_not_align.
12655 (place_field): Call handle_warn_if_not_align.
12656 Copy TYPE_WARN_IF_NOT_ALIGN.
12657 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
12658 (layout_type): Likewise.
12659 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
12661 (tree_decl_common): Add warn_if_not_align.
12662 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
12663 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
12664 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
12665 (DECL_WARN_IF_NOT_ALIGN): Likewise.
12666 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
12667 * doc/extend.texi: Document warn_if_not_aligned attribute.
12668 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
12670 2017-08-17 Martin Liska <mliska@suse.cz>
12673 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
12674 (get_data_dependence): Use it as pointer type.
12675 (distribute_loop): Likewise.
12677 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12679 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
12680 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
12681 (p8_vmrgow_<mode>_direct): New define_insn.
12682 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
12683 handle endianness for vmrgew and vmrgow permute patterns.
12685 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
12687 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
12688 * config/rs6000/rs6000-cpus.def: Remove comment.
12689 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
12690 (POWERPC_MASKS): Likewise.
12691 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
12692 use of TARGET_VSX_TIMODE.
12693 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
12694 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
12695 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
12696 (rs6000_option_override_internal): Remove dead code.
12697 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
12698 (rs6000_legitimize_reload_address): Likewise.
12699 (rs6000_legitimate_address_p): Likewise.
12700 (rs6000_opt_masks): Delete "vsx-timode".
12701 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
12702 from function comment.
12703 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
12704 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
12705 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
12707 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
12708 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
12709 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
12710 (VSX_TI): Likewise.
12712 (define_peephole2): Likewise.
12714 2017-08-17 Martin Sebor <msebor@redhat.com>
12717 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
12718 past the end of an array.
12719 (test_pp_format): Add test cases.
12721 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
12723 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
12724 missing ECF_NOTHROW flags.
12726 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
12729 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
12730 operands residing in integer registers.
12731 (*vsx_le_perm_load_<mode>): Likewise.
12732 (*vsx_le_perm_store_<mode>): Likewise.
12733 (define_peephole2): Add peepholes to optimize the above.
12735 2017-08-17 Marek Polacek <polacek@redhat.com>
12737 PR middle-end/81814
12738 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
12739 to mimic what shorten_compare did. Change the return type to bool.
12740 (fold_cond_expr_with_comparison): Update call to
12741 operand_equal_for_comparison_p.
12742 (fold_ternary_loc): Likewise.
12744 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
12746 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
12748 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
12750 2017-08-17 Richard Biener <rguenther@suse.de>
12752 * tree-ssa-structalias.c (solve_graph): When propagating
12753 to successors update the graphs succ edges and avoid duplicate work.
12755 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
12758 * config/i386/i386.c (ix86_option_override_internal): Save target
12759 specific options after ix86_stack_protector_guard_reg was changed.
12761 2017-08-17 Richard Biener <rguenther@suse.de>
12763 PR tree-optimization/81827
12764 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
12765 (new_var_info): Initialize it conservatively.
12766 (get_call_vi): Mark register vars.
12767 (new_scalar_tmp_constraint_exp): Likewise.
12768 (handle_rhs_call): Likewise.
12769 (handle_const_call): Likewise.
12770 (create_function_info_for): Likewise.
12771 (solve_constraints): Sort varinfos to separate register from
12772 non-register vars to pack points-to solution bitmaps during
12775 2017-08-17 Marek Polacek <polacek@redhat.com>
12777 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
12779 2017-08-17 Richard Biener <rguenther@suse.de>
12781 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
12782 to true when overflow is undefined and we saturated the result.
12784 2017-08-17 Alan Modra <amodra@gmail.com>
12787 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
12788 Don't use store multiple if only one reg needs saving.
12789 (interesting_frame_related_regno): New function.
12790 (rs6000_frame_related): Don't emit frame info for regs that
12792 (rs6000_emit_epilogue): Likewise.
12794 2017-08-16 Nathan Sidwell <nathan@acm.org>
12796 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
12797 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
12798 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
12799 * tree.c (free_lang_data_in_type): Use else-if chain. Always
12800 clear TYPE_LANG_1. Remove obsolete member-function stripping.
12801 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
12802 (verify_type): Adjust for TYPE_BINFO move.
12803 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
12804 process TYPE_BINFO directly.
12805 (hash_tree): Likewise.
12806 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
12808 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
12811 2017-08-16 David Malcolm <dmalcolm@redhat.com>
12813 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
12815 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
12818 * config/i386/i386.md (*anddi_1_btr): Change predicates of
12819 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
12820 Add ix86_binary_operator_ok to insn constraint.
12821 (*iordi_1_bts): Ditto.
12822 (*xordi_1_btc): Ditto.
12823 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
12824 Update corresponding peephole2 pattern.
12828 2017-08-16 Bin Cheng <bin.cheng@arm.com>
12830 PR tree-optimization/81832
12831 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
12832 copy loop header which has IFN_LOOP_DIST_ALIAS call.
12834 2017-08-16 Marek Polacek <polacek@redhat.com>
12837 * fold-const.c (fold_indirect_ref_1): Restore original behavior
12838 regarding size_zero_node.
12840 2017-08-16 Martin Liska <mliska@suse.cz>
12843 * config.gcc: Respect previously set extra_objs in case
12846 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
12848 PR tree-optimization/81835
12849 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
12850 the phi SSA_NAME. Check that the condition in a COND_EXPR does
12851 not depend on the phi.
12853 2017-08-16 Alan Modra <amodra@gmail.com>
12855 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
12858 2017-08-16 Alan Modra <amodra@gmail.com>
12860 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
12861 (save_reg_p): ..into this. Update all callers.
12862 (first_reg_to_save): Simplify.
12864 2017-08-16 Alan Modra <amodra@gmail.com>
12866 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
12869 2017-08-15 Joseph Myers <joseph@codesourcery.com>
12873 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
12874 constant count if that count is less than 32.
12876 2017-08-15 Nathan Sidwell <nathan@acm.org>
12878 * gcc.c (execute): Emit friendlier message if inferior is killed
12879 by an external cause.
12881 2017-08-15 Richard Biener <rguenther@suse.de>
12883 PR tree-optimization/81790
12884 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
12885 CONSTRUCTORs from simplifying and VN.
12887 2017-08-14 Martin Sebor <msebor@redhat.com>
12889 * builtin-attrs.def: Add comments.
12891 2017-08-14 Martin Sebor <msebor@redhat.com>
12894 * doc/extend.texi (attribute nonstring): Document new attribute.
12896 2017-08-14 Martin Sebor <msebor@redhat.com>
12899 * tree-diagnostic.c (default_tree_printer): Handle %G.
12900 * gimple-pretty-print.h (percent_G_format): Declare new function.
12901 * gimple-pretty-print.c (percent_G_format): Define.
12902 * tree-pretty-print.c (percent_K_format): Add argument.
12904 2017-08-14 Martin Sebor <msebor@redhat.com>
12906 PR translation/79998
12907 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
12908 Remove a stray space.
12910 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
12913 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
12914 (*iordi_1_bts): Ditto.
12915 (*xordi_1_btc): Ditto.
12917 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12920 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
12921 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12923 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
12924 quoted strings, and make more translator-friendly.
12925 (darwin_rs6000_override_options): Likewise.
12926 (rs6000_option_override_internal): Likewise.
12927 (rs6000_return_in_memory): Fix overlong line.
12928 (init_cmulative_args): Use quoted strings, and make more
12929 translator-friendly.
12930 (rs6000_pass_by_reference): Fix overlong line.
12931 (def_builtin): Use quoted strings.
12932 (altivec_expand_predicate_builtin): Use quoted strings, and make
12933 more translator-friendly.
12934 (htm_expand_builtin): Use quoted strings.
12935 (cpu_expand_builtin): Use quoted strings, and make more
12936 translator-friendly.
12937 (altivec_expand_builtin): Likewise.
12938 (paired_expand_predicate_builtin): Likewise.
12939 (rs6000_invalid_builtin): Likewise.
12940 (builtin_function_type): Use quoted strings.
12941 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
12942 more translator-friendly.
12943 (rs6000_trampoline_init): Likewise.
12944 (rs6000_handle_altivec_attribute): Likewise.
12945 (rs6000_inner_target_options): Use quoted strings.
12946 (rs6000_disable_incompatible_switches): Likewise.
12947 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
12948 strings, and make more translator-friendly.
12949 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
12951 2017-08-14 Bin Cheng <bin.cheng@arm.com>
12953 PR tree-optimization/81799
12954 * tree-loop-distribution.c (version_loop_by_alias_check): Force
12955 cond_expr to simple gimple operand.
12957 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
12959 PR middle-end/46932
12960 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
12962 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
12966 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
12967 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
12968 TARGET_GASISR_PROLOGUES.
12969 * config/avr/avr.c (avr_option_override): Same.
12970 (avr_pass_pre_proep::execute): Same.
12972 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
12975 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
12976 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
12978 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
12980 * config/i386/i386.md (*load_tp_<mode>): Redefine as
12981 define_insn_and_split. Split to a memory load from 0 in
12982 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
12983 using PTR mode iterator.
12984 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
12985 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
12986 (*add_tp_<mode>): Redefine as define_insn_and_split.
12987 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
12988 address space. Merge with *add_tp_x32 using PTR mode iterator.
12989 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
12990 Split to an add with a memory load from 0 in
12991 DEFAULT_TLS_SEG_REG address space.
12993 2017-08-12 Andrew Pinski <apinski@cavium.com>
12995 * config/aarch64/aarch64-option-extensions.def (rdma):
12996 Fix feature string to what Linux prints out in /proc/cpuinfo.
12998 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
13001 * dwarf2out.c (modified_type_die): For C typedef types that have
13002 an ultimate origin, process the ultimate origin instead of the
13004 (gen_typedef_die): Assert that input DECLs have no ultimate
13006 (gen_type_die_with_usage): For typedef variants that have an
13007 ultimate origin, just call gen_decl_die on the original DECL.
13008 (process_scope_var): Avoid creating DIEs for local typedefs and
13009 concrete static variables.
13011 2017-08-12 Alan Modra <amodra@gmail.com>
13015 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
13016 match gnu-user.h startfile.
13017 (ENDFILE_LINUX_SPEC): Similarly.
13019 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
13022 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
13024 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
13026 2017-08-11 Tamar Christina <tamar.christina@arm.com>
13027 * config/aarch64/aarch64.md (mov<mode>): Change.
13028 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
13029 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
13030 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
13032 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
13034 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
13035 for storage order barriers.
13037 2017-08-11 Martin Liska <mliska@suse.cz>
13040 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
13041 variables of void type.
13043 2017-08-11 Martin Liska <mliska@suse.cz>
13045 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
13046 TARGET_SUPPORTS_ALIASES.
13047 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
13048 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
13049 (optimize_weakref): Likewise.
13050 * symtab.c (symtab_node::noninterposable_alias): Likewise.
13051 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
13052 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
13054 2017-08-11 Martin Liska <mliska@suse.cz>
13057 * config/i386/i386.c (make_resolver_func): Do complete
13058 refactoring of the function.
13060 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13063 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
13064 * config/i386/i386.c (ix86_stack_protect_guard): Use
13065 ix86_stack_protect_guard_symbol_str to generate varible declaration.
13066 * doc/invoke.texi (x86 Options): Document
13067 -mstack-protector-guard-symbol= option.
13069 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13071 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
13072 * config/i386/i386.c (ix86_split_stack_guard): New function.
13073 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
13074 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
13075 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
13076 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
13077 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
13078 (split_stack_space_check): Call ix86_split_stack_guard.
13080 2017-08-10 Martin Sebor <msebor@redhat.com>
13082 * print-tree.c (print_node): Print location using the established
13084 Replace spaces with colons.
13085 (debug_raw, debug): Ditto.
13087 2017-08-10 Martin Sebor <msebor@redhat.com>
13090 * pretty-print.c (pp_format): Correct the handling of %s precision.
13092 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
13095 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13097 (ix86_finalize_stack_frame_flags): This. Also clear
13098 frame_pointer_needed if -fno-omit-frame-pointer is used without
13100 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13101 with ix86_finalize_stack_frame_flags.
13102 (ix86_expand_epilogue): Likewise.
13103 (ix86_expand_split_stack_prologue): Likewise.
13104 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
13106 2017-08-10 Martin Liska <mliska@suse.cz>
13109 * c-attribs.c (handle_target_attribute):
13110 Report warning for an empty string argument of target attribute.
13112 2017-08-09 Jakub Jelinek <jakub@redhat.com>
13115 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
13117 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
13118 or DECL_NONLOCAL labels.
13119 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
13120 or DECL_NONLOCAL labels here.
13122 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
13124 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
13125 to indicate when early gimple folding has been disabled.
13126 (rs6000_gimple_fold_builtin): Add debug content.
13127 (rs6000_invalid_builtin): Fix whitespace.
13128 (rs6000_expand_builtin): Fix whitespace.
13129 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
13131 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
13134 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
13135 SAVE_MULTIPLE if not all the registers that saves, should be saved.
13137 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
13139 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
13140 (qdf24xx): Likewise.
13141 * config/aarch64/aarch64.md: Include falkor.md.
13142 * config/aarch64/falkor.md: New.
13144 2017-08-09 Marek Polacek <polacek@redhat.com>
13147 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
13148 * diagnostic.c (emit_diagnostic): Add a comment.
13149 (emit_diagnostic_valist): New function.
13151 2017-08-09 Marek Polacek <polacek@redhat.com>
13154 * input.c (make_location): New overload.
13155 * input.h (make_location): Declare.
13157 2017-08-08 Alan Modra <amodra@gmail.com>
13158 H.J. Lu <hongjiu.lu@intel.com>
13161 * gcc.c (NO_PIE_SPEC): Delete.
13162 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
13164 (LINK_PIE_SPEC): ..to here.
13165 (LINK_COMMAND_SPEC): Support -no-pie.
13166 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
13167 chain of crtbegin*.o selection, update for PIE_SPEC changes and
13169 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
13170 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
13171 (ENDFILE_CRTEND_SPEC): Similarly.
13173 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
13176 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
13177 (mstack-protector-guard-offset=): Ditto.
13178 * config/i386/i386.c (ix86_option_override): Handle
13179 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
13181 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
13182 ix86_stack_protect_guard_offset variables.
13183 (TARGET_STACK_PROTECT_GUARD): Always define.
13184 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
13185 and -mstack-protector-guard-offset= options.
13187 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13189 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
13190 boundary case for the last candidate.
13192 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13194 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
13195 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
13197 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13199 PR middle-end/19706
13200 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
13201 * config/aarch64/aarch64-builtins.c
13202 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
13203 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
13204 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
13206 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13207 Andrew Pinski <pinskia@gmail.com>
13209 PR middle-end/19706
13210 * internal-fn.def (XORSIGN): New.
13211 * optabs.def (xorsign_optab): New.
13212 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
13213 (convert_expand_mult_copysign): New.
13214 (pass_optimize_widening_mul::execute): Call
13215 convert_expand_mult_copysign.
13217 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13219 PR tree-optimization/81354
13220 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
13221 Insert on edges rather than explicitly creating landing pads.
13222 (analyze_candidates_and_replace): Commit edge inserts.
13224 2017-08-08 Richard Biener <rguenther@suse.de>
13226 PR middle-end/81719
13227 * tree-ssa-loop-niter.c: Include tree-dfa.h.
13228 (expand_simple_operations): Also look through ADDR_EXPRs with
13229 MEM_REF bases treating them as POINTER_PLUS_EXPR.
13231 2017-08-08 Richard Biener <rguenther@suse.de>
13233 PR tree-optimization/81723
13234 * tree-vect-slp.c (struct bst_traits): New hash traits.
13235 (bst_fail): New global.
13236 (vect_build_slp_tree_2): New worker, split out from ...
13237 (vect_build_slp_tree): ... this now wrapping it with using
13238 bst_fail set to cache SLP tree build fails. Properly handle
13240 (vect_analyze_slp_instance): Allocate and free bst_fail.
13242 2017-08-08 Martin Liska <mliska@suse.cz>
13245 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
13246 LABEL_DECLs that can be from a different function.
13248 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13250 PR tree-optimization/81744
13251 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
13252 loop's number of iterations.
13254 2017-08-08 Martin Liska <mliska@suse.cz>
13256 * asan.c: Include header files.
13257 * attribs.c (build_decl_attribute_variant): New function moved
13259 (build_type_attribute_qual_variant): Likewise.
13260 (cmp_attrib_identifiers): Likewise.
13261 (simple_cst_list_equal): Likewise.
13262 (omp_declare_simd_clauses_equal): Likewise.
13263 (attribute_value_equal): Likewise.
13264 (comp_type_attributes): Likewise.
13265 (build_type_attribute_variant): Likewise.
13266 (lookup_ident_attribute): Likewise.
13267 (remove_attribute): Likewise.
13268 (merge_attributes): Likewise.
13269 (merge_type_attributes): Likewise.
13270 (merge_decl_attributes): Likewise.
13271 (merge_dllimport_decl_attributes): Likewise.
13272 (handle_dll_attribute): Likewise.
13273 (attribute_list_equal): Likewise.
13274 (attribute_list_contained): Likewise.
13275 * attribs.h (lookup_attribute): New function moved from tree.[ch].
13276 (lookup_attribute_by_prefix): Likewise.
13277 * bb-reorder.c: Include header files.
13278 * builtins.c: Likewise.
13279 * calls.c: Likewise.
13280 * cfgexpand.c: Likewise.
13281 * cgraph.c: Likewise.
13282 * cgraphunit.c: Likewise.
13283 * convert.c: Likewise.
13284 * dwarf2out.c: Likewise.
13285 * final.c: Likewise.
13286 * fold-const.c: Likewise.
13287 * function.c: Likewise.
13288 * gimple-expr.c: Likewise.
13289 * gimple-fold.c: Likewise.
13290 * gimple-pretty-print.c: Likewise.
13291 * gimple.c: Likewise.
13292 * gimplify.c: Likewise.
13293 * hsa-common.c: Likewise.
13294 * hsa-gen.c: Likewise.
13295 * internal-fn.c: Likewise.
13296 * ipa-chkp.c: Likewise.
13297 * ipa-cp.c: Likewise.
13298 * ipa-devirt.c: Likewise.
13299 * ipa-fnsummary.c: Likewise.
13300 * ipa-inline.c: Likewise.
13301 * ipa-visibility.c: Likewise.
13303 * lto-cgraph.c: Likewise.
13304 * omp-expand.c: Likewise.
13305 * omp-general.c: Likewise.
13306 * omp-low.c: Likewise.
13307 * omp-offload.c: Likewise.
13308 * omp-simd-clone.c: Likewise.
13309 * opts-global.c: Likewise.
13310 * passes.c: Likewise.
13311 * predict.c: Likewise.
13312 * sancov.c: Likewise.
13313 * sanopt.c: Likewise.
13314 * symtab.c: Likewise.
13315 * toplev.c: Likewise.
13316 * trans-mem.c: Likewise.
13317 * tree-chkp.c: Likewise.
13318 * tree-eh.c: Likewise.
13319 * tree-into-ssa.c: Likewise.
13320 * tree-object-size.c: Likewise.
13321 * tree-parloops.c: Likewise.
13322 * tree-profile.c: Likewise.
13323 * tree-ssa-ccp.c: Likewise.
13324 * tree-ssa-live.c: Likewise.
13325 * tree-ssa-loop.c: Likewise.
13326 * tree-ssa-sccvn.c: Likewise.
13327 * tree-ssa-structalias.c: Likewise.
13328 * tree-ssa.c: Likewise.
13329 * tree-streamer-in.c: Likewise.
13330 * tree-vectorizer.c: Likewise.
13331 * tree-vrp.c: Likewise.
13332 * tsan.c: Likewise.
13333 * ubsan.c: Likewise.
13334 * varasm.c: Likewise.
13335 * varpool.c: Likewise.
13336 * tree.c: Remove functions moved to attribs.[ch].
13337 * tree.h: Likewise.
13338 * config/aarch64/aarch64.c: Add attrs.h header file.
13339 * config/alpha/alpha.c: Likewise.
13340 * config/arc/arc.c: Likewise.
13341 * config/arm/arm.c: Likewise.
13342 * config/avr/avr.c: Likewise.
13343 * config/bfin/bfin.c: Likewise.
13344 * config/c6x/c6x.c: Likewise.
13345 * config/cr16/cr16.c: Likewise.
13346 * config/cris/cris.c: Likewise.
13347 * config/darwin.c: Likewise.
13348 * config/epiphany/epiphany.c: Likewise.
13349 * config/fr30/fr30.c: Likewise.
13350 * config/frv/frv.c: Likewise.
13351 * config/ft32/ft32.c: Likewise.
13352 * config/h8300/h8300.c: Likewise.
13353 * config/i386/winnt.c: Likewise.
13354 * config/ia64/ia64.c: Likewise.
13355 * config/iq2000/iq2000.c: Likewise.
13356 * config/lm32/lm32.c: Likewise.
13357 * config/m32c/m32c.c: Likewise.
13358 * config/m32r/m32r.c: Likewise.
13359 * config/m68k/m68k.c: Likewise.
13360 * config/mcore/mcore.c: Likewise.
13361 * config/microblaze/microblaze.c: Likewise.
13362 * config/mips/mips.c: Likewise.
13363 * config/mmix/mmix.c: Likewise.
13364 * config/mn10300/mn10300.c: Likewise.
13365 * config/moxie/moxie.c: Likewise.
13366 * config/msp430/msp430.c: Likewise.
13367 * config/nds32/nds32-isr.c: Likewise.
13368 * config/nds32/nds32.c: Likewise.
13369 * config/nios2/nios2.c: Likewise.
13370 * config/nvptx/nvptx.c: Likewise.
13371 * config/pa/pa.c: Likewise.
13372 * config/pdp11/pdp11.c: Likewise.
13373 * config/powerpcspe/powerpcspe.c: Likewise.
13374 * config/riscv/riscv.c: Likewise.
13375 * config/rl78/rl78.c: Likewise.
13376 * config/rx/rx.c: Likewise.
13377 * config/s390/s390.c: Likewise.
13378 * config/sh/sh.c: Likewise.
13379 * config/sol2.c: Likewise.
13380 * config/sparc/sparc.c: Likewise.
13381 * config/spu/spu.c: Likewise.
13382 * config/stormy16/stormy16.c: Likewise.
13383 * config/tilegx/tilegx.c: Likewise.
13384 * config/tilepro/tilepro.c: Likewise.
13385 * config/v850/v850.c: Likewise.
13386 * config/vax/vax.c: Likewise.
13387 * config/visium/visium.c: Likewise.
13388 * config/xtensa/xtensa.c: Likewise.
13390 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13393 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
13394 constraints since the -mupper-regs-* switches have been
13396 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
13397 into a vector from a double word element that was extracted from
13398 another vector, and eliminate extra XXPERMDI instructions.
13399 (vsx_concat_<mode>_2): Likewise.
13400 (vsx_concat_<mode>_3): Likewise.
13401 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
13402 concat to allow optimizing inserts from previous extracts.
13404 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
13406 * config/i386/i386.c (ix86_stack_protect_guard): Generate
13407 memory reference to a SSP offset in TLS address space.
13408 (ix86_print_operand) <case '@'>: Remove.
13409 (ix86_print_operand_punct_valid_p): Remove '@' code.
13410 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
13411 UNSPEC_SP_TLS_TEST.
13412 (stack_tls_protect_set_<mode>): Remove.
13413 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
13414 (stack_tls_protect_test_<mode>): Remove.
13415 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
13417 2017-08-07 Olivier Hainque <hainque@adacore.com>
13420 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
13422 2017-08-07 Douglas Rupp <rupp@adacore.com>
13424 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
13425 variable was referenced as multidir in command.
13427 2017-08-07 Jakub Jelinek <jakub@redhat.com>
13430 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
13433 2017-08-07 Martin Liska <mliska@suse.cz>
13435 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
13436 * config/rl78/rl78.c: Add include of attribs.h.
13437 * config/sh/sh.c: Likewise.
13438 * config/v850/v850.c: Likewise.
13440 2017-08-07 Tom de Vries <tom@codesourcery.com>
13442 PR middle-end/78266
13443 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
13445 2017-08-07 Martin Liska <mliska@suse.cz>
13447 * config/mips/mips.c: Include attribs.h.
13449 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
13452 * doc/invoke.texi: Document change in behvaior for -Ofast for
13455 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
13457 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
13459 (aarch64_pop_regs): Likewise.
13460 (aarch64_gen_load_pair): Likewise.
13461 (aarch64_save_callee_saves): Likewise.
13462 (aarch64_restore_callee_saves): Likewise.
13464 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13466 * config/i386/i386.c: Revert the last change.
13468 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13471 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13473 (ix86_finalize_stack_frame_flags): This. Also clear
13474 frame_pointer_needed if -fno-omit-frame-pointer is used without
13476 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13477 with ix86_finalize_stack_frame_flags.
13478 (ix86_expand_epilogue): Likewise.
13479 (ix86_expand_split_stack_prologue): Likewise.
13481 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13484 * config/i386/i386.c (get_builtin_code_for_version): Set priority
13485 to P_AES for Westmere.
13487 2017-08-07 Jonathan Yong <10walls@gmail.com>
13489 * config/i386/mingw.opt (fset-stack-executable): Removed.
13490 * config/i386/cygming.opt (fset-stack-executable): Moved
13492 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
13494 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
13496 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
13498 2017-08-07 Marek Polacek <polacek@redhat.com>
13500 PR middle-end/81737
13501 * fold-const.c (fold_indirect_ref_1): Check type_domain.
13503 2017-08-07 Martin Liska <mliska@suse.cz>
13505 * attribs.h (canonicalize_attr_name): New function.
13506 (cmp_attribs): Move from c-format.c and adjusted.
13507 (is_attribute_p): Moved from tree.h.
13508 * tree-inline.c: Add new includes.
13509 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
13510 (private_is_attribute_p): Remove.
13511 (private_lookup_attribute): Likewise.
13512 (private_lookup_attribute_by_prefix): Simplify.
13513 (remove_attribute): Use is_attribute_p.
13514 * tree.h: Remove removed declarations.
13516 2017-08-07 Jakub Jelinek <jakub@redhat.com>
13518 PR middle-end/81698
13519 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
13520 instead of computing it in the function. Formatting fix.
13521 (expand_case): Don't rely on default_edge being the first edge,
13522 clear it if removing it, pass default_edge to
13523 emit_case_dispatch_table.
13524 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
13527 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
13529 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
13530 insn in the function, emit NOP after the insn.
13532 2017-08-06 Tom de Vries <tom@codesourcery.com>
13534 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
13537 2017-08-06 Tom de Vries <tom@codesourcery.com>
13539 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
13542 2017-08-04 Yury Gribov <tetra2005@gmail.com>
13544 PR tree-optimization/57371
13545 * match.pd: New pattern.
13547 2017-08-04 Marek Polacek <polacek@redhat.com>
13549 PR middle-end/81695
13550 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
13551 perform the computation in offset_int.
13553 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13555 PR tree-optimization/81136
13556 * tree-vectorizer.h: Include tree-hash-traits.h.
13557 (vec_base_alignments): New typedef.
13558 (vec_info): Add a base_alignments field.
13559 (vect_record_base_alignments): Declare.
13560 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
13562 (DR_IS_CONDITIONAL_IN_STMT): New macro.
13563 (create_data_ref): Add an is_conditional_in_stmt argument.
13564 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
13565 the is_conditional_in_stmt field.
13566 (data_ref_loc): Add an is_conditional_in_stmt field.
13567 (get_references_in_stmt): Set the is_conditional_in_stmt field.
13568 (find_data_references_in_stmt): Update call to create_data_ref.
13569 (graphite_find_data_references_in_stmt): Likewise.
13570 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
13571 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
13572 (vect_record_base_alignment): New function.
13573 (vect_record_base_alignments): Likewise.
13574 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
13575 for nested statements even if we fail to compute a misalignment.
13576 Use pooled base alignments for unconditional references.
13577 (vect_find_same_alignment_drs): Compare base addresses instead
13579 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
13580 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
13582 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13584 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
13585 Add an explicit name for the enum. Use auto_vec for slp_instances
13586 and grouped_stores.
13587 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
13589 (_bb_vec_info): Add a constructor and destructor.
13590 (vinfo_for_stmt): Return NULL for uids of -1 as well.
13591 (destroy_loop_vec_info): Delete.
13592 (vect_destroy_datarefs): Likewise.
13593 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
13594 (vec_info::vec_info): New function.
13595 (vec_info::~vec_info): Likewise.
13596 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
13597 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
13598 destroy_loop_vec_info.
13599 * tree-vect-loop.c (new_loop_vec_info): Replace with...
13600 (_loop_vec_info::_loop_vec_info): ...this.
13601 (destroy_loop_vec_info): Replace with...
13602 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
13603 the stmt_vec_infos. Leave handling of vec_info information to its
13604 destructor. Remove explicit vector releases.
13605 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
13606 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
13607 * tree-vect-slp.c (new_bb_vec_info): Replace with...
13608 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
13609 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
13610 (destroy_bb_vec_info): Replace with...
13611 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
13612 information to its destructor.
13613 (vect_slp_analyze_bb_1): Use new and delete instead of
13614 new_bb_vec_info and destroy_bb_vec_info.
13615 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
13618 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13620 * tree-data-ref.h (subscript): Add access_fn field.
13621 (data_dependence_relation): Add could_be_independent_p.
13622 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
13623 (same_access_functions): Move to tree-data-ref.c.
13624 * tree-data-ref.c (ref_contains_union_access_p): New function.
13625 (access_fn_component_p): Likewise.
13626 (access_fn_components_comparable_p): Likewise.
13627 (dr_analyze_indices): Add a reference to access_fn_component_p.
13628 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
13630 (constant_access_functions): Likewise.
13631 (add_other_self_distances): Likewise.
13632 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
13633 (initialize_data_dependence_relation): Use XCNEW and remove
13634 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
13635 of access functions that have the same type. Allow the
13636 subsequence to end with different bases in some circumstances.
13637 Record the chosen access functions in SUB_ACCESS_FN.
13638 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
13639 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
13640 (subscript_dependence_tester_1): Likewise dra and drb.
13641 (build_classic_dist_vector): Update calls accordingly.
13642 (subscript_dependence_tester): Likewise.
13643 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
13644 DDR_COULD_BE_INDEPENDENT_P.
13645 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
13646 comp_alias_ddrs instead of may_alias_ddrs.
13647 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
13649 (vect_analyze_data_ref_dependence): Use it if
13650 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
13651 distance vectors if that fails.
13652 (dependence_distance_ge_vf): New function.
13653 (vect_prune_runtime_alias_test_list): Use it. Don't clear
13654 LOOP_VINFO_MAY_ALIAS_DDRS.
13656 2017-08-04 Richard Biener <rguenther@suse.de>
13658 PR middle-end/81705
13659 * fold-const.c (fold_binary_loc): Properly restrict
13660 minus_var0 && minus_var1 case when associating undefined overflow
13663 2017-08-04 Tom de Vries <tom@codesourcery.com>
13665 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
13667 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13669 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13670 Don't start diagnostic messages with a capital letter.
13671 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13673 (rs6000_invalid_builtin): Likewise.
13674 (rs6000_trampoline_init): Likewise.
13676 2017-08-03 Jakub Jelinek <jakub@redhat.com>
13679 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
13680 after setting changeable df flags.
13682 2017-08-03 Richard Biener <rguenther@suse.de>
13684 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
13685 up if the use is in USE - X.
13687 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
13689 * toplev.c (dumpfile.h): New include.
13690 (internal_error_reentered): New static function. Use it...
13691 (internal_error_function): ...here to handle reentered internal_error.
13693 2017-08-03 Richard Biener <rguenther@suse.de>
13695 PR middle-end/81148
13696 * fold-const.c (split_tree): Add minus_var and minus_con
13697 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
13698 here but always use minus_*.
13699 (associate_trees): Assert we never associate with MINUS_EXPR
13700 and NULL first operand. Do not recurse for PLUS_EXPR operands
13701 when associating as MINUS_EXPR either.
13702 (fold_binary_loc): Track minus_var and minus_con.
13704 2017-08-03 Tom de Vries <tom@codesourcery.com>
13707 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
13708 ACCEL_COMPILER, apply finish_options on
13709 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
13711 2017-08-03 Tom de Vries <tom@codesourcery.com>
13714 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
13715 function_entry_patch_area_size > 0.
13717 2017-08-03 Jakub Jelinek <jakub@redhat.com>
13720 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
13721 instead of 10??LU, perform unit multiplication in wide_int,
13722 don't change alloc_object_size_limit if the limit is larger
13725 PR tree-optimization/81655
13726 PR tree-optimization/81588
13727 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
13728 the case when ranges[i].low and high are 1 for unsigned type with
13731 PR middle-end/81052
13732 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
13733 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
13735 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13737 * tree-vrp.h: Add include guard.
13739 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
13742 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
13743 (ud2): New insn pattern.
13744 * config/i386/i386.c (ix86_expand_epilogue):
13745 For naked functions, generate ud2 instead of trap insn.
13747 2017-08-02 Marek Polacek <polacek@redhat.com>
13750 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
13752 2017-08-02 Tom de Vries <tom@codesourcery.com>
13753 Cesar Philippidis <cesar@codesourcery.com>
13755 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
13756 Add missing edge probabilities.
13758 2017-08-02 Tamar Christina <tamar.christina@arm.com>
13760 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
13761 Correct endianness.
13763 2017-08-02 Jakub Jelinek <jakub@redhat.com>
13765 PR middle-end/79499
13766 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
13767 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
13768 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
13769 prologue_seq sequences - if any.
13771 2017-08-02 Richard Biener <rguenther@suse.de>
13773 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
13774 via vectors if supported, integer extracts via punning if supported
13775 or otherwise vector extracts.
13777 2017-08-02 Richard Biener <rguenther@suse.de>
13779 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
13781 (bitmap_insert_into_set): ... this.
13783 2017-08-02 Richard Biener <rguenther@suse.de>
13785 PR tree-optimization/81633
13787 2015-08-17 Alan Hayward <alan.hayward@arm.com>
13789 PR tree-optimization/71752
13790 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
13792 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
13794 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
13795 (machine_function::call_ms2sysv_pad_out): Remove field.
13796 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
13797 (ix86_compute_frame_layout): Likewise.
13799 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
13802 * config/i386/i386.c (ix86_set_func_type): Disallow naked
13803 attribute with interrupt attribute.
13805 2017-08-01 Andrew Pinski <apinski@cavium.com>
13807 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
13808 BIT_INSERT_EXPR's operand 1
13809 to see if the types precision matches.
13811 2017-08-01 Martin Liska <mliska@suse.cz>
13813 PR middle-end/70140
13814 * builtins.c (expand_builtin_memcpy_args): Remove.
13815 (expand_builtin_memcpy): Call newly added function
13816 expand_builtin_memory_copy_args.
13817 (expand_builtin_memcpy_with_bounds): Likewise.
13818 (expand_builtin_mempcpy): Remove last argument.
13819 (expand_builtin_mempcpy_with_bounds): Likewise.
13820 (expand_builtin_memory_copy_args): New function created from
13821 expand_builtin_mempcpy_args with small modifications.
13822 (expand_builtin_mempcpy_args): Remove.
13823 (expand_builtin_stpcpy): Remove unused argument.
13824 (expand_builtin): Likewise.
13825 (expand_builtin_with_bounds): Likewise.
13827 2017-08-01 Martin Liska <mliska@suse.cz>
13830 Make mempcpy more optimal (PR middle-end/70140).
13832 2017-08-01 Jakub Jelinek <jakub@redhat.com>
13835 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
13836 __builtin_vec_cmpne verify both arguments are compatible vectors
13837 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
13838 verify arg1_type is a pointer or array type. For __builtin_vec_st,
13839 move computation of aligned to after checking the argument types.
13843 * config/rs6000/vsx.md (vextract_fp_from_shorth,
13844 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
13847 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
13848 Doug Rupp <rupp@adacore.com>
13849 Olivier Hainque <hainque@adacore.com>
13851 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
13852 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
13854 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
13855 for TARGET_OS_CPP_BUILTIN.
13856 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
13857 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
13859 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
13860 passing required abi options to the assembler for EABI configurations.
13861 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
13862 of .text.hot and .text.unlikely sections for kernel modules when
13863 using ARM style exceptions.
13864 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
13865 options. Add EXTRA_CC1_SPEC.
13866 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
13868 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
13870 (ARM_TARGET2_DWARF_FORMAT): Define.
13871 * config/arm/t-vxworks: Adjust multilib control to removal of the
13872 Diab command line options.
13874 2017-08-01 Martin Liska <mliska@suse.cz>
13876 PR gcov-profile/81561
13877 * gcov.c (unblock): Make unblocking safe as we need to preserve
13878 index correspondence of blocks and block_lists.
13880 2017-08-01 Richard Biener <rguenther@suse.de>
13882 PR tree-optimization/81181
13883 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
13884 (compute_antic): ... end of iteration here.
13886 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
13888 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
13889 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
13890 (ftree-slp-vectorize): Likewise.
13891 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
13892 can no longer be set independent of flag_tree_loop_vectorize.
13893 * omp-general.c (emp_max_vf): Likewise.
13894 * opts.c (enable_fdo_optimizations): Remove references to
13895 flag_tree_vectorize, these are now implicit.
13896 (common_handle_option): Remove handling for OPT_ftree_vectorize,
13897 and leave it for the options machinery.
13899 2017-08-01 Martin Liska <mliska@suse.cz>
13901 PR middle-end/70140
13902 * builtins.c (expand_builtin_memcpy_args): Remove.
13903 (expand_builtin_memcpy): Call newly added function
13904 expand_builtin_memory_copy_args.
13905 (expand_builtin_memcpy_with_bounds): Likewise.
13906 (expand_builtin_mempcpy): Remove last argument.
13907 (expand_builtin_mempcpy_with_bounds): Likewise.
13908 (expand_builtin_memory_copy_args): New function created from
13909 expand_builtin_mempcpy_args with small modifications.
13910 (expand_builtin_mempcpy_args): Remove.
13911 (expand_builtin_stpcpy): Remove unused argument.
13912 (expand_builtin): Likewise.
13913 (expand_builtin_with_bounds): Likewise.
13915 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
13918 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
13919 print "ds:" only for immediates in generic address space.
13921 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
13924 * config/i386/i386.c (ix86_funciton_naked): New prototype.
13925 (ix86_function_ok_for_sibcall): Return false for naked functions.
13927 2017-08-01 Richard Biener <rguenther@suse.de>
13929 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
13930 (compute_antic): Seed worklist with exit block predecessors.
13931 * cfganal.c (dfs_find_deadend): For a cycle return the source
13932 of the edge closing it.
13934 2017-08-01 Tamar Christina <tamar.christina@arm.com>
13936 * config/aarch64/aarch64.c
13937 (aarch64_can_const_movi_rtx_p): Move 0 check.
13939 2017-08-01 Bin Cheng <bin.cheng@arm.com>
13941 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
13942 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
13944 * match.pd: Ditto in address comparison pattern.
13946 2017-08-01 Bin Cheng <bin.cheng@arm.com>
13948 PR tree-optimization/81627
13949 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
13950 closed ssa form for store-store chain.
13952 2017-08-01 Bin Cheng <bin.cheng@arm.com>
13954 PR tree-optimization/81620
13955 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
13956 for store-store chain.
13958 2017-08-01 Jakub Jelinek <jakub@redhat.com>
13960 PR tree-optimization/81588
13961 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
13962 ranges[i].in_p, invert comparison code ccode. For >/>=,
13963 swap rhs1 and rhs2 and comparison code unconditionally,
13964 for </<= don't do that. Don't swap rhs1/rhs2 again if
13965 ranges[i].in_p, instead invert comparison code ccode if
13966 opcode or oe->rank is BIT_IOR_EXPR.
13969 * optabs.def (vec_extract_optab, vec_init_optab): Change from
13970 a direct optab to conversion optab.
13971 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
13972 with GET_MODE_INNER as last argument instead of optab_handler.
13973 * expmed.c (extract_bit_field_1): Likewise. Use vector from
13974 vector extraction if possible and optab is available.
13975 * expr.c (store_constructor): Use convert_optab_handler instead
13976 of optab_handler. Use vector initialization from smaller
13977 vectors if possible and optab is available.
13978 * tree-vect-stmts.c (vectorizable_load): Likewise.
13979 * doc/md.texi (vec_extract, vec_init): Document that the optabs
13980 now have two modes.
13981 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
13982 of vec_init from half-sized vectors with the same element mode.
13983 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
13984 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
13985 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
13986 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
13987 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
13988 after mode in gen_vec_extract* calls.
13989 (vec_extract<mode>): Renamed to ...
13990 (vec_extract<mode><ssescalarmodelower>): ... this.
13991 (vec_extract<mode><ssehalfvecmodelower>): New expander.
13992 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
13993 element mode after mode in gen_vec_init* calls.
13994 (VEC_INIT_HALF_MODE): New mode iterator.
13995 (vec_init<mode>): Renamed to ...
13996 (vec_init<mode><ssescalarmodelower>): ... this.
13997 (vec_init<mode><ssehalfvecmodelower>): New expander.
13998 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
13999 (vec_extractv2sfsf): ... this.
14000 (vec_initv2sf): Renamed to ...
14001 (vec_initv2sfsf): ... this.
14002 (vec_extractv2si): Renamed to ...
14003 (vec_extractv2sisi): ... this.
14004 (vec_initv2si): Renamed to ...
14005 (vec_initv2sisi): ... this.
14006 (vec_extractv4hi): Renamed to ...
14007 (vec_extractv4hihi): ... this.
14008 (vec_initv4hi): Renamed to ...
14009 (vec_initv4hihi): ... this.
14010 (vec_extractv8qi): Renamed to ...
14011 (vec_extractv8qiqi): ... this.
14012 (vec_initv8qi): Renamed to ...
14013 (vec_initv8qiqi): ... this.
14014 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
14015 (vec_init<mode>): Renamed to ...
14016 (vec_init<mode><VEC_base_l>): ... this.
14017 (vec_extract<mode>): Renamed to ...
14018 (vec_extract<mode><VEC_base_l>): ... this.
14019 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
14020 (vec_initv2sfsf): ... this.
14021 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
14022 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14023 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
14024 element mode after mode in gen_vec_init* calls.
14025 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
14026 (vec_init<mode><Vel>): ... this.
14027 (vec_extract<mode>): Renamed to ...
14028 (vec_extract<mode><Vel>): ... this.
14029 * config/aarch64/iterators.md (Vel): New mode attribute.
14030 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
14031 Add element mode after mode in gen_vec_extract* calls.
14032 * config/s390/vector.md (non_vec_l): New mode attribute.
14033 (vec_extract<mode>): Renamed to ...
14034 (vec_extract<mode><non_vec_l>): ... this.
14035 (vec_init<mode>): Renamed to ...
14036 (vec_init<mode><non_vec_l>): ... this.
14037 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
14038 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
14040 * config/arm/iterators.md (V_elem_l): New mode attribute.
14041 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
14042 (vec_extract<mode><V_elem_l>): ... this.
14043 (vec_extractv2di): Renamed to ...
14044 (vec_extractv2didi): ... this.
14045 (vec_init<mode>): Renamed to ...
14046 (vec_init<mode><V_elem_l>): ... this.
14047 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
14048 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
14049 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
14050 Add element mode after gen_vec_extract* calls.
14051 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
14052 (vec_init<mode><unitmode>): ... this.
14053 (vec_extract<mode>): Renamed to ...
14054 (vec_extract<mode><unitmode>): ... this.
14055 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
14056 (vec_init<mode><unitmode>): ... this.
14057 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
14058 (vec_initv2sfsf): ... this.
14059 (vec_extractv2sf): Renamed to ...
14060 (vec_extractv2sfsf): ... this.
14061 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
14062 Add element mode after gen_vec_extract* calls.
14063 * config/mips/mips.md (unitmode): New mode iterator.
14064 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
14065 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
14066 * config/spu/spu.md (inner_l): New mode attribute.
14067 (vec_init<mode>): Renamed to ...
14068 (vec_init<mode><inner_l>): ... this.
14069 (vec_extract<mode>): Renamed to ...
14070 (vec_extract<mode><inner_l>): ... this.
14071 * config/sparc/sparc.md (veltmode): New mode iterator.
14072 (vec_init<VMALL:mode>): Renamed to ...
14073 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
14074 * config/ia64/vect.md (vec_initv2si): Renamed to ...
14075 (vec_initv2sisi): ... this.
14076 (vec_initv2sf): Renamed to ...
14077 (vec_initv2sfsf): ... this.
14078 (vec_extractv2sf): Renamed to ...
14079 (vec_extractv2sfsf): ... this.
14080 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
14081 (vec_init<mode>): Renamed to ...
14082 (vec_init<mode><VEC_base_l>): ... this.
14083 (vec_extract<mode>): Renamed to ...
14084 (vec_extract<mode><VEC_base_l>): ... this.
14085 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
14086 (vec_initv2sfsf): ... this.
14087 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
14088 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14089 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
14090 gen_vec_init* calls.
14092 2017-08-01 Richard Biener <rguenther@suse.de>
14094 PR tree-optimization/81297
14095 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
14096 TREE_OVERFLOW from INTEGER_CSTs.
14098 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
14100 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
14102 2017-07-31 Carl Love <cel@us.ibm.com>
14104 * config/rs6000/rs6000-c: Add support for built-in functions
14105 vector signed char vec_xl_be (signed long long, signed char *);
14106 vector unsigned char vec_xl_be (signed long long, unsigned char *);
14107 vector signed int vec_xl_be (signed long long, signed int *);
14108 vector unsigned int vec_xl_be (signed long long, unsigned int *);
14109 vector signed long long vec_xl_be (signed long long, signed long long *);
14110 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
14111 vector signed short vec_xl_be (signed long long, signed short *);
14112 vector unsigned short vec_xl_be (signed long long, unsigned short *);
14113 vector double vec_xl_be (signed long long, double *);
14114 vector float vec_xl_be (signed long long, float *);
14115 * config/rs6000/altivec.h (vec_xl_be): Add #define.
14116 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
14117 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
14119 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
14120 (altivec_expand_builtin): Add switch statement to call
14121 altivec_expand_xl_be for each builtin.
14122 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
14123 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
14124 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
14125 __builtin_vsx_le_be_v16qi.
14126 * doc/extend.texi: Update the built-in documentation file for the
14127 new built-in functions.
14129 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14132 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
14134 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
14136 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14138 * config.gcc: Add z14.
14139 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
14140 CPU model numbers for z13s and z14.
14141 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
14143 * config/s390/s390-opts.h (enum processor_type): Rename
14144 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14145 * config/s390/s390.c (processor_table): Add field for CPU name to
14146 be passed to Binutils.
14147 (s390_asm_output_machine_for_arch): Use the new field in
14148 processor_table for Binutils.
14149 (s390_expand_builtin): Replace arch12 with z14.
14150 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14151 (s390_get_sched_attrmask): Likewise.
14152 (s390_get_unit_mask): Likewise.
14153 * config/s390/s390.opt: Add z14 to processor_type enum.
14155 2017-07-31 Martin Jambor <mjambor@suse.cz>
14158 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
14159 regardless of optimization level.
14161 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
14162 Martin Liska <mliska@suse.cz>
14164 * predict.def: Remove old comment and adjust probability.
14165 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
14166 PREDICT statements.
14168 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14171 * config/i386/i386.c (ix86_function_naked): New function.
14172 (ix86_can_use_return_insn_p): Return false for naked functions.
14173 (ix86_expand_prologue): Skip prologue for naked functions.
14174 (ix86_expand_epilogue): Skip epilogue for naked functions
14175 and emit trap instruction.
14176 (ix86_warn_func_return): New function.
14177 (ix86_attribute_table): Add "naked" attribute specification.
14178 (TARGET_WARN_FUNC_RETURN): Define.
14179 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
14181 2017-07-31 Martin Liska <mliska@suse.cz>
14183 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
14184 (dump_gimple_bb_header): Always dump BB info.
14185 (pp_cfg_jump): Do not append info about BB when dumping a jump.
14187 2017-07-31 Martin Liska <mliska@suse.cz>
14190 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
14191 also with current_function_decl non-null equality.
14193 2017-07-31 Jakub Jelinek <jakub@redhat.com>
14196 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
14197 change type to the element type, instead add eltype variable and
14198 use it where we are interested in the element type.
14200 PR tree-optimization/81603
14201 * ipa-polymorphic-call.c
14202 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
14203 offset arithmetic in offset_int, bail out if the resulting bit offset
14204 doesn't fit into shwi.
14206 2017-07-31 Martin Liska <mliska@suse.cz>
14208 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
14209 (gimplify_save_expr): Fix comment.
14211 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
14214 * config/i386/i386.c (ix86_function_arg): Update arguments for
14216 (ix86_compute_frame_layout): Set the initial stack offset to
14217 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
14218 INCOMING_FRAME_SP_OFFSET.
14219 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
14220 stack before exception handler returns.
14221 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
14222 the 'ERROR_CODE' for exception handler.
14224 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
14226 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
14227 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
14228 (ASM_OUTPUT_REG_POP): Ditto.
14229 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
14230 instead of asm_fprintf to output pure string.
14232 2017-07-29 Jakub Jelinek <jakub@redhat.com>
14234 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
14235 to imported_module_or_decl hook.
14236 (debug_nothing_tree_tree_tree_bool): Remove.
14237 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
14238 * debug.c (do_nothing_debug_hooks): Use
14239 debug_nothing_tree_tree_tree_bool_bool instead of
14240 debug_nothing_tree_tree_tree_bool.
14241 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14242 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
14243 * sdbout.c (sdb_debug_hooks): Likewise.
14244 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
14245 (gen_namespace_die): Add DW_AT_export_symbols attribute if
14247 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
14248 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
14249 attribute, don't add anything.
14251 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14253 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
14254 (fold_build2_stat_loc): Likewise.
14255 (fold_build3_stat_loc): Likewise.
14256 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
14257 (fold_build1_loc): Remove macro.
14258 (fold_build2_loc): Likewise.
14259 (fold_build3_loc): Likewise.
14261 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14263 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
14264 (gimple_build_debug_bind_source_stat): Likewise.
14265 * gimple.h (gimple_build_debug_bind): Remove macro.
14266 (gimple_build_debug_bind_source): Likewise.
14268 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14270 * bitmap.c (bitmap_alloc): Adjust.
14271 (bitmap_gc_alloc): Likewise.
14272 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
14274 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14276 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
14277 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
14278 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
14279 (bitmap_gc_alloc_stat): Likewise.
14280 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
14282 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14284 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
14285 * rtl.h (shallow_copy_rtx): Remove macro.
14287 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14289 * emit-rtl.c (gen_raw_REG): Adjust.
14290 * gengenrtl.c (gendef): Likewise.
14291 * rtl.c (rtx_alloc_stat): Remove _stat from name.
14292 * rtl.h (rtx_alloc): Remove macro.
14294 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14296 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
14297 (build_tree_list_stat): Likewise.
14298 * tree.h (build_tree_list): Remove macro.
14299 (build_tree_list_vec): Likewise.
14301 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14303 * tree.c (make_vector_stat): Remove _stat from name.
14304 (build_vector_stat): Likewise.
14305 * tree.h (make_vector_stat): Remove macro.
14306 (build_vector_stat): Likewise.
14308 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14310 * tree.h (build_var_debug_value): Remove prototype.
14312 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14314 * tree.c (tree_cons_stat): Remove _stat from name.
14315 * tree.h (tree_cons): Remove macro.
14317 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14319 * tree.c (build_vl_exp_stat): Remove _stat from name.
14320 * tree.h (build_vl_exp): Remove macro.
14322 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14324 * tree.c (build_decl_stat): Remove _stat from name.
14325 * tree.h (build_decl): Remove macro.
14327 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14329 * gimple.c (gimple_build_with_ops_stat): Adjust.
14330 (gimple_alloc_stat): Remove _stat from name.
14331 * gimple.h (gimple_alloc): Remove macro.
14333 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14335 * tree.c (make_tree_vec_stat): Remove _stat from name.
14336 (grow_tree_vec_stat): Likewise.
14337 * tree.h (make_tree_vec_stat): Adjust prototype.
14338 (grow_tree_vec_stat): Likewise.
14339 (make_tree_vec): Remove macro.
14340 (grow_tree_vec): Likewise.
14342 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14344 * fold-const.c (fold_build1_stat_loc): Adjust.
14345 (fold_build2_stat_loc): Likewise.
14346 (fold_build3_stat_loc): Likewise.
14347 * tree.c (build0_stat): Remove _stat from name.
14348 (build1_stat): Likewise.
14349 (build2_stat): Likewise.
14350 (build3_stat): Likewise.
14351 (build4_stat): Likewise.
14352 (build5_stat): Likewise.
14353 * tree.h (build1_loc): Remove macro, and rename _stat function
14355 (build2_loc): Likewise.
14356 (build3_loc): Likewise.
14357 (build4_loc): Likewise.
14358 (build5_loc): Likewise.
14360 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14362 * tree.c (make_int_cst_stat): Remove _stat from name.
14363 * tree.h (make_int_cst_stat): Adjust prototype.
14364 (make_int_cst): Remove macro.
14366 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14368 * tree.c (make_tre_binfo_stat): Remove _stat from name.
14369 * tree.h (make_tree_binfo_stat): Adjust prototype.
14370 (make_tree_binfo): Remove.
14372 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14374 * tree.c (copy_node_stat): Rename to copy_node.
14375 (build_distinct_type_copy): Adjust.
14376 * tree.h (copy_node_stat): Adjust prototype.
14377 (copy_node): Remove macro.
14379 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14381 * tree.c (make_node_stat): rename to make_node.
14382 (build_tree_list_stat): Adjust.
14383 (build0_stat): Likewise.
14384 (build2_stat): Likewise.
14385 (build3_stat): Likewise.
14386 (build4_stat): Likewise.
14387 (build5_stat): Likewise.
14388 (build_decl_stat): Likewise.
14389 * tree.h (make_node_stat): Adjust prototype.
14390 (make_node): remove macro.
14392 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
14394 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
14395 (PPC_FEATURE2_SCV): Likewise.
14396 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
14398 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14400 * config/aarch64/aarch64.c
14401 (aarch64_internal_mov_immediate): Add new special pattern.
14402 * config/aarch64/aarch64.md (*movdi_aarch64):
14403 Add reg/32bit const mov case.
14405 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14406 Richard Sandiford <richard.sandiford@linaro.org>
14408 * config/aarch64/aarch64.md (mov<mode>): Generalize.
14409 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14410 Add integer and movi cases.
14411 (movi-split-hf-df-sf split, fp16): New.
14412 (enabled): Added TARGET_FP_F16INST.
14413 * config/aarch64/iterators.md (GPF_HF): New.
14414 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
14416 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14418 * config/aarch64/aarch64.c
14419 (aarch64_simd_container_mode): Add prototype.
14420 (aarch64_expand_mov_immediate): Add HI support.
14421 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
14422 (aarch64_can_const_movi_rtx_p): New.
14423 (aarch64_preferred_reload_class):
14424 Remove restrictions of using FP registers for certain SIMD operations.
14425 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
14426 (aarch64_valid_floating_const): Add integer move validation.
14427 (aarch64_simd_imm_scalar_p): Remove.
14428 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
14429 (aarch64_legitimate_constant_p): Expand list of supported cases.
14430 * config/aarch64/aarch64-protos.h
14431 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
14432 (aarch64_reinterpret_float_as_int): New.
14433 (aarch64_simd_imm_scalar_p): Remove.
14434 * config/aarch64/constraints.md (Uvi): New.
14435 (Dd): Split into Ds and new Dd.
14436 * config/aarch64/aarch64.md (*movsi_aarch64):
14438 (*movdi_aarch64): Add SIMD mov case.
14440 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14442 * tree-predcom.c: (struct chain): Handle store-store chain in which
14443 stores for elimination only store loop invariant values.
14444 (execute_pred_commoning_chain): Ditto.
14445 (prepare_initializers_chain_store_elim): Ditto.
14446 (prepare_finalizers): Ditto.
14447 (is_inv_store_elimination_chain): New function.
14448 (initialize_root_vars_store_elim_1): New function.
14450 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14452 * tree-predcom.c: Revise general description of the pass.
14453 (enum chain_type): New enum type for store elimination.
14454 (struct chain): New field supporting store elimination.
14455 (struct component): Ditto.
14456 (dump_chain): Dump store-stores chain.
14457 (release_chain): Release resources.
14458 (split_data_refs_to_components): Compute and create component
14459 contains only stores for elimination.
14460 (get_chain_last_ref_at): New function.
14461 (make_invariant_chain): Initialization.
14462 (make_rooted_chain): Specify chain type in parameter and record it.
14463 (add_looparound_copies): Skip for store-stores chain.
14464 (determine_roots_comp): Compute type of chain and pass it to
14466 (initialize_root_vars_store_elim_2): New function.
14467 (finalize_eliminated_stores): New function.
14468 (remove_stmt): Handle store for elimination.
14469 (execute_pred_commoning_chain): Execute predictive commoning on
14470 store-store chains.
14471 (determine_unroll_factor): Skip unroll for store-stores chain.
14472 (prepare_initializers_chain_store_elim): New function.
14473 (prepare_initializers_chain): Hanlde store-store chain.
14474 (prepare_finalizers_chain, prepare_finalizers): New function.
14475 (tree_predictive_commoning_loop): Return integer value indicating
14476 if loop is unrolled or lcssa form is corrupted.
14477 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
14479 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14481 * tree-predcom.c (initialize_root): Delete.
14482 (execute_pred_commoning_chain): Initialize root vars and replace
14483 reference of non-combined chain directly, rather than call above
14486 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14488 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
14489 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
14491 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14493 * tree-predcom.c (struct chain): New field init_seq.
14494 (release_chain): Release init_seq.
14495 (prepare_initializers_chain): Record intialization stmts in above
14497 (insert_init_seqs): New function.
14498 (tree_predictive_commoning_loop): Call insert_init_seqs.
14500 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14502 * tree-predcom.c (determine_roots_comp): Skip trivial components.
14504 2017-07-28 Richard Biener <rguenther@suse.de>
14506 * match.pd: Remove superfluous :c.
14507 * genmatch.c (simplify::id): Add member.
14508 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
14510 (current_id): New global.
14511 (dt_node::parent): Move from ...
14512 (dt_operand::parent): ... here. Add for_id member.
14513 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
14514 (decision_tree::find_node): Relax order requirement when
14515 merging DT_TRUE nodes to ones inbetween the current simplify
14516 and the one we try to merge with. Add diagnostic whenever
14517 we need to enforce pattern order by not merging.
14518 (decision_tree::insert): Set current_id.
14519 (decision_tree::print_node): Dump parent node and for_id.
14520 (parser::last_id): Add member.
14521 (parser::push_simplify): Assign unique id.
14522 (parser::parser): Initialize last_id.
14524 2017-07-28 Martin Liska <mliska@suse.cz>
14527 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
14528 gimple_build_debug_bind.
14530 2017-07-28 Richard Biener <rguenther@suse.de>
14532 PR tree-optimization/81502
14533 * match.pd: Add pattern combining BIT_INSERT_EXPR with
14535 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
14537 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
14538 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
14539 for BIT_FIELD_REF args.
14540 * fold-const.c (make_bit_field_ref): Likewise.
14541 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
14543 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14546 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
14547 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
14548 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
14549 Or it into SANITIZER_UNDEFINED.
14550 * ubsan.c: Include gimple-fold.h and varasm.h.
14551 (ubsan_expand_ptr_ifn): New function.
14552 (instrument_pointer_overflow): New function.
14553 (maybe_instrument_pointer_overflow): New function.
14554 (instrument_object_size): Formatting fix.
14555 (pass_ubsan::execute): Call instrument_pointer_overflow
14556 and maybe_instrument_pointer_overflow.
14557 * internal-fn.c (expand_UBSAN_PTR): New function.
14558 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
14559 * sanitizer.def (__ubsan_handle_pointer_overflow,
14560 __ubsan_handle_pointer_overflow_abort): New builtins.
14561 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
14562 * internal-fn.def (UBSAN_PTR): New internal function.
14563 * opts.c (sanitizer_opts): Add pointer-overflow.
14564 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
14565 * fold-const.c (build_range_check): Compute pointer range check in
14566 integral type if pointer arithmetics would be needed. Formatting
14569 2017-07-28 Martin Liska <mliska@suse.cz>
14572 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
14573 parameters that are of a variable-length.
14575 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
14577 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
14579 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
14580 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
14581 (CRT_CALL_STATIC_FUNCTION): Likewise.
14582 (ASM_DEFAULT_SPEC): New define.
14583 (ASM_SPEC32): Likewise.
14584 (ASM_SPEC64): Likewise.
14585 (ASM_SPEC_COMMON): Likewise.
14586 (ASM_SPEC): Likewise.
14587 (INVALID_64BIT): Likewise.
14588 (LINK_OS_DEFAULT_SPEC): Likewise.
14589 (LINK_OS_SPEC32): Likewise.
14590 (LINK_OS_SPEC64): Likewise.
14591 (POWERPC_LINUX): Likewise.
14592 (PTRDIFF_TYPE): Likewise.
14593 (RESTORE_FP_PREFIX): Likewise.
14594 (RESTORE_FP_SUFFIX): Likewise.
14595 (SAVE_FP_PREFIX): Likewise.
14596 (SAVE_FP_SUFFIX): Likewise.
14597 (SIZE_TYPE): Likewise.
14598 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
14599 (TARGET_64BIT): Likewise.
14600 (TARGET_64BIT): Likewise.
14601 (TARGET_AIX): Likewise.
14602 (WCHAR_TYPE_SIZE): Likewise.
14603 (WCHAR_TYPE): Undefine.
14604 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
14605 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
14606 (CPP_OS_RTEMS_SPEC): Delete.
14607 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
14608 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
14610 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
14612 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14614 PR tree-optimization/81578
14615 * tree-parloops.c (build_new_reduction): Bail out if
14616 reduction_code isn't one of the standard OpenMP reductions.
14617 Move the details printing after that decision.
14619 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14621 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
14622 related to reload_in_progress.
14623 (splat_input_operand): Likewise.
14624 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
14626 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
14628 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
14629 (TARGET_INSTANTIATE_DECLS): Likewise.
14630 (legitimate_indexed_address_p): Delete reload_in_progress code.
14631 (rs6000_debug_legitimate_address_p): Likewise.
14632 (rs6000_eliminate_indexed_memrefs): Likewise.
14633 (rs6000_emit_le_vsx_store): Likewise.
14634 (rs6000_emit_move_si_sf_subreg): Likewise.
14635 (rs6000_emit_move): Likewise.
14636 (register_to_reg_type): Likewise.
14637 (rs6000_pre_atomic_barrier): Likewise.
14638 (rs6000_machopic_legitimize_pic_address): Likewise.
14639 (rs6000_allocate_stack_temp): Likewise.
14640 (rs6000_address_for_fpconvert): Likewise.
14641 (rs6000_address_for_altivec): Likewise.
14642 (rs6000_secondary_memory_needed_rtx): Delete function.
14643 (rs6000_check_sdmode): Likewise.
14644 (rs6000_alloc_sdmode_stack_slot): Likewise.
14645 (rs6000_instantiate_decls): Likewise.
14646 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
14647 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
14648 Delete reload_in_progress.
14649 (*vec_reload_and_plus_<mptrsize>): Likewise.
14650 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
14651 (vsx_div_v2di): Likewise.
14652 (vsx_udiv_v2di): Likewise.
14654 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14656 * config/rs6000/rs6000.opt (mlra): Replace with stub.
14657 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
14658 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
14659 (rs6000_debug_reg_global): Delete print of LRA status.
14660 (rs6000_option_override_internal): Delete dead LRA related code.
14661 (rs6000_lra_p): Delete function.
14662 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
14664 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
14666 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
14667 * config/riscv/rtems.h: New file.
14669 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14670 Sudakshina Das <sudi.das@arm.com>
14672 * config/aarch64/aarch64.md
14673 (define_split for and<mode>3nr_compare): Move
14674 non aarch64_logical_operand to a register.
14675 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
14676 register immediate operand to a register.
14677 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
14679 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14681 PR middle-end/81564
14682 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
14684 2017-07-27 Richard Biener <rguenther@suse.de>
14686 PR tree-optimization/81573
14687 PR tree-optimization/81494
14688 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
14689 multi defuse cycle case.
14691 2017-07-27 Richard Biener <rguenther@suse.de>
14693 PR tree-optimization/81571
14694 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
14697 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
14699 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
14700 earlier and only if MASK_FPU is set. Adjust formatting.
14702 2017-07-27 Martin Liska <mliska@suse.cz>
14704 * opt-functions.awk: Add validation of value of Init.
14705 * optc-gen.awk: Pass new argument.
14707 2017-07-27 Martin Liska <mliska@suse.cz>
14709 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
14710 Fix wrong condition.
14712 2017-07-27 Martin Liska <mliska@suse.cz>
14714 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
14715 BBs and edges seen by autoFDO.
14717 2017-07-27 Richard Biener <rguenther@suse.de>
14719 PR tree-optimization/81502
14720 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
14721 with incompatible but same sized type.
14722 (execute_update_addresses_taken): Likewise.
14724 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
14726 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
14727 flag_tree_loop_vectorize rather than flag_tree_vectorize.
14729 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14732 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
14733 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
14734 Change s_operand to memory_operand.
14736 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
14738 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
14739 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
14740 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
14741 Emit instructions rather than returning an expression. Handle TFmode
14742 and KFmode by casting to TImode.
14743 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
14744 (rs6000_emit_le_vsx_store): Likewise.
14745 * config/rs6000/vsx.md (VSX_TI): New iterator.
14746 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
14747 (*vsx_le_undo_permute_<mode>): Likewise.
14748 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
14749 emit the split sequence.
14750 (*vsx_le_perm_store_<mode>): Likewise.
14752 2017-07-27 Jakub Jelinek <jakub@redhat.com>
14754 PR tree-optimization/81555
14755 PR tree-optimization/81556
14756 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
14757 if true, force CHANGED for the recursive invocation.
14758 (reassociate_bb): Remember original length of ops array, pass
14759 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
14761 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
14762 attributes for noipa attribute. For naked attribute use
14763 lookup_attribute first before lookup_attribute_spec.
14764 * final.c (rest_of_handle_final): Disable IPA RA for functions with
14766 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
14767 for functions with noipa attribute.
14768 (cgraph_externally_visible_p): Return true for functions with noipa
14770 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
14771 for functions with noipa attribute.
14772 * doc/extend.texi: Document noipa function attribute.
14773 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
14774 also for functions with noipa attribute.
14775 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
14777 2017-07-26 Andrew Pinski <apinski@cavium.com>
14779 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
14780 vec_unalign_load_cost and vec_unalign_store_cost.
14782 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14784 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
14785 -mvsx-small-integer option.
14786 (ISA_3_0_MASKS_IEEE): Likewise.
14787 (OTHER_VSX_VECTOR_MASKS): Likewise.
14788 (POWERPC_MASKS): Likewise.
14789 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
14790 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
14791 code, only testing for DImode being allowed in non-VSX floating
14793 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
14794 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
14796 (rs6000_option_override_internal): Delete -mvsx-small-integer.
14797 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
14798 TARGET_P8_VECTOR test.
14799 (rs6000_secondary_reload_simple_move): Likewise.
14800 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
14801 since TARGET_P9_VECTOR was already tested.
14802 (rs6000_opt_masks): Remove -mvsx-small-integer.
14803 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
14804 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14806 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
14807 test for TARGET_VEXTRACTUB was used, and that uses
14809 (p9 extract splitter): Likewise.
14810 (vsx_extract_<mode>_di_p9): Likewise.
14811 (vsx_extract_<mode>_store_p9): Likewise.
14812 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
14813 for TARGET_P9_VECTOR was used. Delete code that is now dead with
14814 the elimination of TARGET_VSX_SMALL_INTEGER.
14815 (vsx_extract_<mode>_p8): Likewise.
14816 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
14817 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
14818 (vsx_set_<mode>_p9): Likewise.
14819 (vsx_set_v4sf_p9): Likewise.
14820 (vsx_set_v4sf_p9_zero): Likewise.
14821 (vsx_insert_extract_v4sf_p9): Likewise.
14822 (vsx_insert_extract_v4sf_p9_2): Likewise.
14823 * config/rs6000/rs6000.md (sign extend splitter): Change
14824 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
14825 (floatsi<mode>2_lfiwax_mem): Likewise.
14826 (floatunssi<mode>2_lfiwzx_mem): Likewise.
14827 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
14828 since a test for TARGET_P9_VECTOR was used.
14829 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14830 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
14831 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14832 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
14833 TARGET_P8_VECTOR test.
14834 (fix_trunc<mode>si2_stfiwx): Likewise.
14835 (fix_trunc<mode>si2_internal): Likewise.
14836 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
14837 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14839 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
14840 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
14841 TARGET_P8_VECTOR test.
14842 (fixuns_trunc<mode>si2_stfiwx): Likewise.
14843 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
14844 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14846 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
14847 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
14848 since a test for TARGET_P9_VECTOR was used.
14849 (splitter for loading small constants): Likewise.
14851 2017-07-26 Andrew Pinski <apinski@cavium.com>
14853 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
14856 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
14859 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
14860 (fp_valid_at): Likewise.
14862 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
14864 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
14865 (qdf24xx_addrcost_table): Likewise.
14866 (cortexa57_tunings): Update to use generic_branch_cost.
14867 (cortexa72_tunings): Likewise.
14868 (cortexa73_tunings): Likewise.
14869 (qdf24xx_tunings): Likewise.
14871 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
14873 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
14874 (thunderx2t99_branch_cost): Likewise.
14875 (cortexa35_tunings): Update to use generic_branch_cost.
14876 (cortexa53_tunings): Likewise.
14877 (cortexa57_tunings): Likewise.
14878 (cortexa72_tunings): Likewise.
14879 (cortexa73_tunings): Likewise.
14880 (thunderx2t99_tunings): Likewise.
14882 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
14884 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
14885 (sparc_option_override): Honour MASK_FSMULD.
14886 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
14887 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
14888 * config/sparc/sparc.opt (mfsmuld): New option.
14889 * doc/invoke.texi (mfsmuld): Document option.
14891 2017-07-26 Marek Polacek <polacek@redhat.com>
14893 PR middle-end/70992
14894 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
14896 2017-07-26 Richard Biener <rguenther@suse.de>
14898 * gimple-match-head.c (do_valueize): Return OP if valueize
14900 (get_def): New helper to get at the def stmt of a SSA name
14901 if valueize allows.
14902 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
14903 do_valueize to get at the def stmt.
14904 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
14906 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
14908 PR middle-end/46932
14909 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
14911 2017-07-26 Martin Liska <mliska@suse.cz>
14914 * function.c (expand_function_start): Make expansion of
14915 nonlocal_goto_save_area after parm_birth_insn.
14917 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
14919 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
14920 from all CPU target flags enable members.
14922 2017-07-26 Richard Biener <rguenther@suse.de>
14924 * genmatch.c (dt_simplify::gen): Make iterator vars const.
14925 (decision_tree::gen): Make 'type' const.
14926 (write_predicate): Likewise.
14928 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
14930 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
14931 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
14932 (rs6000_option_override_internal): Likewise.
14933 (rs6000_expand_vector_set): Likewise.
14934 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
14935 (TARGET_UPPER_REGS_SF): Likewise.
14936 (TARGET_UPPER_REGS_DI): Likewise.
14937 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
14938 (TARGET_DIRECT_MOVE_64BIT): Likewise.
14939 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
14940 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14941 (Splitters for DI constants in Altivec registers): Likewise.
14942 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
14943 (vsx_set_v4sf_p9): Likewise.
14944 (vsx_set_v4sf_p9_zero): Likewise.
14945 (vsx_insert_extract_v4sf_p9): Likewise.
14946 (vsx_insert_extract_v4sf_p9_2): Likewise.
14948 2017-07-25 Carl Love <cel@us.ibm.com>
14950 * doc/extend.texi: Update the built-in documentation file for the
14951 existing built-in functions
14952 vector signed char vec_cnttz (vector signed char);
14953 vector unsigned char vec_cnttz (vector unsigned char);
14954 vector signed short vec_cnttz (vector signed short);
14955 vector unsigned short vec_cnttz (vector unsigned short);
14956 vector signed int vec_cnttz (vector signed int);
14957 vector unsigned int vec_cnttz (vector unsigned int);
14958 vector signed long long vec_cnttz (vector signed long long);
14959 vector unsigned long long vec_cnttz (vector unsigned long long);
14961 2017-07-25 Andrew Pinski <apinski@cavium.com>
14963 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
14964 accesses where the use is for the first operand of a BIT_INSERT.
14966 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
14969 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
14970 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
14972 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
14974 * config/i386/gstabs.h: Delete.
14975 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
14977 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
14979 * config/i386/i386.c (ix86_decompose_address): Do not check for
14980 register RTX when looking at index_reg or base_reg.
14981 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
14983 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
14985 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
14986 to update EH info here.
14988 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
14990 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
14992 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
14994 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
14996 2017-07-25 Torsten Duwe <duwe@suse.de>
14998 * common.opt: Introduce -fpatchable-function-entry
14999 command line option, and its variables function_entry_patch_area_size
15000 and function_entry_patch_area_start.
15001 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
15002 including a two-value parser.
15003 * target.def (print_patchable_function_entry): New target hook.
15004 * targhooks.h (default_print_patchable_function_entry): New function.
15005 * targhooks.c (default_print_patchable_function_entry): Likewise.
15006 * toplev.c (process_options): Switch off IPA-RA if
15007 patchable function entries are being generated.
15008 * varasm.c (assemble_start_function): Look at the
15009 patchable-function-entry command line switch and current
15010 function attributes and maybe generate NOP instructions by
15011 calling the print_patchable_function_entry hook.
15012 * doc/extend.texi: Document patchable_function_entry attribute.
15013 * doc/invoke.texi: Document -fpatchable_function_entry
15014 command line option.
15015 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
15017 * doc/tm.texi: Re-generate.
15019 2017-07-25 Jakub Jelinek <jakub@redhat.com>
15022 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
15023 TARGET_AVX512DQ rather than TARGET_AVX512BW.
15025 2017-07-25 Tamar Christina <tamar.christina@arm.com>
15027 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
15029 2017-07-25 Richard Biener <rguenther@suse.de>
15031 PR tree-optimization/81455
15032 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
15033 not walk in cycles when looking for guards.
15035 2017-07-25 Richard Biener <rguenther@suse.de>
15037 PR tree-optimization/81529
15038 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
15039 when optimizing backedge uses.
15041 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
15043 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
15045 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
15046 to dl_section_ref. On AIX, append an expression to subtract
15047 the size of the section length to dl_section_ref.
15049 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
15051 * configure.ac: If any of the config.* scripts fail, exit 1.
15052 * configure: Regenerate.
15054 2017-07-25 Richard Biener <rguenther@suse.de>
15056 PR middle-end/81546
15057 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
15058 of immediate uses, be more verbose on errors.
15060 2017-07-25 Richard Biener <rguenther@suse.de>
15062 PR tree-optimization/81510
15063 * tree-vect-loop.c (vect_is_simple_reduction): When the
15064 reduction stmt is not inside the loop bail out.
15066 2017-07-25 Richard Biener <rguenther@suse.de>
15068 PR tree-optimization/81303
15069 * tree-vect-loop-manip.c (vect_loop_versioning): Build
15070 profitability check against LOOP_VINFO_NITERSM1.
15072 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15074 * domwalk.c (cmp_bb_postorder): Simplify.
15075 (sort_bbs_postorder): New function. Use it...
15076 (dom_walker::walk): ...here to optimize common cases.
15078 2017-07-25 Martin Liska <mliska@suse.cz>
15081 * ipa-visibility.c (function_and_variable_visibility): Make the
15082 redirection just on target that supports aliasing.
15083 Fix GNU coding style.
15085 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
15088 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
15090 * config/arm/rtems.h: Likewise.
15091 * config/bfin/rtems.h: Likewise.
15092 * config/i386/rtemself.h: Likewise.
15093 * config/lm32/rtems.h: Likewise.
15094 * config/m32c/rtems.h: Likewise.
15095 * config/m68k/rtemself.h: Likewise.
15096 * config/microblaze/rtems.h: Likewise.
15097 * config/mips/rtems.h: Likewise.
15098 * config/moxie/rtems.h: Likewise.
15099 * config/nios2/rtems.h: Likewise.
15100 * config/powerpcspe/rtems.h: Likewise.
15101 * config/rs6000/rtems.h: Likewise.
15102 * config/rtems.h: Likewise.
15103 * config/sh/rtems.h: Likewise.
15104 * config/sh/rtemself.h: Likewise.
15105 * config/sparc/rtemself.h: Likewise.
15107 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15110 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
15111 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
15112 * tree-ssa-structalias.c (alias_get_name): Same.
15114 2017-07-25 Bin Cheng <bin.cheng@arm.com>
15117 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
15118 instructions if no du chain is found.
15120 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15122 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
15124 2017-07-25 Richard Biener <rguenther@suse.de>
15126 PR middle-end/81505
15127 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
15130 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
15132 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
15133 upper-regs options.
15134 (ISA_2_7_MASKS_SERVER): Likewise.
15135 (ISA_3_0_MASKS_IEEE): Likewise.
15136 (OTHER_P8_VECTOR_MASKS): Likewise.
15137 (OTHER_VSX_VECTOR_MASKS): Likewise.
15138 (POWERPC_MASKS): Likewise.
15139 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
15140 duplicate list of options.
15141 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
15142 explicit -mupper-regs options.
15143 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
15144 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
15145 alias for -mupper-regs-df.
15146 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
15147 (rs6000_init_hard_regno_mode_ok): Likewise.
15148 (rs6000_option_override_internal): Likewise.
15149 (rs6000_opt_masks): Likewise.
15150 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
15151 options in terms of whether -mvsx or -mpower8-vector was used.
15152 (TARGET_UPPER_REGS_DI): Likewise.
15153 (TARGET_UPPER_REGS_SF): Likewise.
15154 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
15155 -mupper-regs-* options.
15157 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15159 * passes.c (emergency_dump_function): Print some empty lines and a
15160 header before the RTL dump.
15162 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15164 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
15166 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
15169 * config/aarch64/aarch64.c (aarch64_classify_symbol):
15170 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
15172 2017-07-24 Carl Love <cel@us.ibm.com>
15174 * config/rs6000/rs6000-c.c: Add support for built-in functions
15175 vector float vec_extract_fp32_from_shorth (vector unsigned short);
15176 vector float vec_extract_fp32_from_shortl (vector unsigned short);
15177 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
15178 vec_extract_fp_from_shortl): Add defines for the two builtins.
15179 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
15180 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
15182 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
15183 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
15184 * doc/extend.texi: Update the built-in documentation file for the
15185 new built-in function.
15187 2017-07-24 Jakub Jelinek <jakub@redhat.com>
15190 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
15192 * doc/generic.texi: Likewise.
15193 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
15194 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
15196 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
15198 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
15199 (aarch64_mls_elt_merge<mode>): Likewise.
15201 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
15203 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
15204 having __cxa_atexit.
15206 2017-07-23 Michael Collison <michael.collison@arm.com>
15208 * config/arm/arm.c (arm_option_override): Deprecate
15209 use of -mstructure-size-boundary.
15210 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
15211 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
15213 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15216 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
15217 Reduce cost estimate for direct moves.
15219 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
15222 * config/i386/i386.c (ix86_option_override_internal): Disable
15223 BMI, BMI2 and TBM instructions for -m16.
15225 2017-07-21 Carl Love <cel@us.ibm.com>
15227 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15228 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15229 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15230 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15231 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15232 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15233 VMULOSW): New enum "unspec" values.
15234 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15235 altivec_vmulosw): New patterns.
15236 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15237 VMULOSW): Add definitions.
15239 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
15241 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
15242 (qdf24xx): Likewise.
15243 * config/aarch64/aarch64-options-extensions.def (rdma); New.
15244 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
15245 (AARCH64_FL_V8_1): Renumber.
15246 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
15247 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
15248 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
15249 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
15250 rdma to feature modifiers list.
15252 2017-07-21 Yury Gribov <tetra2005@gmail.com>
15254 PR middle-end/56727
15255 * ipa-visibility (function_and_variable_visibility): Convert
15256 recursive PLT call to direct call if appropriate.
15258 2017-07-21 Andrew Pinski <apinski@cavium.com>
15260 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
15261 operand 1 to see if the types precision matches.
15262 * fold-const.c (operand_equal_p): Likewise.
15264 2017-07-21 Richard Biener <rguenther@suse.de>
15266 PR tree-optimization/81303
15267 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
15268 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
15269 (vect_peeling_hash_get_lowest_cost): Adjust.
15270 (vect_enhance_data_refs_alignment): Likewise. Use
15271 vect_get_peeling_costs_all_drs to compute the penalty for no
15272 peeling to match up costs.
15274 2017-07-21 Richard Biener <rguenther@suse.de>
15276 PR tree-optimization/81500
15277 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
15278 we didn't identify a reduction path.
15280 2017-07-21 Tom de Vries <tom@codesourcery.com>
15281 Cesar Philippidis <cesar@codesourcery.com>
15283 PR gcov-profile/81442
15284 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
15287 2017-07-21 Tom de Vries <tom@codesourcery.com>
15290 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
15292 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
15293 nvptx_override_options_after_change.
15295 2017-07-21 Ulrich Drepper <drepper@redhat.com>
15297 * dwarf2out.c (output_file_names): Avoid double testing for
15298 dwarf_version >= 5.
15300 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
15302 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
15304 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15306 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
15308 (try_crossjump_to_edge): Do not punt on partitioned functions.
15310 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15312 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
15313 Put all BBs reachable only via paths crossing cold region to cold
15315 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
15317 2016-07-21 Richard Biener <rguenther@suse.de>
15319 PR tree-optimization/81303
15320 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
15321 into account prologue and epilogue iterations when raising
15322 min_profitable_iters to sth at least covering one vector iteration.
15324 2017-07-21 Tamar Christina <tamar.christina@arm.com>
15326 * config/arm/arm.c (arm_test_cpu_arch_dat):
15329 2017-07-20 Nathan Sidwell <nathan@acm.org>
15331 Remove TYPE_METHODS.
15332 * tree.h (TYPE_METHODS): Delete.
15333 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
15334 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
15335 (dbxout_type_methods): Scan TYPE_FIELDS.
15336 (dbxout_type): Don't check TYPE_METHODS here.
15337 * function.c (use_register_for_decl): Always ignore register for
15338 class types when not optimizing.
15339 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
15340 * tree.c (free_lang_data_in_type): Stitch out member functions and
15341 templates from TYPE_FIELDS.
15342 (build_distinct_type_copy, verify_type_variant,
15343 verify_type): Member fns are on TYPE_FIELDS.
15344 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
15345 * tree-pretty-print.c (dump_generic_node): Likewise.
15347 2017-07-20 Jakub Jelinek <jakub@redhat.com>
15350 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
15351 V2TImode and V4TImode.
15352 (ix86_expand_vector_extract): Likewise.
15353 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
15354 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
15355 (ssescalarmode): Handle V4TImode and V2TImode.
15356 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
15357 (*vec_extractv2ti, *vec_extractv4ti): New insns.
15358 (VEXTRACTI128_MODE): New mode iterator.
15359 (splitter for *vec_extractv?ti first element): New.
15360 (VEC_INIT_MODE): New mode iterator.
15361 (vec_init<mode>): Consolidate 3 expanders into one using
15362 VEC_INIT_MODE mode iterator.
15364 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
15366 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
15367 non_spilled_static_chain_regno_p.
15369 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
15371 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
15373 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
15375 * bb-reorder.c (connect_traces): Allow copying of blocks within
15378 2017-07-20 Richard Biener <rguenther@suse.de>
15380 * gimple.h (gimple_phi_result): Add gphi * overload.
15381 (gimple_phi_result_ptr): Likewise.
15382 (gimple_phi_arg): Likewise. Adjust index assert to only
15383 allow actual argument accesses rather than all slots available
15385 (gimple_phi_arg_def): Add gphi * overload.
15386 * tree-phinodes.c (make_phi_node): Initialize only actual
15388 (resize_phi_node): Clear memory not covered by old node,
15389 do not initialize excess argument slots.
15390 (reserve_phi_args_for_new_edge): Initialize new argument slot
15393 2017-07-20 Bin Cheng <bin.cheng@arm.com>
15395 PR tree-optimization/81388
15397 2016-07-21 Bin Cheng <bin.cheng@arm.com>
15399 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
15400 by removing computation of may_be_zero.
15402 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15403 Tom de Vries <tom@codesourcery.com>
15405 PR middle-end/81030
15406 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
15407 when gimple level profile disagrees with what RTL expander did.
15409 2017-07-20 Richard Biener <rguenther@suse.de>
15411 PR tree-optimization/61171
15412 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
15413 (vect_analyze_stmt): Add slp instance parameter.
15414 (vectorizable_reduction): Likewise.
15415 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
15416 (vect_is_simple_reduction): Deal with chains not detected
15417 as SLP reduction chain, specifically not properly associated
15418 chains containing a mix of plus/minus.
15419 (get_reduction_op): Remove.
15420 (get_initial_defs_for_reduction): Simplify, pass in whether
15421 this is a reduction chain, pass in the SLP node for the PHIs.
15422 (vect_create_epilog_for_reduction): Get the SLP instance as
15424 (vectorizable_reduction): Get the SLP instance as arg.
15425 During analysis remember the SLP node with the PHIs in the
15426 instance. Simplify getting at the vectorized reduction PHIs.
15427 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
15428 through SLP instance.
15429 (vect_slp_analyze_operations): Likewise.
15430 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
15431 (vect_transform_stmt): Likewise.
15433 2017-07-20 Tom de Vries <tom@codesourcery.com>
15435 PR tree-optimization/81489
15436 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
15437 read of phi arg location to before loop that modifies phi.
15439 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15441 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
15444 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15446 PR middle-end/81331
15447 * except.c (execute): Fix ordering issue.
15449 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
15451 PR rtl-optimization/81423
15452 * combine.c (make_compound_operation_int): Don't try to optimize
15453 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
15455 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
15457 PR rtl-optimization/81423
15458 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
15459 with a constant that is -1 in the truncated to mode.
15461 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15463 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
15464 (determine_unlikely_bbs): ... here.
15465 * predict.h (propagate_unlikely_bbs_forward): Declare.
15466 * cfgexpand.c (pass_expand::execute): Use it.
15467 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
15469 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
15470 propagate_unlikely_bbs_forward.
15472 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15474 PR middle-end/81331
15475 * except.c (maybe_add_nop_after_section_switch): New function.
15478 2017-07-19 Tom de Vries <tom@codesourcery.com>
15480 * gimple.h (gimple_phi_set_arg): Make assert more strict.
15482 2017-07-19 Tom de Vries <tom@codesourcery.com>
15484 * gimple.h (gimple_phi_arg): Make assert more strict.
15486 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
15488 * config.gcc (powerpc*-*-*): Add mmintrin.h.
15489 * config/rs6000/mmintrin.h: New file.
15490 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
15492 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15494 PR tree-optimization/81346
15495 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
15497 2017-07-19 Tom de Vries <tom@codesourcery.com>
15499 * config/nvptx/nvptx.md (VECIM): Add V2DI.
15501 2017-07-19 Tom de Vries <tom@codesourcery.com>
15503 * config/nvptx/nvptx-modes.def: Add V2DImode.
15504 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
15505 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
15506 (nvptx_output_mov_insn): Handle lack of mov.b128.
15507 (nvptx_print_operand): Handle 'H' and 'L' codes.
15508 (nvptx_vector_mode_supported): Allow V2DImode.
15509 (nvptx_preferred_simd_mode): New function.
15510 (nvptx_data_alignment): New function.
15511 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
15512 nvptx_preferred_simd_mode.
15513 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
15515 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
15517 2017-07-19 Tom de Vries <tom@codesourcery.com>
15519 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
15520 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
15521 (nvptx_vector_mode_supported): New function. Allow V2SImode.
15522 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
15523 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
15524 (mov<VECIM>_insn): New define_insn.
15525 (define_expand "mov<VECIM>): New define_expand.
15527 2017-07-19 Tom de Vries <tom@codesourcery.com>
15529 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
15531 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15533 PR tree-optimization/81346
15534 * fold-const.h (fold_div_compare, range_check_type): Declare.
15535 * fold-const.c (range_check_type): New function.
15536 (build_range_check): Use range_check_type.
15537 (fold_div_compare): No longer static, rewritten into
15538 a match.pd helper function.
15539 (fold_comparison): Don't call fold_div_compare here.
15540 * match.pd (X / C1 op C2): New optimization using fold_div_compare
15541 as helper function.
15543 2017-07-19 Nathan Sidwell <nathan@acm.org>
15545 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
15546 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
15547 * tree.c (find_decls_types_r, verify_type): Use
15548 TYPE_{MIN,MAX}_VALUE_RAW.
15549 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
15550 (hash_tree): Likewise.
15551 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
15553 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
15556 2017-07-18 Tom de Vries <tom@codesourcery.com>
15558 PR middle-end/81464
15559 * omp-expand.c (expand_omp_for_static_chunk): Handle
15560 equal-argument loop exit phi.
15562 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
15565 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
15566 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
15567 operand 2 predicate.
15568 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
15569 operand 2 predicate.
15570 (ror,rol -> rorx splitters): Use const_int_operand as
15571 operand 2 predicate.
15573 2017-06-18 Richard Biener <rguenther@suse.de>
15575 PR tree-optimization/81410
15576 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
15577 the gap in the ! slp_perm SLP case after each group.
15579 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15581 PR middle-end/81463
15582 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
15585 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15587 PR middle-end/81462
15588 * predict.c (set_even_probabilities): Cleanup; do not affect
15589 probabilities that are already known.
15590 (combine_predictions_for_bb): Call even when count is set.
15592 2017-07-18 Nathan Sidwell <nathan@acm.org>
15594 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
15597 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15600 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
15601 optimization for loop niter analysis.
15603 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
15606 * config/avr/avr.c (avr_optimize_casesi): Don't use
15607 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
15609 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
15611 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
15612 body_cost_vec from _vect_peel_extended_info.
15613 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
15614 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
15617 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15619 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
15621 2017-07-18 Richard Biener <rguenther@suse.de>
15623 PR tree-optimization/80620
15624 PR tree-optimization/81403
15625 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
15626 info when re-using a VN table entry.
15628 2017-07-18 Richard Biener <rguenther@suse.de>
15630 PR tree-optimization/81418
15631 * tree-vect-loop.c (vectorizable_reduction): Properly compute
15632 vectype_in. Verify that with lane-reducing reduction operations
15633 we have a single def-use cycle.
15635 2017-07-17 Carl Love <cel@us.ibm.com>
15637 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
15639 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15640 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15641 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15642 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15643 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15644 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15645 VMULOSW): New enum "unspec" values.
15646 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
15647 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
15648 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15649 altivec_vmulosw): New patterns.
15650 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15651 VMULOSW): Add definitions.
15652 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
15654 * config/alpha/alpha.c: Include predict.h.
15656 2017-07-17 Yury Gribov <tetra2005@gmail.com>
15658 * tree-vrp.c (compare_assert_loc): Fix comparison function
15659 to return predictable results.
15661 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15663 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
15665 (subdi3): Likewise.
15666 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
15667 * doc/invoke.texi (mexpand-adddi): Update text.
15669 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15671 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
15672 that also clobbers the CC register. The old expand code is moved
15674 (*arc_clzsi2): ... here.
15675 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
15676 the CC register. The old expand code is moved to ...
15677 (arc_ctzsi2): ... here.
15679 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15681 * config/arc/arc.opt (mindexed-loads): Use initial value
15682 TARGET_INDEXED_LOADS_DEFAULT.
15683 (mauto-modify-reg): Use initial value
15684 TARGET_AUTO_MODIFY_REG_DEFAULT.
15685 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
15686 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
15687 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
15688 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
15690 2017-07-17 Martin Liska <mliska@suse.cz>
15693 * opts.c (finish_options): Do not allow -fgnu-tm
15694 w/ -fsanitize={kernel-,}address. Say sorry.
15696 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15699 * tree-loop-distribution.c (classify_partition): Only assert on
15700 numer of iterations.
15701 (merge_dep_scc_partitions): Delete prameter. Update function call.
15702 (distribute_loop): Remove code handling loop with unknown niters.
15703 (pass_loop_distribution::execute): Skip loop with unknown niters.
15705 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15708 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
15709 function sort_partitions_by_post_order.
15711 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15713 PR tree-optimization/81374
15714 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
15715 the max index of basic blocks, rather than number of basic blocks.
15717 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15719 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
15721 (arc_legitimate_pic_operand_p): Likewise.
15722 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
15724 (arc_needs_pcl_p): Likewise.
15725 (arc_legitimate_pc_offset_p): Likewise.
15726 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
15727 function is also used in constrains.md.
15728 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
15729 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
15730 PLUS. Only return true/false in known cases, otherwise assert.
15731 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
15732 is already called in arc_legitimate_constant_p.
15733 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
15735 (LEGITIMATE_PIC_OPERAND_P): Use
15736 arc_raw_symbolic_reference_mentioned_p function.
15737 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
15742 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15743 Andrew Burgess <andrew.burgess@embecosm.com>
15745 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
15746 (arc_return_address_register): New function.
15747 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
15748 (arc_handle_fndecl_attribute): Add naked attribute.
15749 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
15750 (TARGET_WARN_FUNC_RETURN): Likewise.
15751 (arc_allocate_stack_slots_for_args): New function.
15752 (arc_warn_func_return): Likewise.
15753 (machine_function): Change type fn_type.
15754 (arc_compute_function_type): Consider new naked function type,
15755 change function return type.
15756 (arc_must_save_register): Adapt to handle new
15757 arc_compute_function_type's return type.
15758 (arc_expand_prologue): Likewise.
15759 (arc_expand_epilogue): Likewise.
15760 (arc_return_address_regs): Delete.
15761 (arc_return_address_register): New function.
15762 (arc_epilogue_uses): Use above function.
15763 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
15764 (arc_function_type): Change encoding, add naked type.
15765 (ARC_INTERRUPT_P): Change to handle the new encoding.
15766 (ARC_FAST_INTERRUPT_P): Likewise.
15767 (ARC_NORMAL_P): Define.
15768 (ARC_NAKED_P): Likewise.
15769 (arc_compute_function_type): Delete prototype.
15770 * config/arc/arc.md (in_ret_delay_slot): Use
15771 arc_return_address_register function.
15772 (simple_return): Likewise.
15773 (p_return_i): Likewise.
15775 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15777 PR tree-optimization/81428
15778 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
15779 can't be built for those types.
15781 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
15783 Remove stuff dead since r239246.
15785 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
15786 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
15787 (avr_inform_devices): Remove dead stuff.
15789 2017-07-17 Tamar Christina <tamar.christina@arm.com>
15791 * config/arm/arm_neon.h: Fix softp typo.
15793 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15795 PR tree-optimization/81365
15796 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
15797 aggregate moves onto bb predecessor edges, make sure there are no
15798 loads that could alias the lhs in between the start of bb and the
15801 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
15804 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
15805 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
15806 [LSHIFTRT, outer_code = TRUNCATE]: Same.
15808 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15810 PR tree-optimization/81396
15811 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
15812 (init_symbolic_number): Initialize it to 1.
15813 (perform_symbolic_merge): Add n_ops from both operands into the new
15815 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
15816 without base_addr as useless if they need more than one operation.
15817 (bswap_replace): Handle !bswap case for NULL base_addr.
15819 2017-07-17 Tom de Vries <tom@codesourcery.com>
15822 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
15825 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
15827 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
15828 conditional builtin define __FIX_LEON3FT_B2BST.
15830 2017-07-17 Daniel Cederman <cederman@gaisler.com>
15832 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
15833 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
15836 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
15838 PR rtl-optimization/81424
15839 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
15840 to remove potential trapping from operands if -fnon-call-exceptions.
15842 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15844 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
15845 profile_proability for scalling.
15846 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
15848 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15850 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
15852 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15854 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
15855 fixpoint arithmetics.
15857 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15859 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
15860 fixpoint arithmetics.
15862 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15864 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
15865 fixpoint arithmetics.
15867 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15869 * profile-count.h (profile_probability::from_reg_br_prob_note,
15870 profile_probability::to_reg_br_prob_note): New functions.
15871 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
15872 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
15873 * predict.c (probability_reliable_p): Update.
15874 (edge_probability_reliable_p): Update.
15875 (br_prob_note_reliable_p): Update.
15876 (invert_br_probabilities): Update.
15877 (add_reg_br_prob_note): New function.
15878 (combine_predictions_for_insn): Update.
15879 * asan.c (asan_clear_shadow): Update.
15880 * cfgbuild.c (compute_outgoing_frequencies): Update.
15881 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
15882 (update_br_prob_note): Update.
15883 (rtl_verify_edges): Update.
15884 (purge_dead_edges): Update.
15885 (fixup_reorder_chain): Update.
15886 * emit-rtl.c (try_split): Update.
15887 * ifcvt.c (cond_exec_process_insns): Update.
15888 (cond_exec_process_if_block): Update.
15889 (dead_or_predicable): Update.
15890 * internal-fn.c (expand_addsub_overflow): Update.
15891 (expand_neg_overflow): Update.
15892 (expand_mul_overflow): Update.
15893 * loop-doloop.c (doloop_modify): Update.
15894 * loop-unroll.c (compare_and_jump_seq): Update.
15895 * optabs.c (emit_cmp_and_jump_insn_1): Update.
15896 * predict.h: Update.
15897 * reorg.c (mostly_true_jump): Update.
15899 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
15900 * config/alpha/alpha.c (emit_unlikely_jump): Update.
15901 * config/arc/arc.c: (emit_unlikely_jump): Update.
15902 * config/arm/arm.c: (emit_unlikely_jump): Update.
15903 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
15904 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
15905 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
15906 (ix86_print_operand): Update.
15907 (ix86_split_fp_branch): Update.
15908 (predict_jump): Update.
15909 * config/ia64/ia64.c (ia64_print_operand): Update.
15910 * config/mmix/mmix.c (mmix_print_operand): Update.
15911 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
15912 (rs6000_expand_split_stack_prologue): Update.
15913 * config/rs6000/rs6000.c: Update.
15914 * config/s390/s390.c (s390_expand_vec_strlen): Update.
15915 (s390_expand_vec_movstr): Update.
15916 (s390_expand_cs_tdsi): Update.
15917 (s390_expand_split_stack_prologue): Update.
15918 * config/sh/sh.c (sh_print_operand): Update.
15919 (expand_cbranchsi4): Update.
15920 (expand_cbranchdi4): Update.
15921 * config/sparc/sparc.c (output_v9branch): Update.
15922 * config/spu/spu.c (get_branch_target): Update.
15923 (ea_load_store_inline): Update.
15924 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
15925 * config/tilepro/tilepro.c: Update.
15927 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
15929 * gimplify.c (mostly_copy_tree_r): Revert latest change.
15930 (gimplify_save_expr): Likewise.
15932 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15934 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
15936 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15938 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
15940 * timevar.def (TV_IPA_FNSUMMARY): Define.
15942 2017-07-16 Daniel Cederman <cederman@gaisler.com>
15944 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
15945 to back store errata sensitive sequence from being generated.
15946 (sqrtdf2_fix): Likewise.
15948 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15950 * tree-ssa-threadupdate.c (compute_path_counts,
15951 update_joiner_offpath_counts): Use profile_probability.
15953 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
15956 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15958 * config/arm/arm-c.c (arm_cpu_builtins): Define
15959 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
15961 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
15963 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15964 array entries to represent __ieee128 versions of the
15965 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
15966 scalar_extract_sig, and scalar_insert_exp built-in functions.
15967 (altivec_resolve_overloaded_builtin): Add special case handling
15968 for the __builtin_scalar_insert_exp function, as represented by
15969 the P9V_BUILTIN_VEC_VSIEDP constant.
15970 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
15971 exponent support for __ieee128 argument.
15972 (VSESQP): Add scalar extract signature support for __ieee128
15974 (VSTDCNQP): Add scalar test negative support for __ieee128
15976 (VSIEQP): Add scalar insert exponent support for __int128 argument
15977 with __ieee128 result.
15978 (VSIEQPF): Add scalar insert exponent support for __ieee128
15979 argument with __ieee128 result.
15980 (VSTDCQP): Add scalar test data class support for __ieee128
15982 (VSTDCNQP): Add overload support for scalar test negative with
15983 __ieee128 argument.
15984 (VSTDCQP): Add overload support for scalar test data class
15985 __ieee128 argument.
15986 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
15987 UNSPEC_VSX_SXSIGDP.
15988 (UNSPEC_VSX_SIEXPQP): New constant.
15989 (xsxexpqp): New insn for VSX scalar extract exponent quad
15991 (xsxsigqp): New insn for VSX scalar extract significand quad
15993 (xsiexpqpf): New insn for VSX scalar insert exponent quad
15994 precision with floating point argument.
15995 (xststdcqp): New expand for VSX scalar test data class quad
15997 (xststdcnegqp): New expand for VSX scalar test negative quad
15999 (xststdcqp): New insn to match expansions for VSX scalar test data
16000 class quad precision and VSX scalar test negative quad precision.
16001 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
16002 special case operand checking to enforce that second operand of
16003 VSX scalar test data class with quad precision argument is a 7-bit
16005 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
16006 prototypes and descriptions of __ieee128 versions of
16007 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
16008 scalar_test_data_class, and scalar_test_neg built-in functions.
16010 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16012 PR tree-optimization/81162
16013 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
16014 replace a negate with an add.
16016 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
16018 * doc/invoke.texi (arm/-mcpu): Document +crypto.
16020 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16022 * config/arm/arm-c.c (arm_cpu_builtins): Define
16023 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
16025 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16027 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
16028 (armv8-r): Set ARM Cortex-R52 as default CPU.
16029 * config/arm/arm-tables.opt: Regenerate.
16030 * config/arm/arm-tune.md: Regenerate.
16031 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
16033 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
16034 extension for -mcpu=cortex-r52.
16036 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16038 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
16039 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
16040 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
16041 (fp-armv8): Define it as FP_ARMv8 only.
16042 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
16043 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
16045 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
16046 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
16047 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
16048 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
16049 than TARGET_FPU_ARMV8.
16050 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
16051 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
16052 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
16054 * config/arm/neon.md (neon_vrint): Likewise.
16055 (neon_vcvt): Likewise.
16056 (neon_<fmaxmin_op><mode>): Likewise.
16057 (<fmaxmin><mode>3): Likewise.
16058 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
16059 * config/arm/predicates.md (arm_cond_move_operator): Check against
16060 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
16062 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
16064 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
16065 to top of function.
16067 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16069 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
16070 loop in comment with memset.
16072 2017-07-14 Martin Liska <mliska@suse.cz>
16074 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
16075 * dwarf2out.c (is_java): Remove the function.
16076 (output_pubname): Remove usage of the function.
16077 (lower_bound_default): Remove usage of DW_LANG_Java.
16078 (gen_compile_unit_die): Likewise.
16079 * gcc.c: Remove compiler defaults for .java and .zip files.
16080 * gimple-expr.c (remove_suffix): Change as there's no longer
16081 extension than 4-letter one.
16082 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
16083 (gimplify_save_expr): Likewise.
16084 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
16085 as it's possible even for other languages than Java.
16086 * langhooks.h (struct lang_hooks): Remove Java from a comment.
16087 * lto-opts.c (lto_write_options): Remove reference to Java.
16088 * opts.c (strip_off_ending): Update file extension handling.
16089 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
16090 * tree-eh.c (lower_resx): Likewise.
16091 * tree.c (free_lang_data_in_type): Remove dead code.
16092 (find_decls_types_r): Likewise.
16093 (build_common_builtin_nodes): Remove Java from a comment.
16094 (verify_type): Remove dead code.
16095 * varasm.c (assemble_external): Remove Java from a comment.
16097 2017-07-14 Martin Liska <mliska@suse.cz>
16099 * opts.c (finish_options): Add quotes.
16100 (common_handle_option): Likewise.
16102 2017-07-14 Martin Liska <mliska@suse.cz>
16104 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
16105 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
16106 Remove N_SO_PASCAL.
16107 * dwarf2out.c (lower_bound_default): Do not handle
16109 (gen_compile_unit_die): Likewise.
16110 * gcc.c: Remove default extension binding for GNU Pascal.
16111 * stmt.c: Remove Pascal language from a comment.
16112 * xcoffout.c: Likewise.
16114 2017-07-13 David Malcolm <dmalcolm@redhat.com>
16117 * diagnostic-show-locus.c (fixit_cmp): New function.
16118 (layout::layout): Sort m_fixit_hints.
16119 (column_range::column_range): Assert that the values are valid.
16120 (struct char_span): New struct.
16121 (correction::overwrite): New method.
16122 (struct source_line): New struct.
16123 (line_corrections::add_hint): Add assertions. Reimplement memcpy
16124 calls in terms of classes source_line and char_span, and
16125 correction::overwrite.
16126 (selftest::test_overlapped_fixit_printing_2): New function.
16127 (selftest::diagnostic_show_locus_c_tests): Call it.
16129 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
16131 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
16132 early if there is no lhs.
16134 2017-07-13 Martin Liska <mliska@suse.cz>
16136 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
16137 (gen_reference_type_die): Likewise.
16138 * stor-layout.c: Remove Pascal-related comment.
16140 2017-07-13 Martin Liska <mliska@suse.cz>
16142 * opts.c (finish_options): Add quotes to error messages.
16143 (parse_sanitizer_options): Likewise.
16145 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
16147 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
16149 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
16151 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
16153 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
16155 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
16157 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
16159 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
16162 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
16163 provides the hardware capability bits, define the macro
16164 __BUILTIN_CPU_SUPPORTS__.
16165 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
16166 if GLIBC does not provide the hardware capability bits. Add a
16167 gcc_unreachable call if the built-in cpu function is neither
16168 __builtin_cpu_is nor __builtin_cpu_supports.
16169 (rs6000_get_function_versions_dispatcher): Change the warning
16170 that an old GLIBC is used which does not export the capability
16171 bits to be an error.
16172 * doc/extend.texi (target_clones attribute): Document the
16173 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
16174 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
16175 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
16176 the macros defined by GCC if the newer GLIBC is available.
16178 2017-07-12 Jeff Law <law@redhat.com>
16180 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
16181 remaining includes slightly.
16182 * config/riscv/riscv-builtins.c: Include profile-count.h.
16184 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16187 * config/avr/avr.c (avr_set_current_function): In diagnostic
16188 messages: Quote keywords and (parts of) identifiers.
16189 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
16192 2017-07-12 Carl Love <cel@us.ibm.com>
16194 * config/rs6000/rs6000-c.c: Add support for built-in functions
16195 vector bool char vec_revb (vector bool char);
16196 vector bool short vec_revb (vector short char);
16197 vector bool int vec_revb (vector bool int);
16198 vector bool long long vec_revb (vector bool long long);
16199 * doc/extend.texi: Update the built-in documentation file for the
16200 new built-in functions.
16202 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16204 * config/s390/s390.md: Remove movcc splitter.
16206 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16208 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
16209 load/store on condition.
16211 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16214 * config/avr/avr.c (avr_encode_section_info)
16215 [progmem && !TREE_READONLY]: Error if progmem object needs
16218 2017-07-11 Michael Collison <michael.collison@arm.com>
16220 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
16223 2017-07-11 Carl Love <cel@us.ibm.com>
16225 * config/rs6000/rs6000-c.c: Add support for builtins
16226 vector unsigned int vec_parity_lsbb (vector signed int);
16227 vector unsigned int vec_parity_lsbb (vector unsigned int);
16228 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
16229 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
16230 vector unsigned long long vec_parity_lsbb (vector signed long long);
16231 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
16232 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
16233 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
16234 * doc/extend.texi: Update the built-in documentation file for the
16235 new built-in functions.
16237 2017-07-11 David Malcolm <dmalcolm@redhat.com>
16239 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
16240 (layout::m_primary_loc): New field.
16241 (layout::layout): Initialize new field. Move location filtering
16242 logic from here to...
16243 (layout::maybe_add_location_range): ...this new method. Add
16244 support for filtering to just the lines already specified by other
16246 (layout::will_show_line_p): New method.
16247 (gcc_rich_location::add_location_if_nearby): New method.
16248 (selftest::test_add_location_if_nearby): New test function.
16249 (selftest::diagnostic_show_locus_c_tests): Call it.
16250 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
16253 2017-07-11 Tom de Vries <tom@codesourcery.com>
16255 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
16256 (bb_first_real_insn): New function.
16257 (nvptx_single): Add extra initialization of broadcasted condition
16260 2017-07-11 Nathan Sidwell <nathan@acm.org>
16262 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
16264 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
16266 * doc/extend.texi (AVR Function Attributes): Remove weblink to
16267 Binutils doc as TEXI will mess them up.
16268 * doc/invoke.texi (AVR Options): Same here.
16270 2017-07-11 Daniel Cederman <cederman@gaisler.com>
16272 * config/sparc/sparc.opt (mfix-ut700): New option.
16273 (mfix-gr712rc): Likewise.
16274 (sparc_fix_b2bst): New variable.
16275 * doc/invoke.texi (SPARC options): Document them.
16276 (ARM options): Fix warnings.
16277 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
16278 instructions to prevent sequences that can trigger the store-store
16279 errata for certain LEON3FT processors.
16280 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
16281 (sparc_option_override): Set sparc_fix_b2bst appropriately.
16282 * config/sparc/sparc.md (fix_b2bst): New attribute.
16283 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
16285 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
16288 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
16290 (*rsqrtsf2_sse): Ditto.
16292 (div<mode>3): Macroize insn from divdf3 and divsf3
16293 using MODEF mode iterator.
16295 2017-07-10 Martin Sebor <msebor@redhat.com>
16297 PR tree-optimization/80397
16298 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
16299 instead of testing for equality to INTEGER_TYPE.
16301 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
16303 * config.gcc: Remove uclibc from arc target spec.
16305 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
16307 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
16309 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16312 * lto-wrapper.c (remove_option): New function.
16313 (merge_and_complain): Merge PIC/PIE options more realistically.
16315 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16317 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
16321 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
16322 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
16323 * config.in: Regenerate.
16324 * configure: Regenerate.
16325 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
16326 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
16327 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
16328 (TARGET_GASISR_PROLOGUES): ...target mask.
16329 * common/config/avr/avr-common.c
16330 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
16331 Set -mgas-isr-prologues.
16332 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
16333 INSERT_PASS_BEFORE for it.
16334 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
16335 * config/avr/avr.c (avr_option_override)
16336 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
16337 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
16338 (avr_attribute_table) <no_gccisr>: Add new function attribute.
16339 (avr_set_current_function) <is_no_gccisr>: Init machine field.
16340 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
16342 (make_avr_pass_pre_proep): New function.
16343 (emit_push_sfr) <treg>: Add argument to function and use it
16344 instead of TMP_REG.
16345 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
16346 and set machine->gasisr.yes.
16347 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
16348 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
16349 __gcc_isr.n_pushed to .L__stack_usage.
16350 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
16351 (avr_asm_final_postscan_insn): ...this new static function.
16352 * config/avr/avr.h (machine_function)
16353 <is_no_gccisr, use_L__stack_usage>: New fields.
16354 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
16355 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
16356 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
16357 (gasisr, *gasisr): New expander and insn.
16358 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
16359 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
16360 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
16362 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
16364 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
16367 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16369 Move jump-tables out of .text again.
16372 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
16373 (ASM_OUTPUT_ADDR_VEC): New function.
16374 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
16375 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
16376 INSN_ADDRESSes as asm comment.
16377 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
16378 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
16379 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
16380 * config/avr/avr.md (*tablejump): Adjust comment.
16381 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
16382 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
16384 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
16385 (avr_output_addr_vec): New proto.
16386 (avr_log_t) <insn_addresses>: New field.
16388 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
16391 * config/i386/i386.c (ix86_function_arg_advance): Set
16392 outgoing_args_on_stack to true if there are outgoing arguments
16394 (ix86_function_arg): Likewise.
16395 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
16396 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
16397 * config/i386/i386.h (machine_function): Add
16398 outgoing_args_on_stack.
16400 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
16402 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
16403 supporting pthreds.
16404 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
16406 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
16408 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
16409 (REAL_H): Remove $(MACHMODE_H).
16410 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
16412 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
16413 $(MACHMODE_H) and double-int.h.
16414 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
16416 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
16419 2017-07-07 Andrew Pinski <apinski@cavium.com>
16421 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
16422 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
16424 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16426 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
16427 Add warning if GCC was not configured to link against a GLIBC that
16428 exports the hardware capability bits.
16429 (make_resolver_func): Make resolver function private and not a
16430 COMDAT function. Create the name with clone_function_name instead
16431 of make_unique_name.
16434 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
16435 correct operand in doing the split.
16437 2017-07-07 Carl Love <cel@us.ibm.com>
16439 * config/rs6000/rs6000-c: Add support for built-in function
16440 vector unsigned short vec_pack_to_short_fp32 (vector float,
16442 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
16443 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
16444 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
16445 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
16446 (convert_4f32_8i16): Add define_expand.
16447 * doc/extend.texi: Update the built-in documentation file for the
16448 new built-in function.
16450 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16452 * config/sparc/m8.md: New file.
16453 * config/sparc/sparc.md: Include m8.md.
16455 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16457 * config/sparc/sparc.opt: New option -mvis4b.
16458 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
16459 (sparc_option_override): Handle VIS4B.
16460 (enum sparc_builtins): Define
16461 SPARC_BUILTIN_DICTUNPACK{8,16,32},
16462 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
16463 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
16464 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
16465 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
16466 (check_constant_argument): New function.
16467 (sparc_vis_init_builtins): Define builtins
16468 __builtin_vis_dictunpack{8,16,32},
16469 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
16470 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
16471 __builtin_vis_fpcmpde{8,16,32}shl and
16472 __builtin_vis_fpcmpur{8,16,32}shl.
16473 (sparc_expand_builtin): Check that the constant operands to
16474 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
16475 constant and in range.
16476 * config/sparc/sparc-c.c (sparc_target_macros): Handle
16478 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
16479 (SPARC_IMM5_P): Likewise.
16480 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
16481 (enabled): Handle vis4b.
16482 (UNSPEC_DICTUNPACK): New unspec.
16483 (UNSPEC_FPCMPSHL): Likewise.
16484 (UNSPEC_FPUCMPSHL): Likewise.
16485 (UNSPEC_FPCMPDESHL): Likewise.
16486 (UNSPEC_FPCMPURSHL): Likewise.
16487 (cpu_feature): New CPU feature `vis4b'.
16488 (dictunpack{8,16,32}): New insns.
16489 (FPCSMODE): New mode iterator.
16490 (fpcscond): New code iterator.
16491 (fpcsucond): Likewise.
16492 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
16493 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
16494 (fpcmpde{8,16,32}{si,di}shl): Likewise.
16495 (fpcmpur{8,16,32}{si,di}shl): Likewise.
16496 * config/sparc/constraints.md: Define constraints `q' for unsigned
16497 2-bit integer constants and `t' for unsigned 5-bit integer
16499 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
16501 (imm5_operand_dictunpack16): Likewise.
16502 (imm5_operand_dictunpack32): Likewise.
16503 (imm2_operand): Likewise.
16504 * doc/invoke.texi (SPARC Options): Document -mvis4b.
16505 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
16506 ditunpack* and fpcmp*shl builtins.
16508 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16510 * config.gcc: Handle m8 in --with-{cpu,tune} options.
16511 * config.in: Add HAVE_AS_SPARC6 define.
16512 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
16514 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
16516 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
16517 (CPP_CPU_SPEC): Handle m8.
16518 (ASM_CPU_SPEC): Likewise.
16519 * config/sparc/sparc-opts.h (enum processor_type): Add
16521 * config/sparc/sparc.c (m8_costs): New struct.
16522 (sparc_option_override): Handle TARGET_CPU_m8.
16523 (sparc32_initialize_trampoline): Likewise.
16524 (sparc64_initialize_trampoline): Likewise.
16525 (sparc_issue_rate): Likewise.
16526 (sparc_register_move_cost): Likewise.
16527 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
16528 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
16529 (ASM_CPU64_DEFAULT_SPEC): Likewise.
16530 (CPP_CPU_SPEC): Handle M8.
16531 (ASM_CPU_SPEC): Likewise.
16532 (AS_M8_FLAG): Define.
16533 * config/sparc/sparc.md: Add m8 to the cpu attribute.
16534 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
16535 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
16537 * configure: Regenerate.
16538 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
16541 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16543 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
16545 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
16546 ("*movdi_insn_sp32"): Do not set v3pipe.
16547 ("*movsi_insn"): Likewise.
16548 ("*movdi_insn_sp64"): Likewise.
16549 ("*movsf_insn"): Likewise.
16550 ("*movdf_insn_sp32"): Likewise.
16551 ("*movdf_insn_sp64"): Likewise.
16552 ("*zero_extendsidi2_insn_sp64"): Likewise.
16553 ("*sign_extendsidi2_insn"): Likewise.
16554 ("*mov<VM32:mode>_insn"): Likewise.
16555 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16556 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16557 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16558 ("<vlop:code><VL:mode>3"): Likewise.
16559 ("*not_<vlop:code><VL:mode>3"): Likewise.
16560 ("*nand<VL:mode>_vis"): Likewise.
16561 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
16562 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
16563 ("one_cmpl<VL:mode>2"): Likewise.
16564 ("faligndata<VM64:mode>_vis"): Likewise.
16565 ("alignaddrsi_vis"): Likewise.
16566 ("alignaddrdi_vis"): Likweise.
16567 ("alignaddrlsi_vis"): Likewise.
16568 ("alignaddrldi_vis"): Likewise.
16569 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16570 ("bmaskdi_vis"): Likewise.
16571 ("bmasksi_vis"): Likewise.
16572 ("bshuffle<VM64:mode>_vis"): Likewise.
16573 ("cmask8<P:mode>_vis"): Likewise.
16574 ("cmask16<P:mode>_vis"): Likewise.
16575 ("cmask32<P:mode>_vis"): Likewise.
16576 ("pdistn<P:mode>_vis"): Likewise.
16577 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16579 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16581 * config/sparc/sparc.md ("subtype"): New insn attribute.
16582 ("*wrgsr_sp64"): Set insn subtype.
16583 ("*rdgsr_sp64"): Likewise.
16584 ("alignaddrsi_vis"): Likewise.
16585 ("alignaddrdi_vis"): Likewise.
16586 ("alignaddrlsi_vis"): Likewise.
16587 ("alignaddrldi_vis"): Likewise.
16588 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16589 ("fexpand_vis"): Likewise.
16590 ("fpmerge_vis"): Likewise.
16591 ("faligndata<VM64:mode>_vis"): Likewise.
16592 ("bshuffle<VM64:mode>_vis"): Likewise.
16593 ("cmask8<P:mode>_vis"): Likewise.
16594 ("cmask16<P:mode>_vis"): Likewise.
16595 ("cmask32<P:mode>_vis"): Likewise.
16596 ("fchksm16_vis"): Likewise.
16597 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
16598 ("fmean16_vis"): Likewise.
16599 ("fp<plusminus_insn>64_vis"): Likewise.
16600 ("<plusminus_insn>v8qi3"): Likewise.
16601 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16602 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
16603 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
16604 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
16605 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
16606 ("*movqi_insn"): Likewise.
16607 ("*movhi_insn"): Likewise.
16608 ("*movsi_insn"): Likewise.
16609 ("movsi_pic_gotdata_op"): Likewise.
16610 ("*movdi_insn_sp32"): Likewise.
16611 ("*movdi_insn_sp64"): Likewise.
16612 ("movdi_pic_gotdata_op"): Likewise.
16613 ("*movsf_insn"): Likewise.
16614 ("*movdf_insn_sp32"): Likewise.
16615 ("*movdf_insn_sp64"): Likewise.
16616 ("*zero_extendhisi2_insn"): Likewise.
16617 ("*zero_extendqihi2_insn"): Likewise.
16618 ("*zero_extendqisi2_insn"): Likewise.
16619 ("*zero_extendqidi2_insn"): Likewise.
16620 ("*zero_extendhidi2_insn"): Likewise.
16621 ("*zero_extendsidi2_insn_sp64"): Likewise.
16622 ("ldfsr"): Likewise.
16623 ("prefetch_64"): Likewise.
16624 ("prefetch_32"): Likewise.
16625 ("tie_ld32"): Likewise.
16626 ("tie_ld64"): Likewise.
16627 ("*tldo_ldub_sp32"): Likewise.
16628 ("*tldo_ldub1_sp32"): Likewise.
16629 ("*tldo_ldub2_sp32"): Likewise.
16630 ("*tldo_ldub_sp64"): Likewise.
16631 ("*tldo_ldub1_sp64"): Likewise.
16632 ("*tldo_ldub2_sp64"): Likewise.
16633 ("*tldo_ldub3_sp64"): Likewise.
16634 ("*tldo_lduh_sp32"): Likewise.
16635 ("*tldo_lduh1_sp32"): Likewise.
16636 ("*tldo_lduh_sp64"): Likewise.
16637 ("*tldo_lduh1_sp64"): Likewise.
16638 ("*tldo_lduh2_sp64"): Likewise.
16639 ("*tldo_lduw_sp32"): Likewise.
16640 ("*tldo_lduw_sp64"): Likewise.
16641 ("*tldo_lduw1_sp64"): Likewise.
16642 ("*tldo_ldx_sp64"): Likewise.
16643 ("*mov<VM32:mode>_insn"): Likewise.
16644 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16645 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16647 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16649 * config/sparc/sparc.md ("type"): New insn type viscmp.
16650 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
16652 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
16653 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
16654 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16655 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
16657 ("n7_vis_logical_11cycle"): Likewise.
16658 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
16659 * config/sparc/niagara2.md ("niag3_vis": Likewise.
16660 * config/sparc/niagara.md ("niag_vis"): Likewise.
16661 * config/sparc/ultra3.md ("us3_fga"): Likewise.
16662 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
16664 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16666 * config/sparc/sparc.md: New instruction type `bmask'.
16667 (bmaskdi_vis): Use the `bmask' type.
16668 (bmasksi_vis): Likewise.
16669 * config/sparc/ultra3.md (us3_array): Likewise.
16670 * config/sparc/niagara7.md (n7_array): Likewise.
16671 * config/sparc/niagara4.md (n4_array): Likewise.
16672 * config/sparc/niagara2.md (niag2_vis): Likewise.
16673 (niag3_vis): Likewise.
16674 * config/sparc/niagara.md (niag_vis): Likewise.
16676 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16678 * ipa-comdats.c: Remove optimize check from gate.
16679 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
16680 for functions not optimized.
16681 (ipa_fn_summary_read): Skip optimize check.
16682 (ipa_fn_summary_write): Likewise.
16683 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
16685 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
16687 (can_inline_edge_p): Check flag_pcc_struct_return for match.
16688 (check_callers): Give up on caller which is not optimized.
16689 (inline_small_functions): Likewise.
16690 (ipa_inline): Do not give up when not optimizing.
16691 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
16692 away unoptimizes cdtors.
16693 (whole_program_function_and_variable_visibility): Do
16694 ipa_discover_readonly_nonaddressable_vars in LTO mode.
16695 * ipa.c (process_references): Do not check optimize.
16696 (symbol_table::remove_unreachable_nodes): Update optimize check.
16697 (set_writeonly_bit): Update optimize check.
16698 (pass_ipa_cdtor_merge::gate): Do not check optimize.
16699 (pass_ipa_single_use::gate): Remove.
16701 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16703 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
16704 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
16705 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
16706 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
16707 permute_load, permute_store, adjust_extract, adjust_splat,
16708 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
16709 replace_swap_with_copy, dump_swap_insn_table,
16710 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
16711 recombine_lvx_pattern, recombine_stvx_pattern,
16712 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
16713 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
16714 to file rs6000-p8swap.c.
16715 * config/rs6000/rs6000-p8swap.c: New file.
16716 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
16717 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
16718 and rs6000*-*-* targets.
16720 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16722 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
16724 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16726 * lto-wrapper.c (merge_and_complain): Do not merge
16727 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
16728 fsigned_zeros, ftrapping_math, fwrapv.
16729 (append_compiler_options): Do not track these options.
16730 (append_linker_options): Likewie
16732 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16734 * cgraphunit.c (cgraph_node::finalize_function): When
16735 !flag_toplevel_reorde set no_reorder flag.
16736 (varpool_node::finalize_decl): Likewise.
16737 (symbol_table::compile): Drop no toplevel reorder path.
16739 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16741 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
16742 edges; zero probability is not better than uninitialized.
16744 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
16746 * asan.h (asan_sanitize_allocas_p): Declare.
16747 * asan.c (asan_sanitize_allocas_p): New function.
16748 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
16749 (handle_builtin_alloca): Likewise.
16750 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
16751 if !asan_sanitize_allocas_p.
16752 * params.def (asan-instrument-allocas): Add new option.
16753 * params.h (ASAN_PROTECT_ALLOCAS): Define.
16754 * opts.c (common_handle_option): Disable allocas sanitization for
16757 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
16759 * asan.c: Include gimple-fold.h.
16760 (get_last_alloca_addr): New function.
16761 (handle_builtin_stackrestore): Likewise.
16762 (handle_builtin_alloca): Likewise.
16763 (asan_emit_allocas_unpoison): Likewise.
16764 (get_mem_refs_of_builtin_call): Add new parameter, remove const
16765 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
16766 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
16767 (instrument_builtin_call): Pass gimple iterator to
16768 get_mem_refs_of_builtin_call.
16769 (last_alloca_addr): New global.
16770 * asan.h (asan_emit_allocas_unpoison): Declare.
16771 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
16772 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
16773 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
16774 if function calls alloca.
16775 * gimple-fold.c (replace_call_with_value): Remove static keyword.
16776 * gimple-fold.h (replace_call_with_value): Declare.
16777 * internal-fn.c: Include asan.h.
16778 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
16779 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
16781 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16783 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
16784 (C_SELFTEST_FLAGS): New.
16785 (CPP_SELFTEST_FLAGS): New.
16786 (SELFTEST_DEPS): New, from deps of s-selftest.
16787 (C_SELFTEST_DEPS): New, from deps of s-selftest.
16788 (CPP_SELFTEST_DEPS): New.
16789 (selftest): Add dependency on s-selftest-c++.
16790 (s-selftest): Rename to...
16791 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
16792 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
16793 than SELFTEST_FLAGS.
16794 (selftest-gdb): Rename to...
16795 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
16797 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
16798 (selftest-valgrind): Rename to...
16799 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
16801 (selftest-valgrind): Reintroduce as an alias for
16802 selftest-c-valgrind.
16803 (s-selftest-c++): New.
16804 (selftest-c++-gdb): New.
16805 (selftest-c++-valgrind): New.
16807 2017-07-06 Olivier Hainque <hainque@adacore.com>
16809 * gcc.c (process_command): When deciding if undefined variables
16810 should be ignored when processing specs, accept "gcc -v" as well.
16812 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16814 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
16815 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
16817 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
16819 * config/arm/arm-cpus.in (armv8-r): Add new entry.
16820 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
16821 * config/arm/arm-tables.opt: Regenerate.
16822 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
16824 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
16826 2017-07-06 Carl Love <cel@us.ibm.com>
16828 * ChangeLog: Clean up from mid air collision
16830 2017-07-06 Carl Love <cel@us.ibm.com>
16832 * config/rs6000/rs6000-c.c: Add support for built-in functions
16833 vector signed int vec_subc (vector signed int, vector signed int);
16834 vector signed __int128 vec_subc (vector signed __int128,
16835 vector signed __int128);
16836 vector unsigned __int128 vec_subc (vector unsigned __int128,
16837 vector unsigned __int128);
16838 vector signed int vec_sube (vector signed int, vector signed int,
16839 vector signed int);
16840 vector unsigned int vec_sube (vector unsigned int,
16841 vector unsigned int,
16842 vector unsigned int);
16843 vector signed __int128 vec_sube (vector signed __int128,
16844 vector signed __int128,
16845 vector signed__int128);
16846 vector unsigned __int128 vec_sube (vector unsigned __int128,
16847 vector unsigned __int128,
16848 vector unsigned __int128);
16849 vector signed int vec_subec (vector signed int, vector signed int,
16850 vector signed int);
16851 vector unsigned int vec_subec (vector unsigned int,
16852 vector unsigned int,
16853 vector unsigned int);
16854 vector signed __int128 vec_subec (vector signed __int128,
16855 vector signed __int128,
16856 vector signed__int128);
16857 vector unsigned __int128 vec_subec (vector unsigned __int128,
16858 vector unsigned __int128,
16859 vector unsigned __int128);
16860 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
16861 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
16862 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
16863 BU_ALTIVEC_OVERLOAD_X definitions.
16864 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
16865 * doc/extend.texi: Update the built-in documentation file for the new
16866 built-in functions.
16868 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16871 * diagnostic-show-locus.c (layout::layout): Use start and finish
16872 spelling location for the start and finish of each range.
16873 * genmatch.c (linemap_client_expand_location_to_spelling_point):
16874 Add unused aspect param.
16875 * input.c (expand_location_1): Add "aspect" param, and use it
16876 to access the correct part of the location.
16877 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
16879 (expand_location_to_spelling_point): Likewise.
16880 (linemap_client_expand_location_to_spelling_point): Add "aspect"
16881 param, and pass it to expand_location_1.
16883 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
16885 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
16886 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
16887 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
16888 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
16889 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
16890 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
16891 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
16892 _mm_maskz_getmant_ss): New intrinsics.
16893 (__builtin_ia32_getexpss128_mask): Changed to ...
16894 __builtin_ia32_getexpss128_round ... this.
16895 (__builtin_ia32_getexpsd128_mask): Changed to ...
16896 __builtin_ia32_getexpsd128_round ... this.
16897 * config/i386/i386-builtin-types.def
16898 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
16899 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
16900 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
16901 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
16902 __builtin_ia32_getmantss_mask_round): New builtins.
16903 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
16904 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
16905 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
16906 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
16907 * config/i386/sse.md
16908 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
16909 avx512f_sgetexp<mode><mask_scalar_name>
16910 <round_saeonly_scalar_name> ... this.
16911 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
16912 %0, %1, %2<round_saeonly_op3>}): Changed to ...
16913 vgetexp<ssescalarmodesuffix>
16914 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16915 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
16916 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
16917 avx512f_vgetmant<mode><mask_scalar_name>
16918 <round_saeonly_scalar_name> ... this.
16919 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
16920 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
16921 vgetmant<ssescalarmodesuffix>
16922 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
16923 %0<mask_scalar_operand4>, %1, %2
16924 <round_saeonly_scalar_mask_op4>, %3} ... this.
16925 * config/i386/subst.md (mask_scalar_operand4,
16926 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
16927 round_saeonly_scalar_nimm_predicate): New subst attributes.
16929 2017-07-06 Julia Koval <julia.koval@intel.com>
16931 * config/i386/i386.c (ix86_erase_embedded_rounding):
16932 Remove code for old rounding pattern.
16934 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
16936 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
16938 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
16940 * doc/sourcebuild.texi (Test Directives, Variants of
16941 dg-require-support): Add documentation for dg-require-stack-check.
16943 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
16945 * config/i386/subst.md (mask_scalar, round_scalar,
16946 round_saeonly_scalar): New meta-templates.
16947 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
16948 round_scalar_mask_operand3, round_scalar_mask_op3,
16949 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
16950 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
16951 round_saeonly_scalar_constraint,
16952 round_saeonly_scalar_prefix): New subst attribute.
16953 * config/i386/sse.md
16954 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
16955 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
16956 <round_scalar_name> ... this.
16957 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
16958 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
16959 <round_scalar_name> ... this.
16960 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
16961 <sse>_vm<code><mode>3<mask_scalar_name>
16962 <round_saeonly_scalar_name> ... this.
16963 (v<plusminus_mnemonic><ssescalarmodesuffix>
16964 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
16965 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
16966 v<plusminus_mnemonic><ssescalarmodesuffix>
16967 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16968 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
16969 (v<multdiv_mnemonic><ssescalarmodesuffix>
16970 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
16971 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
16972 v<multdiv_mnemonic><ssescalarmodesuffix>
16973 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16974 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
16975 (v<maxmin_float><ssescalarmodesuffix>
16976 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
16977 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
16978 v<maxmin_float><ssescalarmodesuffix>
16979 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16980 %0<mask_scalar_operand3>, %1, %<iptr>2
16981 <round_saeonly_scalar_mask_op3>} ... this.
16983 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
16985 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
16986 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
16988 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
16989 Alan Hayward <alan.hayward@arm.com>
16990 David Sherwood <david.sherwood@arm.com>
16992 * combine.c (simplify_if_then_else): Remove "enum" before
16994 * compare-elim.c (can_eliminate_compare): Likewise.
16995 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
16997 (aarch64_lookup_simd_builtin_type): Likewise.
16998 (aarch64_simd_builtin_type): Likewise.
16999 (aarch64_init_simd_builtin_types): Likewise.
17000 (aarch64_simd_expand_args): Likewise.
17001 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
17003 (aarch64_reverse_mask): Likewise.
17004 (aarch64_simd_emit_reg_reg_move): Likewise.
17005 (aarch64_gen_adjusted_ldpstp): Likewise.
17006 (aarch64_ccmp_mode_to_code): Likewise.
17007 (aarch64_operands_ok_for_ldpstp): Likewise.
17008 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17009 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
17011 (aarch64_min_divisions_for_recip_mul): Likewise.
17012 (aarch64_reassociation_width): Likewise.
17013 (aarch64_get_condition_code_1): Likewise.
17014 (aarch64_simd_emit_reg_reg_move): Likewise.
17015 (aarch64_simd_attr_length_rglist): Likewise.
17016 (aarch64_reverse_mask): Likewise.
17017 (aarch64_operands_ok_for_ldpstp): Likewise.
17018 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17019 (aarch64_gen_adjusted_ldpstp): Likewise.
17020 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
17022 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
17023 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
17024 (arm_lookup_simd_builtin_type): Likewise.
17025 (arm_simd_builtin_type): Likewise.
17026 (arm_init_simd_builtin_types): Likewise.
17027 (arm_expand_builtin_args): Likewise.
17028 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
17029 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
17030 (ft32_setup_incoming_varargs): Likewise.
17031 (ft32_function_arg): Likewise.
17032 (ft32_function_arg_advance): Likewise.
17033 (ft32_pass_by_reference): Likewise.
17034 (ft32_arg_partial_bytes): Likewise.
17035 (ft32_valid_pointer_mode): Likewise.
17036 (ft32_addr_space_pointer_mode): Likewise.
17037 (ft32_addr_space_legitimate_address_p): Likewise.
17038 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
17040 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
17041 (ix86_emit_outlined_ms2sysv_restore): Likewise.
17042 (iamcu_alignment): Likewise.
17043 (canonicalize_vector_int_perm): Likewise.
17044 (ix86_noce_conversion_profitable_p): Likewise.
17045 (ix86_mpx_bound_mode): Likewise.
17046 (ix86_operands_ok_for_move_multiple): Likewise.
17047 * config/microblaze/microblaze-protos.h
17048 (microblaze_expand_conditional_branch_reg): Likewise.
17049 * config/microblaze/microblaze.c
17050 (microblaze_expand_conditional_branch_reg): Likewise.
17051 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
17053 (rs6000_reassociation_width): Likewise.
17054 (rs6000_invalid_binary_op): Likewise.
17055 (fusion_p9_p): Likewise.
17056 (emit_fusion_p9_load): Likewise.
17057 (emit_fusion_p9_store): Likewise.
17058 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
17060 (riscv_hard_regno_mode_ok_p): Likewise.
17061 (riscv_address_insns): Likewise.
17062 (riscv_split_symbol): Likewise.
17063 (riscv_legitimize_move): Likewise.
17064 (riscv_function_value): Likewise.
17065 (riscv_hard_regno_nregs): Likewise.
17066 (riscv_expand_builtin): Likewise.
17067 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
17068 (riscv_build_integer): Likewise.
17069 (riscv_split_integer): Likewise.
17070 (riscv_legitimate_constant_p): Likewise.
17071 (riscv_cannot_force_const_mem): Likewise.
17072 (riscv_regno_mode_ok_for_base_p): Likewise.
17073 (riscv_valid_base_register_p): Likewise.
17074 (riscv_valid_offset_p): Likewise.
17075 (riscv_valid_lo_sum_p): Likewise.
17076 (riscv_classify_address): Likewise.
17077 (riscv_legitimate_address_p): Likewise.
17078 (riscv_address_insns): Likewise.
17079 (riscv_load_store_insns): Likewise.
17080 (riscv_force_binary): Likewise.
17081 (riscv_split_symbol): Likewise.
17082 (riscv_force_address): Likewise.
17083 (riscv_legitimize_address): Likewise.
17084 (riscv_move_integer): Likewise.
17085 (riscv_legitimize_const_move): Likewise.
17086 (riscv_legitimize_move): Likewise.
17087 (riscv_address_cost): Likewise.
17088 (riscv_subword): Likewise.
17089 (riscv_output_move): Likewise.
17090 (riscv_canonicalize_int_order_test): Likewise.
17091 (riscv_emit_int_order_test): Likewise.
17092 (riscv_function_arg_boundary): Likewise.
17093 (riscv_pass_mode_in_fpr_p): Likewise.
17094 (riscv_pass_fpr_single): Likewise.
17095 (riscv_pass_fpr_pair): Likewise.
17096 (riscv_get_arg_info): Likewise.
17097 (riscv_function_arg): Likewise.
17098 (riscv_function_arg_advance): Likewise.
17099 (riscv_arg_partial_bytes): Likewise.
17100 (riscv_function_value): Likewise.
17101 (riscv_pass_by_reference): Likewise.
17102 (riscv_setup_incoming_varargs): Likewise.
17103 (riscv_print_operand): Likewise.
17104 (riscv_elf_select_rtx_section): Likewise.
17105 (riscv_save_restore_reg): Likewise.
17106 (riscv_for_each_saved_reg): Likewise.
17107 (riscv_register_move_cost): Likewise.
17108 (riscv_hard_regno_mode_ok_p): Likewise.
17109 (riscv_hard_regno_nregs): Likewise.
17110 (riscv_class_max_nregs): Likewise.
17111 (riscv_memory_move_cost): Likewise.
17112 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
17113 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
17114 (rl78_addr_space_address_mode): Likewise.
17115 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17117 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17118 (rs6000_reassociation_width): Likewise.
17119 (rs6000_invalid_binary_op): Likewise.
17120 (fusion_p9_p): Likewise.
17121 (emit_fusion_p9_load): Likewise.
17122 (emit_fusion_p9_store): Likewise.
17123 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
17124 (ok_for_simple_move_operands): Likewise.
17125 (ok_for_simple_move_strict_operands): Likewise.
17126 (ok_for_simple_arith_logic_operands): Likewise.
17127 (visium_legitimize_reload_address): Likewise.
17128 (visium_select_cc_mode): Likewise.
17129 (output_cbranch): Likewise.
17130 (visium_split_double_move): Likewise.
17131 (visium_expand_copysign): Likewise.
17132 (visium_expand_int_cstore): Likewise.
17133 (visium_expand_fp_cstore): Likewise.
17134 * config/visium/visium.c (visium_pass_by_reference): Likewise.
17135 (visium_function_arg): Likewise.
17136 (visium_function_arg_advance): Likewise.
17137 (visium_libcall_value): Likewise.
17138 (visium_setup_incoming_varargs): Likewise.
17139 (visium_legitimate_constant_p): Likewise.
17140 (visium_legitimate_address_p): Likewise.
17141 (visium_legitimize_address): Likewise.
17142 (visium_secondary_reload): Likewise.
17143 (visium_register_move_cost): Likewise.
17144 (visium_memory_move_cost): Likewise.
17145 (prepare_move_operands): Likewise.
17146 (ok_for_simple_move_operands): Likewise.
17147 (ok_for_simple_move_strict_operands): Likewise.
17148 (ok_for_simple_arith_logic_operands): Likewise.
17149 (visium_function_value_1): Likewise.
17150 (rtx_ok_for_offset_p): Likewise.
17151 (visium_legitimize_reload_address): Likewise.
17152 (visium_split_double_move): Likewise.
17153 (visium_expand_copysign): Likewise.
17154 (visium_expand_int_cstore): Likewise.
17155 (visium_expand_fp_cstore): Likewise.
17156 (visium_split_cstore): Likewise.
17157 (visium_select_cc_mode): Likewise.
17158 (visium_split_cbranch): Likewise.
17159 (output_cbranch): Likewise.
17160 (visium_print_operand_address): Likewise.
17161 * expmed.c (flip_storage_order): Likewise.
17162 * expmed.h (emit_cstore): Likewise.
17163 (flip_storage_order): Likewise.
17164 * genrecog.c (validate_pattern): Likewise.
17165 * hsa-gen.c (gen_hsa_addr): Likewise.
17166 * internal-fn.c (expand_arith_overflow): Likewise.
17167 * ira-color.c (allocno_copy_cost_saving): Likewise.
17168 * lra-assigns.c (find_hard_regno_for_1): Likewise.
17169 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
17170 (process_invariant_for_inheritance): Likewise.
17171 * lra-eliminations.c (move_plus_up): Likewise.
17172 * omp-low.c (lower_oacc_reductions): Likewise.
17173 * simplify-rtx.c (simplify_subreg): Likewise.
17174 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
17175 (TARGET_CHKP_BOUND_MODE): Likewise..
17176 * targhooks.c (default_chkp_bound_mode): Likewise.
17177 (default_setup_incoming_vararg_bounds): Likewise.
17178 * targhooks.h (default_chkp_bound_mode): Likewise.
17179 (default_setup_incoming_vararg_bounds): Likewise.
17180 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
17181 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
17182 (have_whole_vector_shift): Likewise.
17183 * tree-vect-stmts.c (vectorizable_load): Likewise.
17184 * doc/tm.texi: Regenerate.
17186 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17188 Graceful degrade if Binutils PR21472 is not available.
17191 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
17192 .rodata in flash test fails.
17193 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
17194 * confgure: Regenerate.
17195 * config.in: Regenerate.
17196 * config/avr/avr.c (avr_asm_named_section)
17197 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
17198 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
17199 (avr_asm_init_sections): Same.
17201 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17203 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
17204 (fma<VH:mode>4_intrinsic): Likewise.
17205 (*fmsub<VCVTF:mode>4): Likewise.
17206 (*fmsub<VH:mode>4_intrinsic): Likewise.
17208 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17211 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
17212 Don't depend on "optimize > 0".
17213 (out_movhi_r_mr, out_movqi_mr_r): Same.
17214 (out_movhi_mr_r, out_movqi_r_mr): Same.
17215 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
17216 io_address_operand on "optimize > 0".
17218 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17220 * tree-loop-distribution.c: Add general explanantion on the pass.
17221 (generate_loops_for_partition): Mark distributed loop.
17222 (pg_add_dependence_edges): New parameter. Handle alias data
17223 dependence specially and record it in the parameter if asked.
17224 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
17225 (init_partition_graph_vertices, add_partition_graph_edge): New.
17226 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
17227 (free_partition_graph_vdata, build_partition_graph): New.
17228 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
17229 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
17230 (data_ref_segment_size, latch_dominated_by_data_ref): New.
17231 (compute_alias_check_pairs, version_loop_by_alias_check): New.
17232 (version_for_distribution_p, finalize_partitions): New.
17233 (distribute_loop): Handle alias data dependence specially. Factor
17234 out loop fusion code as functions and call these functions.
17236 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17238 * tree-loop-distribution.c (classify_partition): New parameter and
17239 better handle reduction statement.
17240 (rdg_build_partitions): Revise comment.
17241 (distribute_loop): Compute statements in all partitions and pass it
17242 to classify_partition.
17244 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17246 * tree-loop-distribution.c (enum partition_type): New.
17247 (struct partition): New field type.
17248 (partition_merge_into): Add parameter. Update partition type.
17249 (data_dep_in_cycle_p, update_type_for_merge): New functions.
17250 (build_rdg_partition_for_vertex): Compute partition type.
17251 (rdg_build_partitions): Dump partition type.
17252 (distribute_loop): Update calls to partition_merge_into.
17254 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17256 * tree-loop-distribution.c (struct ddr_hasher): New.
17257 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
17259 (classify_partition): Call get_data_dependence.
17260 (pg_add_dependence_edges): Ditto.
17261 (distribute_loop): Release data dependence hash table.
17263 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17265 * tree-loop-distribution.c (ref_base_address): Delete.
17266 (similar_memory_accesses): Rename ...
17267 (share_memory_accesses): ... to this. Check if partitions access
17268 the same memory reference.
17269 (distribute_loop): Call share_memory_accesses.
17271 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17273 * tree-loop-distribution.c (struct partition): New field recording
17274 its data reference.
17275 (partition_alloc, partition_free): Init and release data refs.
17276 (partition_merge_into): Merge data refs.
17277 (build_rdg_partition_for_vertex): Collect data refs for partition.
17278 (pg_add_dependence_edges): Change parameters from vector to bitmap.
17280 (distribute_loop): Remve data refs from vertice data of partition
17283 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17285 * tree-loop-distribution.c (params.h): Include header file.
17286 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
17287 (datarefs_vec): New global var.
17288 (create_rdg_vertices): Use datarefs_vec directly.
17289 (free_rdg): Don't free data references.
17290 (build_rdg): Update use. Don't free data references.
17291 (distribute_loop): Compute global variable for data references.
17292 Bail out if there are too many data references.
17294 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17296 * tree-loop-distribution.c (loop_nest): New global var.
17297 (build_rdg): Use loop directly, rather than loop nest.
17298 (pg_add_dependence_edges): Remove loop nest parameter. Use global
17300 (distribute_loop): Compute global variable loop nest. Update use.
17302 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17304 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
17305 (partition_merge_into): New parameter. Dump reason for fusion.
17306 (distribute_loop): Update use of partition_merge_into.
17308 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17310 * tree-loop-distribution.c (bb_top_order_index): New.
17311 (bb_top_order_index_size, bb_top_order_cmp): New.
17312 (stmts_from_loop): Use topological order.
17313 (pass_loop_distribution::execute): Compute and release topological
17314 order for basic blocks.
17316 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17318 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
17321 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17323 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
17324 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
17325 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
17326 * internal-fn.def (LOOP_DIST_ALIAS): New.
17327 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
17328 (fold_loop_internal_call): ... this.
17329 (vect_loop_dist_alias_call): New function.
17330 (set_uid_loop_bbs): Call fold_loop_internal_call.
17331 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
17334 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
17337 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
17338 Require dead FLAGS_REG at the beginning of a peephole.
17340 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
17343 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
17344 arguments in the call to __builtin_ia32_sbb_u32.
17345 (_subborrow_u64): Swap _X and _Y arguments in the call to
17346 __builtin_ia32_sbb_u64.
17348 2017-07-04 Jakub Jelinek <jakub@redhat.com>
17351 * tree-vrp.c (compare_assert_loc): Turn into a function template
17352 with stable template parameter. Only test if a->e is NULL,
17353 !a->e == !b->e has been verified already. Use e == NULL or
17354 e != NULL instead of e or ! e tests. If stable is true, don't use
17355 iterative_hash_expr, on the other side allow a or b or both NULL
17356 and sort the NULLs last.
17357 (process_assert_insertions): Sort using compare_assert_loc<false>
17358 instead of compare_assert_loc, later sort using
17359 compare_assert_loc<true> before calling process_assert_insertions_for
17360 in a loop. Use break instead of continue once seen NULL pointer.
17362 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
17364 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17365 Cortex-R7 and Cortex-R8 processors.
17367 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17369 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
17370 uninitialized while src is not.
17372 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
17374 * common/config/arm/arm-common.c: Adjust include path for
17376 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
17377 (arm-cpu.h): Create in build directory. Adjust dependency rules.
17378 (arm-cpu-data.h): Likewise.
17379 (arm-cpu-cdata.h): Likewise.
17380 * config/arm/arm-cpu.h: Delete.
17381 * config/arm/arm-cpu-cdata.h: Delete.
17382 * config/arm/arm-cpu-data.h: Delete.
17384 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
17386 * config/arm/arm-cpus.in (cortex-a55): New.
17387 (cortex-a75): Likewise.
17388 (cortex-a75.cortex-a55): Likewise.
17389 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
17391 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
17392 * config/arm/arm-cpu-cdata.h: Regenerate.
17393 * config/arm/arm-cpu-data.h: Regenerate.
17394 * config/arm/arm-cpu.h: Regenerate.
17395 * config/arm/arm-tables.opt: Regenerate.
17396 * config/arm/arm-tune.md: Regenerate.
17398 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17400 * haifa-sched.c (sched_create_recovery_edges): Update profile.
17402 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17404 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
17407 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
17409 PR tree-optimization/81292
17410 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
17411 full_string_p, also call adjust_related_strinfos if the adjustment
17412 is simple, otherwise invalidate related strinfos.
17414 2017-07-04 Martin Liska <mliska@suse.cz>
17417 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
17418 newly created variable as DECL_IGNORED_P.
17420 2017-07-04 Martin Liska <mliska@suse.cz>
17423 * ipa-inline.c (inline_small_functions):
17424 Use xstrdup_for_dump.
17426 2017-07-04 Tom de Vries <tom@codesourcery.com>
17428 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
17430 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
17433 * config/darwin.c (darwin_function_switched_text_sections):
17436 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
17438 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
17440 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17442 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
17444 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17446 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
17447 min_profitable_iters, and th as inclusive lower bounds.
17448 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
17449 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
17450 for min_profitable_iters and min_profitable_estimate.
17451 (vect_transform_loop): Treat th as an inclusive lower bound.
17452 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
17454 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
17457 * config/darwin.c (darwin_function_switched_text_sections):
17458 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
17459 in two pieces, and suppress the use of buf.
17461 2017-07-03 Nathan Sidwell <nathan@acm.org>
17463 * hash-table.h (hash_table_mod1): Fix indentation.
17465 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17467 PR middle-end/81290
17468 * predict.c (force_edge_cold): Be more careful about propagation
17470 * profile-count.h (profile_probability::guessed,
17471 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
17473 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
17475 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
17477 * doc/invoke.texi (rcpc architecture extension): Document it.
17479 2017-07-03 Richard Biener <rguenther@suse.de>
17481 PR tree-optimization/60510
17482 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
17483 the scalar reduction PHI and use it.
17484 (vectorizable_reduction): Properly guard the single_defuse_cycle
17485 path for non-SLP reduction chains where we cannot use it.
17486 Rework reduc_def/index and vector type deduction. Rework
17487 vector operand gathering during reduction op code-gen.
17488 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
17489 chains dissolve the chain and leave it to non-SLP reduction
17492 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17494 * tree-data-ref.h (dr_alignment): Declare.
17495 * tree-data-ref.c (dr_alignment): New function.
17496 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
17497 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
17499 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
17501 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17503 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
17504 and base_misalignment fields.
17505 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
17506 * tree-data-ref.c: Include builtins.h.
17507 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
17508 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
17509 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
17510 * tree-vect-data-refs.c: Include tree-cfg.h.
17511 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
17512 fields instead of calculating an alignment here.
17513 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
17514 innermost_loop_behavior fields.
17516 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17518 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
17520 (DR_STEP_ALIGNMENT): New macro.
17521 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
17522 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
17523 (create_data_ref): Print it.
17524 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
17525 to tell whether the step preserves vector (mis)alignment.
17526 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17527 Move the check for an integer step and generalise to all INTEGER_CST.
17528 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
17529 Print the outer step alignment.
17531 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17533 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
17534 with offset_alignment.
17535 (DR_ALIGNED_TO): Delete.
17536 (DR_OFFSET_ALIGNMENT): New macro.
17537 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
17538 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
17539 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
17540 (create_data_ref): Likewise.
17541 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17542 (vect_analyze_data_refs): Likewise.
17543 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
17544 creating dummy innermost behavior.
17546 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17548 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
17549 with a "innermost_loop_behavior *" and refeence tree.
17550 * tree-data-ref.c (dr_analyze_innermost): Likewise.
17551 (create_data_ref): Update call accordingly.
17552 * tree-predcom.c (find_looparound_phi): Likewise.
17554 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17556 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
17557 fields with dr_wrt_vec_loop.
17558 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
17559 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
17560 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
17561 (vect_dr_behavior): New function.
17562 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17563 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
17564 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
17565 track whether the step preserves the misalignment.
17566 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17567 Use vect_dr_behavior.
17568 (vect_setup_realignment): Update call accordingly.
17569 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
17570 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
17571 call to vect_create_addr_base_for_vector_ref.
17572 (vect_create_cond_for_align_checks): Likewise.
17573 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
17574 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
17575 (vect_recog_mask_conversion_pattern): Likewise.
17576 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
17577 (new_stmt_vec_info): Remove redundant zeroing.
17579 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17581 * common/config/arm/arm-common.c (arm_be8_option): New function.
17582 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
17583 (ISA_ARMv6): Add isa_bit_be8.
17584 * config/arm/arm.h (arm_be8_option): Add prototype.
17585 (BE8_SPEC_FUNCTION): New define.
17586 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
17587 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
17588 (mlittle-endian): Similarly.
17589 (mbe8, mbe32): New options.
17590 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
17591 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
17593 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17595 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
17597 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17599 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
17600 (cleanup_tree_cfg_bb): Use it.
17601 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
17603 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
17605 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17608 * loop-doloop.c (add_test): Update profile.
17610 2017-07-03 Martin Liska <mliska@suse.cz>
17613 * sanopt.c (rewrite_usage_of_param): New function.
17614 (sanitize_rewrite_addressable_params): Likewise.
17615 (pass_sanopt::execute): Call rewrite_usage_of_param.
17617 2017-07-03 Richard Biener <rguenther@suse.de>
17619 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
17620 back to using VIEW_CONVERT_EXPR.
17622 2017-07-03 Martin Liska <mliska@suse.cz>
17625 * doc/extend.texi: Document when a resolver function is
17626 generated for target_clones.
17628 2017-07-03 Martin Liska <mliska@suse.cz>
17630 * asan.c (asan_emit_stack_protection): Unpoison just red zones
17631 and shadow memory of auto variables which are subject of
17632 use-after-scope sanitization.
17633 (asan_expand_mark_ifn): Add do set only when is_poison.
17635 2016-07-03 Richard Biener <rguenther@suse.de>
17637 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
17639 (vect_force_simple_reduction): Record reduction def -> phi mapping.
17640 (vectorizable_reduction): Perform reduction PHI creation when
17641 visiting a reduction PHI and adjust and simplify code generation
17642 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
17643 (vect_transform_loop): Visit reduction PHIs.
17644 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
17645 defs into the SLP tree.
17646 (vect_build_slp_tree): Reduction defs terminate the recursion.
17647 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
17649 (vect_get_vec_defs_for_stmt_copy): Export.
17650 (vect_get_vec_defs): Likewise.
17651 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
17653 (vect_get_vec_defs_for_stmt_copy): Declare.
17654 (vect_get_vec_defs): Likewise.
17656 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17658 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
17659 parameter with a "loop" parameter and use it instead of the
17660 loop containing DR_STMT. Don't check simple_iv when doing
17661 BB analysis. Describe the two analysis modes in the comment.
17663 2017-07-03 Tom de Vries <tom@codesourcery.com>
17665 PR tree-optimization/69468
17666 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
17667 (find_same_succ_bb): Handle ignore_edge_flags.
17669 2017-07-03 Tom de Vries <tom@codesourcery.com>
17671 PR tree-optimization/81192
17672 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
17674 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
17676 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
17678 2017-07-03 Tom de Vries <tom@codesourcery.com>
17680 PR tree-optimization/81192
17681 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
17682 BB_SAME_SUCC (bb) == NULL.
17684 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17686 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
17689 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17691 * dumpfile.c: Include profile-count.h
17692 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
17694 (insert_cond_bb): Update profile.
17695 * tree-cfg.h (insert_cond_bb): Update prototype.
17696 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
17697 * tree-dump.c: Do not include tree-cfg.
17699 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17701 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
17703 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17705 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
17708 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17710 * tree-complex.c (expand_complex_div_wide): update profile.
17712 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17713 Alan Hayward <alan.hayward@arm.com>
17714 David Sherwood <david.sherwood@arm.com>
17716 * Makefile.in (MACHMODE_H): Remove insn-modes.h
17717 (CORETYPES_H): New define.
17718 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
17719 (insn-modes-inline.h, s-modes-inline-h): New rules.
17720 (generated_files): Add insn-modes-inline.h.
17721 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
17722 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
17723 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
17724 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
17725 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
17726 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
17727 (build/gencodes.o, build/genconditions.o): Likewise.
17728 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
17729 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
17730 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
17731 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
17732 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
17733 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
17734 * coretypes.h: Include everything up to real.h for generators.
17735 Include insn-modes.h first. Include wide-int-print.h after
17736 wide-int.h. Include insn-modes-inline.h and then machmode.h.
17737 * machmode.h: Don't include insn-modes.h here.
17738 * function-tests.c: Remove includes of signop.h, machmode.h,
17739 double-int.h and wide-int.h.
17741 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
17743 * optc-save-gen.awk: Likewise.
17744 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
17745 * godump.c: Remove include of wide-int-print.h.
17746 * pretty-print.h: Likewise.
17747 * wide-int-print.cc: Likewise.
17748 * wide-int.cc: Likewise.
17749 * hash-map-tests.c: Remove include of signop.h.
17750 * hash-set-tests.c: Likewise.
17751 * rtl-tests.c: Likewise.
17752 * mkconfig.sh: Remove include of machmode.h.
17753 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
17755 (emit_insn_modes_inline_h): ...this new function. Emit the code
17756 into an insn-modes-inline.h header file, adding appropriate
17757 include guards and end comments.
17758 (emit_insn_modes_c_header): Remove include of machmode.h.
17759 (emit_min_insn_modes_c_header): Include coretypes.h rather than
17761 (main): Handle -i flag and call emit_insn_modes_inline_h when
17764 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17766 * tree-ssa-strlen.c (strinfo): Rename the length field to
17767 nonzero_chars. Add a full_string_p field.
17768 (compare_nonzero_chars, zero_length_string_p): New functions.
17769 (get_addr_stridx): Add an offset_out parameter.
17770 Use compare_nonzero_chars.
17771 (get_stridx): Update accordingly. Use compare_nonzero_chars.
17772 (new_strinfo): Update after above changes to strinfo.
17773 (set_endptr_and_length): Set full_string_p.
17774 (get_string_length): Update after above changes to strinfo.
17775 (unshare_strinfo): Update call to new_strinfo.
17776 (maybe_invalidate): Likewise.
17777 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
17778 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
17779 as a uhwi instead of an shwi. Update after above changes to
17780 strinfo and new_strinfo.
17781 (zero_length_string): Assert that chainsi contains full strings.
17782 Use zero_length_string_p. Update call to new_strinfo.
17783 (adjust_related_strinfos): Update after above changes to strinfo.
17784 Copy full_string_p from origsi.
17785 (adjust_last_stmt): Use zero_length_string_p.
17786 (handle_builtin_strlen): Update after above changes to strinfo and
17787 new_strinfo. Install the lhs as the string length if the previous
17788 entry didn't describe a full string.
17789 (handle_builtin_strchr): Update after above changes to strinfo
17791 (handle_builtin_strcpy): Likewise.
17792 (handle_builtin_strcat): Likewise.
17793 (handle_builtin_malloc): Likewise.
17794 (handle_pointer_plus): Likewise.
17795 (handle_builtin_memcpy): Likewise. Track nonzero characters
17796 that aren't necessarily followed by a nul terminator.
17797 (handle_char_store): Likewise.
17799 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17801 PR tree-optimization/80769
17802 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
17803 for malloc and calloc. Document the new invariant that all related
17804 strinfos have delayed lengths or none do.
17805 (verify_related_strinfos): Move earlier in file.
17806 (set_endptr_and_length): New function, split out from...
17807 (get_string_length): ...here. Also set the lengths of related
17809 (zero_length_string): Assert that chainsi has known (rather than
17811 (adjust_related_strinfos): Likewise.
17813 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17815 PR tree-optimization/81136
17816 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
17817 assert that two references with the same misalignment have the same
17818 compile-time misalignment if those compile-time misalignments
17821 2017-07-01 Andi Kleen <ak@linux.intel.com>
17823 * print-tree.c (print_node): Print all attributes.
17825 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17827 * cfg.c (scale_bbs_frequencies): New function.
17828 * cfg.h (scale_bbs_frequencies): Declare it.
17829 * cfgloopanal.c (single_likely_exit): Cleanup.
17830 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
17832 (scale_loop_profile): Likewise.
17833 (loop_version): Likewise.
17834 (create_empty_loop_on_edge): Update.
17835 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
17836 scale_loop_frequencies, scale_loop_profile, loopify,
17837 loop_version): Update prototypes.
17838 * modulo-sched.c (sms_schedule): Update.
17839 * predict.c (unlikely_executed_edge_p): Also check probability.
17840 (probably_never_executed_edge_p): Fix typo.
17841 * tree-if-conv.c (version_loop_for_if_conversion): Update.
17842 * tree-parloops.c (gen_parallel_loop): Update.
17843 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
17844 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
17845 * tree-ssa-loop-split.c (split_loop): Update.
17846 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
17847 * tree-vect-loop-manip.c (vect_do_peeling): Update.
17848 (vect_loop_versioning): Update.
17849 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
17851 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17853 * trans-mem.c (split_bb_make_tm_edge): Update profile.
17855 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17857 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
17858 to keep profile consistent.
17860 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17862 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
17863 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
17864 * profile-count.h (max_safe_multiplier): Make unsigned.
17865 (profile_count::guessed_zero): New.
17867 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17869 * bb-reorder.c (fix_up_crossing_landing_pad,
17870 fix_crossing_conditional_branches): Use make_single_succ_edge
17871 to keep profile consistent.
17873 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17875 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
17878 2017-07-01 Jakub Jelinek <jakub@redhat.com>
17881 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
17882 the right scopes, make sure cond_jump isn't preserved between multiple
17883 iterations. Search for fallthru edge whenever there are 3+ edges and
17884 use find_fallthru_edge for it.
17886 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17888 Patch by Alexander Monakov <amonakov@ispras.ru>
17889 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
17890 probabilities consistently.
17892 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17894 * pa.c (pa_expand_compare_and_swap_loop): Update call of
17895 emit_cmp_and_jump_insns.
17897 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17900 * tree-inline.c (expand_call_inline): Combine profile statuses.
17902 2017-06-30 Andrew Pinski <apinski@cavium.com>
17904 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
17905 fold_stmt returned true.
17907 2017-06-30 Nathan Sidwell <nathan@acm.org>
17909 * ggc.h (empty_string): Delete.
17910 * cfgexpand.c (expand_asm_stmt): Use plain "".
17911 * optabs.c (expand_asm_memory_barrier): Likewise.
17912 * stringpool.c (empty_string): Delete.
17913 (digit_vector, digit_string): Delete.
17914 (ggc_alloc_string): Use plain "", don't optimize single digit
17915 strings. Use ggc_alloc_atomic.
17917 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
17919 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
17920 comparison set and one other set, use the cost of the non-comparison
17923 2017-06-30 Nathan Sidwell <nathan@acm.org>
17925 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
17928 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
17930 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
17931 loops. Remove now unneeded calls to gimple_switch_set_label() that
17932 just set removed labels to NULL_TREE.
17934 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
17936 * tree-ssanames.c (set_range_info_raw): Abstract from ...
17937 (set_range_info): ...here. Only call set_range_info_raw if domain
17939 (set_nonzero_bits): Call set_range_info_raw.
17940 * tree-ssanames.h (set_range_info_raw): New.
17942 2017-06-30 Jakub Jelinek <jakub@redhat.com>
17945 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
17946 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
17947 of nonimmediate_operand and <store_mask_constraint> instead of m
17948 for the input operand. For V8FI iterator, always split if input
17949 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
17950 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
17951 <store_mask_predicate> instead of register_operand and
17952 <store_mask_constraint> instead of v for the input operand. Make
17953 sure both operands aren't MEMs for if not <mask_applied>.
17955 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
17957 * lto-wrapper.c (copy_file) Close both file descriptors before
17960 2017-06-30 Martin Liska <mliska@suse.cz>
17963 * multiple_target.c (create_dispatcher_calls): Make ifunc
17964 also for function that don't have calls or are not referenced.
17966 2017-06-30 Richard Biener <rguenther@suse.de>
17968 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
17969 analyze the first scalar stmt. Move vector type computation
17970 for the BB case here from ...
17971 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
17972 live operation processing in the SLP case properly.
17974 2017-06-30 Richard Biener <rguenther@suse.de>
17976 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
17978 2017-06-30 Martin Liska <mliska@suse.cz>
17981 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
17982 before BUILT_IN_UNWIND_RESUME when ASAN is used.
17984 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
17986 * doc/invoke.texi (AArch64): Add missing options and remove redundant
17989 2017-06-30 Richard Biener <rguenther@suse.de>
17991 PR tree-optimization/81249
17992 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
17993 condition reduction result to original scalar type.
17995 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17997 * profile-count.h (enum profile_quality): Fix typos and whitespace
18000 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18002 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
18003 type for branch probabilities.
18005 2017-06-29 Julian Brown <julian@codesourcery.com>
18006 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18008 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
18009 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
18010 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
18011 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
18013 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18015 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
18016 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
18017 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
18018 CC usage from generic code to here.
18019 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
18020 CC usage into the target macros.
18022 2017-06-29 Maya Rashish <coypu@sdf.org>
18024 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
18027 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18029 * arm/arm-builtins.c: Include profile-count.h
18030 * except.c (sjlj_emit_function_enter): Use
18031 profile_probability::unlikely.
18033 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18035 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
18036 and tocrel_offset be pointer args rather than implicitly using
18038 (legitimate_constant_pool_address_p, rs6000_emit_move,
18039 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
18040 tocrel_offset and use in toc_relative_expr_p call.
18041 (print_operand, print_operand_address): Use static tocrel_base_oac
18042 and tocrel_offset_oac.
18043 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
18046 2017-06-29 Maya Rashish <coypu@sdf.org>
18048 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
18050 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
18052 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
18053 objects, take into account only the alignment of 'op0' and 'mode1' if
18056 2017-06-29 Steve Ellcey <sellcey@cavium.com>
18058 * ccmp.c (ccmp_tree_comparison_p): New function.
18059 (ccmp_candidate_p): Update to use above function.
18060 (get_compare_parts): New function.
18061 (expand_ccmp_next): Update to use new functions.
18062 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
18064 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
18065 take mode as argument.
18066 * ccmp.h (expand_ccmp_expr): Add mode as argument.
18067 * expr.c (expand_expr_real_1): Pass mode as argument.
18069 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
18071 * combine.c (combine_instructions): Print insns to dump_file, together
18074 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18076 * asan.c (asan_emit_stack_protection): Update.
18077 (create_cond_insert_point): Update.
18078 * auto-profile.c (afdo_propagate_circuit): Update.
18079 * basic-block.h (struct edge_def): Turn probability to
18080 profile_probability.
18081 (EDGE_FREQUENCY): Update.
18082 * bb-reorder.c (find_traces_1_round): Update.
18083 (better_edge_p): Update.
18084 (sanitize_hot_paths): Update.
18085 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
18086 (make_single_succ_edge): Update.
18087 (check_bb_profile): Update.
18088 (dump_edge_info): Update.
18089 (update_bb_profile_for_threading): Update.
18090 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
18091 probabilitycount to 0.
18092 * cfgbuild.c (compute_outgoing_frequencies): Update.
18093 * cfgcleanup.c (try_forward_edges): Update.
18094 (outgoing_edges_match): Update.
18095 (try_crossjump_to_edge): Update.
18096 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
18097 (expand_gimple_tailcall): Update.
18098 (construct_init_block): Use make_single_succ_edge.
18099 (construct_exit_block): Use make_single_succ_edge.
18100 * cfghooks.c (verify_flow_info): Update.
18101 (redirect_edge_succ_nodup): Update.
18102 (split_edge): Update.
18103 (account_profile_record): Update.
18104 * cfgloopanal.c (single_likely_exit): Update.
18105 * cfgloopmanip.c (scale_loop_profile): Update.
18106 (set_zero_probability): Remove.
18107 (duplicate_loop_to_header_edge): Update.
18108 * cfgloopmanip.h (loop_version): Update prototype.
18109 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
18110 (force_nonfallthru_and_redirect): Update.
18111 (update_br_prob_note): Update.
18112 (rtl_verify_edges): Update.
18113 (purge_dead_edges): Update.
18114 (rtl_lv_add_condition_to_bb): Update.
18115 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
18116 * cgraphunit.c (init_lowered_empty_function): Update.
18117 (cgraph_node::expand_thunk): Update.
18118 * cilk-common.c: Include profile-count.h
18119 * dojump.c (inv): Remove.
18120 (jumpifnot): Update.
18121 (jumpifnot_1): Update.
18122 (do_jump_1): Update.
18124 (do_jump_by_parts_greater_rtx): Update.
18125 (do_compare_rtx_and_jump): Update.
18126 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
18127 do_jump_1. do_compare_rtx_and_jump): Update prototype.
18128 * dwarf2cfi.c: Include profile-count.h
18129 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
18130 (sjlj_emit_dispatch_table): Likewise.
18131 * explow.c: Include profile-count.h
18132 * expmed.c (emit_store_flag_force): Update.
18133 (do_cmp_and_jump): Update.
18134 * expr.c (compare_by_pieces_d::generate): Update.
18135 (compare_by_pieces_d::finish_mode): Update.
18136 (emit_block_move_via_loop): Update.
18137 (store_expr_with_bounds): Update.
18138 (store_constructor): Update.
18139 (expand_expr_real_2): Update.
18140 (expand_expr_real_1): Update.
18141 * expr.h (try_casesi, try_tablejump): Update prototypes.
18142 * gimple-pretty-print.c (dump_probability): Update.
18143 (dump_profile): New.
18144 (dump_gimple_label): Update.
18145 (dump_gimple_bb_header): Update.
18146 * graph.c (draw_cfg_node_succ_edges): Update.
18147 * hsa-gen.c (convert_switch_statements): Update.
18148 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
18149 (find_if_case_1): Update.
18150 (find_if_case_2): Update.
18151 * internal-fn.c (expand_arith_overflow_result_store): Update.
18152 (expand_addsub_overflow): Update.
18153 (expand_neg_overflow): Update.
18154 (expand_mul_overflow): Update.
18155 (expand_vector_ubsan_overflow): Update.
18156 * ipa-cp.c (good_cloning_opportunity_p): Update.
18157 * ipa-split.c (split_function): Use make_single_succ_edge.
18158 * ipa-utils.c (ipa_merge_profiles): Update.
18159 * loop-doloop.c (add_test): Update.
18160 (doloop_modify): Update.
18161 * loop-unroll.c (compare_and_jump_seq): Update.
18162 (unroll_loop_runtime_iterations): Update.
18163 * lra-constraints.c (lra_inheritance): Update.
18164 * lto-streamer-in.c (input_cfg): Update.
18165 * lto-streamer-out.c (output_cfg): Update.
18166 * mcf.c (adjust_cfg_counts): Update.
18167 * modulo-sched.c (sms_schedule): Update.
18168 * omp-expand.c (expand_omp_for_init_counts): Update.
18169 (extract_omp_for_update_vars): Update.
18170 (expand_omp_ordered_sink): Update.
18171 (expand_omp_for_ordered_loops): Update.
18172 (expand_omp_for_generic): Update.
18173 (expand_omp_for_static_nochunk): Update.
18174 (expand_omp_for_static_chunk): Update.
18175 (expand_cilk_for): Update.
18176 (expand_omp_simd): Update.
18177 (expand_omp_taskloop_for_outer): Update.
18178 (expand_omp_taskloop_for_inner): Update.
18179 * omp-simd-clone.c (simd_clone_adjust): Update.
18180 * optabs.c (expand_doubleword_shift): Update.
18181 (expand_abs): Update.
18182 (emit_cmp_and_jump_insn_1): Update.
18183 (expand_compare_and_swap_loop): Update.
18184 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
18185 * predict.c (predictable_edge_p): Update.
18186 (edge_probability_reliable_p): Update.
18187 (set_even_probabilities): Update.
18188 (combine_predictions_for_insn): Update.
18189 (combine_predictions_for_bb): Update.
18190 (propagate_freq): Update.
18191 (estimate_bb_frequencies): Update.
18192 (force_edge_cold): Update.
18193 * profile-count.c (profile_count::dump): Add missing space into dump.
18194 (profile_count::debug): Add newline.
18195 (profile_count::differs_from_p): Explicitly convert to unsigned.
18196 (profile_count::stream_in): Update.
18197 (profile_probability::dump): New member function.
18198 (profile_probability::debug): New member function.
18199 (profile_probability::differs_from_p): New member function.
18200 (profile_probability::differs_lot_from_p): New member function.
18201 (profile_probability::stream_in): New member function.
18202 (profile_probability::stream_out): New member function.
18203 * profile-count.h (profile_count_quality): Rename to ...
18204 (profile_quality): ... this one.
18205 (profile_probability): New.
18206 (profile_count): Update.
18207 * profile.c (compute_branch_probabilities): Update.
18208 * recog.c (peep2_attempt): Update.
18209 * sched-ebb.c (schedule_ebbs): Update.
18210 * sched-rgn.c (find_single_block_region): Update.
18211 (compute_dom_prob_ps): Update.
18212 (schedule_region): Update.
18213 * sel-sched-ir.c (compute_succs_info): Update.
18214 * stmt.c (struct case_node): Update.
18215 (do_jump_if_equal): Update.
18216 (get_outgoing_edge_probs): Update.
18217 (conditional_probability): Update.
18218 (emit_case_dispatch_table): Update.
18219 (expand_case): Update.
18220 (expand_sjlj_dispatch_table): Update.
18221 (emit_case_nodes): Update.
18222 * targhooks.c: Update.
18223 * tracer.c (better_p): Update.
18224 (find_best_successor): Update.
18225 * trans-mem.c (expand_transaction): Update.
18226 * tree-call-cdce.c: Update.
18227 * tree-cfg.c (gimple_split_edge): Upate.
18228 (move_sese_region_to_fn): Upate.
18229 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
18230 * tree-eh.c (lower_resx): Upate.
18231 (cleanup_empty_eh_move_lp): Upate.
18232 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18233 * tree-inline.c (copy_edges_for_bb): Update.
18234 (copy_cfg_body): Update.
18235 * tree-parloops.c (gen_parallel_loop): Update.
18236 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
18237 (gimple_gen_time_profiler): Update.
18238 * tree-ssa-dce.c (remove_dead_stmt): Update.
18239 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
18240 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
18241 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
18242 (unloop_loops): Update.
18243 (try_peel_loop): Update.
18244 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18245 * tree-ssa-loop-split.c (connect_loops): Update.
18246 (split_loop): Update.
18247 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
18248 (hoist_guard): Update.
18249 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
18250 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
18251 (value_replacement): Update.
18252 * tree-ssa-reassoc.c (branch_fixup): Update.
18253 * tree-ssa-tail-merge.c (replace_block_by): Update.
18254 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
18255 (create_edge_and_update_destination_phis): Update.
18256 (compute_path_counts): Update.
18257 (recompute_probabilities): Update.
18258 (update_joiner_offpath_counts): Update.
18259 (freqs_to_counts_path): Update.
18260 (duplicate_thread_path): Update.
18261 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
18262 (struct switch_conv_info): Update.
18263 (gen_inbound_check): Update.
18264 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
18265 (vect_do_peeling): Update.
18266 (vect_loop_versioning): Update.
18267 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18268 (optimize_mask_stores): Update.
18269 * ubsan.c (ubsan_expand_null_ifn): Update.
18270 * value-prof.c (gimple_divmod_fixed_value): Update.
18271 (gimple_divmod_fixed_value_transform): Update.
18272 (gimple_mod_pow2): Update.
18273 (gimple_mod_pow2_value_transform): Update.
18274 (gimple_mod_subtract): Update.
18275 (gimple_mod_subtract_transform): Update.
18276 (gimple_ic): Update.
18277 (gimple_stringop_fixed_value): Update.
18278 (gimple_stringops_transform): Update.
18279 * value-prof.h: Update.
18281 2017-06-29 Carl Love <cel@us.ibm.com>
18283 * config/rs6000/rs6000-c.c: Add support for built-in functions
18284 vector signed int vec_signed (vector float);
18285 vector signed long long vec_signed (vector double);
18286 vector signed int vec_signed2 (vector double, vector double);
18287 vector signed int vec_signede (vector double);
18288 vector signed int vec_signedo (vector double);
18289 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
18290 instruction generator.
18291 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18292 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
18293 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
18295 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
18296 vunsignede_v2df): Add define_expands.
18297 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
18298 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
18299 VEC_UNSIGNEDO): Add definitions.
18300 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18301 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
18302 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
18303 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
18304 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
18305 * config/rs6000/altivec.h (vec_signed, vec_signed2,
18306 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
18307 vec_unsignede, vec_unsignedo): Add builtin defines.
18308 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
18310 * doc/extend.texi: Update the built-in documentation file for the
18311 new built-in functions.
18313 2017-06-29 Richard Biener <rguenther@suse.de>
18315 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
18316 reduction chains to LOOP_VINFO_REDUCTIONS.
18317 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
18318 SLP reductions after processing reduction chains.
18320 2017-06-29 Nathan Sidwell <nathan@acm.org>
18322 * builtins.c (fold_builtin_FUNCTION): Use
18323 lang_hooks.decl_printable_name.
18325 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
18327 PR middle-end/81194
18328 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
18329 with only one label.
18330 * stmt.c (expand_case): Assert NCASES is greater than one.
18332 2017-06-29 Richard Biener <rguenther@suse.de>
18334 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
18336 (group_case_labels): Likewise.
18337 (find_taken_edge): Push sanity checking on val to workers...
18338 (find_taken_edge_cond_expr): ... here
18339 (find_taken_edge_switch_expr): ... and here, handle cases
18340 with just a default label.
18341 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
18342 (group_case_labels): Likewise.
18343 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
18344 group_case_labels does anything cleanup the CFG again.
18346 2017-06-29 Bin Cheng <bin.cheng@arm.com>
18348 PR tree-optimization/81196
18349 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
18350 exit condition comparing two IVs.
18352 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
18354 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
18355 profile to the dummy entry at the end of the list of architectures.
18356 * config/arm/arm-cpu-cdata.h: Regenerated.
18358 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18359 Michael Collison <michael.collison@arm.com>
18362 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
18364 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
18365 (*aarch64_reg_<mode>3_minus_mask): New pattern.
18366 (*aarch64_<optab>_reg_di3_mask2): New pattern.
18367 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
18368 of shift when the shift amount is masked with constant equal to
18369 the size of the mode.
18370 * config/aarch64/predicates.md (subreg_lowpart_operator): New
18373 2017-06-29 Martin Liska <mliska@suse.cz>
18375 * config/i386/i386.opt: Change range from [1,5] to [0,5].
18377 2017-06-29 Yury Gribov <tetra2005@gmail.com>
18380 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
18383 (edge_growth_cache_entry::edge_growth_cache_entry): New
18385 (reset_edge_growth_cache): Update to use constructor.
18387 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18389 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
18390 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
18391 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
18393 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
18395 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
18396 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
18398 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
18400 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
18401 (*-linux-uclibc*): Add t-uclibc tmake_file.
18402 * config/t-musl: New.
18403 * config/t-uclibc: New.
18405 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
18407 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
18409 (gen_comm_data): Emit architectural setting of arch_prof.
18410 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
18412 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
18413 (armv8-m.base, armv8-m.main): Likewise.
18414 * arm-protos.h (arm_build_target): Add profile field.
18415 (arch_option): Likewise.
18416 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
18418 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
18419 arm_active_target.profile.
18421 2017-06-28 Richard Biener <rguenther@suse.de>
18423 PR middle-end/81227
18424 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
18425 TYPE_OVERFLOW_WRAPS.
18426 * match.pd (negate_expr_p): Likewise.
18427 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
18428 fold_build2, not fold_binary.
18430 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
18432 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
18433 Convert memory address to Pmode.
18434 (aarch64_print_operand): Assert MEM operands are always Pmode.
18436 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
18439 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
18440 Remove redundant if.
18441 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
18442 * config/arm/aarch-common-protos.h
18443 (aarch_forward_to_shift_is_not_shifted_re): Remove.
18444 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
18446 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
18449 * multiple_target.c (create_dispatcher_calls): Set the default
18450 clone to be static, not public.
18452 2017-06-28 Richard Biener <rguenther@suse.de>
18454 * tree-vect-loop.c (vectorizable_reduction): Move special
18455 cond reduction IV var creation ...
18456 (vect_create_epilog_for_reduction): ... here. Remove induction_index
18457 parameter. Use STMT_VINFO_VECTYPE.
18458 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
18461 2017-06-28 Martin Liska <mliska@suse.cz>
18464 * ipa-visibility.c (non_local_p): Handle visibility.
18466 2017-06-28 Martin Liska <mliska@suse.cz>
18469 * common.opt: Add IntegerRange to various options.
18470 * opt-functions.awk (integer_range_info): New function.
18471 * optc-gen.awk: Add integer_range_info to cl_options struct.
18472 * opts-common.c (decode_cmdline_option): Handle
18473 CL_ERR_INT_RANGE_ARG.
18474 (cmdline_handle_error): Likewise.
18475 * opts.c (print_filtered_help): Show valid interval in
18476 when --help is provided.
18477 * opts.h (struct cl_option): Add range_min and range_max fields.
18478 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
18480 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
18482 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
18483 (x * C EQ/NE y * C): New transformation.
18485 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
18487 * genmultilib (combination_space): Accept '+' in option names.
18489 2017-06-28 Martin Liska <mliska@suse.cz>
18492 * asan.c (instrument_derefs): Bail out inner references
18493 that are hard register variables.
18495 2017-06-28 Jakub Jelinek <jakub@redhat.com>
18498 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
18499 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
18501 2017-06-28 Richard Biener <rguenther@suse.de>
18503 * tree-vectorizer.h (vect_get_vec_defs): Remove.
18504 (vect_get_slp_defs): Adjust.
18505 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
18507 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
18509 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
18510 get_initial_defs_for_reduction instead of vect_get_vec_defs.
18511 (vectorizable_reduction): Adjust.
18512 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
18514 (vect_get_slp_defs): Likewise.
18515 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
18516 (vectorizable_bswap): Adjust.
18517 (vectorizable_call): Likewise.
18518 (vectorizable_conversion): Likewise.
18519 (vectorizable_assignment): Likewise.
18520 (vectorizable_shift): Likewise.
18521 (vectorizable_operation): Likewise.
18522 (vectorizable_store): Likewise.
18523 (vectorizable_condition): Likewise.
18524 (vectorizable_comparison): Likewise.
18526 2017-06-28 Michael Collison <michael.collison@arm.com>
18529 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
18531 (arm_gen_movmemqi): Removed unused variable 'i'.
18532 Convert 'for' loop into 'while' loop.
18533 (arm_expand_prologue): Remove last unnecessary set of insn.
18534 (thumb_pop): Remove unused variable 'pushed_words'.
18535 (thumb_exit): Remove last unnecessary set of regs_to_pop.
18537 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18539 * config/s390/predicates.md: Use s390_rel_address_ok_p.
18540 * config/s390/s390-protos.h: Add prototype of
18541 s390_rel_address_ok_p.
18542 * config/s390/s390.c (s390_got_symbol): New function.
18543 (s390_rel_address_ok_p): New function.
18544 (legitimize_pic_address): Use s390_rel_address_ok_p.
18545 (s390_load_got): Use s390_got_symbol.
18546 (s390_option_override): Issue error if
18547 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
18548 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
18550 * config/s390/s390.opt: New option mpic-data-is-text-relative.
18552 2017-06-27 Andrew Pinski <apinski@cavium.com>
18554 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
18555 (X * copysign (1.0, X)): New pattern.
18556 (X * copysign (1.0, -X)): New pattern.
18557 (copysign (-1.0, CST)): New pattern.
18559 2017-06-27 Joseph Myers <joseph@codesourcery.com>
18561 * genmultilib (combination_space): Remove variable.
18562 Validate reuse rules against regular expression for any sequence
18563 of multilib options in any order.
18565 2017-06-27 Michael Collison <michael.collison@arm.com>
18567 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
18568 call aarch64_split_simd_combine.
18569 * (aarch64_combine_internal<mode>): Delete pattern.
18570 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
18571 Allow register and subreg operands.
18573 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18575 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
18576 specific need, just fallback on defaults.
18577 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
18579 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18580 Olivier Hainque <hainque@adacore.com>
18582 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
18584 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
18585 targets. Pick a default if no particular attempt applied.
18586 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
18589 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18591 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
18592 (x86_64-wrs-vxworks7): Likewise.
18594 2017-06-27 Marek Polacek <polacek@redhat.com>
18597 * ubsan.c (instrument_null): Check get_base_address's result for null.
18599 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18601 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
18603 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18605 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
18606 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
18607 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
18608 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
18609 New function types.
18610 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
18611 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
18612 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
18613 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
18614 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
18615 BUILT_IN_FEUPDATEENV): New builtins.
18616 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
18617 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
18618 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
18619 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
18621 (builtin_structptr_types): Adjust size.
18622 * tree.c (builtin_structptr_types): Add four entries.
18624 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18625 Olivier Hainque <hainque@adacore.com>
18627 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
18628 (TLS_SYM): New local macro, forcing reference to __tls__ on
18629 link command lines for VxWorks 7 RTPs, triggering initialization
18631 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
18632 OS features TLS support, true for RTPs on VxWorks 7.
18633 * config/vxworks.c (vxworks_override_options): Setup emutls
18636 2017-06-27 Jakub Jelinek <jakub@redhat.com>
18638 * predict.c (test_prediction_value_range): Use -1U instead of -1
18639 to avoid narrowing conversion warning.
18640 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
18641 to avoid narrowing conversion warning.
18642 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
18644 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
18646 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18648 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
18649 64bit configurations.
18650 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
18651 (SIZE_TYPE): Likewise.
18652 * config/vxworks.c (vxworks_emutls_var_fields): Use
18653 long_unsigned_type_node instead of unsigned_type_node as the offset
18654 field type, which is "pointer" mode in emutls.c.
18656 2017-06-27 Jakub Jelinek <jakub@redhat.com>
18659 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
18661 PR middle-end/81207
18662 * gimple-fold.c (replace_call_with_call_and_fold): Handle
18663 gimple_vuse copying separately from gimple_vdef copying.
18665 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18667 * value-prof.c (free_hist): Remove call to memset and the enclosing if
18670 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
18671 Olivier Hainque <hainque@adacore.com>
18673 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
18674 for all vxworks7 targets.
18675 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
18676 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
18677 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
18678 variations for VX6/VX7 and 32/64bits later on in ...
18679 (VXWORKS_LIB_SPEC): Leverage new macros.
18680 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
18681 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
18683 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
18685 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
18686 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
18688 2017-06-26 Carl Love <cel@us.ibm.com>
18690 * config/rs6000/rs6000-c.c: Add support for built-in functions
18691 vector bool char vec_reve (vector bool char);
18692 vector signed char vec_reve (vector signed char);
18693 vector unsigned char vec_reve (vector unsigned char);
18694 vector bool int vec_reve (vector bool int);
18695 vector signed int vec_reve (vector signed int);
18696 vector unsigned int vec_reve (vector unsigned int);
18697 vector bool long long vec_reve (vector bool long long);
18698 vector signed long long vec_reve (vector signed long long);
18699 vector unsigned long long vec_reve (vector unsigned long long);
18700 vector bool short vec_reve (vector bool short);
18701 vector signed short vec_reve (vector signed short);
18702 vector double vec_reve (vector double);
18703 vector float vec_reve (vector float);
18704 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
18705 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
18706 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
18707 (altivec_vreve): New pattern.
18708 * config/rs6000/altivec.h (vec_reve): New define.
18709 * doc/extend.texi (vec_rev): Update the built-in documentation file
18710 for the new built-in functions.
18712 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18714 PR tree-optimization/71815
18715 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
18717 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
18719 (slsr_process_phi): Likewise.
18720 (replace_uncond_cands_and_profitable_phis): Don't replace a
18721 multiply candidate with a stride of 1 (copy or cast).
18722 (phi_incr_cost): Call uses_consumed_by_stmt rather than
18724 (lowest_cost_path): Likewise.
18725 (total_savings): Likewise.
18727 2017-06-26 Richard Biener <rguenther@suse.de>
18730 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
18731 Use def_builtin_pure for all gather builtins.
18733 2017-06-26 Richard Biener <rguenther@suse.de>
18735 PR tree-optimization/81203
18736 * tree-tailcall.c (find_tail_calls): Do not move stmts into
18737 non-dominating BBs.
18739 2017-06-26 Marek Polacek <polacek@redhat.com>
18742 * doc/invoke.texi: Document -Wmultistatement-macros.
18744 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
18746 * doc/sourcebuild.texi (ARM-specific attributes): Document new
18747 arm_neon_ok_no_float_abi effective target.
18749 2017-06-26 Richard Biener <rguenther@suse.de>
18751 PR tree-optimization/80928
18752 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
18753 (copy_bbs): Set BB_DUPLICATED flag early.
18754 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
18756 (execute_on_shrinking_pred): Likewise.
18757 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
18758 BB_DUPLICATED blocks.
18759 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
18760 iterate over all PHIs considering removal of *gsi.
18762 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
18764 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
18767 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18769 * config/rs6000/rs6000-string.c: (expand_block_clear,
18770 do_load_for_compare, select_block_compare_mode,
18771 compute_current_alignment, expand_block_compare,
18772 expand_strncmp_align_check, expand_strn_compare,
18773 expand_block_move, rs6000_output_load_multiple)
18774 Move functions related to string/block move/compare
18775 to a separate file.
18776 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
18777 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
18778 for this function which is now used in two files.
18779 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
18780 * config.gcc: Add rs6000-string.o to extra_objs for
18781 targets powerpc*-*-* and rs6000*-*-*.
18783 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
18786 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
18787 32-bit, since indexed is not valid for DImode.
18788 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
18789 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
18790 (define_peephole2 for Altivec d-form load): Add 32-bit support.
18791 (define_peephole2 for Altivec d-form store): Likewise.
18794 * multiple_target.c (create_dispatcher_calls): Only create the
18795 dispatcher call if the function is the default clone of a
18796 versioned function.
18798 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
18800 PR middle-end/80902
18801 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
18802 a call, force the call to not be a tail call.
18804 2017-06-23 Jeff Law <law@redhat.com>
18806 * doc/contrib.texi: Add entry for Steven Pemberton's work on
18809 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
18811 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
18812 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
18813 handling for early expansion of vector shifts (sl,sr,sra,rl).
18814 (builtin_function_type): Add vector shift right instructions
18815 to the unsigned argument list.
18817 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
18819 rtl-optimizatoin/79286
18820 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
18821 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
18822 trap. PIC register plus a const unspec without offset can never trap.
18824 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
18826 * tree.h (builtin_structptr_type): New type.
18827 (builtin_structptr_types): Declare new array.
18828 * tree.c (builtin_structptr_types): New array.
18829 (free_lang_data, build_common_tree_nodes): Use it.
18831 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
18834 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
18837 2017-06-22 Matt Turner <mattst88@gmail.com>
18839 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
18840 Lake models to skylake case. Assume skylake for unknown
18841 models with clflushopt.
18843 2017-06-22 Jeff Law <law@redhat.com>
18845 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
18846 frame sizes that do not satisfy aarch64_uimm12_shift.
18848 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
18850 * profile-count.h (apply_probability,
18851 apply_scale, probability_in): Fix checks for zero.
18853 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18855 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
18856 * doc/cppdiropts.texi (-I @var{dir}): Document it.
18858 2016-06-22 Richard Biener <rguenther@suse.de>
18860 * tree-vect-loop.c (vect_model_reduction_cost): Handle
18861 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
18862 REDUC_MAX_EXPR support.
18863 (vectorizable_reduction): Likewise.
18864 (vect_create_epilog_for_reduction): Likewise.
18866 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
18868 * match.pd (A / (1 << B) -> A >> B): New.
18869 * generic-match-head.c: Include optabs-tree.h.
18870 * gimple-match-head.c: Likewise.
18871 * optabs-tree.h (target_supports_op_p): New.
18872 * optabs-tree.c (target_supports_op_p): New.
18874 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18876 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
18877 $gcc_cv_ld --help output.
18878 (gcc_cv_ld_demangle): Likewise.
18879 (gcc_cv_ld_eh_frame_hdr): Likewise.
18880 (gcc_cv_ld_pie): Likewise.
18881 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
18882 (gcc_cv_ld_buildid): Likewise.
18883 (gcc_cv_ld_sysroot): Likewise.
18884 (ld_bndplt_support): Likewise.
18885 (ld_pushpopstate_support): Likewise.
18886 * configure: Regenerate.
18887 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
18889 2017-06-21 Jakub Jelinek <jakub@redhat.com>
18892 * config/i386/sse.md (round<mode>2): Renumber match_dup and
18893 operands indexes to avoid gap between operands and match_dups.
18895 2017-06-21 Andrew Pinski <apinski@cavium.com>
18897 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
18898 Increment Arith_shift and Arith_shift_reg by 1.
18899 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
18901 * config/aarch64/aarch64.c (thunderx_tunings): Enable
18902 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
18903 (aarch64_strip_extend): Add new argument and test for it.
18904 (aarch64_cheap_mult_shift_p): New function.
18905 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
18906 add a cost if it is true.
18907 Update calls to aarch64_strip_extend.
18908 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
18910 2017-06-21 Andrew Pinski <apinski@cavium.com>
18912 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
18914 (thunderxt88): Likewise.
18915 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
18916 (thunderx_prefetch_tune): New variable.
18917 (thunderx2t99_prefetch_tune): Update for the correct values.
18918 (thunderxt88_tunings): New variable.
18919 (thunderx_tunings): Use thunderx_prefetch_tune instead of
18920 generic_prefetch_tune.
18921 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
18923 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18925 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
18926 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
18927 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
18928 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
18929 (aarch64_atomic_cas<mode>, GPI): Likewise.
18931 2017-06-21 Martin Liska <mliska@suse.cz>
18933 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
18934 statements on cold and hot labels.
18935 * predict.c (tree_estimate_probability_bb): Remove the
18936 prediction from this place.
18938 2017-06-21 Martin Liska <mliska@suse.cz>
18940 PR tree-optimization/79489
18941 * gimplify.c (maybe_add_early_return_predict_stmt): New
18943 (gimplify_return_expr): Call the function.
18944 * predict.c (tree_estimate_probability_bb): Remove handling
18946 * predict.def: Update comment about early return predictor.
18947 * gimple-predict.h (is_gimple_predict): New function.
18948 * predict.def: Change default value of early return to 66.
18949 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
18951 * passes.def: Put pass_strip_predict_hints to the beginning of
18954 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
18956 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
18957 FUNCTION_DECL declarations.
18958 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
18960 (dwaf2out_decl): Likewise.
18961 * godump.c (go_early_global_decl): Skip call to the real debug hook
18962 for FUNCTION_DECL declarations.
18963 * passes.c (rest_of_decl_compilation): Skip call to the
18964 early_global_decl debug hook for FUNCTION_DECL declarations, unless
18965 -fdump-go-spec is passed.
18967 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
18969 * config/i386/i386.c (struct builtin_isa): New field pure_p.
18970 Reorder for compactness.
18971 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
18972 (def_builtin_pure, def_builtin_pure2): New functions.
18973 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
18975 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
18977 * match.pd (nop_convert): New predicate.
18978 ((A +- CST1) +- CST2): Allow some NOP conversions.
18980 2017-06-21 Jakub Jelinek <jakub@redhat.com>
18983 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
18984 with ctors/dtors if GOVD_SHARED is set.
18986 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
18988 * config/aarch64/aarch64.md (movti_aarch64):
18989 Emit mov rather than orr.
18990 (movtf_aarch64): Likewise.
18991 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
18992 Emit mov rather than orr.
18994 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
18996 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
18997 Swap alternatives, make integer dup more expensive.
18999 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19001 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
19002 Return true for non-tls symbols.
19004 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
19006 * config/aarch64/aarch64-cores.def (cortex-a55): New.
19007 (cortex-a75): Likewise.
19008 (cortex-a75.cortex-a55): Likewise.
19009 * config/aarch64/aarch64-tune.md: Regenerate.
19010 * doc/invoke.texi (-mtune): Document new values for -mtune.
19012 2017-06-21 Tom de Vries <tom@codesourcery.com>
19014 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
19015 stack_size feature.
19016 (Effective-Target Keywords, Other attributes): Suggest using
19017 dg-add-options stack_size feature to get stack limit in stack_size
19018 effective target documentation.
19020 2017-06-21 Julian Brown <julian@codesourcery.com>
19021 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19023 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
19024 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
19025 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
19027 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
19028 attribute type list for neon_multiply.
19029 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
19030 attribute type list for neon_multiply.
19031 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
19032 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
19033 attribute type list for neon_multiply.
19034 * config/arm/types.md (crypto_pmull): Add.
19035 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
19036 attribute type list.
19038 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
19040 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
19043 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19045 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
19046 to make sure not to dereference a NULL cost_classes_ptr pointer.
19048 2017-06-20 Carl Love <cel@us.ibm.com>
19050 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19051 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19052 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
19053 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
19054 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
19055 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
19056 VMULOSW): New enum "unspec" values.
19057 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
19058 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
19059 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
19060 altivec_vmulosw): New patterns.
19061 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
19062 VMULOSW): Add definitions.
19064 2017-06-20 Julia Koval <julia.koval@intel.com>
19066 * config/i386/i386.c: Fix rounding expand for new pattern.
19067 * config/i386/subst.md: Fix pattern (parallel -> unspec).
19069 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19071 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
19072 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
19074 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19076 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
19079 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19081 * config/aarch64/aarch64-cores.def: Rearrange to sort by
19082 architecture, then by implementer ID.
19083 * config/aarch64/aarch64-tune.md: Regenerate.
19085 2017-06-20 Richard Biener <rguenther@suse.de>
19087 PR middle-end/81097
19088 * fold-const.c (split_tree): Fold to type before negating.
19090 2017-06-20 David Malcolm <dmalcolm@redhat.com>
19092 * diagnostic-show-locus.c
19093 (selftest::test_fixit_deletion_affecting_newline): New function.
19094 (selftest::diagnostic_show_locus_c_tests): Call it.
19096 2017-06-20 Andreas Schwab <schwab@suse.de>
19099 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
19102 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19104 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
19105 __ARM_FEATURE_COPROC according to support.
19107 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19109 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
19110 Rewritten to avoid overflow for > 32-bit pointers.
19113 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
19114 by removing enum keyword.
19115 (ubsan_type_descriptor): Likewise. Formatting fix.
19118 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
19119 splitter): Require TARGET_SSE2 in the condition.
19121 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
19124 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
19125 for doing vector set of SFmode on ISA 3.0.
19126 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
19127 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
19129 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
19130 SFmode value into a V4SF variable that was extracted from another
19131 V4SF variable without converting the element to double precision
19132 and back to single precision vector format.
19133 (vsx_insert_extract_v4sf_p9_2): Likewise.
19135 2017-06-19 Jakub Jelinek <jakub@redhat.com>
19137 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
19138 in UWHI to avoid undefined overflow.
19141 * ubsan.h (enum ubsan_encode_value_phase): New.
19142 (ubsan_encode_value): Change second argument to
19143 enum ubsan_encode_value_phase with default value of
19144 UBSAN_ENCODE_VALUE_GENERIC.
19145 * ubsan.c (ubsan_encode_value): Change second argument to
19146 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
19147 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
19148 create_tmp_var_raw instead of create_tmp_var and use a
19150 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
19151 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
19152 ubsan_encode_value callers.
19155 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
19156 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
19157 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
19159 2017-06-19 Richard Biener <rguenther@suse.de>
19161 PR middle-end/81118
19162 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
19163 estimates if we changed anything.
19165 2017-06-19 Richard Biener <rguenther@suse.de>
19167 PR tree-optimization/80887
19168 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
19169 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
19170 simplified lookups, then reset mprts_hook.
19171 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
19173 (try_to_simplify): Likewise.
19175 2017-06-19 Martin Liska <mliska@suse.cz>
19178 * gimplify.c (gimplify_switch_expr):
19179 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
19181 2017-06-19 Martin Liska <mliska@suse.cz>
19183 * doc/install.texi: Document that PGO runs in 4 stages.
19185 2017-06-19 Martin Liska <mliska@suse.cz>
19188 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
19189 to dispatcher function name.
19190 * multiple_target.c (replace_function_decl): New function.
19191 (create_dispatcher_calls): Redirect both edges and references.
19193 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
19195 * profile-count.c (profile_count::dump): Dump quality.
19196 (profile_count::differs_from_p): Update for unsigned val.
19197 * profile-count.h (profile_count_quality): New enum.
19198 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
19200 2017-06-19 Richard Biener <rguenther@suse.de>
19202 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
19203 struct function as arg.
19204 (estimate_numbers_of_iterations): Export overload with loop arg.
19205 (free_numbers_of_iterations_estimates_loop): Use an overload of
19206 free_numbers_of_iterations_estimates instead.
19207 * tree-cfg.c (remove_bb): Adjust.
19208 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
19209 * tree-parloops.c (gen_parallel_loop): Likewise.
19210 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
19212 (tree_unroll_loops_completely): Likewise.
19213 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
19214 Use an overload instead and export.
19215 (estimated_loop_iterations): Adjust.
19216 (max_loop_iterations): Likewise.
19217 (likely_max_loop_iterations): Likewise.
19218 (estimate_numbers_of_iterations): Take struct function as arg
19220 (loop_exits_before_overflow): Adjust.
19221 (free_numbers_of_iterations_estimates_loop): Use an overload.
19222 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
19223 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
19225 2017-06-19 Richard Biener <rguenther@suse.de>
19228 * ipa-prop.c (find_constructor_constant_at_offset): Handle
19229 RANGE_EXPR conservatively.
19231 2017-06-16 Carl Love <cel@us.ibm.com>
19233 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19234 definitions for vec_float, vec_float2, vec_floato,
19235 vec_floate built-ins.
19236 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
19237 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
19239 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
19240 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
19241 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
19242 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
19243 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
19244 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
19245 vec_floato): Add builtin defines.
19246 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
19247 Update the built-in documentation file for the new built-in
19250 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19252 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
19253 (mthumb): Mark as the negative of -marm.
19255 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19257 * doc/invoke.texi (ARM Options, -mcpu): Document supported
19259 (ARM Options, -mtune): Document that this accepts the same
19260 extension options as -mcpu.
19261 (ARM Options, -mfpu): Document addition of -mfpu=auto.
19263 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19265 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
19266 permitted extensions.
19268 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19270 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
19271 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
19272 (armv8-m.main): Add option +nodsp.
19273 * config/arm/arm-cpu-cdata.h: Regenerated.
19275 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19277 * config/arm/t-fuchsia: New file.
19278 * config.gcc (arm*-*-fuchsia*): Use it.
19280 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19282 * config/arm/t-symbian: Rewrite for new option infrastructure.
19284 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19286 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
19287 (MULTILIB_REQUIRED): Likewise.
19289 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19291 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
19292 (MULTILIB_RESUE): Likewise.
19293 (MULTILIB_MATCHES): Likewise.
19294 (MULTLIB_REQUIRED): Likewise.
19296 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19298 * config/arm/t-rtems: Rewrite for new option framework.
19300 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19302 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
19303 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
19304 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
19305 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
19306 * config/arm/t-multilib: ... here.
19307 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
19308 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
19309 armv7-a and armv8*-a when A-profile libraries have not been built.
19310 * config/arm/t-rmprofile: Rewrite.
19312 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19314 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
19315 with a backslash. Remove the backslash after substituting unescaped
19317 * doc/fragments.texi (MULTILIB_REUSE): Document it.
19319 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19321 * config.gcc: (arm*-*-*): When building a-profile libraries, force
19322 the driver to pass through the default setting of -mfloat-abi.
19323 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
19325 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
19326 (all_feat_combs): New rule.
19327 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
19329 * config/arm/t-aprofile: Rewrite.
19331 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19333 * config/arm/arm.h (FPUTYPE_AUTO): Define.
19334 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
19335 fpu is not specified by the user/command-line.
19336 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
19337 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
19338 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
19339 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
19340 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
19341 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
19343 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19345 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
19346 * config/arm/t-arm-elf: Rewritten.
19348 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19350 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
19351 have some floating-point instructions.
19352 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
19353 (TARGET_MAYBE_HARD_FLOAT): New macro.
19354 * config/arm/arm-builtins.c (arm_init_builtins): Use
19355 TARGET_MAYBE_HARD_FLOAT.
19356 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
19358 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19360 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
19361 (configargs.h): Include it.
19362 (arm_print_hint_for_fpu_option): New function.
19363 (arm_parse_fpu_option): New function.
19364 (candidate_extension): New class.
19365 (arm_canon_for_multilib): New function.
19366 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
19367 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
19368 (ARCH_CANONICAL_SPECS): New macro.
19369 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
19371 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19373 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
19374 are set after handling multilib fragments. Set target_cpu_default2
19377 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19379 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
19381 (arm*-*-*): Set target_cpu_default2 to a quoted string.
19382 * config/arm/parsecpu.awk (check_cpu): Validate any extension
19384 (check_arch): Likewise.
19385 * config/arm/arm.c (arm_configure_build_target): Handle
19386 TARGET_CPU_DEFAULT being a string constant. Scan any feature
19387 options in the default.
19389 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19391 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
19392 when an option is an alias of another.
19393 * config/arm/parsecpu.awk (optalias): New parser token.
19394 (gen_comm_data): Mark non-alias options as such. Emit entries
19395 for extension aliases.
19396 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
19397 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
19398 (armv6kz, armv6zk, armv6t2): Likewise.
19399 (armv7): Make vfpv3-d16 an alias.
19400 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
19402 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
19403 Sort in canonical order.
19404 (armv8-a): Sort in canonical order.
19405 (armv8.1-a, armv8.2-a): Likewise.
19406 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
19408 (cortex-a9): Sort in canonical order.
19409 * config/arm/arm.c (selftests.h): Include it.
19410 (arm_test_cpu_arch_data): New function.
19411 (arm_run_self_tests): New function.
19412 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
19413 (targetm): Move declaration to the end of the file.
19414 * arm-cpu-cdata.h: Regenerated.
19416 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19418 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
19419 call to target_mode_check describing the type of option passed.
19420 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
19421 (arm_target_thumb_only): Use arm_parse_arch_option_name or
19422 arm_parse_cpu_option_name to match parameters against list of
19424 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
19425 arm_arch_core_flags data structure.
19426 * config/arm/arm-cpu_cdata.h: Regenerated.
19428 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19430 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
19432 (arm_print_hint_for_cpu_option): Likewise.
19433 (arm_print_hint_for_arch_option): Likewise.
19434 (arm_parse_cpu_option_name): Likewise.
19435 (arm_parse_arch_option_name): Likewise.
19436 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
19437 of entries in the all_fpus list.
19438 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
19439 (arm_parse_cpu_option_name): Declare.
19440 (arm_parse_arch_option_name): Declare.
19441 (arm_parse_option_features): Declare.
19442 (arm_intialize_isa): Declare.
19443 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
19445 (gen_comm_data): ... here. Make definitions non-static.
19446 * config/arm/arm-cpu-data.h: Regenerated.
19447 * config/arm/arm-cpu-cdata.h: Regenerated.
19449 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19451 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
19452 (cpu_arch_extension): New structure.
19453 (cpu_arch_option, arch_option, cpu_option): New structures.
19454 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
19455 architecture types.
19456 (gen_data): Generate new format data tables.
19457 * config/arm/arm.c (cpu_tune): New structure.
19458 (cpu_option, processors): Delete.
19459 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
19460 (arm_print_hint_for_cpu_option): ... this and ...
19461 (arm_print_hint_for_arch_option): ... this.
19462 (arm_parse_arch_cpu_name): Delete. Replace with ...
19463 (arm_parse_cpu_option_name): ... this and ...
19464 (arm_parse_arch_option_name): ... this.
19465 (arm_unrecognized_feature): Change type of target parameter to
19467 (arm_parse_arch_cpu_features): Delete. Replace with ...
19468 (arm_parse_option_features): ... this.
19469 (arm_configure_build_target): Rework to use new configuration data
19471 (arm_print_tune_info): Rework for new configuration data tables.
19472 * config/arm/arm-cpu-data.h: Regenerated.
19473 * config/arm/arm-cpu.h: Regenerated.
19475 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19477 * Makefile.in (OBJS): Move sbitmap.o from here ...
19478 (OBJS-libcommon): ... to here.
19480 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19482 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
19483 (ISA_ALL_CRYPTO): New macro.
19484 (ISA_ALL_SIMD): New macro
19485 (ISA_ALL_FP): New macro.
19486 * config/arm/arm.c (fpu_bitlist): Update initializer.
19487 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
19489 (arm9e): Add fpu. Add option for nofp
19490 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
19491 (arm926ej-s, arm1026ej-s): Likewise.
19492 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
19493 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
19494 neon-fp16, neon-vfpv4, nofp and nosimd.
19495 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
19496 (cortex-a8): Add fpu. Add option for nofp.
19497 (cortex-a9): Add fpu. Add options for nosimd and nofp.
19498 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
19499 (cortex-r4f): Add fpu.
19500 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
19501 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
19503 (cortex-r8): Likewise.
19504 (cortex-m4): Add fpu. Add option for nofp.
19505 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
19506 (cortex-a17.cortex-a7): Likewise.
19507 (cortex-a32): Add fpu. Add options for crypto and nofp.
19508 (cortex-a35, cortex-a53): Likewise.
19509 (cortex-a57): Add fpu. Add option for crypto.
19510 (cortex-a72, cortex-a73): Likewise.
19511 (exynos-m1): Likewise.
19512 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
19513 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
19514 (cortex-m33): Add fpu. Add option for nofp.
19515 * config/arm/arm-cpu-cdata.h: Regenerated
19516 * config/arm/arm-cpu-data.h: Regenerated.
19518 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19520 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
19521 (armv5te, armv5tej): Likewise.
19522 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
19523 (armv7): Add options fp and vfpv3-d16.
19524 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
19525 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
19527 (armv7ve): Likewise.
19528 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
19529 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
19530 (armv8-a): Add nocrypto option.
19531 (armv8.1-a, armv8.2-a): Likewise.
19532 (armv8-m.main): add options fp, fp.dp and nofp.
19534 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19536 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
19538 (armv8-a+crc): Delete.
19539 (armv8.1-a): Add options simd, crypto and nofp.
19540 (armv8.2-a): Add options fp16, simd, crypto and nofp.
19541 (armv8.2-a+fp16): Delete.
19542 (armv8-m.main): Add option dsp.
19543 (armv8-m.main+dsp): Delete.
19544 (cortex-a8): Add fpu. Add nofp option.
19545 (cortex-a9): Add fpu. Add nofp and nosimd options.
19546 * config/arm/parsecpu.awk (gen_data): Generate option tables and
19547 link to main cpu and architecture data structures.
19548 (gen_comm_data): Only put isa attributes from the main architecture
19550 (option): New statement for architecture and CPU entries.
19551 * arm.c (struct cpu_option): New structure.
19552 (struct processors): Add entry for options.
19553 (arm_unrecognized_feature): New function.
19554 (arm_parse_arch_cpu_name): Ignore any characters after the first
19556 (arm_parse_arch_cpu_feature): New function.
19557 (arm_configure_build_target): Separate out any CPU and architecture
19558 features and parse separately. Don't error out if -mfpu=auto is
19559 used with only an architecture string.
19560 (arm_print_asm_arch_directives): New function.
19561 (arm_file_start): Call it.
19562 * config/arm/arm-cpu-cdata.h: Regenerated.
19563 * config/arm/arm-cpu-data.h: Likewise.
19564 * config/arm/arm-tables.opt: Likewise.
19566 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19568 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
19569 assembler when it is not -mfpu=auto.
19571 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19573 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
19574 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
19575 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
19576 (ASM_CPU_SPEC): Rewrite.
19577 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
19578 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
19579 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
19580 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
19581 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
19582 copied string is NUL-terminated. Also strip any characters prefixed
19584 (arm_rewrite_selected_arch): New function.
19585 (arm_rewrite_march): New function.
19587 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19589 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
19590 (x_arm_cpu_string, x_arm_tune_string): Likewise.
19591 (march, mcpu, mtune): Convert to string-based options.
19592 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
19593 (arm_parse_arch_cpu_name): New function.
19594 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
19595 identify selected architecture or CPU.
19596 (arm_option_save): New function.
19597 (TARGET_OPTION_SAVE): Redefine.
19598 (arm_option_restore): Restore string options.
19599 (arm_option_print): Print string options.
19601 2017-06-16 Martin Sebor <msebor@redhat.com>
19603 PR tree-optimization/80933
19604 PR tree-optimization/80934
19605 * builtins.c (fold_builtin_3): Do not handle bcmp here.
19606 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
19607 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
19608 (gimple_fold_builtin): Call them.
19610 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19612 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
19613 as unlikely; update profile.
19615 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19617 * predict.c (force_edge_cold): Handle declaring edges impossible
19620 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19622 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
19624 (try_unroll_loop_completely): Fix reporting.
19626 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19628 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
19630 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
19633 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
19634 if given a non-constant argument for an intrinsic which requires a
19637 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19639 * profile.c (compare_freqs): New function.
19640 (branch_prob): Sort edge list.
19641 (find_spanning_tree): Assume that the list is priority sorted.
19643 2017-06-16 Richard Biener <rguenther@suse.de>
19645 PR tree-optimization/81090
19646 * passes.def (pass_record_bounds): Remove.
19647 * tree-pass.h (make_pass_record_bounds): Likewise.
19648 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
19649 make_pass_record_bounds): Likewise.
19650 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
19651 not free niter estimates at the beginning but at the end.
19652 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
19654 2017-06-16 Richard Biener <rguenther@suse.de>
19656 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
19657 initializer to workaround ICE in host GCC 4.8.
19659 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19661 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
19663 (clone_inlined_nodes): Update.
19665 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19667 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
19668 prefetch settings, and enable prefetching by default at -O3.
19670 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19672 * config/aarch64/aarch64.c (aarch64_override_options_internal):
19673 Set flag_prefetch_loop_arrays according to tuning data.
19675 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19677 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
19678 New tune structure.
19679 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
19680 [Unrelated to main purpose of the patch] Place the pointer field last
19681 to enable type checking errors when tune structure are wrongly merged.
19682 * config/aarch64/aarch64.c (generic_prefetch_tune,)
19683 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
19684 (thunderx2t99_prefetch_tune): New tune constants.
19685 (tune_params *_tunings): Update all tunings (no functional change).
19686 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
19687 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
19688 from tunings structures.
19690 2017-06-16 Jakub Jelinek <jakub@redhat.com>
19693 * ubsan.c (instrument_null): Add T argument, use it instead
19694 of computing it based on IS_LHS.
19695 (instrument_object_size): Likewise.
19696 (pass_ubsan::execute): Adjust instrument_null and
19697 instrument_object_size callers to pass gimple_get_lhs or
19698 gimple_assign_rhs1 result to it. Use instrument_null instead of
19699 calling get_base_address and instrument_mem_ref. Handle
19700 aggregate call arguments for object-size sanitization.
19702 2017-06-16 Yury Gribov <tetra2005@gmail.com>
19704 PR tree-optimization/81089
19705 * tree-vrp.c (is_masked_range_test): Validate operands of
19708 2017-06-15 Martin Sebor <msebor@redhat.com>
19711 * dumpfile.c (dump_register): Avoid calling memset to initialize
19712 a class with a default ctor.
19713 * gcc.c (struct compiler): Remove const qualification.
19714 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
19715 * hash-table.h: Ditto.
19716 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
19718 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
19719 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
19721 * params.h (struct param_info): Make struct members non-const.
19722 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
19723 with default initialization.
19724 * vec.h (vec_copy_construct, vec_default_construct): New helper
19726 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
19727 with vec_copy_construct.
19728 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
19729 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
19730 * doc/invoke.texi (-Wclass-memaccess): Document.
19732 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19734 * emit-rtl.h (is_leaf): Update comment about local
19735 register allocator.
19737 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19740 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
19741 for a variable to have a section before checking if the section has a
19743 Set section to.persistent if persistent attribute is set.
19744 Warn if .persistent attribute is used on an automatic variable.
19746 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
19748 PR rtl-optimization/80474
19749 * reorg.c (update_block): Do not ignore instructions in a delay slot.
19751 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
19753 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
19756 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
19758 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
19759 (casesi): Emit bounds checking as RTL.
19760 (casesi_internal_mips16_<mode>): Remove bounds checking.
19762 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
19764 * config/xtensa/xtensa.c (xtensa_option_override): Append
19765 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
19766 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
19767 xtensa_doloop_hooks): Define unconditionally.
19768 (xtensa_reorg_loops): Only call reorg_loops in the presence of
19770 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
19771 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
19772 for it in xtensa_option_override.
19773 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
19774 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
19776 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
19778 * doc/cppopts.texi: Document '-' special value to -MF.
19780 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
19782 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
19783 (cortex_a53_fconst): Likewise.
19784 (cortex_a53_fpmul): Likewise.
19785 (cortex_a53_f_load_64): Likewise.
19786 (cortex_a53_f_load_many): Likewise.
19787 (cortex_a53_advsimd_alu): Likewise.
19788 (cortex_a53_advsimd_alu_q): Likewise.
19789 (cortex_a53_advsimd_mul): Likewise.
19790 (cortex_a53_advsimd_mul_q): Likewise.
19791 (fpmac bypass): Add new bypass for fpmac-fpmac case.
19792 Add missing fmul, r2f_cvt and fconst cases.
19794 2017-06-14 Richard Biener <rguenther@suse.de>
19796 PR middle-end/81088
19797 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
19799 (fold_binary_loc): When associating do not treat pre-existing
19800 TREE_OVERFLOW on literal constants as a reason to allow
19801 TREE_OVERFLOW on associated literal constants.
19803 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19805 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
19806 (MASK_FEATURES): New macro.
19807 * config/sparc/sparc.c (sparc_option_override): Remove the special
19808 handling of -mfpu and generalize it to all MASK_FEATURES switches.
19810 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19812 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
19813 a division of 0 if non-call exceptions are enabled.
19815 2017-06-14 Andrew Pinski <apinski@cavium.com>
19816 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19819 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
19820 Improve vector initialization code gen for only variable case.
19822 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19824 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
19826 2017-06-14 Richard Biener <rguenther@suse.de>
19828 PR tree-optimization/81083
19829 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
19832 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19834 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
19835 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
19836 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
19837 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
19838 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
19839 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
19841 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19843 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
19844 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
19846 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19848 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
19850 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19852 * config/rs6000/t-rtems: Don't handle SPE.
19854 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19856 * config/rs6000/t-linux: Don't handle SPE.
19858 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19860 * config/rs6000/eabispe.h: Delete file.
19862 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19864 * config/rs6000/t-spe: Delete file.
19866 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19868 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
19869 (rs6000_legitimate_offset_address_p): Return false for anything in
19870 V2SImode or V2SFmode.
19872 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19874 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
19875 except V2SF and V2SI. Rearrange the vector modes, and add comments.
19876 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
19878 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
19879 (rs6000_legitimate_offset_address_p): Ditto.
19880 (rs6000_emit_move): Ditto.
19881 (rs6000_init_builtins): Remove V4HI_type_node.
19883 2017-06-13 Martin Liska <mliska@suse.cz>
19886 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
19887 (gate_asan): Likewise.
19888 * asan.h (asan_no_sanitize_address_p): Remove the function.
19889 (sanitize_flags_p): New function.
19890 * builtins.def: Fix coding style.
19891 * common.opt: Use renamed enum value.
19892 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
19893 * doc/extend.texi: Document no_sanitize attribute.
19894 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
19895 to SANITIZE_UNDEFINED_NONDEFAULT.
19896 * gcc.c (sanitize_spec_function): Use the renamed enum value.
19897 * gimple-fold.c (optimize_atomic_compare_exchange_p):
19898 Use sanitize_flags_p.
19899 * gimplify.c (gimplify_function_tree): Likewise.
19900 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
19901 * opts.c (parse_no_sanitize_attribute): New function.
19902 (common_handle_option): Use renamed enum value.
19903 * opts.h (parse_no_sanitize_attribute): Declare.
19904 * tree.c (sanitize_flags_p): New function.
19905 * tree.h: Declared here.
19906 * tsan.c: Use sanitize_flags_p.
19907 * ubsan.c (ubsan_expand_null_ifn): Likewise.
19908 (instrument_mem_ref): Likewise.
19909 (instrument_bool_enum_load): Likewise.
19910 (do_ubsan_in_current_function): Remove the function.
19911 (pass_ubsan::execute): Use sanitize_flags_p.
19912 * ubsan.h: Remove do_ubsan_in_current_function
19913 * tree-cfg.c (print_no_sanitize_attr_value): New function.
19914 (dump_function_to_file): Use it here.
19916 2017-06-13 Martin Jambor <mjambor@suse.cz>
19918 PR tree-optimization/80803
19919 PR tree-optimization/81063
19920 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
19921 (propagate_subaccesses_across_link): Enqueue subtree whenever
19922 necessary instead of relying on the caller.
19924 2017-06-13 Martin Jambor <mjambor@suse.cz>
19926 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
19927 that have a first_link.
19928 (sort_and_splice_var_accesses): Do not check first_link before
19930 (subtree_mark_written_and_enqueue): Likewise.
19931 (propagate_all_subaccesses): Likewise and do not stop at first
19932 parent with a first_link.
19934 2017-06-13 Martin Jambor <mjambor@suse.cz>
19936 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
19939 2017-06-13 Yury Gribov <tetra2005@gmail.com>
19941 * match.pd: New pattern.
19943 2017-06-13 Yury Gribov <tetra2005@gmail.com>
19945 * tree-vrp.c (is_masked_range_test): New function.
19946 (register_edge_assert_for): Determine ranges for
19949 2017-06-13 Yury Gribov <tetra2005@gmail.com>
19951 PR tree-optimization/67328
19952 * fold-const.c (maskable_range_p): New function.
19953 (build_range_check): Generate bittests if possible.
19955 2017-06-13 Martin Liska <mliska@suse.cz>
19957 * gimple-pretty-print.c (dump_probability): Add new argument.
19958 (dump_edge_probability): Dump both probability and count.
19959 (dump_gimple_label): Likewise.
19960 (dump_gimple_bb_header): Likewise.
19962 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
19965 * config/avr/avr-devices.c: Fix indentation.
19966 * config/avr/gen-avr-mmcu-specs.c: Dito.
19968 2017-06-13 Richard Biener <rguenther@suse.de>
19970 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
19971 instead get vector type from stmt_info.
19972 (vectorizable_reduction): Adjust. Remove dead code.
19974 2017-06-13 Richard Biener <rguenther@suse.de>
19976 PR middle-end/81065
19977 * fold-const.c (extract_muldiv_1): Remove bogus distribution
19978 case of C * (x * C2 + C3).
19979 (fold_addr_of_array_ref_difference): Properly fold index difference.
19981 2017-06-12 David S. Miller <davem@davemloft.net>
19984 * config/sparc/sparc.md (return expander): Emit frame blockage if
19985 function uses alloca.
19987 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
19989 * combine.c (make_field_assignment): Check len rather than the mode
19990 precision when calling force_to_mode.
19992 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
19994 Support multilibs and devices that see flash in RAM address range.
19997 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
19998 (avr_mcu_t) <flash_pm_offset>: New field.
19999 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
20000 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
20001 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
20002 (AVR_TINY_PM_OFFSET): Remove macro.
20003 * config/avr/avr.opt (-mshort-calls): New option.
20004 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
20005 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
20006 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
20007 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
20008 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
20009 instead of avr_arch->have_jmp_call.
20010 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
20011 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
20012 avr_arch->flash_pm_offset to define.
20013 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
20014 new field flash_pm_offset. Add entry for avrxmega3.
20015 (avr_texinfo): Add entry for avrxmega3.
20016 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
20017 attiny212, attiny214,
20018 attiny412, attiny414, attiny416, attiny417,
20019 attiny814, attiny816, attiny817,
20020 attiny1614, attiny1616, attiny1617,
20021 attiny3214, attiny3216, attiny3217.
20022 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
20023 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
20024 (avr_print_operand_address) [AVR_TINY]: Same.
20025 (avr_asm_init_sections) <readonly_data_section>: Only patch
20026 callback if avr_arch->flash_pm_offset = 0.
20027 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
20028 for rodata if avr_arch->flash_pm_offset != 0.
20029 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
20030 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
20031 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
20032 (m_options): Append opt_rcall.
20033 (m_dirnames): Append dir_rcall.
20034 * config/avr/t-multilib: Regenerate.
20036 * configure.ac [target=avr]: Check whether avrxmega3 default
20037 linker description file works as needed.
20038 * configure: Regenerate.
20039 * doc/avr-mmcu.texi: Regenerate.
20040 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
20041 <__AVR_ARCH__>: Document avrxmega3 and 103.
20042 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
20043 <__AVR_SHORT_CALLS__>: Document it.
20044 <__AVR_PM_BASE_ADDRESS__>: Document it.
20045 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
20046 (AVR Variable Attributes) <progmem>: Document this is
20047 not needed for avrxmega3.
20048 (AVR Named Address Spaces) <__flash>: Dito.
20050 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
20052 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
20054 2017-06-12 Doug Rupp <rupp@adacore.com>
20056 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
20057 Append vxworks-stdint.h to the tm_file list.
20058 * config/vxworks-stdint.h: New file.
20060 2017-06-12 Martin Liska <mliska@suse.cz>
20062 PR tree-optimization/81041
20063 * tree-profile.c (gimple_gen_ic_func_profiler):
20064 Create an extra BB in profile-generate
20065 (gimple_gen_time_profiler): Likewise.
20067 2017-06-12 Jakub Jelinek <jakub@redhat.com>
20069 PR tree-optimization/81003
20070 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
20071 (update_range_test): Use it instead of force_gimple_operand_gsi.
20073 2017-06-12 Richard Biener <rguenther@suse.de>
20075 PR tree-optimization/81053
20076 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
20077 with backedge value not defined in loop. Simplify def stmt
20080 2017-06-11 Tom de Vries <tom@codesourcery.com>
20083 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
20085 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
20086 nvptx_cannot_force_const_mem.
20088 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20090 * opts.c (finish_options): Move test for flag_split_stack after
20091 it has been initialized.
20093 2017-06-11 Jason Merrill <jason@redhat.com>
20095 * tree.h (id_equal): New.
20096 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
20097 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
20098 instead of strcmp of IDENTIFIER_POINTER.
20100 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20102 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
20103 (mark_all_inlined_calls_cdtor): Fix formating.
20104 (inline_transform): Rescale profile before inlining.
20106 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20108 * cgraph.h (cgraph_edge::clone): Update prototype.
20109 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
20110 (cgraph_node::create_clone): Update.
20111 (cgraph_node::create_version_clone): Update.
20112 * tree-inline.c (copy_bb): Update.
20113 (expand_call_inline): Update.
20115 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20117 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
20118 factored out from ...
20119 (rs6000_emit_prologue): ... here.
20121 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20123 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
20124 factored out from ...
20125 (rs6000_emit_prologue): ... here.
20127 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20129 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
20131 (handle_missing_profiles): Fix computation of tp_first_run.
20132 (counts_to_freqs): Do not touch freqs when count is 0.
20134 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20136 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
20139 2017-06-10 Tom de Vries <tom@codesourcery.com>
20141 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
20142 attributes): Document signal effective target.
20144 2017-06-10 Tom de Vries <tom@codesourcery.com>
20146 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
20147 Document effective target stack_size.
20149 2017-06-09 David Malcolm <dmalcolm@redhat.com>
20151 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
20152 to the edit_context if they can be auto-applied.
20154 2017-06-9 Ian Lance Taylor <iant@golang.org>
20156 * opts.c (finish_options): If -fsplit-stack, disable implicit
20157 -forder-blocks-and-partition.
20158 * doc/invoke.texi (Optimize Options): Document that when using
20159 -fsplit-stack -forder-blocks-and-partition is not implicitly
20162 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20164 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
20165 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
20166 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
20167 * builtins.def (abort, trap, unreachable): Declare cold.
20168 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
20169 * tree-core.h (ECF_COLD): New.
20170 * tree.c (set_call_expr_flags): Handle ECF_COLD.
20171 (build_common_builtin_nodes): Mark unreachable and abort as cold.
20173 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20175 * predict.c (unlikely_executed_stmt_p): Cleanup.
20177 2017-06-09 Richard Biener <rguenther@suse.de>
20179 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
20180 model if the ref is always written to.
20182 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20184 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
20186 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20188 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
20191 2017-06-09 Tom de Vries <tom@codesourcery.com>
20194 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
20195 "target cannot support label values" when encountering LABEL_REF.
20197 2017-06-09 Martin Liska <mliska@suse.cz>
20199 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
20200 (gimple_gen_ic_func_profiler): Emit direct comparison
20201 of __gcov_indirect_call_callee with NULL.
20202 (gimple_gen_time_profiler): Change probability from
20203 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
20205 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20207 * profile.c (edge_gcov_counts): Turn to pointer.
20208 (compute_branch_probabilities, compute_branch_probabilities): Update.
20209 (branch_prob): Do not clear edge_gcov_count.
20210 * profile.h (edge_gcov_counts): Turn to pointer.
20211 (edge_gcov_count): Update.
20213 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20215 * gimple.h (gimple_check_failed): Mark cold.
20217 2017-06-09 Richard Biener <rguenther@suse.de>
20219 PR tree-optimization/66623
20220 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
20221 refactor check_reduction into two parts, properly computing
20222 whether we have to check reduction validity for outer loop
20225 2017-06-09 Richard Biener <rguenther@suse.de>
20227 PR tree-optimization/79483
20228 * graphite-scop-detection.c (order): New global.
20229 (get_order): Compute bb to order mapping that satisfies code
20230 generation constraints.
20231 (cmp_pbbs): New helper.
20232 (build_scops): Start domwalk at entry block, sort generated
20235 2017-06-09 Richard Biener <rguenther@suse.de>
20237 PR middle-end/81007
20238 * ipa-polymorphic-call.c
20239 (ipa_polymorphic_call_context::restrict_to_inner_class):
20240 Skip FIELD_DECLs with error_mark_node type.
20241 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
20244 2017-06-09 Martin Liska <mliska@suse.cz>
20246 * predict.c (struct branch_predictor): New struct.
20247 (test_prediction_value_range): New test.
20248 (predict_c_tests): New function.
20249 * selftest-run-tests.c (selftest::run_tests): Run the function.
20250 * selftest.h: Declare new tests.
20252 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
20255 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
20256 gen_add3_insn did not fail.
20257 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
20258 r0, construct that number in a temporary reg and add that reg to r0.
20259 If asked to put the result in r0 as well, fail.
20261 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
20263 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
20264 for early expansion of vec_eqv.
20266 2017-06-08 Jakub Jelinek <jakub@redhat.com>
20268 PR middle-end/81005
20269 * ubsan.c (instrument_null): Avoid pointless code temporary.
20270 (pass_ubsan::execute): Instrument aggregate arguments of calls.
20272 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
20276 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
20279 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
20282 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20284 * predict.c (unlikely_executed_edge_p): Move ahead.
20285 (probably_never_executed_edge_p): Use it.
20287 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20289 PR middle-end/79988
20290 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
20291 gimple_call_builtin_p call.
20293 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20295 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
20296 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
20297 rtl_check_failed_type2, rtl_check_failed_code1,
20298 rtl_check_failed_code2, rtl_check_failed_code_mode,
20299 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
20300 rtvec_check_failed_bounds, rtl_check_failed_flag,
20301 _fatal_insn_not_found, _fatal_insn): Likewise.
20302 * tree.h (tree_contains_struct_check_failed,
20303 tree_check_failed, tree_not_check_failed,
20304 tree_class_check_failed, tree_range_check_failed,
20305 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
20306 tree_vec_elt_check_failed, phi_node_elt_check_failed,
20307 tree_operand_check_failed, omp_clause_check_failed,
20308 omp_clause_operand_check_failed, omp_clause_range_check_failed):
20311 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20313 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
20314 flag_branch_probabilities.
20315 * ipa-inline.c (edge_badness): Likewise.
20316 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
20317 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
20318 * predict.c (maybe_hot_frequency_p): Likewise.
20319 (probably_never_executed): Likewise.
20320 * sched-ebb.c (schedule_ebbs): Likewise.
20321 * sched-rgn.c (find_single_block_region): Likewise.
20322 * tracer.c (tail_duplicate): Likewise.
20324 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20326 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
20327 longer requires x_flag_profile_use.
20329 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20331 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
20332 instead of flag_reorder_blocks_and_partition.
20333 * dbxout.c (dbxout_function_end): Likewise.
20334 * dwarf2out.c (gen_subprogram_die): Likewise.
20335 * haifa-sched.c (sched_create_recovery_edges): Likewise.
20336 * hw-doloop.c (reorg_loops): Likewise.
20337 * varasm.c (assemble_start_function,
20338 assemble_end_function): Likewise.
20339 (decide_function_section): Do not check for
20340 flag_reorder_blocks_and_partition.
20342 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20344 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20346 (chkp_get_hard_register_fake_addr_expr): Ditto.
20347 (chkp_build_addr_expr): Add check for hard reg case.
20348 (chkp_parse_array_and_component_ref): Ditto.
20349 (chkp_find_bounds_1): Ditto.
20350 (chkp_process_stmt): Don't generate bounds store for
20353 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20355 * predict.c (maybe_hot_bb_p): Do not check profile status.
20356 (maybe_hot_edge_p): Likewise.
20357 (probably_never_executed): Check for zero counts even if profile
20359 (unlikely_executed_edge_p): New function.
20360 (unlikely_executed_stmt_p): New function.
20361 (unlikely_executed_bb_p): New function.
20362 (set_even_probabilities): Use unlikely predicates.
20363 (combine_predictions_for_bb): Likewise.
20364 (predict_paths_for_bb): Likewise.
20365 (predict_paths_leading_to_edge): Likewise.
20366 (determine_unlikely_bbs): New function.
20367 (estimate_bb_frequencies): Use it.
20368 (compute_function_frequency): Use zero counts even if profile is
20370 * profile-count.h: Fix typo.
20372 2017-08-08 Julia Koval <julia.koval@intel.com>
20374 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
20375 _mm512_mask_cvtsepi16_storeu_epi8,
20376 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
20377 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
20378 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
20379 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
20380 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
20381 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
20382 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
20383 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
20384 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
20385 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
20386 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
20387 __builtin_ia32_pmovuswb256mem_mask,
20388 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
20389 __builtin_ia32_pmovwb512mem_mask): New builtins.
20391 2017-08-08 Julia Koval <julia.koval@intel.com>
20393 PR target/73350,80862
20394 * config/i386/subst.md (round): Fix round pattern.
20395 * config/i386/i386.c (ix86_erase_embedded_rounding):
20396 Fix erasing rounding for the fixed pattern.
20398 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20400 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
20402 2017-06-08 Martin Liska <mliska@suse.cz>
20404 PR gcov-profile/80911
20405 * gcov.c (block_info::block_info): New constructor.
20407 2017-06-07 Carl Love <cel@us.ibm.com>
20409 * config/rs6000/rs6000-c: The return type of the following
20410 built-in functions was implemented as int not long long. Fix sign
20411 of return value for the unsigned version of vec_mulo and vec_mule.
20412 vector unsigned long long vec_bperm (vector unsigned long long,
20413 vector unsigned char)
20414 vector signed long long vec_mule (vector signed int,
20416 vector unsigned long long vec_mule (vector unsigned int,
20417 vector unsigned int)
20418 vector signed long long vec_mulo (vector signed int,
20420 vector unsigned long long vec_mulo (vector unsigned int,
20421 vector unsigned int)
20422 * doc/extend.texi: Fix the documentation for the built-in
20425 2017-06-07 Carl Love <cel@us.ibm.com>
20428 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
20431 2017-06-07 Carl Love <cel@us.ibm.com>
20433 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
20434 support, Generate doublehv for signed int/float for BE case only.
20436 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
20438 * doc/invoke.texi (mcx16): Rewrite.
20440 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20442 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
20443 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
20444 *mov<mode>_softfloat, and an anonymous splitter): Use
20445 nonimmediate_operand instead of rs6000_nonimmediate_operand.
20447 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20449 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
20451 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
20452 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
20453 (rs6000_debug_reg_global): Adjust.
20454 (rs6000_init_hard_regno_mode_ok): Adjust.
20455 (rs6000_dbx_register_number): Adjust.
20456 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
20457 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
20458 Remove SPE_ACC and SPEFSCR.
20459 (REG_ALLOC_ORDER): Ditto.
20460 (FRAME_POINTER_REGNUM): Change to 111.
20461 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
20462 (REG_CLASS_NAMES): Ditto.
20463 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
20464 (REGISTER_NAMES): Ditto.
20465 (ADDITIONAL_REG_NAMES): Ditto.
20466 (rs6000_reg_names): Ditto.
20467 * config/rs6000/rs6000.md: Renumber some register number
20470 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20472 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
20474 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
20475 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
20477 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
20478 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
20479 Delete the SPE high registers.
20480 (REG_ALLOC_ORDER): Ditto.
20481 (enum reg_class): Remove SPE_HIGH_REGS.
20482 (REG_CLASS_NAMES): Ditto.
20483 (REG_CLASS_CONTENTS): Delete the SPE high registers.
20484 (REGISTER_NAMES): Ditto.
20485 (rs6000_reg_names): Ditto.
20486 * doc/tm.texi.in: Remove SPE as example.
20487 * doc/tm.texi: Regenerate.
20489 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20491 * config/rs6000/8540.md (ppc8540_brinc): Delete.
20492 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
20493 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
20494 * config/rs6000/rs6000.md (type): Remove "brinc".
20496 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20498 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
20499 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
20500 * config/rs6000/linuxspe.h: Delete file.
20501 * config/rs6000/rs6000.md: Don't include spe.md.
20502 * config/rs6000/spe.h: Delete file.
20503 * config/rs6000/spe.md: Delete file.
20504 * config/rs6000/t-rs6000: Remove spe.md.
20506 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20508 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
20509 (reg_or_none500mem_operand): Delete.
20510 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
20511 instead of reg_or_none500mem_operand.
20513 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20515 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
20516 handling of SPE flags.
20517 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
20519 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20521 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
20523 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
20524 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
20525 paired_divv2sf3): Similar.
20526 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
20527 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
20528 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
20530 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
20531 Rename the paired_* instruction patterns.
20532 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
20534 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
20535 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
20536 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
20537 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
20538 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
20539 PAIRED_VECTOR_MODE.
20540 (struct machine_function): Delete field spe_insn_chain_scanned_p.
20541 (spe_func_has_64bit_regs_p): Delete.
20542 (spe_expand_predicate_builtin): Delete.
20543 (spe_expand_evsel_builtin): Delete.
20544 (TARGET_DWARF_REGISTER_SPAN): Do not define.
20545 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
20546 (invalid_e500_subreg): Delete.
20547 (rs6000_legitimize_address): Always force_reg op2 as well, for
20548 paired single memory accesses.
20549 (rs6000_member_type_forces_blk): Delete.
20550 (rs6000_spe_function_arg): Delete.
20551 (rs6000_expand_unop_builtin): Delete SPE handling.
20552 (rs6000_expand_binop_builtin): Ditto.
20553 (spe_expand_stv_builtin): Delete.
20554 (bdesc_2arg_spe): Delete.
20555 (spe_expand_builtin): Delete.
20556 (spe_expand_predicate_builtin): Delete.
20557 (spe_expand_evsel_builtin): Delete.
20558 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
20559 (spe_init_builtins): Delete.
20560 (spe_func_has_64bit_regs_p): Delete.
20561 (savres_routine_name): Delete "info" parameter. Adjust callers.
20562 (rs6000_emit_stack_reset): Ditto.
20563 (rs6000_dwarf_register_span): Delete.
20564 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
20565 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
20566 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
20568 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
20570 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
20571 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
20572 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
20573 mulv2sf3, divv2sf3): Delete expanders.
20575 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20577 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
20579 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20581 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
20582 * config/rs6000/rs6000.c: Ditto.
20584 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20586 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
20587 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
20588 comparison_operator.
20590 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20592 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
20593 * config/rs6000/rs6000.opt: Ditto.
20594 * config/rs6000/t-rtems: Ditto.
20596 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20598 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
20599 TARGET_E500_SINGLE by 0, simplify.
20600 * config/rs6000/rs6000.c: Ditto.
20601 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
20602 (spe_build_register_parallel): Delete.
20603 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
20604 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
20605 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
20606 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
20607 (E500_CONVERT): Delete.
20608 * config/rs6000/spe.md: Remove many patterns and all define_constants.
20610 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20612 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
20613 * config/rs6000/dfp.md: Ditto.
20614 (negdd2, *negdd2_fpr): Merge.
20615 (absdd2, *absdd2_fpr): Merge.
20616 (negtd2, *negtd2_fpr): Merge.
20617 (abstd2, *abstd2_fpr): Merge.
20618 * config/rs6000/e500.h: Delete file.
20619 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
20620 TARGET_FPRS by 1 and simplify.
20621 * config/rs6000/rs6000-c.c: Ditto.
20622 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
20623 TARGET_DF_SPE by 0.
20624 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
20626 * config/rs6000/rs6000.md: Ditto.
20627 (floatdidf2, *floatdidf2_fpr): Merge.
20628 (move_from_CR_gt_bit): Delete.
20629 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
20630 (E500_CR_IOR_COMPARE): Delete.
20631 (All patterns that require !TARGET_FPRS): Delete.
20632 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
20634 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20636 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
20638 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20640 * graphds.c (add_edge): Intitialize edge's attached data.
20641 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
20642 pointer parameter. Call pointed function on each edge during
20643 graph traversing. Skip traversing the edge when the function
20645 (graphds_dfs, graphds_scc): Ditto.
20646 (for_each_edge): New parameter. Pass the new parameter to callback
20648 * graphds.h (skip_edge_callback): New function pointer type.
20649 (graphds_dfs, graphds_scc): New function pointer parameter.
20650 (graphds_edge_callback, for_each_edge): New parameter.
20652 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20654 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
20655 out code checking if runtime alias check is possible to below ...
20656 Call the new function.
20657 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
20658 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
20660 2017-06-07 Marek Polacek <polacek@redhat.com>
20663 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
20664 TYPE_OVERFLOW_WRAPS check.
20666 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20668 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
20669 if versioning is required.
20670 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
20671 peeling with the check for versioning.
20673 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20675 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
20676 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
20677 Set true to new parameter if new ssa variable is defined.
20678 (vect_gen_vector_loop_niters): Refactor. Set range information
20679 for the new vector loop bound variable.
20680 (vect_do_peeling): Ditto.
20682 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20684 * tree-affine.c (ssa.h): Include header file.
20685 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
20686 has wrapping overflow behavior.
20688 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20690 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
20692 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20694 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
20695 (tree_to_aff_combination): ... here.
20697 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20699 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
20700 reg_pressure model function.
20701 (ivopts_global_cost_for_size): Delete.
20702 (determine_set_costs, iv_ca_recount_cost): Call new model function
20703 ivopts_estimate_reg_pressure.
20705 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20707 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
20708 expensive than udiv. Remove floating point cases from mod.
20710 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20712 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
20713 Increase idiv cost.
20715 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20717 * config/aarch64/aarch64.md
20718 (copysignsf3): Fix mask generation.
20720 2017-06-07 Jakub Jelinek <jakub@redhat.com>
20722 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
20724 (class dump_manager): Add register_dumps method.
20725 * dumpfile.c: Include langhooks.h.
20726 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
20727 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
20728 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
20729 (dump_manager::dump_register): Start with 512 entries instead of 32.
20730 (dump_manager::register_dumps): New method.
20731 * toplev.c (general_init): Instead of invoking register_dumps
20732 langhook, invoke register_dumps method on the dump manager.
20733 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
20736 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
20738 * doc/md.texi: Clarify the restrictions on a define_insn condition.
20739 Say that # requires an associated define_split to exist, and that
20740 the define_split must be suitable for use after register allocation.
20742 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20744 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
20745 (compute_outgoing_frequencies): Also initialize zero counts.
20746 (find_many_sub_basic_blocks): Do not produce uninitialized profile
20747 around loops; preserve more of profile when nothing changes.
20749 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
20751 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
20753 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
20754 * config/arm/arm-cpu-cdata.h: Regenerate.
20755 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
20756 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
20757 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
20758 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
20759 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
20761 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
20763 * config/arm/t-rmprofile: Likewise.
20764 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
20766 2017-06-06 David S. Miller <davem@davemloft.net>
20769 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
20770 blockage if function uses alloca.
20772 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20774 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
20775 New "uid" fields to hold pretty-print IDs of group and ref.
20776 Memory references are now identified as <group_id>:<ref_id>
20777 instead of using [random] addresses.
20778 (dump_mem_details): Simplify, no functional change.
20779 (dump_mem_ref): Simplify and make output more concise.
20780 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
20781 (find_or_create_group): Initialize group uid.
20782 (record_ref): Initialize ref uid. Improve debug output.
20783 (prune_group_by_reuse, should_issue_prefetch_p,)
20784 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
20785 (mark_nontemporal_store, determine_loop_nest_reuse):
20786 Improve debug output.
20788 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20790 * dbgcnt.def (prefetch): New debug counter.
20791 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
20792 (schedule_prefetches): Stop issueing prefetches if debug counter
20795 2017-06-06 Tom de Vries <tom@codesourcery.com>
20797 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
20798 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
20800 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20802 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
20803 Use aarch64_reg_or_zero predicate for operand 4.
20804 (aarch64_compare_and_swap<mode> define_insn_and_split):
20805 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
20806 (aarch64_store_exclusive<mode>): Likewise for operand 2.
20808 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
20810 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
20811 (arm_compute_save_core_reg_mask): This.
20812 (thumb1_compute_save_reg_mask): Rename into ...
20813 (thumb1_compute_save_core_reg_mask): This.
20814 (arm_compute_save_reg0_reg12_mask): Adapt comment.
20815 (arm_compute_frame_layout): Likewise.
20817 2017-06-06 Richard Biener <rguenther@suse.de>
20819 PR tree-optimization/80974
20820 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
20821 keep or clear leaders SSA info.
20823 2017-06-06 Tom de Vries <tom@codesourcery.com>
20825 * config/nvptx/nvptx.c (split_mode_p): New function.
20826 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
20828 2017-06-06 Tom de Vries <tom@codesourcery.com>
20830 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
20832 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20835 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
20838 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20840 * shrink-wrap.c (handle_simple_exit): Update profile.
20841 (try_shrink_wrapping): Upate profile.
20843 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20845 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
20846 (tree_guess_outgoing_edge_probabilities): New.
20847 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
20848 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
20850 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20852 * ipa-split.c (split_function): Initialize return bb profile.
20854 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20856 * profile.c (compute_branch_probabilities): Also initialize
20857 EXIT_BLOCK profile.
20859 2017-06-06 Richard Biener <rguenther@suse.de>
20861 PR tree-optimization/80928
20862 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
20863 (vect_analyze_loop_operations): Properly guard analysis for
20865 (vect_transform_loop): Likewise.
20866 (vect_analyze_loop_2): Also reset SLP type on PHIs.
20867 (vect_model_induction_cost): Do not cost for pure SLP.
20868 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
20869 of induction in inner loop vectorization.
20870 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
20871 (vect_get_and_check_slp_defs): Handle vect_induction_def.
20872 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
20874 (vect_analyze_slp_cost_1): Cost induction.
20875 (vect_detect_hybrid_slp_stmts): Handle PHIs.
20876 (vect_get_slp_vect_defs): Likewise.
20877 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
20878 (vect_transform_stmt): Handle SLP reductions.
20879 * tree-vectorizer.h (vectorizable_induction): Adjust.
20881 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
20883 * config/rs6000/rs6000.c (make_resolver_func): Update
20884 init_lowered_empty_function call.
20886 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
20888 * doc/invoke.texi: Document the -fprofile-abs-path option.
20889 * common.opt (fprofile-abs-path): New option.
20890 * gcov-io.h (gcov_write_filename): Declare.
20891 * gcov-io.c (gcov_write_filename): New function.
20892 * coverage.c (coverage_begin_function): Use gcov_write_filename.
20893 * profile.c (output_location): Likewise.
20895 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20897 * shring-wrap.c: Revert accidental commit.
20899 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
20901 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
20903 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20905 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
20907 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
20908 profile in callgraph edge.
20909 * profile-count.h (apply_probability): If THIS is 0, then result is 0
20910 (apply_scale): Likewise.
20911 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
20912 Also scale profile when inlining function with zero profile.
20913 (initialize_cfun): Update exit block profile even when it is zero.
20914 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
20915 when profile is read.
20917 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
20919 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
20920 (CLONE_*): New constants to define the processors we can generate
20921 code for with the target_clone attribute.
20922 (rs6000_clone_map): New array to identify which clone processors
20923 the current program is running on.
20924 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
20925 target_clone attribute.
20926 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
20927 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
20928 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
20929 (cpu_expand_builtin): Add support for target_clone attribute.
20930 (rs6000_valid_attribute_p): Allow "default" attribute.
20931 (get_decl_name): New debug function to simplify printing the
20932 current function name in debugging statements.
20933 (rs6000_clone_priority): New functions to support the target_clone
20934 attribute, and be able to generate code to switch between ISA 2.05
20935 through ISA 3.0 (power6 through power9).
20936 (rs6000_compare_version_priority): Likewise.
20937 (rs6000_get_function_versions_dispatcher): Likewise.
20938 (make_resolver_func): Likewise.
20939 (add_condition_to_bb): Likewise.
20940 (dispatch_function_versions): Likewise.
20941 (rs6000_generate_version_dispatcher_body): Likewise.
20942 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
20943 (fusion_gpr_load_p): Fix a spacing issue.
20944 * doc/extend.texi (Common Function Attributes): Document that the
20945 PowerPC supports the target_clone attribute.
20947 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
20949 * config/arm/arm.h: explain F symbol found in description of ARM
20950 register allocation in its legend.
20952 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20954 * config/mips/frame-header-opt.c: Include profile-count.h.
20955 * config/riscv/riscv.c: Include profile-count.h
20957 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20959 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
20961 (sm_set_flag_if_changed): Add bbs field.
20962 (execute_sm_if_changed_flag_set): Pass BBS.
20963 (execute_sm): Update.
20965 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20967 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
20970 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20972 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
20973 (peephole2): New peephole2 to emit the above.
20974 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
20976 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20978 * config/aarch64/aarch64.c (define_peephole2 above
20979 *sub_<shift>_<mode>): New peephole.
20981 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
20983 * config/i386/i386.c (make_resolver_func): Update.
20984 * Makefile.in: Add profile-count.h and profile-count.o
20985 * auto-profile.c (afdo_indirect_call): Update to new API.
20986 (afdo_set_bb_count): Update.
20987 (afdo_propagate_edge): Update.
20988 (afdo_propagate_circuit): Update.
20989 (afdo_calculate_branch_prob): Update.
20990 (afdo_annotate_cfg): Update.
20991 * basic-block.h: Include profile-count.h
20992 (struct edge_def): Turn count to profile_count.
20993 (struct basic_block_def): Likewie.
20994 (REG_BR_PROB_BASE): Move to profile-count.h
20995 (RDIV): Move to profile-count.h
20996 * bb-reorder.c (max_entry_count): Turn to profile_count.
20997 (find_traces): Update.
20998 (rotate_loop):Update.
20999 (connect_traces):Update.
21000 (sanitize_hot_paths):Update.
21001 * bt-load.c (migrate_btr_defs): Update.
21002 * cfg.c (RDIV): Remove.
21003 (init_flow): Use alloc_block.
21004 (alloc_block): Uninitialize count.
21005 (unchecked_make_edge): Uninitialize count.
21006 (check_bb_profile): Update.
21007 (dump_edge_info): Update.
21008 (dump_bb_info): Update.
21009 (update_bb_profile_for_threading): Update.
21010 (scale_bbs_frequencies_int): Update.
21011 (scale_bbs_frequencies_gcov_type): Update.
21012 (scale_bbs_frequencies_profile_count): New.
21013 * cfg.h (update_bb_profile_for_threading): Update.
21014 (scale_bbs_frequencies_profile_count): Declare.
21015 * cfgbuild.c (compute_outgoing_frequencies): Update.
21016 (find_many_sub_basic_blocks): Update.
21017 * cfgcleanup.c (try_forward_edges): Update.
21018 (try_crossjump_to_edge): Update.
21019 * cfgexpand.c (expand_gimple_tailcall): Update.
21020 (construct_exit_block): Update.
21021 * cfghooks.c (verify_flow_info): Update.
21022 (dump_bb_for_graph): Update.
21023 (split_edge): Update.
21024 (make_forwarder_block): Update.
21025 (duplicate_block): Update.
21026 (account_profile_record): Update.
21027 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
21028 (get_estimated_loop_iterations): Update.
21029 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
21030 (single_likely_exit): Update.
21031 * cfgloopmanip.c (scale_loop_profile): Update.
21033 (set_zero_probability): Update.
21034 (lv_adjust_loop_entry_edge): Update.
21035 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
21036 (purge_dead_edges): Update.
21037 (rtl_account_profile_record): Update.
21038 * cgraph.c (cgraph_node::create): Uninitialize count.
21039 (symbol_table::create_edge): Uninitialize count.
21040 (cgraph_update_edges_for_call_stmt_node): Update.
21041 (cgraph_edge::dump_edge_flags): Update.
21042 (cgraph_node::dump): Update.
21043 (cgraph_edge::maybe_hot_p): Update.
21044 * cgraph.h: Include profile-count.h
21045 (create_clone), create_edge, create_indirect_edge): Update.
21046 (cgraph_node): Turn count to profile_count.
21047 (cgraph_edge0: Likewise.
21048 (make_speculative, clone): Update.
21049 (create_edge): Update.
21050 (init_lowered_empty_function): Update.
21051 * cgraphclones.c (cgraph_edge::clone): Update.
21052 (duplicate_thunk_for_node): Update.
21053 (cgraph_node::create_clone): Update.
21054 * cgraphunit.c (cgraph_node::analyze): Update.
21055 (cgraph_node::expand_thunk): Update.
21056 * final.c (dump_basic_block_info): Update.
21057 * gimple-streamer-in.c (input_bb): Update.
21058 * gimple-streamer-out.c (output_bb): Update.
21059 * graphite.c (print_global_statistics): Update.
21060 (print_graphite_scop_statistics): Update.
21061 * hsa-brig.c: Include basic-block.h.
21062 * hsa-dump.c: Include basic-block.h.
21063 * hsa-gen.c (T sum_slice): Update.
21064 (convert_switch_statements):Update.
21065 * hsa-regalloc.c: Include basic-block.h.
21066 * ipa-chkp.c (chkp_produce_thunks): Update.
21067 * ipa-cp.c (struct caller_statistics): Update.
21068 (init_caller_stats): Update.
21069 (gather_caller_stats): Update.
21070 (ipcp_cloning_candidate_p): Update.
21071 (good_cloning_opportunity_p): Update.
21072 (get_info_about_necessary_edges): Update.
21073 (dump_profile_updates): Update.
21074 (update_profiling_info): Update.
21075 (update_specialized_profile): Update.
21076 (perhaps_add_new_callers): Update.
21077 (decide_about_value): Update.
21078 (ipa_cp_c_finalize): Update.
21079 * ipa-devirt.c (struct odr_type_warn_count): Update.
21080 (struct decl_warn_count): Update.
21081 (struct final_warning_record): Update.
21082 (possible_polymorphic_call_targets): Update.
21083 (ipa_devirt): Update.
21084 * ipa-fnsummary.c (redirect_to_unreachable): Update.
21085 * ipa-icf.c (sem_function::merge): Update.
21086 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
21087 * ipa-inline.c (compute_uninlined_call_time): Update.
21088 (compute_inlined_call_time): Update.
21089 (want_inline_small_function_p): Update.
21090 (want_inline_self_recursive_call_p): Update.
21091 (edge_badness): Update.
21092 (lookup_recursive_calls): Update.
21093 (recursive_inlining): Update.
21094 (inline_small_functions): Update.
21095 (dump_overall_stats): Update.
21096 (dump_inline_stats): Update.
21097 * ipa-profile.c (ipa_profile_generate_summary): Update.
21098 (ipa_propagate_frequency): Update.
21099 (ipa_profile): Update.
21100 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
21101 * ipa-utils.c (ipa_merge_profiles): Update.
21102 * loop-doloop.c (doloop_modify): Update.
21103 * loop-unroll.c (report_unroll): Update.
21104 (unroll_loop_runtime_iterations): Update.
21105 * lto-cgraph.c (lto_output_edge): Update.
21106 (lto_output_node): Update.
21107 (input_node): Update.
21108 (input_edge): Update.
21109 (merge_profile_summaries): Update.
21110 * lto-streamer-in.c (input_cfg): Update.
21111 * lto-streamer-out.c (output_cfg): Update.
21112 * mcf.c (create_fixup_graph): Update.
21113 (adjust_cfg_counts): Update.
21114 (sum_edge_counts): Update.
21115 * modulo-sched.c (sms_schedule): Update.
21116 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
21117 * predict.c (maybe_hot_count_p): Update.
21118 (probably_never_executed): Update.
21119 (dump_prediction): Update.
21120 (combine_predictions_for_bb): Update.
21121 (propagate_freq): Update.
21122 (handle_missing_profiles): Update.
21123 (counts_to_freqs): Update.
21124 (rebuild_frequencies): Update.
21125 (force_edge_cold): Update.
21126 * predict.h: Include profile-count.h
21127 (maybe_hot_count_p, counts_to_freqs): UPdate.
21128 * print-rtl-function.c: Do not include cfg.h
21129 * print-rtl.c: Include basic-block.h
21130 * profile-count.c: New file.
21131 * profile-count.h: New file.
21132 * profile.c (is_edge_inconsistent): Update.
21133 (correct_negative_edge_counts): Update.
21134 (is_inconsistent): Update.
21135 (set_bb_counts): Update.
21136 (read_profile_edge_counts): Update.
21137 (compute_frequency_overlap): Update.
21138 (compute_branch_probabilities): Update; Initialize and deinitialize
21140 (branch_prob): Update.
21141 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
21142 (edge_gcov_count): New.
21143 (bb_gcov_count): New.
21144 * shrink-wrap.c (try_shrink_wrapping): Update.
21145 * tracer.c (better_p): Update.
21146 * trans-mem.c (expand_transaction): Update.
21147 (ipa_tm_insert_irr_call): Update.
21148 (ipa_tm_insert_gettmclone_call): Update.
21149 * tree-call-cdce.c: Update.
21150 * tree-cfg.c (gimple_duplicate_sese_region): Update.
21151 (gimple_duplicate_sese_tail): Update.
21152 (gimple_account_profile_record): Update.
21153 (execute_fixup_cfg): Update.
21154 * tree-inline.c (copy_bb): Update.
21155 (copy_edges_for_bb): Update.
21156 (initialize_cfun): Update.
21157 (freqs_to_counts): Update.
21158 (copy_cfg_body): Update.
21159 (expand_call_inline): Update.
21160 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
21161 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
21162 (try_unroll_loop_completely): Update.
21163 (try_peel_loop): Update.
21164 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
21165 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
21166 * tree-ssa-loop-split.c (connect_loops): Update.
21167 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
21168 * tree-ssa-reassoc.c (branch_fixup): Update.
21169 * tree-ssa-tail-merge.c (replace_block_by): Update.
21170 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
21171 (compute_path_counts): Update.
21172 (update_profile): Update.
21173 (recompute_probabilities): Update.
21174 (update_joiner_offpath_counts): Update.
21175 (estimated_freqs_path): Update.
21176 (freqs_to_counts_path): Update.
21177 (clear_counts_path): Update.
21178 (ssa_fix_duplicate_block_edges): Update.
21179 (duplicate_thread_path): Update.
21180 * tree-switch-conversion.c (case_bit_test_cmp): Update.
21181 (struct switch_conv_info): Update.
21182 * tree-tailcall.c (decrease_profile): Update.
21183 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
21184 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
21185 * value-prof.c (check_counter): Update.
21186 (gimple_divmod_fixed_value): Update.
21187 (gimple_mod_pow2): Update.
21188 (gimple_mod_subtract): Update.
21189 (gimple_ic_transform): Update.
21190 (gimple_stringop_fixed_value): Update.
21191 * value-prof.h (gimple_ic): Update.
21193 2017-06-02 Carl Love <cel@us.ibm.com>
21195 * config/rs6000/rs6000-c: Add support for built-in functions
21196 vector double vec_doublee (vector signed int);
21197 vector double vec_doublee (vector unsigned int);
21198 vector double vec_doublee (vector float);
21199 vector double vec_doubleh (vector signed int);
21200 vector double vec_doubleh (vector unsigned int);
21201 vector double vec_doubleh (vector float);
21202 vector double vec_doublel (vector signed int);
21203 vector double vec_doublel (vector unsigned int);
21204 vector double vec_doublel (vector float);
21205 vector double vec_doubleo (vector signed int);
21206 vector double vec_doubleo (vector unsigned int);
21207 vector double vec_doubleo (vector float);.
21208 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
21209 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
21211 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
21212 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
21213 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
21215 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
21216 vec_doublel, vec_doubleh.
21217 * doc/extend.texi: Update the built-in documentation file for the
21218 new built-in functions.
21220 2017-06-02 David Malcolm <dmalcolm@redhat.com>
21223 * ipa-inline-analysis.c (free_growth_caches): Set
21224 edge_removal_hook_holder to NULL after removing it.
21226 2017-06-02 Sudakshina Das <sudi.das@arm.com>
21228 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
21229 comparision with zero.
21231 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21232 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21233 for early expansion of vec_min and vec_max builtins.
21234 (builtin_function_type): Add min/max unsigned variants to those
21235 identified as having unsigned arguments.
21237 2017-06-02 Olivier Hainque <hainque@adacore.com>
21239 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
21241 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21243 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
21244 Use VALL_F16 iterator rather than VALL.
21246 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21248 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
21249 Emit CBNZ inside loop when doing a strong exchange and comparing
21250 against zero. Generate the CC flags after the loop.
21252 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
21254 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
21255 (dl_section_ref): New.
21256 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
21257 On AIX, append an expression to subtract the size of the
21258 section length to dl_section_ref.
21260 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21262 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21263 for early expansion of vector absolute builtins.
21265 2017-06-02 Richard Biener <rguenther@suse.de>
21267 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
21268 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
21270 2017-06-02 Richard Biener <rguenther@suse.de>
21272 PR tree-optimization/80948
21273 * tree-tailcall.c (find_tail_calls): Track stmts to move in
21274 stmt order as well.
21276 2017-06-02 Richard Biener <rguenther@suse.de>
21278 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
21280 * tree-vect-stmts.c (process_use): Do not mark backedge defs
21281 for inductions as relevant.
21283 2017-06-02 Richard Biener <rguenther@suse.de>
21285 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
21286 (vectorizable_induction): ... this. Remove dead code.
21288 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
21290 * builtins. (expand_builtin_alloca): Remove second parameter and
21291 infer its value from the first parameter instead.
21292 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
21294 2017-06-02 Jakub Jelinek <jakub@redhat.com>
21296 PR rtl-optimization/80903
21297 * loop-doloop.c (add_test): Unshare sequence.
21299 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
21301 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
21303 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
21305 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
21307 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
21308 xlogue_layout::get_instance, logue_layout::xlogue_layout,
21309 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
21310 (xlogue_layout::get_stub_rtx): Make static.
21311 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
21312 (xlogue_layout::compute_stub_managed_regs): Rename to...
21313 (xlogue_layout::count_stub_managed_regs): ...this.
21314 (xlogue_layout::is_stub_managed_reg): New function.
21315 (xlogue_layout::m_stub_names): Rename to...
21316 (xlogue_layout::s_stub_names): ...this, make static.
21317 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
21318 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
21319 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
21320 xlogue_layout::s_stub_names): Instantiate statics.
21321 (stub_managed_regs): Remove.
21322 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
21323 (disable_call_ms2sysv_xlogues): Rename to...
21324 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
21325 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
21327 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
21328 change after reload_completed.
21329 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
21331 (ix86_expand_prologue): Likewise.
21332 (ix86_expand_epilogue): Likewise.
21333 (ix86_expand_split_stack_prologue): Likewise.
21334 (ix86_compute_frame_layout): Remove frame parameter ...
21335 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
21336 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
21338 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
21339 (ix86_frame): Move from here ...
21340 * config/i386/i386.h (ix86_frame): ... to here.
21341 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
21342 complete ix86_frame data structure instead. Remove some_ld_name.
21344 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
21346 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
21347 symbols that hold a DECL_VALUE_EXPR.
21349 2017-06-01 Martin Jambor <mjambor@suse.cz>
21351 PR tree-optimization/80898
21352 * tree-sra.c (process_subtree_disqualification): Removed.
21353 (disqualify_candidate): Do not acll
21354 process_subtree_disqualification.
21355 (subtree_mark_written_and_enqueue): New function.
21356 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
21357 RHS has been disqualified and re-queue LHS if necessary. Apart
21358 from that, ignore disqualified RHS.
21360 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21362 * config/s390/s390.c (s390_emit_epilogue): Disable early return
21363 address fetch for z10 or later.
21365 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21367 * config/arc/arc.md (tst_movb): Add guard when splitting.
21369 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21371 * config/arc/arc.c (arc_can_eliminate): Test against
21372 arc_frame_pointer_needed.
21374 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21376 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
21377 to prevent store reordering.
21378 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
21379 (type): Add block type.
21380 (stack_tie): Define special instruction to be used in
21383 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21385 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
21386 constraint. It is not valid for the pattern.
21387 (noncommutative_binary_comparison): Likewise.
21389 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21391 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
21394 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21396 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
21397 be used by the reg-alloc.
21399 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21401 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
21402 reg-alloc when having mul64 or mul32x16 instructions.
21403 (mulsidi3): Likewise.
21404 (umulsidi3): Likewise.
21405 (mulsi32x16): New pattern.
21406 (mulsi64): Likewise.
21407 (mulsidi64): Likewise.
21408 (umulsidi64): Likewise.
21409 (MUL32x16_REG): Define.
21410 (mul64_600): Use MUL32x16_REG.
21411 (mac64_600): Likewise.
21412 (umul64_600): Likewise.
21413 (umac64_600): Likewise.
21415 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21417 * config/arc/arc.md (mulsi3_700): Make it commutative.
21419 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
21421 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
21423 (*sign_extendsidi2_insn): Likewise for movstosw.
21425 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
21427 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
21428 the type of the input discriminant value. Convert the
21429 discriminant value of signedness vary.
21431 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
21433 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
21434 Add to -Wall section.
21436 2017-06-01 Richard Biener <rguenther@suse.de>
21438 PR middle-end/66313
21439 * fold-const.c (fold_plusminus_mult_expr): If the factored
21440 factor may be zero use a wrapping type for the inner operation.
21441 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
21442 and handle moved defs.
21443 (process_assignment): Properly guard the unary op case. Return a
21444 tri-state indicating that moving the stmt before the call may allow
21445 to continue. Pass through to_move.
21446 (find_tail_calls): Handle moving unrelated defs before
21449 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
21452 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
21453 splitter result in the canonical way.
21455 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
21457 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
21458 also for 32bit target. Update insn attributes.
21459 (zero-extendsidi2 splitter): Allow all registers for operand 1.
21461 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
21463 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
21464 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
21465 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
21466 (_mm_maskz_min_ss): New intrinsics.
21468 2017-05-31 Martin Liska <mliska@suse.cz>
21470 * tree-vect-loop.c (vect_create_epilog_for_reduction):
21471 Change comment style to one we normally use.
21472 (vectorizable_reduction): Likewise.
21473 (vectorizable_induction): Likewise.
21474 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
21475 (vectorizable_call): Likewise.
21476 (vectorizable_simd_clone_call): Likewise.
21477 (vectorizable_conversion): Likewise.
21478 (vectorizable_assignment): Likewise.
21479 (vectorizable_shift): Likewise.
21480 (vectorizable_operation): Likewise.
21481 (vectorizable_store): Likewise.
21482 (vectorizable_load): Likewise.
21483 * tree-vectorizer.h: Likewise.
21485 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
21487 * passes.c (emergency_dump_function): New.
21488 * tree-pass.h (emergency_dump_function): Declare.
21489 * plugin.c (plugins_internal_error_function): Remove.
21490 * plugin.h (plugins_internal_error_function): Remove declaration.
21491 * toplev.c (internal_error_function): New static function. Use it...
21492 (general_init): ...here.
21494 2017-05-31 Graham Markall <graham.markall@embecosm.com>
21496 * config/arc/arc.c (arc_print_operand): Handle constant operands.
21497 (arc_rtx_costs): Add costs for new patterns.
21498 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
21499 * config/arc/predicates.md: Add _1_2_3_operand predicate.
21501 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21503 * tree-ssa-strlen.c (get_next_strinfo): New function.
21504 (get_stridx_plus_constant): Use it.
21505 (zero_length_string): Likewise.
21506 (adjust_related_strinfos): Likewise.
21507 (adjust_last_stmt): Likewise.
21509 2017-05-31 Richard Biener <rguenther@suse.de>
21512 * config/i386/i386.c (ix86_expand_builtin): Remove assert
21513 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
21515 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21517 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
21518 loop_vinfo argument and use of dependence distance vectors.
21519 Check instead whether the two references differ only in their
21520 initial value and assume that they have the same alignment if the
21521 difference is a multiple of the vector alignment.
21522 (vect_analyze_data_refs_alignment): Update call accordingly.
21524 2017-05-31 Martin Liska <mliska@suse.cz>
21527 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
21529 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21531 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
21532 (create_intersect_range_checks): Move from ...
21533 * tree-data-ref.c (create_intersect_range_checks_index)
21534 (create_intersect_range_checks): ... to here.
21535 (create_runtime_alias_checks): New function factored from ...
21536 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
21537 here. Call above function.
21538 * tree-data-ref.h (create_runtime_alias_checks): New function.
21540 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21542 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
21543 segment length for dr_b and compute it in wide_int.
21545 2017-05-31 Richard Biener <rguenther@suse.de>
21547 PR tree-optimization/80906
21548 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
21549 and pass through iv_map.
21550 (copy_bb_and_scalar_dependences): Adjust.
21551 (translate_pending_phi_nodes): Likewise.
21552 (copy_loop_close_phi_args): Handle code-generating IVs instead
21555 2017-05-30 David Malcolm <dmalcolm@redhat.com>
21557 * diagnostic-color.c (color_dict): Add "type-diff".
21558 (parse_gcc_colors): Update comment.
21559 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
21560 -fdiagnostics-show-template-tree and -fno-elide-type.
21561 (GCC_COLORS): Add type-diff to example.
21563 (-fdiagnostics-show-template-tree): New.
21564 (-fno-elide-type): New.
21565 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
21566 the pp_format_decoder callback. Call any m_format_postprocessor's
21568 (pretty_printer::pretty_printer): Initialize
21569 m_format_postprocessor.
21570 (pretty_printer::~pretty_printer): Delete any
21571 m_format_postprocessor.
21572 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
21573 (class format_postprocessor): New class.
21574 (struct pretty_printer::format_decoder): Document the new parameters.
21575 (struct pretty_printer::m_format_postprocessor): New field.
21576 * tree-diagnostic.c (default_tree_printer): Update for new
21577 bool and const char ** params.
21578 * tree-diagnostic.h (default_tree_printer): Likewise.
21580 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
21582 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
21583 (lwa_operand): Delete rs6000_gen_cell_microcode test.
21584 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
21585 rs6000_gen_cell_microcode code.
21586 (rs6000_final_prescan_insn): Delete.
21587 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
21588 "warn-cell-microcode" entries.
21589 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
21590 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
21591 throughout. Change cc_reg_not_micro_cr0_operand to
21592 cc_reg_not_cr0_operand throughout.
21593 (*extendhi<mode>2_noload): Delete.
21594 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
21595 (mwarn-cell-microcode): Delete.
21596 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
21597 -mgen-cell-microcode and -mwarn-cell-microcode.
21599 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
21602 * config/i386/constraints.md (Yd): New constraint.
21604 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
21605 and (?Yd, r) alternatives. Update insn attributes.
21606 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
21607 and (?*Yd, r) alternatives. Update insn attributes.
21608 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
21610 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
21612 * gimplify.c (gimplify_modify_expr): Don't create a
21613 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
21616 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
21618 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
21620 2017-05-30 Richard Biener <rguenther@suse.de>
21622 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
21623 and reduc_def fields.
21624 (STMT_VINFO_REDUC_TYPE): New define.
21625 (STMT_VINFO_REDUC_DEF): Likewise.
21626 (vect_force_simple_reduction): Adjust prototype.
21627 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
21628 (vect_is_simple_reduction): Remove check_reduction argument.
21629 (vect_force_simple_reduction): Adjust and set
21630 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21631 (vectorizable_reduction): Do not re-do reduction analysis
21632 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21633 * tree-parloops.c (gather_scalar_reductions): Adjust.
21635 2017-05-30 Richard Biener <rguenther@suse.de>
21637 PR middle-end/80901
21638 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
21641 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
21643 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
21644 Introduce unknown_misalignment parameter and remove vf.
21645 (vect_peeling_hash_get_lowest_cost):
21646 Pass unknown_misalignment parameter.
21647 (vect_enhance_data_refs_alignment):
21648 Fix unsupportable data ref treatment.
21650 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21652 * tree-vect-data-refs.c (vect_get_data_access_cost):
21653 Workaround for SLP handling.
21654 (vect_enhance_data_refs_alignment):
21655 Compute costs for doing no peeling at all, compare to the best
21656 peeling costs so far and avoid peeling if cheaper.
21658 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21660 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
21661 Return peeling info and set costs to zero for unlimited cost
21663 (vect_enhance_data_refs_alignment): Also inspect all datarefs
21664 with unknown misalignment. Compute and costs for unknown
21665 misalignment, compare them to the costs for known misalignment
21666 and choose the cheapest for peeling.
21668 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21670 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
21671 (vect_get_peeling_costs_all_drs): Create function.
21672 (vect_peeling_hash_get_lowest_cost):
21673 Use vect_get_peeling_costs_all_drs.
21674 (vect_peeling_supportable): Create function.
21675 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
21677 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21679 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
21680 DR_HAS_NEGATIVE_STEP.
21681 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
21682 (vect_enhance_data_refs_alignment): Use.
21683 (vect_duplicate_ssa_name_ptr_info): Use.
21684 * tree-vectorizer.h (dr_misalignment): Use.
21685 (known_alignment_for_access_p): Use.
21687 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21690 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
21692 (has_section_name): New function.
21694 2017-05-30 Martin Liska <mliska@suse.cz>
21697 * auto-profile.c (get_function_decl_from_block): Fix
21700 2017-05-30 Richard Biener <rguenther@suse.de>
21702 PR middle-end/80876
21703 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
21705 2017-05-30 Martin Liska <mliska@suse.cz>
21707 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
21708 * dumpfile.h (struct dump_file_info): Remove ctors.
21710 2017-05-30 Martin Liska <mliska@suse.cz>
21712 * predict.def: Fix GNU coding style.
21714 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
21716 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
21717 Mark 'to' argument with ATTRIBUTE_UNUSED.
21719 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
21721 * config/xtensa/xtensa.c (xtensa_emit_call): Use
21722 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
21723 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
21726 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
21728 * doc/install.texi (Options specification): Restore entry of
21729 --enable-sjlj-exceptions.
21731 2017-05-27 Michael Eager <eager@eagercon.com>
21734 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
21736 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
21738 * config/microblaze/microblaze.h
21739 (FIXED_REGISTERS): Update in macro.
21740 (CALL_USED_REGISTERS): Update in macro.
21742 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
21744 * doc/install.texi: Add links to macOS binary distributions.
21746 2017-05-27 Jakub Jelinek <jakub@redhat.com>
21750 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21752 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
21754 2017-05-26 Martin Liska <mliska@suse.cz>
21756 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
21758 2017-05-26 Martin Liska <mliska@suse.cz>
21760 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
21761 always leading ';; '.
21762 (dump_bb_info): Likewise.
21763 (brief_dump_cfg): Likewise.
21764 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
21765 * dumpfile.c: Remove usage of TDF_VERBOSE.
21766 * dumpfile.h (enum dump_kind): Likewise.
21767 (dump_gimple_bb_header): Do not use TDF_COMMENT.
21768 * print-tree.c (debug_verbose): Remove.
21769 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
21770 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
21771 * tree-diagnostic.c (default_tree_printer): Replace
21772 TDF_DIAGNOSTIC with TDF_SLIM.
21774 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21776 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
21777 in parameter loop, rather than loop_vinfo.
21778 (create_intersect_range_checks): Ditto.
21779 (vect_create_cond_for_alias_checks): Update call to above functions.
21781 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21783 PR tree-optimization/80815
21784 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
21785 for merging runtime alias checks. Handle negative DR_STEPs.
21787 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21789 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
21791 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
21792 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
21793 out code pruning runtime alias checks.
21794 * tree-data-ref.c (prune_runtime_alias_test_list): New function
21795 factored out from above.
21796 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
21798 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
21800 (prune_runtime_alias_test_list): New decalaration.
21802 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21804 * tree-vect-data-refs.c (compare_tree): Rename and move ...
21805 * tree-data-ref.c (data_ref_compare_tree): ... to here.
21806 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
21807 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
21808 (operator==, comp_dr_with_seg_len_pair): Ditto.
21809 (vect_prune_runtime_alias_test_list): Ditto.
21811 2017-05-26 Martin Liska <mliska@suse.cz>
21814 * params.def: Bound partial-inlining-entry-probability param.
21816 2017-05-26 Marek Polacek <polacek@redhat.com>
21819 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
21822 2017-05-26 Richard Biener <rguenther@suse.de>
21824 PR tree-optimization/80842
21825 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
21828 2017-05-26 Richard Biener <rguenther@suse.de>
21830 PR tree-optimization/80844
21831 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
21833 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
21835 * doc/md.texi (Machine Constraints): Update x86 family
21836 machine constraints section to match 'config/i386/constraints.md'.
21838 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
21840 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
21842 2017-05-25 Nathan Sidwell <nathan@acm.org>
21844 * doc/invoke.texi (--enable-languages): Update documentation.
21846 2017-05-25 Martin Liska <mliska@suse.cz>
21848 * dumpfile.c: Add TDF_FOLDING.
21849 * dumpfile.h (enum dump_kind): Likewise.
21850 * genmatch.c (dt_simplify::gen_1): Use it.
21852 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21854 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
21856 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21858 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
21859 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
21861 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21863 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
21864 * match.pd (X == C): Rewrite it here.
21865 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
21866 with_certain_nonzero_bits2): New predicates.
21867 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
21869 2017-05-24 Nathan Sidwell <nathan@acm.org>
21871 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
21874 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
21877 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
21879 * config/powerpcspe: New port. Files are copied from the rs6000
21880 port, with "rs6000" in filenames replaced by "powerpcspe".
21882 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
21884 PR rtl-optimization/80754
21885 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
21887 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
21889 * config/sparc/sparc.md (length): Return the correct value for -mflat
21890 sibcalls to match output_sibcall.
21892 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
21896 * config/rs6000/rs6000.c (struct machine_function): Add new field
21898 (rs6000_get_separate_components): Init that field, use it.
21899 (rs6000_components_for_bb): Use the field.
21901 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
21903 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
21905 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
21907 PR middle-end/80823
21908 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
21910 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21913 * config/s390/s390.c (s390_check_qrst_address): Check incoming
21914 address against address_operand predicate.
21915 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
21917 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
21919 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
21920 parameters passed indirectly.
21922 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
21924 * config/i386/i386.md (*movdi_internal): Remove SSE4
21925 alternative 18 (?r, *v). Update insn attributes.
21926 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
21927 Update insn attributes.
21928 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
21929 Update insn attributes.
21930 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
21931 alternative 1 (r, v). Remove isa attribute.
21932 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
21933 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
21934 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
21936 2017-05-23 Tom de Vries <tom@codesourcery.com>
21938 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
21941 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21943 * cgraphunit.c (symbol_table::process_new_functions): Update.
21944 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
21945 (inline_generate_summary): Rename to ...
21946 (ipa_fn_summary_generate): ... this one.
21947 (inline_read_summary): Rename to ...
21948 (ipa_fn_summary_read): ... this one.
21949 (inline_write_summary): Rename to ...
21950 (ipa_fn_summary_write): ... this one.
21951 (inline_free_summary): Rename to ...
21952 (ipa_free_fn_summary): ... this one.
21953 (pass_data_local_fn_summary, pass_local_fn_summary,
21954 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
21955 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
21956 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
21957 make_pass_ipa_fn_summary): New.
21958 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
21959 inline_write_summary, inline_free_summary): Remove.
21960 (ipa_free_fn_summary) : New.
21961 * ipa-inline.c (ipa_inline): Update.
21962 (pass_ipa_inline): Do not generate summaries.
21963 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
21965 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
21966 and add pass_ipa_fn_summary.
21967 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
21969 (make_pass_inline_parameters): Remove.
21971 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
21973 * omp-low.c (struct omp_context): Remove "default_kind" member.
21976 * omp-offload.c (execute_oacc_device_lower): Remove the
21977 parallelism dimensions function attributes for unparallelized
21978 OpenACC kernels constructs.
21980 2017-05-23 Martin Liska <mliska@suse.cz>
21982 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
21984 (cgraph_edge::make_speculative): Likewise.
21985 (cgraph_edge::resolve_speculation): Likewise.
21986 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
21987 (cgraph_node::dump): Likewise.
21988 * cgraph.h: Likewise.
21989 * cgraphunit.c (analyze_functions): Likewise.
21990 (symbol_table::compile): Likewise.
21991 * ipa-cp.c (print_all_lattices): Likewise.
21992 (determine_versionability): Likewise.
21993 (initialize_node_lattices): Likewise.
21994 (ipcp_verify_propagated_values): Likewise.
21995 (estimate_local_effects): Likewise.
21996 (update_profiling_info): Likewise.
21997 (create_specialized_node): Likewise.
21998 (perhaps_add_new_callers): Likewise.
21999 (decide_about_value): Likewise.
22000 (decide_whether_version_node): Likewise.
22001 (identify_dead_nodes): Likewise.
22002 (ipcp_store_bits_results): Likewise.
22003 * ipa-devirt.c (dump_targets): Likewise.
22004 (ipa_devirt): Likewise.
22005 * ipa-icf.c (sem_item::dump): Likewise.
22006 (sem_function::equals): Likewise.
22007 (sem_variable::equals): Likewise.
22008 (sem_item_optimizer::read_section): Likewise.
22009 (sem_item_optimizer::execute): Likewise.
22010 (congruence_class::dump): Likewise.
22011 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
22012 (dump_inline_summary): Likewise.
22013 (estimate_node_size_and_time): Likewise.
22014 (inline_analyze_function): Likewise.
22015 * ipa-inline-transform.c (inline_call): Likewise.
22016 * ipa-inline.c (report_inline_failed_reason): Likewise.
22017 (want_early_inline_function_p): Likewise.
22018 (edge_badness): Likewise.
22019 (update_edge_key): Likewise.
22020 (inline_small_functions): Likewise.
22021 * ipa-profile.c (ipa_profile): Likewise.
22022 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
22023 (ipa_make_edge_direct_to_target): Likewise.
22024 (remove_described_reference): Likewise.
22025 (ipa_impossible_devirt_target): Likewise.
22026 (propagate_controlled_uses): Likewise.
22027 (ipa_print_node_params): Likewise.
22028 (ipcp_transform_function): Likewise.
22029 * ipa-pure-const.c (pure_const_read_summary): Likewise.
22030 (propagate_pure_const): Likewise.
22031 * ipa-reference.c (generate_summary): Likewise.
22032 (read_write_all_from_decl): Likewise.
22033 (propagate): Likewise.
22034 (ipa_reference_read_optimization_summary): Likewise.
22035 * ipa-utils.c (ipa_merge_profiles): Likewise.
22036 * ipa.c (walk_polymorphic_call_targets): Likewise.
22037 (symbol_table::remove_unreachable_nodes): Likewise.
22038 (ipa_single_use): Likewise.
22039 * passes.c (execute_todo): Likewise.
22040 * predict.c (drop_profile): Likewise.
22041 * symtab.c (symtab_node::get_dump_name): New function.
22042 (symtab_node::dump_name): Likewise.
22043 (symtab_node::dump_asm_name): Likewise.
22044 (symtab_node::dump_references): Likewise.
22045 (symtab_node::dump_referring): Likewise.
22046 (symtab_node::dump_base): Likewise.
22047 (symtab_node::debug_symtab): Likewise.
22048 * tree-sra.c (convert_callers_for_node): Likewise.
22049 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
22050 * value-prof.c (init_node_map): Likewise.
22052 2017-05-23 Martin Liska <mliska@suse.cz>
22054 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
22055 and symtab_node::debug_symtab to symbol_table::debug.
22056 * cgraphunit.c (analyze_functions): Use the renamed function.
22057 (symbol_table::compile): Likewise.
22058 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
22059 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
22060 * passes.c (execute_todo): Likewise.
22061 * symtab.c (symbol_table::dump): New function.
22062 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
22064 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22066 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
22067 that nonconst implies exec.
22069 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22071 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22072 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22073 (inline_edge_summary_vec): Turn into ...
22074 (ipa_call_summaries): ... this one.
22075 (redirect_to_unreachable, edge_set_predicate,
22076 evaluate_properties_for_edge, inline_summary_alloc,
22077 reset_ipa_call_summary, reset_inline_summary,
22078 inline_summary_t::duplicate): Update.
22079 (inline_edge_duplication_hook): Turn to ...
22080 (ipa_call_summary_t::duplicate): ... this one.
22081 (inline_edge_removal_hook): Turn to ...
22082 (ipa_call_summary_t::remove): ... this one.
22083 (dump_inline_edge_summary): Turn to ...
22084 (dump_ipa_call_summary): ... this one.
22085 (estimate_function_body_sizes): Update.
22086 (inline_update_callee_summaries): Update.
22087 (remap_edge_change_prob): Update.
22088 (remap_edge_summaries): Update.
22089 (inline_merge_summary): Update.
22090 (do_estimate_edge_time): Update.
22091 (inline_generate_summary): Update.
22092 (inline_read_section): Update.
22093 (inline_read_summary): Update.
22094 (inline_free_summary): Update.
22095 * ipa-inline.c (can_inline_edge_p): Update.
22096 (compute_inlined_call_time): Update.
22097 (want_inline_small_function_p): Update.
22098 (edge_badness): Update.
22099 (early_inliner): Update.
22100 * ipa-inline.h (inline_edge_summary): Turn to ...
22101 (ipa_call_summary): ... this one.
22102 (ipa_call_summary_t): New class.
22103 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22104 (ipa_call_summaries): New.
22105 (inline_edge_summary): Remove.
22106 (estimate_edge_growth): Update.
22107 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22108 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22109 * ipa-split.c (execute_split_functions): Update.
22110 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22112 2017-05-23 Tom de Vries <tom@codesourcery.com>
22114 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22115 attributes): Document rdrand effective target.
22117 2017-05-23 Tom de Vries <tom@codesourcery.com>
22119 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22120 attributes): Sort alphabetically.
22122 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
22124 * config/avr/genmultilib.awk: Use gsub instead of gensub.
22126 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
22129 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
22130 V2DF/V2DI splat into two separate patterns, one that handles
22131 registers, and the other that only handles memory. Drop support
22132 for splatting from a GPR on ISA 2.07 and then splitting the
22133 splat into direct move and splat.
22134 (vsx_splat_<mode>_reg): Likewise.
22135 (vsx_splat_<mode>_mem): Likewise.
22137 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
22139 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
22141 2017-05-22 Jakub Jelinek <jakub@redhat.com>
22143 PR middle-end/80809
22144 * omp-low.c (finish_taskreg_remap): New function.
22145 (finish_taskreg_scan): If unit size of ctx->record_type
22146 is non-constant, unshare the size expression and replace
22147 decls in it with possible outer var refs.
22149 PR middle-end/80809
22150 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
22151 GOVD_SHARED rather than GOVD_PRIVATE with it.
22152 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
22153 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
22155 PR middle-end/80853
22156 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
22157 as last argument to build_outer_var_ref for pointer bases of array
22158 section reductions.
22160 2017-05-19 Martin Sebor <msebor@redhat.com>
22162 * print-tree.c (print_node): Print DECL_READ_P flag.
22164 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22166 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
22167 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
22168 * cgraph.c: Likewise.
22169 * cgraphunit.c: Likewise.
22170 * gengtype.c: Likewise.
22171 * ipa-cp.c: Likewise.
22172 * ipa-devirt.c: Likewise.
22173 * ipa-icf.c: Likewise.
22174 * ipa-predicate.c: Likewise.
22175 * ipa-profile.c: Likewise.
22176 * ipa-prop.c: Likewise.
22177 * ipa-split.c: Likewise.
22179 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
22180 edge_predicate_pool, dump_inline_hints,
22181 inline_summary::account_size_time, redirect_to_unreachable,
22182 edge_set_predicate, set_hint_predicate,
22183 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
22184 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
22185 inline_summary_t::remove, remap_hint_predicate_after_duplication,
22186 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
22187 ipa_call_summary_t::remove, initialize_growth_caches,
22188 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
22189 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
22190 mark_modified, unmodified_parm_1, unmodified_parm,
22191 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
22192 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
22193 compute_bb_predicates, will_be_nonconstant_expr_predicate,
22194 will_be_nonconstant_predicate, record_modified_bb_info,
22195 get_minimal_bb, record_modified, param_change_prob,
22196 phi_result_unknown_predicate, predicate_for_phi_result,
22197 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
22198 estimate_function_body_sizes, compute_inline_parameters,
22199 compute_inline_parameters_for_curren, pass_data_inline_parameters,
22200 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
22201 inline_update_callee_summaries, remap_edge_change_prob,
22202 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
22203 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
22204 inline_analyze_function, inline_summary_t::insert,
22205 inline_generate_summary, read_ipa_call_summary, inline_read_section,
22206 inline_read_summary, write_ipa_call_summary, inline_write_summary,
22207 inline_free_summary): Move to ipa-fnsummary.h
22208 (predicate_t): Remove.
22209 * ipa-fnsummary.c: New file.
22210 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
22211 (enum inline_hints_vals, inline_hints, agg_position_info,
22212 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
22213 inline_summaries, ipa_call_summary, ipa_call_summary_t,
22214 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
22215 dump_inline_summary, dump_inline_hints, inline_generate_summary,
22216 inline_read_summary, inline_write_summary, inline_free_summary,
22217 inline_analyze_function, initialize_inline_failed,
22218 inline_merge_summary, inline_update_overall_summary,
22219 compute_inline_parameters): Move to ipa-fnsummary.h
22220 * ipa-fnsummary.h: New file.
22221 * ipa-inline-transform.h: Include ipa-inline.h.
22222 * ipa-inline.c: LIkewise.
22224 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22226 * ipa-inline.c (edge_badness): Use inlined_time instead of
22227 inline_summaries->get.
22229 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22231 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
22233 2017-05-22 Nathan Sidwell <nathan@acm.org>
22235 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
22236 (fdump-lang): Document 'raw' option.
22237 * dumpfile.h (TDI_tu): Delete.
22238 * dumpfile.c (dump_files): Remove translation-unit.
22239 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
22241 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
22243 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
22244 command option from $(AWK) call.
22245 * config/avr/genmultilib.awk: Simplify and rewrite so that it
22246 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
22247 [FORMAT]: Remove handling of variable.
22248 * config/avr/t-multilib: Regenerate.
22250 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22252 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
22254 (dump_inline_summary): Do not print self_time.
22255 (estimate_function_body_sizes): Do not set self_time.
22256 (compute_inline_parameters): Likewise.
22257 (inline_read_section, inline_write_summary): Do not stream self_time.
22258 * ipa-inline.h (inline_summary): Drop self_time.
22260 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22262 * ipa-inline-analysis.c (account_size_time): Rename to ...
22263 (inline_summary::account_size_time): ... this one.
22264 (reset_ipa_call_summary): Turn to ...
22265 (ipa_call_summary::reset): ... this one.
22266 (reset_inline_summary): Turn to ...
22267 (inline_summary::reset): ... this one.
22268 (inline_summary_t::remove): Update.
22269 (inline_summary_t::duplicate): Update.
22270 (ipa_call_summary_t::remove): Update.
22271 (dump_inline_summary): Update.
22272 (estimate_function_body_sizes): Update.
22273 (compute_inline_parameters): Update.
22274 (estimate_node_size_and_time): Update.
22275 (inline_merge_summary): Update.
22276 (inline_update_overall_summary): Update.
22277 (inline_read_section): Update.
22278 (inline_write_summary): Update.
22279 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
22280 add account_size_time and reset member functions.
22281 (ipa_call_summary): Add reset function.
22282 * ipa-predicate.h (predicate::operator &): Constify.
22284 2017-05-22 Richard Biener <rguenther@suse.de>
22286 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
22288 2017-05-19 Jason Merrill <jason@redhat.com>
22290 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
22292 2017-05-19 Marek Polacek <polacek@redhat.com>
22295 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
22296 TYPE_OVERFLOW_WRAPS checks.
22298 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
22300 * tree-core.h (enum omp_clause_default_kind): Add
22301 "OMP_CLAUSE_DEFAULT_PRESENT".
22302 * tree-pretty-print.c (dump_omp_clause): Handle it.
22303 * gimplify.c (enum gimplify_omp_var_data): Add
22304 "GOVD_MAP_FORCE_PRESENT".
22305 (gimplify_adjust_omp_clauses_1): Map it to
22306 "GOMP_MAP_FORCE_PRESENT".
22307 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
22309 * gimplify.c (oacc_default_clause): Clarify.
22311 2017-05-19 Nathan Sidwell <nathan@acm.org>
22313 LANG_HOOK_REGISTER_DUMPS
22314 * toplev.c (general_init): Call register dump lang hook.
22315 * doc/invoke.texi: Document -fdump-lang option family.
22316 * dumpfile.c (dump_files): Remove class dump here.
22317 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
22318 * dumpfile.h (tree_dump_index): Remove TDI_class.
22319 * langhooks-def.h (lhd_register_dumps): Declare.
22320 (LANG_HOOKS_REGISTER_DUMPS): Define.
22321 (LANG_HOOKS_INITIALIZER): Add it.
22322 * langhooks.c (lhd_register_dumps): Define.
22323 * langhooks.h (struct lang_hooks): Add register_dumps.
22325 2017-05-19 Nathan Sidwell <nathan@acm.org>
22327 * context.h (context::set_passes): New.
22328 * context.c (context::context): Do not create pass manager.
22329 * toplev.c (general_init): Create pass manager here.
22331 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
22333 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
22334 use this splitter if two add or or instructions would also work for
22335 the constant we want to generate.
22337 2017-05-19 Richard Biener <rguenther@suse.de>
22340 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
22341 predicate evaluation.
22343 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
22345 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
22347 * ipa-inline.c (want_inline_small_function_p): Do not cast to
22350 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
22352 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22353 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22354 (inline_edge_summary_vec): Turn into ...
22355 (ipa_call_summaries): ... this one.
22356 (redirect_to_unreachable, edge_set_predicate,
22357 evaluate_properties_for_edge, inline_summary_alloc,
22358 reset_ipa_call_summary, reset_inline_summary,
22359 inline_summary_t::duplicate): Update.
22360 (inline_edge_duplication_hook): Turn to ...
22361 (ipa_call_summary_t::duplicate): ... this one.
22362 (inline_edge_removal_hook): Turn to ...
22363 (ipa_call_summary_t::remove): ... this one.
22364 (dump_inline_edge_summary): Turn to ...
22365 (dump_ipa_call_summary): ... this one.
22366 (estimate_function_body_sizes): Update.
22367 (inline_update_callee_summaries): Update.
22368 (remap_edge_change_prob): Update.
22369 (remap_edge_summaries): Update.
22370 (inline_merge_summary): Update.
22371 (do_estimate_edge_time): Update.
22372 (inline_generate_summary): Update.
22373 (inline_read_section): Update.
22374 (inline_read_summary): Update.
22375 (inline_free_summary): Update.
22376 * ipa-inline.c (can_inline_edge_p): Update.
22377 (compute_inlined_call_time): Update.
22378 (want_inline_small_function_p): Update.
22379 (edge_badness): Update.
22380 (early_inliner): Update.
22381 * ipa-inline.h (inline_edge_summary): Turn to ...
22382 (ipa_call_summary): ... this one.
22383 (ipa_call_summary_t): New class.
22384 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22385 (ipa_call_summaries): New.
22386 (inline_edge_summary): Remove.
22387 (estimate_edge_growth): Update.
22388 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22389 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22390 * ipa-split.c (execute_split_functions): Update.
22391 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22393 2017-05-19 Richard Biener <rguenther@suse.de>
22395 PR middle-end/80764
22396 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
22398 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
22400 * config/rs6000/rs6000.c (struct machine_function): Add field
22401 fpr_is_wrapped_separately.
22402 (rs6000_get_separate_components): Use 64 components. Handle the
22403 new FPR components.
22404 (rs6000_components_for_bb): Handle the FPR components.
22405 (rs6000_emit_prologue_components): Handle the FPR components.
22406 (rs6000_emit_epilogue_components): Handle the FPR components.
22407 (rs6000_set_handled_components): Handle the FPR components.
22408 (rs6000_emit_prologue): Don't output prologue code for those FPRs
22409 that are already separately shrink-wrapped.
22410 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
22411 that are already separately shrink-wrapped.
22413 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
22416 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
22419 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
22420 (define_peephole2 for Altivec d-form load): Add peepholes to catch
22421 cases where the register allocator uses a move and an offsettable
22422 memory operation to/from a FPR register on ISA 2.06/2.07.
22423 (define_peephole2 for Altivec d-form store): Likewise.
22425 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
22428 * config/i386/mmx.md (*mov<mode>_internal): Enable
22429 alternatives 11, 12, 13 and 14 also for 32bit targets.
22430 Remove alternatives 15, 16, 17 and 18.
22431 * config/i386/sse.md (vec_concatv2di): Change
22432 alternative (!x, *y) to (x, ?!*Yn).
22434 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
22436 * dumpfile.h (enum dump_kind): Remove stray comma.
22438 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
22440 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
22441 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
22442 predicate::num_conditions
22443 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
22444 (CHANGED): turn into predicate::changed.
22445 (agg_position_info): Move to ipa-predicate.h
22446 (add_condition, predicate::add_clause, predicate::operator &=,
22447 predicate::or_with, predicate::evaluate, predicate::probability,
22448 dump_condition, dump_clause, predicate::dump,
22449 predicate::remap_after_duplication, predicate::remap_after_inlining,
22450 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
22451 (evaluate_conditions_for_known_args): Update.
22452 (set_cond_stmt_execution_predicate): Update.
22453 * ipa-inline.h: Include ipa-predicate.h
22454 (condition, inline_param_summary, conditions, agg_position_info,
22455 predicate): Move to ipa-predicate.h
22456 * ipa-predicate.c: New file.
22457 * ipa-predicate.h: New file.
22459 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
22461 * final.c (leaf_function_p): Check we are not in a sequence.
22463 2017-05-18 Martin Liska <mliska@suse.cz>
22465 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
22466 * dumpfile.c (dump_register): Use new enum dump_kind.
22467 (get_dump_file_name): Likewise.
22468 (dump_enable_all): Likewise.
22469 (dump_switch_p_1): Likewise.
22470 (enable_rtl_dump_file): Remove usage of TDF_RTL.
22471 * dumpfile.h (enum dump_kind): New enum type.
22472 (struct dump_file_info): Create constructor and
22473 format fields and comments.
22474 * passes.c (pass_manager::register_one_dump_file):
22476 * statistics.c (statistics_early_init): Likewise.
22477 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
22478 TDF_TREE with TDF_SLIM.
22479 (gather_memory_references_ref): Likewise.
22481 2017-05-18 Martin Liska <mliska@suse.cz>
22483 * vec.h (struct vnull): Use it.
22485 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
22487 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
22488 (true_predicate, false_predicate, true_predicate_p,
22489 false_predicate_p): Remove.
22490 (single_cond_predicate, not_inlined_predicate): Turn to member function
22492 (add_condition): Update.
22493 (add_clause): Turn to...
22494 (predicate::add_clause): ... this one; update; allow passing NULL
22496 (and_predicates): Turn to ...
22497 (predicate::operator &=): ... this one.
22498 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
22499 (or_predicates): Turn to ...
22500 (predicate::or_with): ... this one.
22501 (evaluate_predicate): Turn to ...
22502 (predicate::evaluate): ... this one.
22503 (predicate_probability): Turn to ...
22504 (predicate::probability): ... this one.
22505 (dump_condition): Update.
22506 (dump_predicate): Turn to ...
22507 (predicate::dump): ... this one.
22508 (account_size_time): Update.
22509 (edge_set_predicate): Update.
22510 (set_hint_predicate): UPdate.
22511 (evaluate_conditions_for_known_args): Update.
22512 (evaluate_properties_for_edge): Update.
22513 (remap_predicate_after_duplication): Turn to...
22514 (predicate::remap_after_duplication): ... this one.
22515 (remap_hint_predicate_after_duplication): Update.
22516 (inline_summary_t::duplicate): UPdate.
22517 (dump_inline_edge_summary): Update.
22518 (dump_inline_summary): Update.
22519 (set_cond_stmt_execution_predicate): Update.
22520 (set_switch_stmt_execution_predicate): Update.
22521 (compute_bb_predicates): Update.
22522 (will_be_nonconstant_expr_predicate): Update.
22523 (will_be_nonconstant_predicate): Update.
22524 (phi_result_unknown_predicate): Update.
22525 (predicate_for_phi_result): Update.
22526 (array_index_predicate): Update.
22527 (estimate_function_body_sizes): Update.
22528 (estimate_node_size_and_time): Update.
22529 (estimate_ipcp_clone_size_and_time): Update.
22530 (remap_predicate): Rename to ...
22531 (predicate::remap_after_inlining): ... this one.
22532 (remap_hint_predicate): Update.
22533 (inline_merge_summary): Update.
22534 (inline_update_overall_summary): Update.
22535 (estimate_size_after_inlining): Update.
22536 (read_predicate): Rename to ...
22537 (predicate::stream_in): ... this one.
22538 (read_inline_edge_summary): Update.
22539 (write_predicate): Rename to ...
22540 (predicate::stream_out): ... this one.
22541 (write_inline_edge_summary): Update.
22542 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
22543 (clause_t): Turn to uint32_t
22544 (predicate): Turn to class; implement constructor and operators
22546 (size_time_entry): Update.
22547 (inline_summary): Update.
22548 (inline_edge_summary): Update.
22550 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
22552 * fold-const.c (fold_binary_loc): Move transformation...
22553 * match.pd (C - X CMP X): ... here.
22555 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
22557 * config/sparc/sparc.c (sparc_option_override): Set function
22558 alignment for -mcpu=niagara7 to 64 to match the I$ line.
22559 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
22561 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
22563 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
22565 2017-05-18 Marek Polacek <polacek@redhat.com>
22568 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
22569 (pass_ubsan::execute): Call gimple_assign_single_p instead of
22570 gimple_assign_load_p.
22572 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
22574 PR middle-end/80692
22575 * real.c (do_compare): Give decimal_do_compare preference over
22576 comparing just the signs.
22578 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
22580 * doc/md.texi (Canonicalization of Instructions): Describe the
22581 canonical form of instructions that inherently set a condition
22584 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
22586 PR middle-end/80775
22587 * tree-cfg.c: Move deletion of unreachable case statements to after
22588 the merging of consecutive case labels.
22590 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
22592 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
22593 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
22594 restoring of callee-saved registers.
22596 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
22598 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
22599 * config/visium/visium.c (single_set_and_flags): Likewise.
22600 * config/visium/visium.md (Substitutions): Likewise.
22602 2017-05-17 Martin Liska <mliska@suse.cz>
22604 * cfg.c: Introduce dump_flags_t type and
22605 use it instead of int type.
22607 * cfghooks.c: Likewise.
22608 * cfghooks.h (struct cfg_hooks): Likewise.
22609 * cfgrtl.c: Likewise.
22610 * cfgrtl.h: Likewise.
22611 * cgraph.c (cgraph_node::get_body): Likewise.
22612 * coretypes.h: Likewise.
22613 * domwalk.c: Likewise.
22614 * domwalk.h: Likewise.
22615 * dumpfile.c (struct dump_option_value_info): Likewise.
22616 (dump_enable_all): Likewise.
22617 (dump_switch_p_1): Likewise.
22618 (opt_info_switch_p): Likewise.
22619 * dumpfile.h (enum tree_dump_index): Likewise.
22620 (struct dump_file_info): Likewise.
22621 * genemit.c: Likewise.
22622 * generic-match-head.c: Likewise.
22623 * gengtype.c (open_base_files): Likewise.
22624 * gimple-pretty-print.c: Likewise.
22625 * gimple-pretty-print.h: Likewise.
22626 * graph.c (print_graph_cfg): Likewise.
22627 * graphite-scop-detection.c (dot_all_sese): Likewise.
22628 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
22629 * loop-unroll.c (report_unroll): Likewise.
22630 * passes.c (pass_manager::register_one_dump_file): Likewise.
22631 * print-tree.c: Likewise.
22632 * statistics.c: Likewise.
22633 * tree-cfg.c: Likewise.
22634 * tree-cfg.h: Likewise.
22635 * tree-dfa.c: Likewise.
22636 * tree-dfa.h: Likewise.
22637 * tree-dump.c (dump_function): Likewise.
22638 * tree-dump.h (struct dump_info): Likewise.
22639 * tree-pretty-print.c: Likewise.
22640 * tree-pretty-print.h: Likewise.
22641 * tree-ssa-live.c: Likewise.
22642 * tree-ssa-live.h: Likewise.
22643 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
22644 * tree-vect-loop.c: Likewise.
22645 * tree-vect-slp.c: Likewise.
22647 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
22648 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22650 PR tree-optimization/80457
22651 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
22652 of all arguments to a statement as scalar_to_vec operations.
22653 (vectorizable_call): Adjust call to vect_model_simple_cost for
22655 (vectorizable_conversion): Likewise.
22656 (vectorizable_assignment): Likewise.
22657 (vectorizable_shift): Likewise.
22658 (vectorizable_operation): Likewise.
22659 (vectorizable_comparison): Likewise.
22660 (vect_is_simple_cond): Record the def types for operands.
22661 (vectorizable_condition): Likewise, call vect_model_simple_cost.
22662 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
22663 for statement argument count.
22665 2017-05-16 Carl Love <cel@us.ibm.com>
22667 * config/rs6000/rs6000-c: Add support for built-in functions
22668 vector unsigned long long vec_bperm (vector unsigned long long,
22669 vector unsigned char)
22670 vector signed long long vec_mule (vector signed int,
22672 vector unsigned long long vec_mule (vector unsigned int,
22673 vector unsigned int)
22674 vector signed long long vec_mulo (vector signed int,
22676 vector unsigned long long vec_mulo (vector unsigned int,
22677 vector unsigned int)
22678 vector signed char vec_sldw (vector signed char,
22679 vector signed char,
22681 vector unsigned char vec_sldw (vector unsigned char,
22682 vector unsigned char,
22684 vector signed short vec_sldw (vector signed short,
22685 vector signed short,
22687 vector unsigned short vec_sldw (vector unsigned short,
22688 vector unsigned short,
22690 vector signed int vec_sldw (vector signed int,
22693 vector unsigned int vec_sldw (vector unsigned int,
22694 vector unsigned int,
22696 vector signed long long vec_sldw (vector signed long long,
22697 vector signed long long,
22699 vector unsigned long long vec_sldw (vector unsigned long long,
22700 vector unsigned long long,
22702 * config/rs6000/rs6000-c: Add support for built-in functions
22703 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
22704 * config/rs6000/altivec.h: Add defintion for vec_sldw.
22705 * doc/extend.texi: Update the built-in documentation for the
22706 new built-in functions.
22708 2017-05-16 Marek Polacek <polacek@redhat.com>
22712 * tree.c (save_expr): Don't fold the expression.
22714 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
22716 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
22717 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
22718 and (?*y,m). Update insn attributes.
22720 2017-05-16 Martin Liska <mliska@suse.cz>
22722 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
22723 flags argument of print_gimple_stmt, print_gimple_expr,
22724 print_generic_stmt and print_generic_expr.
22725 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
22726 * coretypes.h: Likewise.
22727 * except.c (dump_eh_tree): Likewise.
22728 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
22729 * gimple-pretty-print.h: Likewise.
22730 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
22731 (backprop::push_to_worklist): Likewise.
22732 (backprop::pop_from_worklist): Likewise.
22733 (backprop::process_use): Likewise.
22734 (backprop::intersect_uses): Likewise.
22735 (note_replacement): Likewise.
22736 * gimple-ssa-store-merging.c
22737 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
22738 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
22739 (pass_store_merging::execute): Likewise.
22740 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
22741 (ssa_base_cand_dump_callback): Likewise.
22742 (dump_incr_vec): Likewise.
22743 (replace_refs): Likewise.
22744 (replace_mult_candidate): Likewise.
22745 (create_add_on_incoming_edge): Likewise.
22746 (create_phi_basis): Likewise.
22747 (insert_initializers): Likewise.
22748 (all_phi_incrs_profitable): Likewise.
22749 (introduce_cast_before_cand): Likewise.
22750 (replace_one_candidate): Likewise.
22751 * gimplify.c (gimplify_expr): Likewise.
22752 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
22753 (set_rename): Likewise.
22754 (rename_uses): Likewise.
22755 (copy_loop_phi_nodes): Likewise.
22756 (add_close_phis_to_merge_points): Likewise.
22757 (copy_loop_close_phi_args): Likewise.
22758 (copy_cond_phi_args): Likewise.
22759 (graphite_copy_stmts_from_block): Likewise.
22760 (translate_pending_phi_nodes): Likewise.
22761 * graphite-poly.c (print_pdr): Likewise.
22762 (dump_gbb_cases): Likewise.
22763 (dump_gbb_conditions): Likewise.
22764 (print_scop_params): Likewise.
22765 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
22766 (build_cross_bb_scalars_use): Likewise.
22767 (gather_bbs::before_dom_children): Likewise.
22768 * hsa-dump.c (dump_hsa_immed): Likewise.
22769 * ipa-cp.c (print_ipcp_constant_value): Likewise.
22770 (get_replacement_map): Likewise.
22771 * ipa-inline-analysis.c (dump_condition): Likewise.
22772 (estimate_function_body_sizes): Likewise.
22773 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
22774 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
22775 * ipa-prop.c (ipa_dump_param): Likewise.
22776 (ipa_print_node_jump_functions_for_edge): Likewise.
22777 (ipa_modify_call_arguments): Likewise.
22778 (ipa_modify_expr): Likewise.
22779 (ipa_dump_param_adjustments): Likewise.
22780 (ipa_dump_agg_replacement_values): Likewise.
22781 (ipcp_modif_dom_walker::before_dom_children): Likewise.
22782 * ipa-pure-const.c (check_stmt): Likewise.
22783 (pass_nothrow::execute): Likewise.
22784 * ipa-split.c (execute_split_functions): Likewise.
22785 * omp-offload.c (dump_oacc_loop_part): Likewise.
22786 (dump_oacc_loop): Likewise.
22787 * trans-mem.c (tm_log_emit): Likewise.
22788 (tm_memopt_accumulate_memops): Likewise.
22789 (dump_tm_memopt_set): Likewise.
22790 (dump_tm_memopt_transform): Likewise.
22791 * tree-cfg.c (gimple_verify_flow_info): Likewise.
22792 (print_loop): Likewise.
22793 * tree-chkp-opt.c (chkp_print_addr): Likewise.
22794 (chkp_gather_checks_info): Likewise.
22795 (chkp_get_check_result): Likewise.
22796 (chkp_remove_check_if_pass): Likewise.
22797 (chkp_use_outer_bounds_if_possible): Likewise.
22798 (chkp_reduce_bounds_lifetime): Likewise.
22799 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
22800 (chkp_mark_completed_bounds): Likewise.
22801 (chkp_register_incomplete_bounds): Likewise.
22802 (chkp_mark_invalid_bounds): Likewise.
22803 (chkp_maybe_copy_and_register_bounds): Likewise.
22804 (chkp_build_returned_bound): Likewise.
22805 (chkp_get_bound_for_parm): Likewise.
22806 (chkp_build_bndldx): Likewise.
22807 (chkp_get_bounds_by_definition): Likewise.
22808 (chkp_generate_extern_var_bounds): Likewise.
22809 (chkp_get_bounds_for_decl_addr): Likewise.
22810 * tree-chrec.c (chrec_apply): Likewise.
22811 * tree-data-ref.c (dump_data_reference): Likewise.
22812 (dump_subscript): Likewise.
22813 (dump_data_dependence_relation): Likewise.
22814 (analyze_overlapping_iterations): Likewise.
22815 * tree-inline.c (expand_call_inline): Likewise.
22816 (tree_function_versioning): Likewise.
22817 * tree-into-ssa.c (dump_defs_stack): Likewise.
22818 (dump_currdefs): Likewise.
22819 (dump_names_replaced_by): Likewise.
22820 (dump_update_ssa): Likewise.
22821 (update_ssa): Likewise.
22822 * tree-object-size.c (pass_object_sizes::execute): Likewise.
22823 * tree-parloops.c (build_new_reduction): Likewise.
22824 (try_create_reduction_list): Likewise.
22825 (ref_conflicts_with_region): Likewise.
22826 (oacc_entry_exit_ok_1): Likewise.
22827 (oacc_entry_exit_single_gang): Likewise.
22828 * tree-pretty-print.h: Likewise.
22829 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
22830 (get_scalar_evolution): Likewise.
22831 (add_to_evolution): Likewise.
22832 (get_loop_exit_condition): Likewise.
22833 (analyze_evolution_in_loop): Likewise.
22834 (analyze_initial_condition): Likewise.
22835 (analyze_scalar_evolution): Likewise.
22836 (instantiate_scev): Likewise.
22837 (number_of_latch_executions): Likewise.
22838 (gather_chrec_stats): Likewise.
22839 (final_value_replacement_loop): Likewise.
22840 (scev_const_prop): Likewise.
22841 * tree-sra.c (dump_access): Likewise.
22842 (disqualify_candidate): Likewise.
22843 (create_access): Likewise.
22844 (reject): Likewise.
22845 (maybe_add_sra_candidate): Likewise.
22846 (create_access_replacement): Likewise.
22847 (analyze_access_subtree): Likewise.
22848 (analyze_all_variable_accesses): Likewise.
22849 (sra_modify_assign): Likewise.
22850 (initialize_constant_pool_replacements): Likewise.
22851 (find_param_candidates): Likewise.
22852 (decide_one_param_reduction): Likewise.
22853 (replace_removed_params_ssa_names): Likewise.
22854 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
22855 * tree-ssa-copy.c (dump_copy_of): Likewise.
22856 (copy_prop_visit_cond_stmt): Likewise.
22857 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
22858 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
22859 (record_equivalences_from_stmt): Likewise.
22860 * tree-ssa-dse.c (compute_trims): Likewise.
22861 (delete_dead_call): Likewise.
22862 (delete_dead_assignment): Likewise.
22863 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
22864 (forward_propagate_into_cond): Likewise.
22865 (pass_forwprop::execute): Likewise.
22866 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
22867 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
22869 (move_computations_worker): Likewise.
22870 (execute_sm): Likewise.
22871 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
22872 (remove_exits_and_undefined_stmts): Likewise.
22873 (remove_redundant_iv_tests): Likewise.
22874 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
22875 (adjust_iv_update_pos): Likewise.
22876 * tree-ssa-math-opts.c (bswap_replace): Likewise.
22877 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
22878 (value_replacement): Likewise.
22879 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
22880 * tree-ssa-pre.c (print_pre_expr): Likewise.
22881 (get_representative_for): Likewise.
22882 (create_expression_by_pieces): Likewise.
22883 (insert_into_preds_of_block): Likewise.
22884 (eliminate_insert): Likewise.
22885 (eliminate_dom_walker::before_dom_children): Likewise.
22886 (eliminate): Likewise.
22887 (remove_dead_inserted_code): Likewise.
22888 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
22889 * tree-ssa-reassoc.c (get_rank): Likewise.
22890 (eliminate_duplicate_pair): Likewise.
22891 (eliminate_plus_minus_pair): Likewise.
22892 (eliminate_not_pairs): Likewise.
22893 (undistribute_ops_list): Likewise.
22894 (eliminate_redundant_comparison): Likewise.
22895 (update_range_test): Likewise.
22896 (optimize_range_tests_var_bound): Likewise.
22897 (optimize_vec_cond_expr): Likewise.
22898 (rewrite_expr_tree): Likewise.
22899 (rewrite_expr_tree_parallel): Likewise.
22900 (linearize_expr): Likewise.
22901 (break_up_subtract): Likewise.
22902 (linearize_expr_tree): Likewise.
22903 (attempt_builtin_powi): Likewise.
22904 (attempt_builtin_copysign): Likewise.
22905 (transform_stmt_to_copy): Likewise.
22906 (transform_stmt_to_multiply): Likewise.
22907 (dump_ops_vector): Likewise.
22908 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
22909 (print_scc): Likewise.
22910 (set_ssa_val_to): Likewise.
22911 (visit_reference_op_store): Likewise.
22912 (visit_use): Likewise.
22913 (sccvn_dom_walker::before_dom_children): Likewise.
22914 (run_scc_vn): Likewise.
22915 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
22917 (expr_hash_elt::print): Likewise.
22918 (const_and_copies::pop_to_marker): Likewise.
22919 (const_and_copies::record_const_or_copy_raw): Likewise.
22920 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
22921 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
22922 (dump_predicates): Likewise.
22923 (find_uninit_use): Likewise.
22924 (warn_uninitialized_phi): Likewise.
22925 (pass_late_warn_uninitialized::execute): Likewise.
22926 * tree-ssa.c (verify_vssa): Likewise.
22927 (verify_ssa): Likewise.
22928 (maybe_optimize_var): Likewise.
22929 * tree-vrp.c (dump_value_range): Likewise.
22930 (dump_all_value_ranges): Likewise.
22931 (dump_asserts_for): Likewise.
22932 (register_edge_assert_for_2): Likewise.
22933 (vrp_visit_cond_stmt): Likewise.
22934 (vrp_visit_switch_stmt): Likewise.
22935 (vrp_visit_stmt): Likewise.
22936 (vrp_visit_phi_node): Likewise.
22937 (simplify_cond_using_ranges_1): Likewise.
22938 (fold_predicate_in): Likewise.
22939 (evrp_dom_walker::before_dom_children): Likewise.
22940 (evrp_dom_walker::push_value_range): Likewise.
22941 (evrp_dom_walker::pop_value_range): Likewise.
22942 (execute_early_vrp): Likewise.
22944 2017-05-16 Richard Biener <rguenther@suse.de>
22946 * dwarf2out.c (loc_list_from_tree_1): Do not create
22947 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
22949 2017-05-16 Richard Biener <rguenther@suse.de>
22951 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
22953 (note_variable_value_in_expr): If we resolved the decl ref
22954 do not push to the stack.
22956 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
22958 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
22959 operations in fast-math mode.
22960 (vaddq_f16): Likewise.
22961 (vmul_f16): Likewise.
22962 (vmulq_f16): Likewise.
22963 (vsub_f16): Likewise.
22964 (vsubq_f16): Likewise.
22965 * config/arm/neon.md (add<mode>3): New.
22967 (fma:<VH:mode>3): New. Also remove outdated comment.
22970 2017-05-16 Martin Liska <mliska@suse.cz>
22974 * ipa-devirt.c (warn_types_mismatch): Fix typo.
22975 (odr_types_equivalent_p): Likewise.
22977 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
22979 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
22981 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
22984 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
22985 non-interunit SSE move alternatives with '?'.
22986 (zero-extendsidi peephole2): New peephole to skip intermediate
22987 general register in SSE zero-extend sequence.
22989 2017-05-15 Jeff Law <law@redhat.com>
22991 * reorg.c (relax_delay_slots): Create a new variable to hold
22992 the temporary target rather than clobbering TARGET_LABEL.
22994 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
22995 missing argument to extract_bit_field call.
22996 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
22998 2017-05-15 Martin Liska <mliska@suse.cz>
23001 * gcc.c (process_command): Do not allow empty argument of -o option.
23003 2017-05-15 Renlin Li <renlin.li@arm.com>
23005 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
23006 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
23007 * config/aarch64/constraints.md (Usf): Add long call check.
23008 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
23009 (call_value): Likewise.
23010 (sibcall): Likewise.
23011 (sibcall_value): Likewise.
23013 (call_value_insn): New.
23014 (sibcall_insn): Update rtx pattern.
23015 (sibcall_value_insn): Likewise.
23016 (call_internal): Remove.
23017 (call_value_internal): Likewise.
23018 (sibcall_internal): Likewise.
23019 (sibcall_value_internal): Likewise.
23020 (call_reg): Likewise.
23021 (call_symbol): Likewise.
23022 (call_value_reg): Likewise.
23023 (call_value_symbol): Likewise.
23025 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
23028 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
23030 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
23032 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
23033 compatible with CCGOCmode and with CCZmode.
23035 2017-05-14 Martin Sebor <msebor@redhat.com>
23037 PR middle-end/77671
23038 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
23039 (gimple_fold_builtin_snprintf): Same.
23040 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
23041 (gimple_fold_builtin_snprintf): Same.
23042 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
23043 of character types.
23044 (is_call_safe): New function.
23045 (try_substitute_return_value): Call it.
23046 (try_simplify_call): New function.
23047 (pass_sprintf_length::handle_gimple_call): Call it.
23049 2017-05-14 Martin Sebor <msebor@redhat.com>
23051 PR middle-end/80669
23052 * builtins.c (expand_builtin_stpncpy): Simplify.
23054 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
23056 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
23057 * config/i386/i386.h
23058 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
23059 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
23060 (struct machine_function): Add new members call_ms2sysv,
23061 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
23062 (struct machine_frame_state): New fields sp_realigned and
23063 sp_realigned_offset.
23064 * config/i386/i386.c
23065 (enum xlogue_stub): New enum.
23066 (enum xlogue_stub_sets): New enum.
23067 (class xlogue_layout): New class.
23068 (struct ix86_frame): New fields stack_realign_allocate_offset,
23069 stack_realign_offset and outlined_save_offset. Modify comments to
23070 detail stack layout when using out-of-line stubs.
23071 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
23072 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
23073 -mcall-ms2sysv-xlogues.
23074 (stub_managed_regs): New static variable.
23075 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
23076 registers managed by out-of-line stub.
23077 (disable_call_ms2sysv_xlogues): New function.
23078 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
23079 m->call_ms2sysv when appropriate and compute frame layout for
23081 (sp_valid_at, fp_valid_at): New inline functions.
23082 (choose_basereg): New function.
23083 (choose_baseaddr): Add align parameter, use choose_basereg and modify
23085 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
23086 Use align parameter of choose_baseaddr to generated aligned SSE movs
23088 (pro_epilogue_adjust_stack): Modify to track
23089 machine_frame_state::sp_realigned.
23090 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
23091 (ix86_nsaved_sseregs): Likewise.
23092 (ix86_emit_save_regs): Likewise.
23093 (ix86_emit_save_regs_using_mov): Likewise.
23094 (ix86_emit_save_sse_regs_using_mov): Likewise.
23095 (get_scratch_register_on_entry): Likewise.
23096 (gen_frame_set): New function.
23097 (gen_frame_load): Likewise.
23098 (gen_frame_store): Likewise.
23099 (emit_outlined_ms2sysv_save): Likewise.
23100 (emit_outlined_ms2sysv_restore): Likewise.
23101 (ix86_expand_prologue): Modify stack re-alignment code and call
23102 emit_outlined_ms2sysv_save when appropriate.
23103 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
23104 parameter rtx_insn *insn, which allows the function to be used to only
23105 generate the notes.
23106 (ix86_expand_epilogue): Modify validity checks of frame and stack
23107 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
23108 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
23109 * config/i386/predicates.md
23110 (save_multiple): New predicate.
23111 (restore_multiple): Likewise.
23112 * config/i386/sse.md
23113 (save_multiple<mode>): New pattern.
23114 (save_multiple_realign<mode>): Likewise.
23115 (restore_multiple<mode>): Likewise.
23116 (restore_multiple_and_return<mode>): Likewise.
23117 (restore_multiple_leave_return<mode>): Likewise.
23118 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
23120 2017-05-14 Julia Koval <julia.koval@intel.com>
23122 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
23123 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
23124 (__builtin_ia32_xsetbv): New builtins.
23125 * config/i386/i386.c (ix86_expand_special_args_builtin):
23127 (ix86_expand_builtin): Special expand for new intrinsics.
23128 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
23129 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
23130 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
23132 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23134 * cfganal.c (inverted_post_order_compute): Change argument type
23136 * cfganal.h (inverted_post_order_compute): Adjust prototype.
23137 * df-core.c (rest_of_handle_df_initialize): Adjust.
23138 (rest_of_handle_df_finish): Likewise.
23139 (df_analyze_1): Likewise.
23140 (df_analyze): Likewise.
23141 (loop_inverted_post_order_compute): Change argument to be a vec *.
23142 (df_analyze_loop): Adjust.
23143 (df_get_n_blocks): Likewise.
23144 (df_get_postorder): Likewise.
23145 * df.h (struct df_d): Change field to be a vec.
23146 * lcm.c (compute_laterin): Adjust.
23147 (compute_available): Likewise.
23148 * lra-lives.c (lra_create_live_ranges_1): Likewise.
23149 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
23150 * tree-ssa-pre.c (compute_antic): Likewise.
23152 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23154 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
23155 (depth_first_search::depth_first_search): Change structure init
23156 function to this constructor.
23157 (depth_first_search::add_bb): Rename function to this member.
23158 (depth_first_search::execute): Likewise.
23159 (flow_dfs_compute_reverse_finish): Adjust.
23161 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23163 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
23164 (longest_simple_path): Likewise.
23165 * shrink-wrap.c (spread_components): Likewise.
23166 (disqualify_problematic_components): Likewise.
23167 (emit_common_heads_for_components): Likewise.
23168 (emit_common_tails_for_components): Likewise.
23169 (insert_prologue_epilogue_for_components): Likewise.
23171 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23173 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
23176 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23178 * df-core.c (df_set_blocks): Start using auto_bitmap.
23179 (df_compact_blocks): Likewise.
23180 * df-problems.c (df_rd_confluence_n): Likewise.
23181 * df-scan.c (df_insn_rescan_all): Likewise.
23182 (df_process_deferred_rescans): Likewise.
23183 (df_update_entry_block_defs): Likewise.
23184 (df_update_exit_block_uses): Likewise.
23185 (df_entry_block_bitmap_verify): Likewise.
23186 (df_exit_block_bitmap_verify): Likewise.
23187 (df_scan_verify): Likewise.
23188 * lra-constraints.c (lra_constraints): Likewise.
23189 (undo_optional_reloads): Likewise.
23190 (lra_undo_inheritance): Likewise.
23191 * lra-remat.c (calculate_gen_cands): Likewise.
23192 (do_remat): Likewise.
23193 * lra-spills.c (assign_spill_hard_regs): Likewise.
23194 (spill_pseudos): Likewise.
23195 * tree-ssa-pre.c (bitmap_set_and): Likewise.
23196 (bitmap_set_subtract_values): Likewise.
23198 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23200 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
23201 management with auto_bitmap.
23202 (fix_inter_tick): Likewise.
23203 (fix_recovery_deps): Likewise.
23204 * ira.c (add_store_equivs): Likewise.
23205 (find_moveable_pseudos): Likewise.
23206 (split_live_ranges_for_shrink_wrap): Likewise.
23207 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
23208 (rtx_reuse_manager::seen_def_p): Likewise.
23209 (rtx_reuse_manager::set_seen_def): Likewise.
23210 * print-rtl.h (class rtx_reuse_manager): Likewise.
23212 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23214 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
23216 (migrate_btr_def): Likewise.
23217 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
23218 * df-core.c (loop_post_order_compute): Likewise.
23219 (loop_inverted_post_order_compute): Likewise.
23220 * hsa-common.h: Likewise.
23221 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
23222 * init-regs.c (initialize_uninitialized_regs): Likewise.
23223 * ipa-inline.c (resolve_noninline_speculation): Likewise.
23224 (inline_small_functions): Likewise.
23225 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
23226 * ira.c (combine_and_move_insns): Likewise.
23227 (build_insn_chain): Likewise.
23228 * loop-invariant.c (find_invariants): Likewise.
23229 * lower-subreg.c (propagate_pseudo_copies): Likewise.
23230 * predict.c (tree_predict_by_opcode): Likewise.
23231 (predict_paths_leading_to): Likewise.
23232 (predict_paths_leading_to_edge): Likewise.
23233 (estimate_loops_at_level): Likewise.
23234 (estimate_loops): Likewise.
23235 * shrink-wrap.c (try_shrink_wrapping): Likewise.
23236 (spread_components): Likewise.
23237 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
23238 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
23239 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
23240 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
23241 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
23242 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
23243 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
23244 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
23245 (mark_threaded_blocks): Likewise.
23246 (thread_through_all_blocks): Likewise.
23247 * tree-ssa.c (verify_ssa): Likewise.
23248 (execute_update_addresses_taken): Likewise.
23249 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
23251 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23253 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
23255 (post_order_compute): Likewise.
23256 (inverted_post_order_compute): Likewise.
23257 (pre_and_rev_post_order_compute_fn): Likewise.
23259 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23261 * genrecog.c (int_set::int_set): Explicitly construct our
23262 auto_vec base class.
23263 * vec.h (auto_vec::auto_vec): New constructor.
23265 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23267 * bitmap.h (class auto_bitmap): New constructor taking
23268 bitmap_obstack * argument.
23270 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23272 * bitmap.h (class auto_bitmap): Change type of m_bits to
23273 bitmap_head, and adjust ctor / dtor and member operators.
23275 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
23277 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
23278 when returned register mode doesn't match original mode.
23280 2017-05-12 Jeff Law <law@redhat.com>
23281 Jakub Jelinek <jakub@redhat.com>
23283 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
23284 we look for cc setter after the compare-elim changes.
23285 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
23286 within the vector to match what compare-elim now expects.
23287 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
23288 (xorsi3_flags, one_cmplsi2_flags): Likewise.
23290 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
23291 after the compare-elim changes.
23292 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
23293 the vector to match what compare-elim now expects.
23294 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
23295 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
23296 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
23297 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
23298 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
23300 * config/visium/visium.c (single_set_and_flags): Fix where
23301 we look for cc setter after the compare-elim changes.
23302 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
23303 with the vector to match what compare-elim now expects.
23304 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
23305 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
23306 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
23307 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
23308 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
23309 (neg<mode>2_insn_set_overflow): Likewise.
23311 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
23313 PR middle-end/79794
23314 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
23315 maybe_expand_insn call, set ops[0].target. If still set after call,
23316 set alt_rtl. Add extra arg to recursive calls.
23317 (extract_bit_field): Add alt_rtl argument. Pass to
23319 * expmed.h (extract_bit_field): Fix prototype.
23320 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
23321 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
23322 to extract_bit_field_calls.
23323 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
23324 Pass alt_rtl to extract_bit_field calls.
23325 * calls.c (store_unaligned_arguments_into_psuedos)
23326 load_register_parameters): Pass extra NULL to extract_bit_field calls.
23327 * optabs.c (maybe_legitimize_operand): Clear op->target when call
23329 * optabs.h (struct expand_operand): Add target bitfield.
23331 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
23333 * compare-elim.c (try_eliminate_compare): Canonicalize
23334 operation with embedded compare to
23335 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
23336 (set (reg) (operation)].
23338 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
23340 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
23343 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
23344 cost of adding a carry flag for ADC instruction.
23345 [case MINUS]: Ignore the cost of subtracting a carry flag
23346 for SBB instruction.
23348 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
23350 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
23352 * config/rs6000/bmiintrin.h: New file.
23353 * config/rs6000/bmi2intrin.h: New file.
23354 * config/rs6000/x86intrin.h: New file.
23356 2017-05-12 Jeff Law <law@redhat.com>
23358 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
23361 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
23363 PR middle-end/80707
23364 * tree-cfg.c: Remove cfg edges of unreachable case statements.
23366 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23368 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
23369 early expansion of vector divide builtins.
23370 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
23371 builtins identified as having unsigned arguments.
23373 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23375 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
23376 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
23377 expansion of vector logical operations (and, andc, or, xor,
23380 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23382 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
23383 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
23385 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23387 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
23388 early GIMPLE expansion of vector multiplies.
23390 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23392 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
23393 TARGET_HAVE_MOVT conditional.
23394 (movt splitter): Likewise.
23396 2017-05-12 Richard Biener <rguenther@suse.de>
23398 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
23399 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
23400 Fold all stmts not inplace.
23402 2017-05-12 Richard Biener <rguenther@suse.de>
23404 PR tree-optimization/80713
23405 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
23406 inserted_exprs bit for not removed stmts.
23408 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
23410 PR middle-end/69921
23411 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
23412 parallelized" attribute for parallelized OpenACC kernels.
23413 * omp-offload.c (execute_oacc_device_lower): Use it.
23415 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
23416 Set "oacc kernels" attribute.
23417 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
23418 parameter. Adjust all users.
23419 (oacc_fn_attrib_kernels_p): Remove function.
23420 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
23421 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
23422 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
23423 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
23424 assert "oacc kernels" attribute is set.
23426 2017-05-11 Carl Love <cel@us.ibm.com>
23428 * config/rs6000/rs6000-c: Add support for built-in functions
23429 vector unsigned char vec_popcnt (vector signed char)
23430 vector unsigned char vec_popcnt (vector unsigned char)
23431 vector unsigned short vec_popcnt (vector signed short)
23432 vector unsigned short vec_popcnt (vector unsigned short)
23433 vector unsigned int vec_popcnt (vector signed int)
23434 vector unsigned int vec_popcnt (vector unsigned int)
23435 vector unsigned long long vec_popcnt (vector signed long long)
23436 vector unsigned long long vec_popcnt (vector unsigned long long)
23437 vector signed long long vec_slo (vector signed long long,
23438 vector signed char)
23439 vector signed long long vec_slo (vector signed long long,
23440 vector unsigned char)
23441 vector unsigned long long vec_slo (vector unsigned long long,
23442 vector signed char)
23443 vector unsigned long long vec_slo (vector unsigned long long,
23444 vector unsigned char)
23445 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
23446 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
23447 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
23448 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
23449 * doc/extend.texi: Update the built-in documentation file for the
23450 new built-in functions.
23452 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
23454 * attribs.h (sorted_attr_string): Move machine independent
23455 functions for target clone support from the i386 port to common
23456 code. Rename ix86_function_versions to common_function_versions.
23457 Rename make_name to make_unique_name.
23458 (common_function_versions): Likewise.
23459 (make_unique_name): Likewise.
23460 (make_dispatcher_decl): Likewise.
23461 (is_function_default_version): Likewise.
23462 * attribs.c (attr_strcmp): Likewise.
23463 (sorted_attr_string): Likewise.
23464 (common_function_versions): Likewise.
23465 (make_unique_name): Likewise.
23466 (make_dispatcher_decl): Likewise.
23467 (is_function_default_version): Likewise.
23468 * config/i386/i386.c (attr_strcmp): Likewise.
23469 (sorted_attr_string): Likewise.
23470 (ix86_function_versions): Likewise.
23471 (make_name): Likewise.
23472 (make_dispatcher_decl): Likewise.
23473 (is_function_default_version): Likewise.
23474 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
23476 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23479 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23480 Account for direct move costs for vec_construct of integer
23483 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
23486 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
23487 (UNSPEC_STX_ATOMIC): Ditto.
23488 (loaddi_via_sse): New insn.
23489 (storedi_via_sse): Ditto.
23490 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
23491 Update corresponding peephole2 patterns.
23492 (atomic_storedi_fpu): Ditto.
23494 2017-05-11 Julia Koval <julia.koval@intel.com>
23496 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
23497 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
23499 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
23500 (__builtin_ia32_rsqrt14ss_mask): New builtins.
23501 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
23503 2017-05-11 Nathan Sidwell <nathan@acm.org>
23505 * graphite-poly.c: Include dumpfile.h.
23507 * dumpfle.h (dump_function): Declare here ...
23508 * tree-dump.h (dump_function): ... not here.
23509 * dumpfile.c: #include tree-cfg.h.
23510 (dump_function): Move here from ...
23511 * tree-dump.c (dump_function): ... here.
23512 * gimplify.c: #include splay-tree.h, not tree-dump.h.
23513 * graphite-poly.c: Don't include tree-dump.h.
23514 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
23515 * print-tree.c: Likewise.
23516 * stor-layout.c: Likewise.
23517 * tree-nested.c: Likewise.
23519 * dumpfile.c (dump_start): Use TDF_FLAGS.
23520 (dump_enable_all): Fix TDF_KIND check thinko.
23522 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
23524 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23525 array entries to represent two legal parameterizations of the
23526 overloaded __builtin_cmpb function, as represented by the
23527 P6_OV_BUILTIN_CMPB constant.
23528 (altivec_resolve_overloaded_builtin): Add special case handling
23529 for the __builtin_cmpb function, as represented by the
23530 P6_OV_BUILTIN_CMPB constant.
23531 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
23532 (BU_P6_64BIT_2): New macro.
23533 (BU_P6_OVERLOAD_2): New macro
23534 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
23535 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
23536 (CMPB): Add overload support to represent both 32-bit and 64-bit
23537 compare-bytes function.
23538 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
23539 support for TARGET_CMPB.
23540 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
23541 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23542 documentation of the __builtin_cmpb overloaded built-in function.
23544 2017-05-11 Richard Biener <rguenther@suse.de>
23546 PR tree-optimization/80705
23547 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
23548 bases are not vectorizable.
23550 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23552 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
23553 when counting register pressure.
23555 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23557 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
23558 (iv_ca_more_deps): Renamed to ...
23559 (iv_ca_compare_deps): ... this.
23560 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
23562 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23564 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
23566 (determine_group_iv_costs): ... here.
23567 (find_inv_vars_cb): Record inv var if it's not recorded before.
23569 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23571 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
23572 (get_shiftadd_cost): Ditto.
23574 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23576 * tree-ssa-address.c: Include header file.
23577 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
23579 (add_to_parts): Refactor.
23580 (addr_to_parts): New parameter. Update use of move_hint_to_base.
23581 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
23584 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23586 PR tree-optimization/53090
23587 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
23589 (extract_cond_operands): Detect condition with IV on both sides
23590 and return COMP_IV_EXPR_2.
23591 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
23592 (rewrite_use_compare): Simplify by removing call to function
23593 extract_cond_operands.
23595 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23597 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
23598 (extract_cond_operands): Detect condition comparing against non-
23599 invariant bound and return appropriate enum value.
23600 (find_interesting_uses_cond): Update use of extract_cond_operands.
23601 Handle its return value accordingly.
23602 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
23604 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23606 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
23607 nonlinear iv_use computation in loop invariant sensitive way.
23609 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23611 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
23612 (find_iv_candidates): Call relate_compare_use_with_all_cands.
23614 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23616 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
23617 (dump_cand): Support iv_cand.inv_exprs.
23618 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
23620 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
23623 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23625 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
23627 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
23628 as local function. Include necessary header files.
23629 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
23631 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23633 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
23635 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23637 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
23638 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
23639 RSHIFT_EXPR and BIT_NOT_EXPR.
23641 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23643 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
23644 (adjust_setup_cost): New parameter supporting round up adjustment.
23645 (struct address_cost_data): Delete.
23646 (force_expr_to_var_cost): Don't bound cost with spill_cost.
23647 (split_address_cost, ptr_difference_cost): Delete.
23648 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
23649 (struct ainc_cost_data): New struct.
23650 (get_address_cost_ainc): New function.
23651 (get_address_cost, get_computation_cost): Reimplement.
23652 (determine_group_iv_cost_address): Record inv_expr for all uses of
23654 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
23655 (iv_ca_has_deps): Reimplemented to ...
23656 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
23658 (iv_ca_extend): Call iv_ca_more_deps.
23660 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23662 * tree-ssa-address.c (struct mem_address): Move to header file.
23663 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
23664 * tree-ssa-address.h (struct mem_address): Move from C file.
23665 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
23667 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23669 * tree-affine.h (aff_combination_type): New interface.
23670 (aff_combination_zero_p): Remove static.
23671 (aff_combination_const_p): New interface.
23672 (aff_combination_singleton_var_p): New interfaces.
23674 2017-05-11 Richard Biener <rguenther@suse.de>
23676 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
23677 Skip unreachable blocks and destinations.
23678 (eliminate): Move stmt removal and fixup ...
23679 (fini_eliminate): ... here. Skip inserted exprs.
23680 (pass_pre::execute): Move fini_pre after fini_eliminate.
23681 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
23682 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
23683 PRE to get rid of dead code that has invalid SSA form and
23684 split critical edges again.
23686 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23688 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
23690 2017-05-11 Richard Biener <rguenther@suse.de>
23692 * passes.c (execute_function_todo): Verify loops if they are
23693 said to be up-to-date.
23694 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
23695 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
23697 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
23700 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
23701 handle calling assemble_external ourself.
23704 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
23705 modes with zero size. Enhance comment.
23707 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23709 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
23710 built-ins for vec_xl and vec_xst with short and char pointer
23713 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
23715 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
23716 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
23717 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
23718 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
23719 (_mm_maskz_min_round_ss): New intrinsics.
23720 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
23721 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
23722 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
23723 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
23724 (__builtin_ia32_minss_mask_round): New builtins.
23725 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23726 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23727 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
23729 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
23730 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
23732 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
23735 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
23737 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
23738 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
23739 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
23740 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
23741 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
23742 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
23743 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
23744 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23745 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
23746 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
23747 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
23748 (__builtin_ia32_mulss_mask_round): New builtins.
23749 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23750 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23751 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
23753 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
23754 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
23756 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
23759 2017-05-10 Julia Koval <julia.koval@intel.com>
23761 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
23762 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
23763 (_mm256_setr_m128i): New intrinsics.
23765 2017-05-10 Julia Koval <julia.koval@intel.com>
23767 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
23768 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
23769 (_mm_maskz_rcp14_ss): New intrinsics.
23770 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
23771 (__builtin_ia32_rcp14ss_mask): New builtins.
23772 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
23774 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
23776 PR tree-optimization/51513
23777 * tree-cfg.c (gimple_seq_unreachable_p): New function.
23778 (assert_unreachable_fallthru_edge_p): Use it.
23779 (group_case_labels_stmt): Likewise.
23780 * tree-cfg.h: Prototype it.
23781 * stmt.c: Include cfghooks.h and tree-cfg.h.
23782 (emit_case_dispatch_table) <gap_label>: New local variable.
23783 Use it to fill dispatch table gaps.
23784 Test for default_label before updating probabilities.
23785 (expand_case) <default_label>: Remove unneeded initialization.
23786 Test for unreachable default case statement and remove its edge.
23787 Set default_label accordingly.
23788 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
23790 2017-05-10 Carl Love <cel@us.ibm.com>
23792 * config/rs6000/rs6000-c: Add support for built-in functions
23793 vector signed char vec_neg (vector signed char)
23794 vector signed short int vec_neg (vector short int)
23795 vector signed int vec_neg (vector signed int)
23796 vector signed long long vec_neg (vector signed long long)
23797 vector float vec_neg (vector float)
23798 vector double vec_neg (vector double)
23799 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
23801 * config/rs6000/altivec.h: Add define for vec_neg
23802 * doc/extend.texi: Update the built-in documentation for the
23803 new built-in functions.
23805 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23807 PR tree-optimization/77644
23808 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
23810 2017-05-10 Nathan Sidwell <nathan@acm.org>
23812 * dumpfile.h (TDI_lang_all): New.
23813 (TDF_KIND): New. Renumber others
23814 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
23816 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
23818 (get_dump_file_name): Adjust suffix generation.
23819 (dump_enable_all): Use TDF_KIND.
23820 * doc/invoke.texi (-fdump-lang-all): Document.
23822 * dumpfile.h: Tabify.
23824 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
23827 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
23828 Move member access before delete.
23830 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
23832 * tree-inline.c (expand_call_inline): Split block at stmt
23835 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
23838 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
23839 are now unused after splitting mov{sf,sd}_hardfloat.
23840 (f32_lr2): Likewise.
23841 (f32_lm): Likewise.
23842 (f32_lm2): Likewise.
23843 (f32_li): Likewise.
23844 (f32_li2): Likewise.
23845 (f32_lv): Likewise.
23846 (f32_sr): Likewise.
23847 (f32_sr2): Likewise.
23848 (f32_sm): Likewise.
23849 (f32_sm2): Likewise.
23850 (f32_si): Likewise.
23851 (f32_si2): Likewise.
23852 (f32_sv): Likewise.
23853 (f32_dm): Likewise.
23854 (f32_vsx): Likewise.
23855 (f32_av): Likewise.
23856 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
23857 For movsf, order stores so the VSX stores occur before the GPR
23858 store which encourages the register allocator to use a traditional
23859 FPR instead of a GPR. For movsd, order the stores so that the GPR
23860 store comes before the VSX stores to allow the power6 to work.
23861 This is due to the power6 not having a 32-bit integer store
23862 instruction from a FPR.
23863 (movsf_hardfloat): Likewise.
23864 (movsd_hardfloat): Likewise.
23866 2017-05-09 Martin Sebor <msebor@redhat.com>
23868 PR translation/80280
23869 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
23872 PR translation/80280
23873 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
23874 data member added in r247778.
23875 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
23877 2017-05-09 Nathan Sidwell <nathan@acm.org>
23879 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
23881 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
23884 2017-05-09 Marek Polacek <polacek@redhat.com>
23886 * doc/invoke.texi: Fix typo.
23888 2017-05-09 Richard Biener <rguenther@suse.de>
23890 * tree-vrp.c (vrp_val_is_max): Adjust comment.
23891 (vrp_val_is_min): Likewise.
23892 (set_value_range_to_value): Likewise.
23893 (set_value_range_to_nonnegative): Likewise.
23894 (gimple_assign_nonzero_p): Likewise.
23895 (gimple_stmt_nonzero_p): Likewise.
23896 (vrp_int_const_binop): Likewise. Remove unreachable case.
23897 (adjust_range_with_scev): Adjust comments.
23898 (compare_range_with_value): Likewise.
23899 (extract_range_from_phi_node): Likewise.
23900 (test_for_singularity): Likewise.
23902 2017-05-09 Richard Biener <rguenther@suse.de>
23904 * tree-vrp.c (get_single_symbol): Add assert that we don't
23905 get overflowed constants as invariant part.
23906 (compare_values_warnv): Add comment before the TREE_NO_WARNING
23907 checks. Use wi::cmp instead of recursing for integer constants.
23908 (compare_values): Just ignore whether we assumed undefined
23909 overflow instead of failing the compare.
23910 (extract_range_for_var_from_comparison_expr): Add comment before the
23911 TREE_NO_WARNING sets.
23912 (test_for_singularity): Likewise.
23913 (extract_range_from_comparison): Do not disable optimization
23914 when we assumed undefined overflow.
23915 (extract_range_basic): Remove init of unused var.
23917 2017-05-09 Richard Biener <rguenther@suse.de>
23919 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
23920 (extract_range_from_multiplicative_op_1): Adjust.
23921 (extract_range_from_binary_expr_1): Use int_const_binop.
23923 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
23926 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
23927 rs6000_store_data_bypass_p in seven define_bypass directives and
23928 in several comments.
23929 * config/rs6000/rs6000-protos.h: Add prototype for
23930 rs6000_store_data_bypass_p function.
23931 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
23932 function implements slightly different (rs6000-specific) semantics
23933 than store_data_bypass_p, returning false rather than aborting
23934 with assertion error when arguments do not satisfy the
23935 requirements of store data bypass.
23936 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
23937 rs6000_store_data_bypass_p.
23939 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
23941 * config/xtensa/xtensa-protos.h
23942 (xtensa_initial_elimination_offset): New declaration.
23943 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
23944 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
23945 macro definition, add case for FRAME_POINTER_REGNUM when
23946 FRAME_GROWS_DOWNWARD.
23947 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
23948 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
23949 xtensa_initial_elimination_offset.
23951 2017-05-08 Nathan Sidwell <nathan@acm.org>
23953 * doc/invoke.texi: Alphabetize -fdump options.
23955 2017-05-08 Martin Sebor <msebor@redhat.com>
23957 PR translation/80280
23958 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
23960 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
23962 * target.def (compute_frame_layout): New optional target hook.
23963 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
23964 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
23965 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
23967 * reload1.c (verify_initial_elim_offsets): Likewise.
23968 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
23969 (use_simple_return_p): Call arm_compute_frame_layout if needed.
23970 (arm_get_frame_offsets): Split up into this ...
23971 (arm_compute_frame_layout): ... and this function.
23973 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
23975 * config/aarch64/constraints.md (Usa): New constraint.
23976 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
23978 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
23980 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
23981 with_multilib_list after it has been checked.
23983 2017-05-08 Richard Biener <rguenther@suse.de>
23985 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
23986 (bitmap_set_subtract_values): Likewise.
23988 2017-05-08 Richard Biener <rguenther@suse.de>
23990 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
23991 (gimple_assign_nonzero): ... this and remove strict_overflow_p
23993 (gimple_stmt_nonzero_warnv_p): Rename to ...
23994 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
23996 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
23997 (extract_range_basic): Adjust, do not disable propagation on
23998 strict overflow sensitive simplification.
23999 (vrp_visit_cond_stmt): Likewise.
24001 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
24003 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
24004 body size unconditionally.
24006 2017-05-07 Jeff Law <law@redhat.com>
24009 2017-05-06 Jeff Law <law@redhat.com>
24010 PR tree-optimization/78496
24011 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24014 PR tree-optimization/78496
24015 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24016 (simplify_stmt_using_ranges): Call it.
24017 (vrp_dom_walker::before_dom_children): Extract equivalences
24018 from an ASSERT_EXPR with an equality comparison against a
24021 2017-05-06 Jeff Law <law@redhat.com>
24023 PR tree-optimization/78496
24024 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24027 PR tree-optimization/78496
24028 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24029 (simplify_stmt_using_ranges): Call it.
24030 (vrp_dom_walker::before_dom_children): Extract equivalences
24031 from an ASSERT_EXPR with an equality comparison against a
24034 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24036 * lra-constraints.c (lra_copy_reg_equiv): New function.
24037 (split_reg): Use it to copy equivalence information from the
24038 original register to the spill register.
24040 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24042 PR rtl-optimization/75964
24043 * simplify-rtx.c (simplify_const_relational_operation): Remove
24044 invalid handling of comparisons of integer ABS.
24046 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
24048 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
24049 initialize to zero.
24050 (init_regs): Remove declaration.
24051 (function_arg_advance_32): Initialize error_p as boolean variable.
24053 2017-05-05 Nathan Sidwell <nathan@acm.org>
24055 * store-motion.c (remove_reachable_equiv_notes): Reformat long
24056 lines. Use for (;;).
24058 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24060 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
24061 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
24062 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
24063 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
24064 VF=2 that require versioning.
24066 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24068 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
24071 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24073 * diagnostic.h (diagnostic_override_option_index): Convert from
24074 macro to inline function.
24076 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24078 * diagnostic.c (last_module_changed_p): New function.
24079 (set_last_module): New function.
24080 (diagnostic_report_current_module): Convert macro usage to
24081 the above functions.
24082 * diagnostic.h (diagnostic_context::last_module): Strengthen
24083 from const line_map * to const line_map_ordinary *.
24084 (diagnostic_last_module_changed): Delete macro.
24085 (diagnostic_set_last_module): Delete macro.
24087 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24089 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
24090 with diagnostic_report_diagnostic.
24091 (diagnostic_n_impl_richloc): Likewise.
24092 * diagnostic.h (report_diagnostic): Delete macro.
24093 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
24094 with diagnostic_report_diagnostic.
24095 * substring-locations.c (format_warning_va): Likewise.
24097 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24099 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
24100 save/restor of format_spec. Move option-printing code to...
24101 (print_option_information): ...this new function, and
24102 reimplement by simply printing to the pretty_printer,
24103 rather than appending to the format string.
24105 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24107 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
24108 handling logic into...
24109 (update_effective_level_from_pragmas): ...this new function.
24111 2017-05-04 Andrew Waterman <andrew@sifive.com>
24113 * config/riscv/riscv.opt (mstrict-align): New option.
24114 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
24115 (SLOW_UNALIGNED_ACCESS): Define.
24116 (riscv_slow_unaligned_access): Declare.
24117 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
24119 (riscv_slow_unaligned_access): New variable.
24120 (rocket_tune_info): Set slow_unaligned_access to true.
24121 (optimize_size_tune_info): Set slow_unaligned_access to false.
24122 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
24123 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
24124 (riscv_option_override): Set riscv_slow_unaligned_access.
24125 * doc/invoke.texi: Add -mstrict-align to RISC-V.
24127 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
24129 * config/riscv/riscv.md: Unify indentation.
24131 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
24136 * config/rs6000/rs6000.md (u code attribute): Add FIX and
24138 (extendsi<mode>2): Add support for doing sign extension via
24139 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
24140 don't have ISA 3.0 instructions.
24141 (extendsi<mode>2 splitter): Likewise.
24142 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
24143 generate the normal insns since SImode can now go in vector
24144 registers. Disallow the special UNSPECs needed for previous
24145 machines to hide SImode being used. Add new insns
24146 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
24147 (fix_trunc<mode>si2_stfiwx): Likewise.
24148 (fix_trunc<mode>si2_internal): Likewise.
24149 (fixuns_trunc<mode>si2): Likewise.
24150 (fixuns_trunc<mode>si2_stfiwx): Likewise.
24151 (fctiw<u>z_<mode>_smallint): Likewise.
24152 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
24153 of floating point to 32-bit integer from doing a direct move to
24154 the GPR registers to do a store.
24155 (fctiwz_<mode>): Break long line.
24157 2017-05-05 Bin Cheng <bin.cheng@arm.com>
24159 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
24160 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
24161 (addr_list, addr_offset_valid_p): New.
24162 (split_address_groups): Check offset validity with above function.
24163 (gt-tree-ssa-loop-ivopts.h): Include header file.
24165 2017-05-05 Nathan Sidwell <nathan@acm.org>
24167 * config.gcc (arm*-*-*): Add missing 'fi'.
24169 2017-05-05 Steve Ellcey <sellcey@cavium.com>
24171 * doc/invoke.texi (-fopt-info): Explicitly say order of options
24172 included in -fopt-info does not matter.
24173 * doc/optinfo.texi (-fopt-info): Fix description of default
24174 behavour. Explicitly say order of options included in -fopt-info
24177 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
24179 * config.gcc: Allow combinations of aprofile and rmprofile values for
24180 --with-multilib-list.
24181 * config/arm/t-multilib: New file.
24182 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
24183 variables. Remove setting of ISA and floating-point ABI in
24184 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
24185 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
24186 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
24187 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
24189 * config/arm/t-rmprofile: Likewise except for the matches changes.
24190 * doc/install.texi (--with-multilib-list): Document the combination of
24191 aprofile and rmprofile values and warn about pitfalls in doing that.
24193 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24195 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
24196 (movdi_aarch64): Likewise.
24198 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24200 PR tree-optimization/80632
24201 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
24203 (build_arrays): Initialize it for virtual phis.
24204 (fix_phi_nodes): Use it for virtual phis.
24206 PR tree-optimization/80558
24207 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
24208 [x, y] op z into [x op, y op z] for op & or | if conditions
24211 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
24212 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24215 * config/arm/arm.md (use_literal_pool): Remove.
24216 (64-bit immediate split): No longer takes cost into consideration
24217 if arm_disable_literal_pool is enabled.
24218 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
24219 used when arm_disable_literal_pool is enabled.
24220 (arm_max_const_double_inline_cost): Remove use of
24221 arm_disable_literal_pool.
24222 (push_minipool_fix): Add assert.
24223 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
24224 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
24225 (no_literal_pool_sf_immediate): New.
24227 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24229 PR tree-optimization/80613
24230 * tree-ssa-dce.c (propagate_necessity): Remove cases for
24231 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
24233 2017-05-05 Richard Biener <rguenther@suse.de>
24235 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
24237 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
24239 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
24240 of this flag from insn conditions due to removal from r247495.
24242 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24244 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
24246 (arm_early_store_addr_dep_ptr): Likewise.
24247 * config/arm/aarch-common-protos.h
24248 (arm_early_load_addr_dep_ptr): Add prototype.
24249 (arm_early_store_addr_dep_ptr): Likewise.
24250 * config/arm/cortex-a53.md: Add new bypasses.
24252 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24254 * tree.c (next_type_uid): Change type to unsigned.
24255 (type_hash_canon): Decrement back next_type_uid if
24256 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
24257 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
24260 2017-05-04 Martin Sebor <msebor@redhat.com>
24262 * builtins.c: Fix a trivial typo in a comment.
24264 PR middle-end/79234
24265 * builtins.c (check_sizes): Adjust to handle reading past the end.
24266 Avoid printing excessive upper bound of ranges. Use %E to print
24267 tree nodes instead of converting them to %wu.
24268 (expand_builtin_memchr): New function.
24269 (compute_dest_size): Rename...
24270 (compute_objsize): ...to this.
24271 (expand_builtin_memcpy): Adjust.
24272 (expand_builtin_mempcpy): Adjust.
24273 (expand_builtin_strcat): Adjust.
24274 (expand_builtin_strcpy): Adjust.
24275 (check_strncat_sizes): Adjust.
24276 (expand_builtin_strncat): Adjust.
24277 (expand_builtin_strncpy): Adjust and simplify.
24278 (expand_builtin_memset): Adjust.
24279 (expand_builtin_bzero): Adjust.
24280 (expand_builtin_memcmp): Adjust.
24281 (expand_builtin): Handle memcmp.
24282 (maybe_emit_chk_warning): Check strncat just once.
24284 2017-05-04 Martin Sebor <msebor@redhat.com>
24286 PR preprocessor/79214
24287 PR middle-end/79222
24288 PR middle-end/79223
24289 * builtins.c (check_sizes): Add inlining context and issue
24290 warnings even when -Wno-system-headers is set.
24291 (check_strncat_sizes): Same.
24292 (expand_builtin_strncat): Same.
24293 (expand_builtin_memmove): New function.
24294 (expand_builtin_stpncpy): Same.
24295 (expand_builtin): Handle memmove and stpncpy.
24297 2017-05-04 Bin Cheng <bin.cheng@arm.com>
24299 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
24300 which is not used any more.
24302 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24304 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
24306 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24308 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
24309 (cortexa53_tunings): Likewise.
24310 (cortexa57_tunings): Likewise.
24311 (cortexa72_tunings): Likewise.
24312 (cortexa73_tunings): Likewise.
24314 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24316 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
24317 Set loop alignment to 8.
24319 2017-05-04 Martin Sebor <msebor@redhat.com>
24321 PR translation/80280
24322 * builtins.c (expand_builtin_object_size): Add missing quoting to
24323 %D and like directives.
24324 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
24325 (hsa_type_for_tree_type): Same.
24326 (verify_function_arguments): Same.
24327 * symtab.c (symbol_table::change_decl_assembler_name): Same.
24328 * varasm.c (get_section): Same.
24331 2017-05-04 Martin Sebor <msebor@redhat.com>
24333 PR translation/80280
24334 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
24336 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24338 * config/aarch64/aarch64.c (generic_addrcost_table):
24339 Change HI/TI mode setting.
24341 2017-05-04 Martin Jambor <mjambor@suse.cz>
24343 PR tree-optimization/80622
24344 * tree-sra.c (comes_initialized_p): New function.
24345 (build_accesses_from_assign): Only set write lazily when
24346 comes_initialized_p is false.
24347 (analyze_access_subtree): Use comes_initialized_p.
24348 (propagate_subaccesses_across_link): Assert !comes_initialized_p
24349 instead of testing for PARM_DECL.
24351 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24353 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
24354 constraint on operand 0 to allow more general addressing modes.
24355 Adjust output template.
24356 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
24358 * config/aarch64/aarch64-protos.h
24359 (aarch64_address_valid_for_prefetch_p): Declare prototype.
24360 * config/aarch64/constraints.md (Dp): New address constraint.
24361 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
24364 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
24366 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
24367 update use of estimate_ipcp_clone_size_and_time.
24368 (estimate_local_effects): Update use of
24369 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
24370 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
24371 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
24372 Return nonspecialized time.
24374 2017-05-04 Richard Biener <rguenther@suse.de>
24376 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
24377 for the last VUSE which def dominates the PHI. Directly call
24379 (get_continuation_for_phi_1): Remove.
24381 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
24383 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
24384 to explain the use of truncating division. Cap the number of
24385 iterations to the maximum given by nb_iterations_upper_bound,
24388 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24390 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
24391 * configure: Regenerate.
24392 * config.in: Regenerate.
24393 * config/i386/driver-mingw32.c: new file.
24394 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
24395 * config.host: Link driver-mingw32.o on MinGW host.
24396 * doc/install.texi: Document new --enable-mingw-wildcard configure
24399 2017-05-04 Marek Polacek <polacek@redhat.com>
24401 PR tree-optimization/80612
24402 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
24404 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24405 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
24407 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
24408 (movt splitter): Likewise.
24409 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
24410 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
24411 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
24412 block for Thumb-1 with MOVT.
24413 (thumb2_legitimate_address_p): Move code block ...
24414 (can_avoid_literal_pool_for_label_p): ... into this new function.
24415 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
24417 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
24418 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
24419 "M-profile targets with the MOVT instruction".
24421 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24423 * config/arm/arm-builtins.c (arm_init_builtins): Rename
24424 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
24425 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
24427 2017-05-04 Martin Liska <mliska@suse.cz>
24429 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
24430 variable cond_code.
24432 2017-05-04 Richard Biener <rguenther@suse.de>
24434 * tree.c (array_at_struct_end_p): Handle arrays at struct
24435 end with flexarrays more conservatively. Refactor and treat
24436 arrays of arrays or aggregates more strict. Fix
24437 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
24438 * tree.c (array_at_struct_end_p): Adjust prototype.
24439 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
24440 * gimple-fold.c (get_range_strlen): Likewise.
24441 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
24443 2017-05-04 Richard Biener <rguenther@suse.de>
24445 PR tree-optimization/31130
24446 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
24448 (supports_overflow_infinity): Likewise.
24449 (is_negative_overflow_infinity): Likewise.
24450 (is_positive_overflow_infinity): Likewise.
24451 (is_overflow_infinity): Likewise.
24452 (stmt_overflow_infinity): Likewise.
24453 (overflow_infinity_range_p): Likewise.
24454 (usable_range_p): Remove as always returning true.
24455 (make_overflow_infinity): Remove.
24456 (negative_overflow_infinity): Likewise.
24457 (positive_overflow_infinity): Likewise.
24458 (avoid_overflow_infinity): Likewise.
24459 (set_value_range): Adjust accordingly.
24460 (set_value_range_to_nonnegative): Likewise, remove now unused
24461 overflow_infinity arg.
24462 (vrp_operand_equal_p): Adjust.
24463 (update_value_range): Likewise.
24464 (range_int_cst_singleton_p): Likewise.
24465 (operand_less_p): Likewise.
24466 (compare_values_warnv): Likewise.
24467 (extract_range_for_var_from_comparison_expr): Likewise.
24468 (vrp_int_const_binop): Likewise.
24469 (zero_nonzero_bits_from_vr): Likewise.
24470 (extract_range_from_multiplicative_op_1): Likewise.
24471 (extract_range_from_binary_expr_1): Likewise.
24472 (extract_range_from_unary_expr): Likewise.
24473 (extract_range_from_comparison): Likewise.
24474 (extract_range_basic): Likewise.
24475 (adjust_range_with_scev): Likewise.
24476 (compare_ranges): Likewise.
24477 (compare_range_with_value): Likewise.
24478 (dump_value_range): Likewise.
24479 (test_for_singularity): Likewise, remove strict_overflow_p parameter
24481 (simplify_cond_using_ranges): Adjust.
24483 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
24485 * brig-builtins.def: Added a builtin for class_f64.
24486 * builtin-types.def: Added a builtin type needed by class_f64.
24488 2017-05-03 Jason Merrill <jason@redhat.com>
24490 * timevar.def: Add TV_CONSTEXPR.
24492 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24494 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
24496 2017-05-03 Martin Jambor <mjambor@suse.cz>
24498 * ipa-prop.c (ipa_update_after_lto_read): Removed.
24499 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
24500 * ipa-cp.c (ipcp_propagate_stage): Do not call
24501 ipa_update_after_lto_read.
24502 * ipa-inline.c (ipa_inline): Likewise.
24504 2017-05-03 Martin Jambor <mjambor@suse.cz>
24506 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
24507 tag. Added a default constructor and a destructor.
24508 (ipa_edge_args_sum_t): New class;
24509 (ipa_edge_args_sum): Declare.
24510 (ipa_edge_args_vector): Remove declaration.
24511 (IPA_EDGE_REF): Use ipa_edge_args_sum.
24512 (ipa_free_edge_args_substructures): Remove declaration.
24513 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
24514 (ipa_edge_args_info_available_for_edge_p): Likewise.
24515 * ipa-prop.c (ipa_edge_args_vector): Removed.
24516 (edge_removal_hook_holder): Likewise.
24517 (edge_duplication_hook_holder): Likewise.
24518 (ipa_edge_args_sum): New variable.
24519 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
24520 ipa_edge_args_vector.
24521 (ipa_free_edge_args_substructures): Likewise.
24522 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
24523 ipa_edge_args_vector.
24524 (ipa_edge_removal_hook): Turned into method
24525 ipa_edge_args_sum_t::remove.
24526 (ipa_edge_duplication_hook): Turned into method
24527 ipa_edge_args_sum_t::duplicate.
24528 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
24529 registering edge hooks.
24530 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
24531 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
24532 ipa_edge_args_sum instead of ipa_edge_args_vector.
24533 * ipa-profile.c (ipa_profile): Likewise.
24535 2017-05-03 Martin Jambor <mjambor@suse.cz>
24537 * symbol-summary.h (function_summary): New method exists.
24538 (function_summary::symtab_removal): Deallocate through release.
24539 (call_summary): New class.
24540 (gt_ggc_mx): New overload.
24541 (gt_pch_nx): Likewise.
24542 (gt_pch_nx): Likewise.
24544 2017-05-03 Jeff Law <law@redhat.com>
24546 PR tree-optimization/78496
24547 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
24548 from simplify_cond_using_ranges. Split off code to walk
24549 backwards through casts into ...
24550 (simplify_cond_using_ranges_2): New function.
24551 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
24552 (execute_vrp): After identifying jump threads, call
24553 simplify_cond_using_ranges_2.
24555 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
24558 * ipa-inline.h (inline_summary): Add ctor.
24559 (create_ggc): Do not use ggc_cleared_alloc.
24561 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
24562 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24564 * gcc.c (handle_braces): Support escaping in switch matching
24566 * doc/invoke.texi (Spec Files): Document it.
24567 Remove superfluous @code markup in items.
24569 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24571 * diagnostic-show-locus.c (struct column_range): New struct.
24572 (get_affected_columns): New function.
24573 (get_printed_columns): New function.
24574 (struct correction): New struct.
24575 (correction::ensure_capacity): New function.
24576 (correction::ensure_terminated): New function.
24577 (struct line_corrections): New struct.
24578 (line_corrections::~line_corrections): New dtor.
24579 (line_corrections::add_hint): New function.
24580 (layout::print_trailing_fixits): Reimplement in terms of the new
24582 (selftest::test_overlapped_fixit_printing): New function.
24583 (selftest::diagnostic_show_locus_c_tests): Call it.
24585 2017-05-03 Nathan Sidwell <nathan@acm.org>
24587 Canonicalize canonical type hashing
24588 * tree.h (type_hash_canon_hash): Declare.
24589 * tree.c (type_hash_list, attribute_hash_list): Move into
24590 type_hash_canon_hash.
24591 (build_type_attribute_qual_variant): Break out hash code calc into
24592 type_hash_canon_hash.
24593 (type_hash_canon_hash): New. Generic type hash computation.
24594 (build_range_type_1, build_array_type_1, build_function_type,
24595 build_method_type_directly, build_offset_type, build_complex_type,
24596 make_vector_type): Call it.
24598 2017-05-03 Richard Biener <rguenther@suse.de>
24600 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
24601 When all DRs have unknown misaligned do not always peel
24602 when there is a store but apply the same costing model as if
24603 there were only loads.
24605 2017-05-03 Richard Biener <rguenther@suse.de>
24608 PR tree-optimization/80492
24609 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
24610 compare_base_decls returning dont-know properly.
24612 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
24614 * config/arm/iterators.md (CCSI): New mode iterator.
24615 (arch): New mode attribute.
24616 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
24617 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
24618 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
24619 code iterator for success result mode.
24620 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
24621 the corresponding new insn generators.
24623 2017-05-03 Bin Cheng <bin.cheng@arm.com>
24626 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24627 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24629 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
24631 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
24632 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
24633 (DDR_A): Wrap DDR argument in brackets.
24634 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
24635 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
24636 (DDR_REVERSED_P): Likewise.
24638 2017-05-03 Jakub Jelinek <jakub@redhat.com>
24640 PR tree-optimization/79472
24641 * tree-switch-conversion.c (struct switch_conv_info): Add
24642 contiguous_range and default_case_nonstandard fields.
24643 (collect_switch_conv_info): Compute contiguous_range and
24644 default_case_nonstandard fields, don't clear final_bb if
24645 contiguous_range and only the default case doesn't have the required
24647 (check_all_empty_except_final): Set default_case_nonstandard instead
24648 of failing if contiguous_range and the default case doesn't have empty
24650 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
24651 and only the default case doesn't have the required constants. Skip
24653 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
24654 if default_case_nonstandard.
24655 (build_constructors): Build constant 1 just once. Assert that default
24656 values aren't inserted in between cases if contiguous_range. Skip
24658 (build_arrays): Skip virtual phis.
24659 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
24660 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
24661 Handle virtual phis.
24662 (gen_inbound_check): Handle default_case_nonstandard case.
24663 (process_switch): Adjust check_final_bb caller. Call
24664 gather_default_values with the first non-default case instead of
24665 default case if default_case_nonstandard.
24667 2017-05-02 Nathan Sidwell <nathan@acm.org>
24669 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
24670 check. Fix formatting.
24672 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
24674 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
24675 errors when comparing specialized and unspecialized times.
24677 2017-05-02 David Malcolm <dmalcolm@redhat.com>
24679 * diagnostic-show-locus.c
24680 (layout::should_print_annotation_line_p): Make private.
24681 (layout::print_annotation_line): Make private.
24682 (layout::annotation_line_showed_range_p): Make private.
24683 (layout::show_ruler): Make private.
24684 (layout::print_source_line): Make private. Pass in line and
24685 line_width, rather than calling location_get_source_line. Drop
24687 (layout::print_leading_fixits): New method.
24688 (layout::print_any_fixits): Rename to...
24689 (layout::print_trailing_fixits): ...this, and make private.
24690 Don't print newline fixits.
24691 (diagnostic_show_locus): Move logic for printing one row into...
24692 (layout::print_line): ...this new function. Move the
24693 location_get_source_line call and error-handling from
24694 print_source_line to here. Call print_leading_fixits, and rename
24695 print_any_fixits to print_trailing_fixits.
24696 (selftest::test_fixit_insert_containing_newline): Update now that
24697 newlines are partially supported.
24698 (selftest::test_fixit_insert_containing_newline_2): New test.
24699 (selftest::test_fixit_replace_containing_newline): Update comments.
24700 (selftest::diagnostic_show_locus_c_tests): Call the new test.
24701 * edit-context.c (class added_line): New class.
24702 (class edited_line): Describe newline handling in comment.
24703 (edited_line::actually_edited_p): New method.
24704 (edited_line::print_content): Delete redundant decl.
24705 (edited_line::m_predecessors): New field.
24706 (edited_file::print_content): Call edited_line::print_content.
24707 (edited_file::print_diff): Update to support newlines.
24708 (edited_file::print_diff_hunk): Likewise.
24709 (edited_file::print_run_of_changed_lines): New function.
24710 (edited_file::print_diff_line): Convert to...
24711 (print_diff_line): ...this.
24712 (edited_file::get_effective_line_count): New function.
24713 (edited_line::edited_line): Initialize new field m_predecessors.
24714 (edited_line::~edited_line): Clean up m_predecessors.
24715 (edited_line::apply_fixit): Handle newlines.
24716 (edited_line::get_effective_line_count): New function.
24717 (edited_line::print_content): New function.
24718 (edited_line::print_diff_lines): New function.
24719 (selftest::test_applying_fixits_insert_containing_newline): New
24721 (selftest::test_applying_fixits_replace_containing_newline): New
24723 (selftest::insert_line): New function.
24724 (selftest::test_applying_fixits_multiple_lines): Add example of
24726 (selftest::edit_context_c_tests): Call the new tests.
24728 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24730 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
24731 parameter cand. Update dump information.
24732 (get_computation_cost): Update uses.
24734 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24736 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
24737 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
24738 (get_computation_at, rewrite_use_address): Update use of
24739 get_computation_aff.
24741 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24743 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
24744 (get_computation): Delete.
24745 (get_computation_cost): Implement like get_computation_cost_at.
24746 Use get_computation_at.
24747 (get_computation_cost_at): Delete.
24748 (rewrite_use_nonlinear_expr): Use get_computation_at.
24749 (rewrite_use_compare, remove_unused_ivs): Ditto.
24751 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24753 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
24755 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24757 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
24758 (ivopts_global_cost_for_size): Rename parameter and update uses.
24759 (iv_ca_recount_cost): Update uses.
24760 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
24761 candidates seperately in n_invs and n_cands.
24762 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
24764 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24766 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
24767 (find_inv_vars_cb): New.
24768 (find_depends): Renamed to ...
24769 (find_inv_vars): ... this.
24770 (add_candidate_1, force_var_cost): Call find_inv_vars.
24771 (split_address_cost, determine_group_iv_cost_cond): Ditto.
24773 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24775 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
24776 inv_vars. Add inv_exprs.
24777 (struct iv_cand): Rename depends_on to inv_vars.
24778 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
24779 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
24780 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
24781 (dump_cand): Dump inv_vars.
24782 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
24783 (record_invariant, find_depends, add_candidate_1): Ditto.
24784 (set_group_iv_cost, force_var_cost): Ditto.
24785 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
24786 (get_computation_cost_at, get_computation_cost): Ditto.
24787 (determine_group_iv_cost_generic): Ditto.
24788 (determine_group_iv_cost_address): Ditto.
24789 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
24790 (determine_group_iv_costs): Ditto.
24791 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
24792 (iv_ca_set_remove_invariants): Renamed to ...
24793 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
24794 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
24795 (iv_ca_set_add_invariants): Renamed to ...
24796 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
24797 (iv_ca_set_cp): Use iv_ca_set_add_invs.
24798 (iv_ca_has_deps): Support inv_vars and inv_exprs.
24799 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
24800 (create_new_ivs): Remove useless dump.
24802 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24804 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
24806 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
24807 (iv_ca_set_no_cp, create_new_iv): Ditto.
24809 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24811 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24813 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
24815 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
24816 function tree_check2.
24818 2017-05-02 Martin Liska <mliska@suse.cz>
24820 * doc/gcov.texi: Add missing preposition.
24821 * gcov.c (function_info::function_info): Properly fill up
24822 all member variables.
24824 2017-05-02 Tamar Christina <tamar.christina@arm.com>
24826 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
24828 2017-05-02 Tamar Christina <tamar.christina@arm.com>
24830 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
24832 2017-05-02 Martin Liska <mliska@suse.cz>
24835 * lto-streamer-in.c (lto_read_tree_1): Remove
24836 LTO_STREAMER_DEBUG.
24837 * lto-streamer.c (struct tree_hash_entry): Likewise.
24838 (struct tree_entry_hasher): Likewise.
24839 (tree_entry_hasher::hash): Likewise.
24840 (tree_entry_hasher::equal): Likewise.
24841 (lto_streamer_init): Likewise.
24842 (lto_orig_address_map): Likewise.
24843 (lto_orig_address_get): Likewise.
24844 (lto_orig_address_remove): Likewise.
24845 * lto-streamer.h: Likewise.
24846 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
24847 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
24849 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
24851 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
24852 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
24853 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
24854 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
24855 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
24856 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
24857 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
24858 (mm_maskz_sub_ss): New intrinsics.
24859 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24860 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
24861 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
24862 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
24863 (__builtin_ia32_subss_mask_round): New builtins.
24864 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24865 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24866 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
24868 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
24869 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
24871 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
24874 2017-05-02 Martin Jambor <mjambor@suse.cz>
24876 PR tree-optimization/78687
24877 * tree-sra.c (access): New field parent.
24878 (process_subtree_disqualification): New function.
24879 (disqualify_candidate): Call it.
24880 (build_accesses_from_assign): Reset write flag if creating an
24882 (build_access_subtree): Fill in parent field and also prpagate
24883 down grp_write flag.
24884 (create_artificial_child_access): New parameter set_grp_write, set
24885 grp_write to its value.
24886 (propagate_subaccesses_across_link): Also propagate grp_write flag
24888 (propagate_all_subaccesses): Push the closest parent back to work
24889 queue if add_access_to_work_queue returned true.
24891 2017-05-02 Richard Biener <rguenther@suse.de>
24893 * common.opt (fstrict-overflow): Alias negative to fwrapv.
24894 * doc/invoke.texi (fstrict-overflow): Remove all traces of
24895 -fstrict-overflow documentation.
24896 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
24897 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
24898 flag_strict_overflow.
24899 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
24900 * lto-opts.c (lto_write_options): Do not stream it.
24901 * lto-wrapper.c (merge_and_complain): Do not handle it.
24902 * opts.c (default_options_table): Do not set -fstrict-overflow.
24903 (finish_options): Likewise do not clear it when sanitizing.
24904 * simplify-rtx.c (simplify_const_relational_operation): Do not
24905 test flag_strict_overflow.
24907 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
24909 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
24910 using enabled attribute.
24911 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
24912 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
24913 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
24914 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
24915 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
24916 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
24917 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
24918 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
24919 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
24920 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
24922 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
24924 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
24926 2017-05-02 Richard Biener <rguenther@suse.de>
24928 PR tree-optimization/80591
24930 2017-04-10 Richard Biener <rguenther@suse.de>
24932 * tree-ssa-structalias.c (find_func_aliases): Properly handle
24935 2017-05-02 Richard Biener <rguenther@suse.de>
24937 PR tree-optimization/80549
24938 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
24939 (cleanup_tree_cfg_noloop): Create forwarders to known loop
24940 headers if they do not have a preheader.
24942 2017-05-02 Martin Liska <mliska@suse.cz>
24945 * common.opt: Fix typo.
24946 * doc/invoke.texi: Likewise.
24948 2017-05-01 Jan Beulich <jbeulich@suse.com>
24950 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
24951 swapping, add (x,x,m,x,n) alternative.
24953 2017-05-01 Nathan Sidwell <nathan@acm.org>
24955 * calls.c (combine_pending_stack_adjustment_and_call): Remove
24956 unnecessary unadjusted_alignment check.
24958 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
24961 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
24963 * gimplify.c (gimplify_cilk_detach): New function.
24964 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
24965 * tree-core.h: Document EXPR_CILK_SPAWN.
24966 * tree.h (EXPR_CILK_SPAWN): Define.
24968 2017-05-01 David Malcolm <dmalcolm@redhat.com>
24970 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
24971 to use new fixit_hint representation, using the "replace" logic.
24972 (get_line_span_for_fixit_hint): Likewise.
24973 (layout::print_any_fixits): Likewise.
24974 (selftest::test_one_liner_many_fixits): Rename to...
24975 (selftest::test_one_liner_many_fixits_1): ...this, and update
24976 comment and expected output to reflect that the multiple fix-it
24977 hints are now consolidated into one insertion.
24978 (selftest::test_one_liner_many_fixits_2): New test.
24979 (selftest::test_diagnostic_show_locus_one_liner): Update for
24981 (selftest::test_fixit_consolidation): Update for fix-it API
24983 * diagnostic.c (print_parseable_fixits): Likewise.
24984 * edit-context.c (edited_line::m_line_events): Convert from
24985 auto_vec <line_event *> to auto_vec <line_event>.
24986 (class line_event): Convert from abstract base class to a concrete
24987 class, taking over the role of replace_event.
24988 (class insert_event): Delete.
24989 (class replace_event): Rename to class line_event. Convert to
24991 (edit_context::add_fixits): Reimplement.
24992 (edit_context::apply_insert): Delete.
24993 (edit_context::apply_replace): Rename to...
24994 (edit_context::apply_fixit): ...this. Convert to half-open range.
24995 (edited_file::apply_insert): Delete.
24996 (edited_file::apply_replace): Rename to...
24997 (edited_file::apply_fixit): ...this.
24998 (edited_line::~edited_line): Drop deletion of events.
24999 (edited_line::apply_insert): Delete.
25000 (edited_line::apply_replace): Rename to...
25001 (edited_line::apply_fixit): ...this. Convert to half-open range.
25002 Update for change to type of m_line_events.
25003 * edit-context.h (edit_context::apply_insert): Delete.
25004 (edit_context::apply_replace): Rename to...
25005 (edit_context::apply_fixit): ...this.
25007 2017-05-01 Martin Sebor <msebor@redhat.com>
25009 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
25012 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
25015 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
25016 __get_cpuid_max returns 0.
25017 (__get_cpuid_count): Ditto.
25019 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
25021 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
25022 replacement expression is another instance of one of its arguments.
25024 2017-05-01 Jakub Jelinek <jakub@redhat.com>
25027 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
25028 check for stack push/pop autoinc.
25029 * config/i386/i386.c (ix86_agi_dependent): Return false
25030 if the only reason why modified_in_p returned true is that
25031 addr is SP based and set_insn is a push or pop.
25033 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25035 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
25038 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25041 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
25042 (account_size_time): Use two predicates - exec_pred and
25044 (evaluate_conditions_for_known_args): Compute both clause and
25046 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
25047 (inline_summary_t::duplicate): Update.
25048 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
25050 (compute_inline_parameters): Likewise.
25051 (estimate_edge_size_and_time): Update caluclation of time.
25052 (estimate_node_size_and_time): Compute both time and nonspecialized
25054 (estimate_ipcp_clone_size_and_time): Update.
25055 (inline_merge_summary): Update.
25056 (do_estimate_edge_time): Update.
25057 (do_estimate_edge_size): Update.
25058 (do_estimate_edge_hints): Update.
25059 (inline_read_section, inline_write_summary): Stream both new predicates.
25060 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
25062 (compute_inlined_call_time): Cleanup.
25063 (big_speedup_p): Update.
25064 (edge_badness): Update.
25065 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
25066 (size_time_entry): Replace predicate by exec_predicate and
25067 nonconst_predicate.
25068 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
25069 (estimate_edge_time): Return also nonspec_time.
25070 (reset_edge_growth_cache): Update.
25072 2017-04-29 Jakub Jelinek <jakub@redhat.com>
25074 PR rtl-optimization/80491
25075 * ifcvt.c (noce_process_if_block): When looking for x setter
25076 with missing else_bb, don't check only the insn right before
25077 cond_earliest, but look for the last insn that x is modified in
25078 within the same bb.
25080 PR rtl-optimization/80491
25081 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
25083 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
25085 PR tree-optimization/80487
25086 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
25088 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25090 PR tree-optimization/79697
25091 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
25092 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
25093 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
25095 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
25096 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
25098 2017-04-28 Martin Sebor <msebor@redhat.com>
25100 PR tree-optimization/80523
25101 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
25102 (init_target_to_host_charmap, target_to_host, target_strtol10): New
25104 (maybe_warn, format_directive, parse_directive): Use new functions.
25105 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
25107 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
25109 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
25111 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
25113 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
25114 target_header_dir): Set correctly.
25115 * configure: Regenerated.
25116 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25117 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25118 instead of SYSTEM_HEADER_DIR.
25120 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
25122 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
25123 (estimate_local_effects): Likewise.
25124 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
25125 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
25126 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
25127 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
25128 do_estimate_edge_time, estimate_edge_time): Likewise.
25129 * ipa-inline-analysis.c (estimate_node_size_and_time,
25130 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
25131 (estimate_time_after_inlining): Remove.
25133 2017-04-28 Martin Liska <mliska@suse.cz>
25135 * doc/gcov.texi: Enhance documentation of gcov.
25137 2017-04-28 Martin Liska <mliska@suse.cz>
25139 * doc/gcov.texi: Sort options in alphabetic order.
25140 * doc/gcov-dump.texi: Likewise.
25141 * doc/gcov-tool.texi: Likewise.
25142 * gcov.c (print_usage): Likewise.
25143 * gcov-dump.c (print_usage): Likewise.
25144 * gcov-tool.c (print_merge_usage_message): Likewise.
25145 (print_rewrite_usage_message): Likewise.
25146 (print_overlap_usage_message): Likewise.
25148 2017-04-28 Martin Liska <mliska@suse.cz>
25150 PR gcov-profile/53915
25151 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
25153 2017-04-28 Martin Liska <mliska@suse.cz>
25155 PR gcov-profile/79891
25156 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
25157 is marked by compiler as living on a line.
25158 (get_cycles_count): Remove usage of the union.
25159 (output_intermediate_file): Likewise.
25160 (find_source): Fix GNU coding style.
25161 (accumulate_line_counts): Remove old non-all block mode.
25162 (output_lines): Remove usage of the union.
25163 * profile.c (output_location): Include all BBs, even if
25164 belonging to a same line (and file) as a previous BB.
25166 2017-04-28 Martin Liska <mliska@suse.cz>
25168 * gcov.c (process_args): Handle new argument 'w'.
25169 (read_graph_file): Assign ID to BBs.
25170 (output_branch_count): Display BB # if verbose flag is set.
25171 (output_lines): Likewise for arcs.
25172 (print_usage): Add '--verbose' option help.
25173 * doc/gcov.texi: Document --verbose (-w) option.
25175 2017-04-28 Martin Liska <mliska@suse.cz>
25177 * gcov.c (struct block_location_info): New struct.
25178 (process_file): Fill up the new structure.
25179 (read_graph_file): Replace usage of encoding by the newly added
25181 (add_line_counts): Likewise.
25182 (accumulate_line_counts): Remove usage of the union.
25183 (function_info::function_info): New function.
25184 (function_info::~function_info): Likewise.
25185 (process_file): Call delete instead of release_function.
25186 (release_function): Release the function.
25187 (release_structures): Call delete instead of release_function.
25188 (solve_flow_graph): Replace usage of num_blocks.
25189 (find_exception_blocks): Likewise.
25190 (output_lines): Fix GNU coding style.
25192 2017-04-28 Martin Liska <mliska@suse.cz>
25195 * coverage.c (coverage_remove_note_file): New function.
25196 * coverage.h: Declare the function.
25197 * toplev.c (finalize): Clean if an error has been seen.
25199 2017-04-28 Martin Liska <mliska@suse.cz>
25201 PR gcov-profile/80031
25202 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
25203 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
25204 * gcov.c (read_graph_file): Read just number of blocks.
25205 * profile.c (branch_prob): Do not stream 0 flags per a basic
25208 2017-04-28 Martin Liska <mliska@suse.cz>
25210 * gcov-dump.c (tag_*): Add new argument to declarations.
25211 (dump_gcov_file): Likewise.
25212 (tag_blocks): Add and use new argument depth.
25213 (tag_arcs): Likewise.
25214 (tag_lines): Likewise.
25215 (tag_counters): Likewise.
25216 (tag_summary): Likewise.
25217 (dump_working_sets): Use depth to do a proper indentation.
25219 2017-04-28 Jakub Jelinek <jakub@redhat.com>
25222 * cgraph.h (symtab_node::debug_symtab): No longer inline.
25223 * symtab.c (symtab_node::debug_symtab): Move definition here.
25225 2017-04-28 Richard Biener <rguenther@suse.de>
25227 * lto-streamer.h (LTO_major_version): Bump to 7.
25229 2017-04-28 Richard Biener <rguenther@suse.de>
25231 * tree-vrp.c (assert_info): New struct.
25232 (add_assert_info): New helper.
25233 (register_edge_assert_for_2): Refactor to add asserts to a vector
25235 (register_edge_assert_for_1): Likewise.
25236 (register_edge_assert_for): Likewise.
25237 (finish_register_edge_assert_for): New helper actually registering
25238 asserts where live on edge.
25239 (find_conditional_asserts): Adjust.
25240 (find_switch_asserts): Likewise.
25241 (evrp_dom_walker::try_find_new_range): Generalize.
25242 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
25244 2017-04-27 Marek Polacek <polacek@redhat.com>
25247 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
25248 arg10 and arg11 to itype.
25250 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
25252 * doc/extend.texi (Object Size Checking): Improve grammar.
25254 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
25257 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
25258 that the logic for permitting reciprocal estimates matches that
25261 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25264 * tree.c (type_cache_hasher::equal): Only compare
25265 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
25266 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
25267 non-aggregate element types.
25268 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
25269 about the flag on ARRAY_TYPEs in the comment, formatting fix.
25271 2017-04-27 Richard Biener <rguenther@suse.de>
25273 PR middle-end/80533
25274 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
25275 stripping ARRAY_REFs from MEM_EXPR make sure we're not
25276 keeping a reference to a trailing array.
25278 2017-04-27 Richard Biener <rguenther@suse.de>
25280 PR middle-end/80539
25281 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
25282 being in loop-closed SSA form conservatively.
25283 (chrec_fold_multiply_poly_poly): Likewise.
25285 2017-04-27 Tamar Christina <tamar.christina@arm.com>
25287 PR middle-end/79665
25288 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
25289 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
25291 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25294 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
25295 (aarch64_function_arg_alignment): Return unsigned int again, but still
25296 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
25297 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
25298 Don't emit -Wpsabi note.
25299 (aarch64_function_arg_boundary): Likewise.
25300 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
25303 2017-04-26 Nathan Sidwell <nathan@acm.org>
25305 * tree.h (crc32_unsigned_n): Declare.
25306 (crc32_unsigned, crc32_unsigned): Make inline.
25307 * tree.c (crc32_unsigned_bits): Replace with ...
25308 (crc32_unsigned_n): ... this.
25309 (crc32_unsigned, crc32_byte): Remove.
25310 (crc32_string): Remove unnecessary braces.
25312 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25314 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
25315 * ipa-inline-analysis.c (MAX_TIME): Remove.
25316 (account_size_time): Use sreal for time.
25317 (dump_inline_summary): Update.
25318 (estimate_function_body_sizes): Update.
25319 (estimate_edge_size_and_time): Update.
25320 (estimate_calls_size_and_time): Update.
25321 (estimate_node_size_and_time): Update.
25322 (inline_merge_summary): Update.
25323 (inline_update_overall_summary): Update.
25324 (estimate_time_after_inlining): Update.
25325 (inline_read_section): Update.
25326 (inline_write_summary): Update.
25327 * ipa-inline.c (compute_uninlined_call_time): Update.
25328 (compute_inlined_call_time): Update.
25329 (recursive_inlining): Update.
25330 (inline_small_functions): Update.
25331 (dump_overall_stats): Update.
25332 * ipa-inline.h: Include sreal.h.
25333 (size_time_entry): Turn time to sreal.
25334 (inline_summary): Turn self_time nad time to sreal.
25336 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25338 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
25340 (sreal::stream_out, sreal::stream_in): New.
25341 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
25343 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25345 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
25348 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
25351 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
25353 (dimode_scalar_chain::compute_convert_gain): Ditto.
25354 (dimode_scalar_chain::make_vector_copies): Ditto.
25355 (dimode_scalar_chain::convert_reg): Ditto.
25356 (dimode_scalar_chain::convert_insn): Ditto.
25357 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
25358 (VI248_AVX512BW_1): New mode iterator.
25359 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
25360 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
25363 2017-04-25 Martin Sebor <msebor@redhat.com>
25365 PR tree-optimization/80497
25366 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
25367 constants are representable in HOST_WIDE_INT.
25368 (parse_directive): Ditto.
25370 2017-04-25 Martin Sebor <msebor@redhat.com>
25373 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
25374 (new_zero_array): Adjust signature.
25375 (dom_info::dom_init): Used unsigned rather that size_t.
25376 (dom_info::dom_info): Same.
25378 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25379 Jakub Jelinek <jakub@redhat.com>
25382 * config/arm/arm.c: Include gimple.h.
25383 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
25384 returns negative, increment ncrn only if it returned positive.
25385 (arm_needs_doubleword_align): Return int instead of bool,
25386 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
25387 members, but if there is any such non-FIELD_DECL
25388 > PARM_BOUNDARY aligned decl, return -1 instead of false.
25389 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
25390 returns negative, increment nregs only if it returned positive.
25391 (arm_setup_incoming_varargs): Likewise.
25392 (arm_function_arg_boundary): Emit -Wpsabi note if
25393 arm_needs_doubleword_align returns negative, return
25394 DOUBLEWORD_ALIGNMENT only if it returned positive.
25396 2017-04-25 Marek Polacek <polacek@redhat.com>
25399 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
25400 first argument to type.
25402 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
25405 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
25406 type checks to test for compatibility instead of equality.
25408 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25409 Jakub Jelinek <jakub@redhat.com>
25412 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
25414 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
25415 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
25416 the alignment computation, but return their maximum in warn_alignment.
25417 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
25418 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
25420 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
25421 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
25424 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25426 * config/arc/simdext.md (dmpyh): Fix typo.
25428 2017-04-25 Richard Biener <rguenther@suse.de>
25430 PR tree-optimization/80492
25431 * alias.c (compare_base_decls): Handle registers with asm
25432 specification conservatively.
25433 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
25434 compare_base_decls returning dont-know properly.
25436 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25438 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
25439 (legitimate_offset_address_p): New function.
25440 (arc_legitimate_address_p): Use above function.
25442 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25444 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
25446 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25448 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
25449 ACCH registers whenever they are available.
25451 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25453 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
25454 double regs fix when not used.
25456 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25458 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
25460 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
25461 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
25463 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25465 * config/arc/arc.c (arc_output_addsi): Check for h-register class
25466 when emitting short ADD instructions.
25468 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25470 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
25472 (cmpsi_cc_c_insn): Likewise.
25473 (cbranchsi4_scratch): Compute proper instruction length using
25474 compact_hreg_operand.
25475 * config/arc/predicates.md (compact_hreg_operand): New predicate.
25477 2017-04-25 Richard Biener <rguenther@suse.de>
25479 PR middle-end/80509
25480 * passes.c (pass_manager::pass_manager): Initialize
25481 m_name_to_pass_map.
25483 2017-04-25 Richard Biener <rguenther@suse.de>
25485 PR tree-optimization/79201
25486 * tree-ssa-sink.c (statement_sink_location): Handle calls.
25488 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25491 * config/s390/vector.md: Split MEM->GPR vector moves for
25492 non-s_operand addresses.
25494 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25497 * config/s390/predicates.md (reload_const_wide_int_operand): New
25499 * config/s390/s390.md ("movti"): Remove d/P alternative.
25500 ("movti_bigconst"): New pattern definition.
25502 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25505 * s390-protos.h (s390_expand_cs_hqi): Removed.
25506 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
25507 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
25508 modes as well as CCZ1mode and CCZmode.
25509 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
25510 signature of s390_emit_compare_and_swap.
25511 (s390_expand_cs_hqi): Likewise, make static.
25512 (s390_expand_cs_tdsi): Generate an explicit compare before trying
25513 compare-and-swap, in some cases.
25514 (s390_expand_cs): Wrapper function.
25515 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
25517 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
25518 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
25519 patterns for small and large integers. Forbid symref memory operands.
25520 Move expander to s390.c. Require cc register.
25521 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
25522 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
25523 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
25524 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
25525 symref memory operands. Remove CC mode and call s390_match_ccmode
25527 ("atomic_exchange<mode>"): Allow and implement all integer modes.
25529 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25531 * config/s390/s390.md (define_peephole2): New peephole to help
25532 combining the load-and-test pattern with volatile memory.
25534 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25536 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
25537 with CCZmode for TARGET_Z196.
25539 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25541 PR rtl-optimization/80501
25542 * combine.c (make_compound_operation_int): Set subreg_code to SET
25543 even for AND with mask of the sign bit of mode.
25545 PR rtl-optimization/80500
25546 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
25547 sum's initial value.
25549 2017-04-25 Julian Brown <julian@codesourcery.com>
25550 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25552 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
25554 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
25556 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
25558 2017-04-25 Julian Brown <julian@codesourcery.com>
25559 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25561 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
25562 (thunderx2t99_sha): New Reservation.
25564 2017-04-25 Julian Brown <julian@codesourcery.com>
25565 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25567 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
25568 type for 1-element load.
25570 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
25572 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
25574 2017-04-24 Martin Jambor <mjambor@suse.cz>
25576 PR tree-optimization/80293
25577 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
25578 char arrays not totally scalarizable if it is false.
25579 (analyze_all_variable_accesses): Pass correct value in the new
25580 parameter. Add a statistics counter.
25582 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
25584 PR middle-end/79931
25585 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
25587 2017-04-24 Richard Biener <rguenther@suse.de>
25589 PR tree-optimization/80494
25590 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
25591 out for complex types.
25593 2017-04-24 Richard Biener <rguenther@suse.de>
25595 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
25596 * tree-ssa-sccvn.c (print_scc): Print SCC size.
25597 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
25598 (DFS): Adjust and never fail.
25599 (sccvn_dom_walker::fail): Remove.
25600 (sccvn_dom_walker::before_dom_children): Adjust.
25601 (run_scc_vn): Likewise and never fail.
25602 * tree-ssa-pre.c (pass_pre::execute): Adjust.
25603 (pass_fre::execute): Likewise.
25605 2017-04-24 Richard Biener <rguenther@suse.de>
25607 PR tree-optimization/79725
25608 * tree-ssa-sink.c (statement_sink_location): Return whether
25609 failure reason was zero uses. Move that check later.
25610 (sink_code_in_bb): Deal with zero uses by removing the stmt
25613 2017-04-24 Richard Biener <rguenther@suse.de>
25616 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
25617 pointer-based references.
25619 2017-04-24 Richard Biener <rguenther@suse.de>
25622 * pass_manager.h (pass_manager::operator new): Remove.
25623 (pass_manager::operator delete): Likewise.
25624 * passes.c (pass_manager::operator new): Remove.
25625 (pass_manager::operator delete): Likewise.
25626 (pass_manager::pass_manager): Zero individual pass members.
25628 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
25631 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
25632 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
25633 Check "XEXP (src, 1)" operand here.
25634 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
25635 Check "XEXP (src, 1)" operand here.
25636 (dimode_scalar_chain::make_vector_copies): Detect count register
25637 of a shift instruction. Zero extend count register from QImode
25638 to DImode to satisfy vector shift pattern count operand predicate.
25639 Substitute vector shift count operand with a DImode copy.
25640 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
25643 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
25645 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
25646 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
25647 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
25648 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
25649 (UNSPEC_NOREX_MEM): Remove definition.
25651 2017-04-21 Richard Biener <rguenther@suse.de>
25653 PR tree-optimization/79547
25654 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
25655 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
25656 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
25657 without any constraints.
25659 2017-04-21 Richard Biener <rguenther@suse.de>
25661 PR tree-optimization/78847
25662 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
25664 2017-04-21 Richard Biener <rguenther@suse.de>
25666 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
25667 (build_distinct_type_copy): Likewise.
25668 (build_variant_type_copy): Likewise.
25669 * tree.c (build_qualified_type): Pass down mem-stat info.
25670 (build_distinct_type_copy): Likewise.
25671 (build_variant_type_copy): Likewise.
25673 2017-04-21 Richard Biener <rguenther@suse.de>
25675 PR tree-optimization/80237
25676 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
25678 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
25679 for a simplified result.
25681 2016-04-21 Richard Biener <rguenther@suse.de>
25683 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
25684 sth as strict as a simple_iv but a chrec without symbols and an
25685 operand defined in the loop we are peeling (and not some subloop).
25686 (propagate_constants_for_unrolling): Propagate all constants.
25688 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25691 * config/i386/i386.c (print_reg): Remove assert for disalowed
25692 regno values, call output_operand_lossage instead.
25694 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25697 * config/i386/constraints.md (Yc): New register constraint.
25698 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
25699 Use Yc constraint for alternative 2 of operand 0. Remove
25700 preferred_for_speed attribute.
25702 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
25704 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
25705 lastprivate clauses in SIMT case.
25707 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
25709 * doc/invoke.texi (-Wextra-semi): Document new warning option.
25711 2017-04-20 Richard Biener <rguenther@suse.de>
25713 PR tree-optimization/57796
25714 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
25715 as N scalar stores.
25716 (vect_model_load_cost): Cost gathers as N scalar loads.
25718 2017-04-20 Richard Biener <rguenther@suse.de>
25720 * ggc-page.c (ggc_allocated_p): Rename to ...
25721 (safe_lookup_page_table_entry): ... this and return the lookup
25723 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
25725 2017-04-20 Richard Biener <rguenther@suse.de>
25727 PR tree-optimization/80453
25728 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
25729 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
25730 from the conditions.
25731 (vn_phi_eq): Pass them down.
25732 (vn_phi_lookup): Record them.
25733 (vn_phi_insert): Likewise.
25735 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
25737 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
25738 uninitialized variable warning to avoid buffer overrun.
25740 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
25743 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
25744 is suppressed for '{ 0 }' in C.
25746 2017-04-20 Jakub Jelinek <jakub@redhat.com>
25748 * BASE-VER: Set to 8.0.0.
25750 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
25752 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
25753 priority .init_array and .fini_array section with SECTION_NOTYPE
25756 2017-04-20 Jakub Jelinek <jakub@redhat.com>
25758 PR middle-end/80423
25759 * tree.h (build_array_type): Add typeless_storage default argument.
25760 * tree.c (type_cache_hasher::equal): Also compare
25761 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
25762 (build_array_type): Add typeless_storage argument, set
25763 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
25765 (build_nonshared_array_type): Adjust build_array_type_1 caller.
25766 (build_array_type): Likewise. Add typeless_storage argument.
25768 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
25769 Jakub Jelinek <jakub@redhat.com>
25771 PR tree-optimization/80426
25772 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
25773 operation on symbolic operands, also compute the overflow for the
25774 invariant part when the operation degenerates into a negation.
25776 2017-04-19 Jakub Jelinek <jakub@redhat.com>
25779 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
25780 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
25783 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
25785 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
25788 * config/avr/avr.c (tree.h): Include it.
25789 (cgraph.h): Include it.
25790 (avr_encode_section_info): Don't warn for uninitialized progmem
25791 variable if it's just an alias.
25793 2017-04-19 Richard Biener <rguenther@suse.de>
25796 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
25797 when needed by AutoPGO.
25799 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
25802 * doc/lto.texi: Remove an extra 'that'.
25804 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
25806 PR rtl-optimization/80429
25807 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
25808 are only used in debug insns.
25810 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
25811 Vladimir Makarov <vmakarov@redhat.com>
25813 * config/sparc/predicates.md (input_operand): Add comment. Return
25814 true for any memory operand when LRA is in progress.
25815 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
25817 2017-04-18 Jeff Law <law@redhat.com>
25820 * mips.md ({return,simple_return}_internal): Do not overwrite
25823 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25825 PR tree-optimization/80443
25826 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
25827 instead of adding 1, subtract -1 and similarly instead of subtracting
25830 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
25832 PR rtl-optimization/80357
25833 * haifa-sched.c (tmp_bitmap): New variable.
25834 (model_recompute): Handle duplicate use records.
25835 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
25836 (free_global_sched_pressure_data): Free it.
25838 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
25841 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
25842 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25843 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25844 instead of SYSTEM_HEADER_DIR.
25846 2017-04-18 Jeff Law <law@redhat.com>
25848 PR middle-end/80422
25849 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
25850 predecessors after walking up the insn chain.
25852 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25855 * dwarf2out.c (modified_type_die): Try harder not to emit internal
25856 sizetype type into debug info.
25858 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
25861 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
25862 unneeded test for TARGET_UPPER_REGS_SF.
25863 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
25865 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25868 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
25869 instead of gsi_after_labels.
25871 2017-04-18 Jeff Law <law@redhat.com>
25873 * regcprop.c (maybe_mode_change): Avoid creating copies of the
25877 2017-04-13 Jeff Law <law@redhat.com>
25878 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
25879 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
25881 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
25884 * config/avr/avr.c (intl.h): Include it.
25885 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
25887 2017-04-18 Martin Liska <mliska@suse.cz>
25889 PR gcov-profile/78783
25890 * gcov-tool.c (gcov_output_files): Validate that destination
25891 file is either removed by the tool or by a user.
25893 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
25894 Guy Benyei <guybe@mellanox.com>
25896 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
25897 block, and do not negate it, the stored id is already negative.
25899 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
25901 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
25903 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
25906 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
25907 masks of options that should be turned off if the VSX vector
25908 options are turned off.
25909 (OTHER_P8_VECTOR_MASKS): Likewise.
25910 (OTHER_VSX_VECTOR_MASKS): Likewise.
25911 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
25912 rs6000_disable_incompatible_switches to validate no type switches
25914 (rs6000_incompatible_switch): New function to disallow turning on
25915 other vector options if -mno-vsx, -mno-power8-vector, or
25916 -mno-power9-vector are specified.
25918 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
25920 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
25922 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
25924 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
25925 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
25926 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
25927 (ARG_POINTER_CFA_OFFSET): Likewise.
25929 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
25931 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
25932 conditions to take advantage of various optimizations.
25934 2017-04-13 Jeff Law <law@redhat.com>
25936 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
25937 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
25938 (zero_extendsidi2_dext): Likewise.
25940 2017-04-13 Jakub Jelinek <jakub@redhat.com>
25943 * fold-const.c (fold_ternary_loc): Revert
25944 use op0 instead of fold_convert_loc (loc, type, arg0) part of
25947 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
25949 PR rtl-optimization/80343
25950 * lra-remat.c (update_scratch_ops): Assign original hard reg to
25951 new scratch pseudo.
25953 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
25956 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
25957 to ubsan_encode_value.
25959 2017-04-13 Jeff Law <law@redhat.com>
25961 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
25962 appearing in DEBUG_INSNs.
25964 2017-04-13 Martin Liska <mliska@suse.cz>
25966 PR gcov-profile/80413
25967 * gcov-io.c (gcov_write_string): Copy to buffer just when
25968 allocated size is greater than zero.
25970 2017-04-13 Jakub Jelinek <jakub@redhat.com>
25973 * dwarf2out.c (decls_for_scope): Ignore declarations of
25974 current_function_decl in BLOCK_NONLOCALIZED_VARS.
25976 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
25979 * ipa-visibility.c (non_local_p): Fix typos.
25980 (localize_node): When localizing symbol in same comdat group,
25981 dissolve the group only when we know external symbols are going
25983 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
25985 2017-04-12 Jakub Jelinek <jakub@redhat.com>
25987 PR tree-optimization/79390
25988 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
25989 order does not result in usable sequence, retry with reversed operand
25995 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
25996 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
25997 op0 instead of fold_convert_loc (loc, type, arg0).
25999 2017-04-12 Jeff Law <law@redhat.com>
26001 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
26002 has a delay slot in the generated code.
26004 * config/cris/cris.md (cris_preferred_reload_class): Return
26005 GENNONACR_REGS rather than GENERAL_REGS.
26007 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26010 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
26011 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
26012 signedness of the result type.
26014 2017-04-12 Richard Biener <rguenther@suse.de>
26015 Jeff Law <law@redhat.com>
26017 PR tree-optimization/80359
26018 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
26019 trim stores to TARGET_MEM_REFs.
26021 2017-04-12 Richard Biener <rguenther@suse.de>
26023 PR tree-optimization/79390
26024 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
26025 threading case even more.
26027 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
26030 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
26031 for quad_address_p for TImode, instead of just not indexed_address.
26033 2017-04-12 Richard Biener <rguenther@suse.de>
26034 Bernd Edlinger <bernd.edlinger@hotmail.de>
26036 PR middle-end/79671
26037 * alias.c (component_uses_parent_alias_set_from): Handle
26038 TYPE_TYPELESS_STORAGE.
26039 (get_alias_set): Likewise.
26040 * tree-core.h (tree_type_common): Add typeless_storage flag.
26041 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
26042 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
26043 for types containing members with TYPE_TYPELESS_STORAGE.
26044 (place_field): Likewise.
26045 (layout_type): Likewise for ARRAY_TYPE.
26046 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
26047 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
26048 TYPE_TYPELESS_STORAGE.
26049 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
26051 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26054 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
26055 first argument to type.
26057 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26061 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
26062 CONST0_RTX (mode) rather than const0_rtx where appropriate.
26063 (rs6000_expand_binop_builtin): Likewise.
26064 (rs6000_expand_ternop_builtin): Likewise; also add missing
26065 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
26066 vshasigma built-ins.
26067 * doc/extend.texi: Document that vec_xxpermdi's third argument
26068 must be a constant.
26070 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
26072 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
26073 Use shift_const cost parameter when calculating gain of STV shifts.
26075 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
26077 PR rtl-optimization/70478
26078 * lra-constraints.c (process_alt_operands): Check memory for
26079 disfavoring memory insn operand.
26081 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26083 PR middle-end/80100
26084 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
26085 left shift in unsigned HOST_WIDE_INT type.
26087 PR rtl-optimization/80385
26088 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
26089 (not (neg X)) into (plus X -1) for complex or non-integral modes.
26092 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
26093 if they have any depend clauses.
26095 2017-04-11 Martin Liska <mliska@suse.cz>
26098 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
26099 * ipa-split.c (split_function): Create a local comdat symbol
26100 if caller is in a comdat group.
26102 2017-04-11 Martin Liska <mliska@suse.cz>
26105 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
26108 2017-04-11 Martin Sebor <msebor@redhat.com>
26110 PR middle-end/80364
26111 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
26112 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
26114 (directive::set_width, directive::set_precision, format_character):
26116 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
26119 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
26122 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
26123 conflict, set target->arch_name instead of target->cpu_name.
26125 2017-04-11 Richard Biener <rguenther@suse.de>
26127 PR tree-optimization/80374
26128 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
26129 build_zero_cst, remove fold_convertible_p check again.
26131 2017-04-11 Martin Liska <mliska@suse.cz>
26134 * ubsan.c (instrument_object_size): Do not instrument register
26137 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26140 * config/i386/i386-builtin-types.def
26141 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
26142 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
26143 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
26144 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
26145 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
26146 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
26147 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
26148 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
26149 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
26150 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
26151 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
26152 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
26153 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
26154 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
26155 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
26156 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
26157 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
26158 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
26159 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
26160 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
26161 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
26162 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
26163 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
26164 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
26165 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
26166 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
26167 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
26168 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
26169 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
26170 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
26171 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
26172 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
26173 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
26174 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
26175 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
26176 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
26177 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
26178 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
26179 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
26180 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
26181 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
26182 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
26183 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
26184 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
26185 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
26186 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
26188 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
26189 flag to second_arg_count, handle 4 argument function type _COUNT
26190 aliases, handle second_arg_count on second argument rather than last.
26192 2017-04-10 Jeff Law <law@redhat.com>
26194 PR tree-optimization/80374
26195 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
26196 record anything if we can not convert integer_zero_node to the
26199 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
26202 * config/rs6000/rs6000.c (rs6000_option_override_internal):
26203 Enhance special handling given to the TARGET_P9_MINMAX option in
26204 relation to certain other options.
26206 2017-04-10 Bin Cheng <bin.cheng@arm.com>
26208 PR tree-optimization/80153
26209 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
26210 remove POINTER_PLUS_EXPR's base part directly, rather than through
26213 2017-04-10 Richard Biener <rguenther@suse.de>
26214 Bin Cheng <bin.cheng@arm.com>
26216 PR tree-optimization/80153
26217 * tree-affine.c (aff_combination_to_tree): Get base pointer from
26218 the first element of pointer type aff_tree. Build result expr in
26220 (add_elt_to_tree): Convert to type unconditionally. Remove other
26221 fold_convert calls.
26222 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
26223 (rewrite_use_nonlinear_expr): Check invariant using iv information.
26225 2017-04-10 Richard Biener <rguenther@suse.de>
26227 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26230 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
26232 PR rtl-optimization/70478
26233 * lra-constraints.c (curr_small_class_check): New.
26234 (update_and_check_small_class_inputs): New.
26235 (process_alt_operands): Update curr_small_class_check. Disfavor
26236 alternative insn memory operands. Check available regs for small
26239 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26242 * config/mips/mips.opt (-mvirt): Update description.
26243 * doc/invoke.texi (-mvirt): Likewise.
26245 2017-04-10 Richard Biener <rguenther@suse.de>
26247 PR middle-end/80362
26248 * fold-const.c (fold_binary_loc): Look at unstripped ops when
26249 looking for NEGATE_EXPR in -A / -B to A / B folding.
26251 2017-04-10 Martin Liska <mliska@suse.cz>
26253 PR gcov-profile/80224
26254 * gcov.c (print_usage): Fix usage string.
26255 (get_gcov_intermediate_filename): Remove.
26256 (output_gcov_file): Use both for normal and intermediate format.
26257 (generate_results): Do not initialize special file for
26258 intermediate format.
26260 2017-04-10 Richard Biener <rguenther@suse.de>
26262 PR tree-optimization/80304
26263 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
26266 2017-04-10 Nathan Sidwell <nathan@acm.org>
26269 * config/rs6000/rs6000.c (rs6000_vector_type): New.
26270 (rs6000_init_builtins): Use it.
26272 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26274 * config/arm/arm.md (<mrc>): Add mode to SET source.
26275 (<mrrc>): Likewise.
26277 2017-04-10 Richard Biener <rguenther@suse.de>
26279 PR middle-end/80344
26280 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
26282 2017-04-10 Jakub Jelinek <jakub@redhat.com>
26285 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
26286 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
26287 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
26288 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
26289 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
26290 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
26291 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
26292 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
26293 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
26294 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
26295 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
26296 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
26297 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
26298 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
26299 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
26300 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
26301 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
26302 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
26303 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
26304 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
26305 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
26306 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
26307 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
26309 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
26311 PR rtl-optimization/70478
26312 * lra-constraints.c: Reverse the last patch.
26314 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
26316 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
26317 Add comment for WCHAR_T.
26319 2017-04-08 Martin Liska <mliska@suse.cz>
26322 2017-04-07 Martin Liska <mliska@suse.cz>
26325 * ipa-split.c (split_function): Add function part to a same comdat
26328 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26331 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
26333 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
26335 * rs6000/rs6000.c (vec_load_pendulum): Rename...
26336 (vec_pairing): ...to this.
26337 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
26338 (rs6000_sched_init): Adjust for name change.
26339 (struct rs6000_sched_context): Likewise.
26340 (rs6000_init_sched_context): Likewise.
26341 (rs6000_set_sched_context): Likewise.
26343 2017-04-07 Jakub Jelinek <jakub@redhat.com>
26349 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
26351 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
26352 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
26353 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
26355 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
26357 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
26359 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
26361 PR rtl-optimization/70703
26362 * ira-color.c (update_conflict_hard_regno_costs): Use
26363 int64_t instead of HOST_WIDE_INT.
26365 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
26367 PR rtl-optimization/70478
26368 * lra-constraints.c (process_alt_operands): Disfavor alternative
26369 insn memory operands.
26371 2017-04-07 Jeff Law <law@redhat.com>
26373 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
26374 CALL and NOTE_INSN_CALL_ARG_LOCATION.
26376 2017-04-07 Martin Liska <mliska@suse.cz>
26379 * config/aarch64/aarch64.c (aarch64_process_target_attr):
26380 Show error message instead of an ICE.
26382 2017-04-07 Martin Liska <mliska@suse.cz>
26385 * ipa-split.c (split_function): Add function part to a same comdat
26388 2017-04-07 Richard Biener <rguenther@suse.de>
26390 PR middle-end/80341
26391 * tree.c (get_unwidened): Also handle ! for_type case for
26393 * convert.c (do_narrow): Split out from ...
26394 (convert_to_integer_1): ... here. Do not pass final truncation
26395 type to get_unwidened for TRUNC_DIV_EXPR.
26397 2017-04-07 Richard Biener <rguenther@suse.de>
26399 * tree-affine.c (wide_int_ext_for_comb): Take type rather
26401 (aff_combination_const): Adjust.
26402 (aff_combination_scale): Likewise.
26403 (aff_combination_add_elt): Likewise.
26404 (aff_combination_add_cst): Likewise.
26405 (aff_combination_convert): Likewise.
26406 (add_elt_to_tree): Likewise. Remove unused argument.
26407 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
26409 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
26411 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
26413 * config/arm/arm.c (arm_default_short_enums): Use
26414 ARM_DEFAULT_SHORT_ENUMS.
26415 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
26417 2017-04-06 Jakub Jelinek <jakub@redhat.com>
26420 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
26421 members with redundant out-of-class redeclaration.
26423 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
26426 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
26427 * config/i386/i386.md (*zero_extendsidi2):
26428 Add (?*x,*x) and (?*v,*v) alternatives.
26430 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
26433 * config/i386/i386.c (ix86_expand_builtin)
26434 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
26435 mode from insn data. Convert operands to insn operand mode.
26436 Copy operands that don't satisfy insn predicate to a register.
26438 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
26440 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
26443 2017-04-06 Richard Biener <rguenther@suse.de>
26445 PR tree-optimization/80334
26446 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
26447 preserve alignment of accesses.
26449 2017-04-06 Richard Biener <rguenther@suse.de>
26451 PR tree-optimization/80262
26452 * tree-sra.c (build_ref_for_offset): Preserve address-space
26454 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
26455 Drop useless address-space information on MEM_REF offsets.
26457 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
26459 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
26461 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
26463 PR rtl-optimization/70703
26464 * ira-color.c (update_conflict_hard_regno_costs): Use
26465 HOST_WIDE_INT instead of long.
26467 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
26470 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
26471 not defined for x86_64 target. Add -mmmx target option when __SSE2__
26473 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
26474 for x86_64 target. Handle -m3dnowa option.
26476 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
26478 PR rtl-optimization/70703
26479 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
26480 (update_conflict_hard_regno_costs): Use long instead of unsigned
26481 arithmetic for cost calculation.
26483 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26484 Bernd Edlinger <bernd.edlinger@hotmail.de>
26487 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
26490 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
26493 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
26494 ptr_mode with Pmode throughout.
26495 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
26496 into probe_stack_range and use DImode.
26498 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
26501 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
26502 call_eh_return is true.
26504 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26506 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
26507 Initialize last_match_fntype_index.
26509 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26512 * tree-nvr.c: Include internal-fn.h.
26513 (pass_return_slot::execute): Ignore internal calls without
26516 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26517 Richard Biener <rguenther@suse.de>
26520 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
26521 captures used multiple times, except for the last use.
26522 * generic-match-head.c: Include gimplify.h.
26524 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26526 PR tree-optimization/79390
26527 * target.h (struct noce_if_info): Declare.
26528 * targhooks.h (default_noce_conversion_profitable_p): Declare.
26529 * target.def (noce_conversion_profitable_p): New target hook.
26530 * ifcvt.h (struct noce_if_info): New type, moved from ...
26531 * ifcvt.c (struct noce_if_info): ... here.
26532 (noce_conversion_profitable_p): Renamed to ...
26533 (default_noce_conversion_profitable_p): ... this. No longer
26535 (noce_try_store_flag_constants, noce_try_addcc,
26536 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
26537 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
26538 instead of noce_conversion_profitable_p.
26539 * config/i386/i386.c: Include ifcvt.h.
26540 (ix86_option_override_internal): Don't override
26541 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
26542 (ix86_noce_conversion_profitable_p): New function.
26543 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
26544 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
26545 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
26546 * doc/tm.texi: Regenerated.
26548 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26550 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
26553 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26556 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
26557 instructions for small multiply cores.
26559 2017-04-04 Jeff Law <law@redhat.com>
26561 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
26563 (mips_expand_vec_perm_const): Initialize elements in orig_perm
26564 that are not set by the loop over the elements.
26566 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26569 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
26570 int mode, convert_modes it to mode as unsigned, otherwise use
26571 lowpart_subreg to mode rather than SImode.
26572 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
26573 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
26574 Use DImode instead of SImode for the shift count operand.
26575 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
26578 2017-04-04 Richard Biener <rguenther@suse.de>
26580 PR middle-end/80281
26581 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
26582 arithmetic done for the negate or the plus. Simplify.
26583 (A - (-B) -> A + B): Likewise.
26584 * fold-const.c (split_tree): Make sure to not negate pointers.
26586 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
26588 PR rtl-optimization/60818
26589 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
26590 a compare of comparisons with the thing compared if this results
26591 in a different machine mode.
26593 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
26595 * alias.c (base_alias_check): Fix typo in comment.
26596 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
26597 * cgraphunit.c (symbol_table::compile): Likewise.
26598 * collect2.c (maybe_run_lto_and_relink): Likewise.
26599 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
26600 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
26601 * config/avr/avr.c (avr_map_op_t): Likewise.
26602 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
26603 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
26604 * config/epiphany/epiphany.md (movcc): Likewise.
26605 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
26606 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
26608 * config/mips/mips.c (mips_save_restore_reg): Likewise.
26609 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
26610 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
26611 * config/sh/sh.c (sh_rtx_costs): Likewise.
26612 * fold-const.c (fold_truth_andor): Likewise.
26613 * genautomata.c (collapse_flag): Likewise.
26614 * gengtype.h (struct type::u::s): Likewise.
26615 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
26616 * input.c (FORMAT_AMOUNT): Likewise.
26617 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
26618 (known_aggs_to_agg_replacement_list): Likewise.
26619 * ipa-inline-analysis.c: Likewise.
26620 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
26621 * ipa-polymorphic-call.c
26622 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
26623 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
26624 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
26626 * modulo-sched.c (apply_reg_moves): Likewise.
26627 * omp-expand.c (build_omp_regions_1): Likewise.
26628 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
26629 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
26630 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
26631 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
26632 * value-prof.c: Likewise.
26633 * var-tracking.c (val_reset): Likewise.
26635 2017-04-03 Richard Biener <rguenther@suse.de>
26637 PR tree-optimization/80275
26638 * fold-const.c (split_address_to_core_and_offset): Handle
26641 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
26643 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
26644 descriptors is at least equal to that of functions.
26646 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
26648 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
26650 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
26653 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
26654 (mov<IMOD4:mode>): New expander.
26655 (*mov<IMOD4:mode>_internal): New insn and split pattern.
26657 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
26659 PR rtl-optimization/79405
26660 * fwprop.c (propagations_left): New variable.
26661 (forward_propagate_into): Decrement it.
26662 (fwprop_init): Initialize it.
26663 (fw_prop): If the variable has reached zero, stop propagating.
26664 (fwprop_addr): Ditto.
26666 2017-03-31 Jakub Jelinek <jakub@redhat.com>
26669 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
26670 a FUNCTION_DECL, pass it as decl instead of origin to
26673 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
26675 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
26678 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
26681 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
26682 TARGET_VSX_SMALL_INTEGER.
26684 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26686 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
26687 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
26689 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26691 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
26692 extraction from odd-numbered MSA register.
26694 2017-03-31 Jakub Jelinek <jakub@redhat.com>
26696 PR middle-end/80173
26697 * expmed.c (store_bit_field_1): Don't attempt to create
26698 a word subreg out of hard registers wider than word if they
26699 have HARD_REGNO_NREGS of 1 for their mode.
26701 PR middle-end/80163
26702 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
26703 conversions to integer types wider than word and pointer.
26706 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
26707 (rtx_equal_for_cselib_p): Pass 0 to it.
26708 * cselib.c (cselib_hasher::equal): Likewise.
26709 (rtx_equal_for_cselib_1): Add depth argument. If depth
26710 is 128, don't look up VALUE locs and punt. Increment
26711 depth in recursive calls when walking VALUE locs.
26713 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
26715 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
26716 (make_gcov_file_name): Use the canonical path name for generating
26718 (read_line): Fix handling of files with ascii null bytes.
26720 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
26722 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
26723 to initialise a vector register instead
26724 of using a const_int.
26726 2017-03-30 Jakub Jelinek <jakub@redhat.com>
26728 PR translation/80189
26729 * gimplify.c (omp_default_clause): Use %qs instead of %s in
26730 diagnostic messages.
26732 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
26735 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
26736 (dfp_diex_<mode>): Update mode of operand 1.
26737 * doc/extend.texi (dxex, dxexq): Document change to return type.
26738 (diex, diexq): Document change to argument type.
26740 2017-03-30 Martin Jambor <mjambor@suse.cz>
26743 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
26744 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
26745 it reflects the signature changes performed at the callee side.
26746 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
26747 to cgraph_build_function_type_skip_args.
26748 (build_function_decl_skip_args): Adjust call to the above function.
26750 2017-03-30 Jakub Jelinek <jakub@redhat.com>
26753 * config/i386/sse.md
26754 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
26755 register as dest whenever it is a MEM not rtx_equal_p to the
26756 corresponding dup operand, and when forcing into reg move the
26757 reg into the memory afterwards.
26758 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
26759 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
26760 for the force_reg mode.
26761 (avx512vl_vextractf128<mode>): Use register as dest either
26762 always when a MEM, or when it is a MEM not rtx_equal_p to the
26763 corresponding dup operand, or even not when it is a CONST_VECTOR
26764 depending on the mode and lo vs. hi.
26765 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
26767 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
26768 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
26769 Likewise. Require that operands[2] is even.
26770 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
26771 Remove extraneous parens. Require that operands[2] is a multiple
26773 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
26774 operands[0] is a MEM if <mask_applied>, the predicates/constraints
26775 disallow memory then.
26777 2017-03-30 Richard Biener <rguenther@suse.de>
26779 PR tree-optimization/77498
26780 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
26781 to non-constants over backedges.
26783 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
26785 PR rtl-optimization/80233
26786 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
26787 as last_combined_insn. Do not test for BARRIER_P separately.
26789 2017-03-29 Andreas Schwab <schwab@suse.de>
26792 * calls.c (prepare_call_address): Convert funexp to Pmode before
26793 copying to temp reg.
26795 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26797 PR tree-optimization/80158
26798 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
26799 Handle possible future case of more than one alternate
26801 (replace_rhs_if_not_dup): Likewise.
26802 (replace_one_candidate): Likewise.
26804 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
26806 PR rtl-optimization/80193
26807 * ira.c (ira): Do not check allocation for LRA.
26809 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
26811 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
26812 (nvptx_output_simt_exit): Declare.
26813 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
26814 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
26815 (init_softstack_frame): Move initialization of crtl->is_leaf to...
26816 (nvptx_declare_function_name): ...here. Emit declaration of local
26817 memory space buffer for omp_simt_enter insn.
26818 (nvptx_output_unisimt_switch): New.
26819 (nvptx_output_softstack_switch): New.
26820 (nvptx_output_simt_enter): New.
26821 (nvptx_output_simt_exit): New.
26822 * config/nvptx/nvptx.h (struct machine_function): New fields
26823 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
26824 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
26825 (UNSPECV_SIMT_EXIT): Ditto.
26826 (omp_simt_enter_insn): New insn.
26827 (omp_simt_enter): New expansion.
26828 (omp_simt_exit): New insn.
26829 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
26831 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
26832 (expand_GOMP_SIMT_ENTER_ALLOC): New.
26833 (expand_GOMP_SIMT_EXIT): New.
26834 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
26835 (GOMP_SIMT_ENTER_ALLOC): Ditto.
26836 (GOMP_SIMT_EXIT): Ditto.
26837 * target-insns.def (omp_simt_enter): New insn.
26838 (omp_simt_exit): Ditto.
26839 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
26841 (lower_rec_simd_input_clauses): Implement SIMT privatization.
26842 (lower_rec_input_clauses): Likewise.
26843 (lower_lastprivate_clauses): Handle SIMT privatization.
26845 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
26846 (ompdevlow_adjust_simt_enter): New.
26847 (find_simtpriv_var_op): New.
26848 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
26849 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
26851 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
26852 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
26853 (copy_decl_for_dup_finish): Ditto.
26855 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
26857 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
26860 * config/i386/i386.c (ix86_option_override_internal): Always
26861 allow -mpreferred-stack-boundary=3 for 64-bit targets.
26863 2017-03-28 Bin Cheng <bin.cheng@arm.com>
26865 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
26867 2017-03-28 Bin Cheng <bin.cheng@arm.com>
26869 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
26870 mark new edge's irreducible flag accordign to it.
26871 (vect_do_peeling): Check loop preheader edge's irreducible flag
26872 and pass it to function slpeel_add_loop_guard.
26874 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
26876 PR tree-optimization/80218
26877 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
26878 Update block frequencies and counts.
26880 2017-03-28 Richard Biener <rguenther@suse.de>
26882 PR tree-optimization/78644
26883 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
26884 of a simplification result we may not use it at all.
26886 2017-03-28 Richard Biener <rguenther@suse.de>
26889 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
26890 without arguments, generate default definition of a SSA name.
26892 2017-03-28 Richard Biener <rguenther@suse.de>
26894 PR middle-end/80222
26895 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
26896 TYPE_REF_CAN_ALIAS_ALL references.
26897 * fold-const.c (fold_indirect_ref_1): Likewise.
26899 2017-03-28 Martin Liska <mliska@suse.cz>
26902 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
26903 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
26905 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
26906 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
26908 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
26909 (EXTRA_SPECS): Define.
26910 (SUBTARGET_EXTRA_SPECS): Likewise.
26911 (SUBTARGET_CPP_SPEC): Likewise.
26912 * config/arc/elf.h (EXTRA_SPECS): Renamed to
26913 SUBTARGET_EXTRA_SPECS.
26914 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
26916 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
26918 * config/arc/simdext.md (vst64_insn): Update pattern.
26919 (vld32wh_insn): Likewise.
26920 (vld32wl_insn): Likewise.
26921 (vld64_insn): Likewise.
26922 (vld32_insn): Likewise.
26924 2017-03-28 Marek Polacek <polacek@redhat.com>
26927 * fold-const.c (fold_comparison): Use protected_set_expr_location
26928 instead of SET_EXPR_LOCATION.
26930 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
26932 * tree.c (add_expr): Avoid name lookup warning.
26934 2017-03-27 Jeff Law <law@redhat.com>
26936 PR tree-optimization/80216
26937 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
26938 function name. Limit recursion depth.
26939 (record_temporary_equivalences): Corresponding changes.
26941 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
26943 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
26946 2017-03-27 Jakub Jelinek <jakub@redhat.com>
26949 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
26951 * cfgcleanup.c (reg_note_cfa_p): New array.
26952 (insns_have_identical_cfa_notes): New function.
26953 (old_insns_match_p): Don't cross-jump in between /f
26954 and non-/f instructions. If both i1 and i2 are frame related,
26955 verify all CFA notes, their order and content.
26957 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
26960 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
26961 HImode and SImode with zero extend to DImode to one insn.
26962 (bswap<mode>2_extenddi): Likewise.
26963 (bswapsi2_extenddi): Likewise.
26964 (bswaphi2_extendsi): Likewise.
26965 (bswaphi2): Combine bswap HImode and SImode into one insn.
26966 Separate memory insns from swapping register.
26967 (bswapsi2): Likewise.
26968 (bswap<mode>2): Likewise.
26969 (bswaphi2_internal): Delete, no longer used.
26970 (bswapsi2_internal): Likewise.
26971 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
26972 store, and gpr<-gpr swap insns.
26973 (bswap<mode>2_store): Likewise.
26974 (bswaphi2_reg): Register only splitter, combine with the splitter.
26975 (bswaphi2 splitter): Likewise.
26976 (bswapsi2_reg): Likewise.
26977 (bswapsi2 splitter): Likewise.
26978 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
26979 the insns into load, store, and register/register insns.
26980 (bswapdi2_ldbrx): Likewise.
26981 (bswapdi2_load): Likewise.
26982 (bswapdi2_store): Likewise.
26983 (bswapdi2_reg): Likewise.
26985 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
26987 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
26988 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
26990 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
26993 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
26995 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
26996 special handling for target option conflicts between dform
26997 options (-mpower9-dform, -mpower9-dform-vector,
26998 -mpower9-dform-scalar) and -mno-direct-move.
27000 2017-03-27 Richard Biener <rguenther@suse.de>
27002 PR tree-optimization/80181
27003 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
27005 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27007 * config/arc/predicates.md (move_double_src_operand): Replace the
27008 call to move_double_src_operand with a call to address_operand.
27010 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27012 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
27013 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
27014 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
27016 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27018 * config/arc/predicates.md (long_immediate_loadstore_operand):
27019 Consider scaled addresses cases.
27021 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27023 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
27024 restored when in interrupt.
27025 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
27026 doesn't have delay slot.
27028 2017-03-27 Richard Biener <rguenther@suse.de>
27031 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
27032 inlined thunk clones.
27034 2017-03-27 Jakub Jelinek <jakub@redhat.com>
27037 * asan.c (instrument_derefs): Copy over last operand from
27038 original COMPONENT_REF to the new COMPONENT_REF with
27039 DECL_BIT_FIELD_REPRESENTATIVE.
27040 * ubsan.c (instrument_object_size): Likewise.
27042 2017-03-27 Richard Biener <rguenther@suse.de>
27044 PR tree-optimization/80170
27045 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
27046 sure DR/SCEV didnt fold in constants we do not see when looking
27047 at the reference base alignment.
27049 2017-03-27 Richard Biener <rguenther@suse.de>
27051 PR middle-end/80171
27052 * gimple-fold.c (fold_ctor_reference): Properly guard against
27053 NULL return value from canonicalize_constructor_val.
27055 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
27058 * config/i386/i386.c (ix86_expand_builtin)
27059 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
27060 flags reg setting and flags reg using instructions.
27061 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
27062 clobbering instructions to zero extend op2.
27064 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
27066 * doc/install.texi (Configuration) <--with-aix-soname>:
27067 Update link to AIX ld.
27069 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
27071 PR rtl-optimization/80160
27072 PR rtl-optimization/80159
27073 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
27074 reg_alternate_class into account.
27076 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
27079 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
27080 to consider in curr_insn_transform.
27082 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27084 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
27085 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
27086 and emit_mode_inner.
27088 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27090 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
27091 argument to the overloaded builtin variants. Use the new flag to
27092 deprecate certain builtin variants.
27093 * config/s390/s390-builtin-types.def: Add new builtin types.
27094 * config/s390/s390-builtins.h: Support new flags field for
27095 overloaded builtins.
27096 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
27097 (s390_macro_to_expand): Enable vector float data type.
27098 (s390_cpu_cpp_builtins_internal): Indicate support of the new
27099 builtins by incrementing the __VEC__ version number.
27100 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
27102 (s390_resolve_overloaded_builtin): Emit error messages depending
27103 on the builtin flags.
27104 * config/s390/s390.c (s390_expand_builtin): Support additional
27105 flags argument. Change error message to match the messages
27106 emitted in s390-c.c.
27107 * config/s390/s390.md: New UNSPEC_* constants.
27108 (op_type): Add new instruction types.
27109 * config/s390/vecintrin.h: Add new builtins and test data class
27111 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
27112 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
27113 (VEC_INEXACT, VEC_NOINEXACT): New constants.
27114 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
27115 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
27116 ("vec_mergel<mode>"): V_HW -> VEC_HW.
27118 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
27119 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
27120 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
27121 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
27123 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
27124 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
27125 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
27126 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
27128 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
27129 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
27130 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
27131 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
27132 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
27133 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
27134 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
27136 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
27137 ("vec_scatter_element<V_HW_4:mode>_DI")
27138 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
27139 ("vec_fpint<mode>", "vflls")
27140 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
27141 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
27142 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
27143 ("*vec_cmphe<mode>_cc"): ... these.
27145 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
27146 mode constant instead of magic value.
27148 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27150 * config/s390/s390.c (s390_expand_vec_compare): Support other
27151 vector floating point modes than just V2DF.
27152 (s390_expand_vcond): Likewise.
27153 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
27154 (s390_cannot_change_mode_class): Prevent mode changes between TF
27155 and V1TF in vector registers.
27156 * config/s390/s390.md (DF, SF): New mode attributes.
27157 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
27158 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
27159 SFmode support for VRs.
27160 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
27162 (VFT, VF_HW): New mode iterators.
27163 (vw, sdx): New mode attributes.
27164 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
27165 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
27166 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
27167 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
27168 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
27169 also the new vector floating point modes. Renaming to ...
27171 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
27172 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
27173 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
27174 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
27175 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
27176 ("vec_unordered<mode>"): ... these.
27178 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
27179 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
27180 ("*vec_extendv2df"): New insn definitions.
27182 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27184 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
27185 ("mulditi3_2", "*muldi3_sign"): New patterns.
27186 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
27187 rename the pattern definition.
27189 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27191 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
27193 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
27195 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27197 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
27198 instruction if possible.
27199 * config/s390/vector.md (vec_halfnumelts): New mode
27201 ("*vec_vllezlf<mode>"): New pattern.
27203 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27205 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
27206 ("popcountv4si2", "popcountv2di2"): Rename to ...
27207 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
27208 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
27210 ("popcount<mode>2_vxe"): New pattern.
27212 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27214 * common/config/s390/s390-common.c (processor_flags_table): Add
27216 * config.gcc: Add arch12.
27217 * config/s390/driver-native.c (s390_host_detect_local_cpu):
27218 Default to arch12 for unknown CPU model numbers.
27219 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
27220 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
27221 PROCESSOR_max sanity check.
27222 * config/s390/s390-opts.h (enum processor_type): Add
27224 * config/s390/s390.c (processor_table): Add arch12.
27225 (s390_expand_builtin): Add check for B_VXE flag.
27226 (s390_issue_rate): Add PROCESSOR_ARCH12.
27227 (s390_get_sched_attrmask): Likewise.
27228 (s390_get_unit_mask): Likewise.
27229 (s390_sched_score): Enable z13 scheduling for arch12.
27230 (s390_sched_reorder): Likewise.
27231 (s390_sched_variable_issue): Likewise.
27232 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
27234 (s390_tune_attr): Use z13 scheduling also for arch12.
27235 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
27236 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
27237 (TARGET_VXE_P): New macros.
27238 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
27239 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
27240 * config/s390/s390.opt: Add arch12 as processor_type.
27242 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27244 * config/s390/s390.md
27245 ("fixuns_truncdddi2", "fixuns_trunctddi2")
27246 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
27247 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
27249 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
27250 Rename expanders to ...
27252 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
27253 ("fixuns_truncdddi2_emu"): ... these.
27255 ("fixuns_trunc<mode>si2_emu"): New expander.
27257 ("*fixuns_truncdfdi2_z13"): Rename to ...
27258 ("*fixuns_truncdfdi2_vx"): ... this.
27260 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27262 * config/s390/2964.md: Remove the single element vector compare
27263 instructions which are no longer used.
27264 * config/s390/s390.c (s390_select_ccmode): Remove handling of
27266 (s390_canonicalize_comparison): Remove handling of DFmode
27268 (s390_expand_vec_compare_scalar): Remove function.
27269 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
27270 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
27272 ("*cmp<mode>_ccs"): Add wfcdb instruction.
27274 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27276 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
27278 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
27279 will anyway by matched by mov<mode>_64dfp.
27281 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27283 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
27284 vlef/vstef. Add missing operand to vleif.
27286 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27288 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
27289 pair for all vector types with 64 bit elements.
27290 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
27291 * config/s390/vector.md (V_HW_64): ... here.
27292 (V_128_NOSINGLE): New mode iterator.
27293 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
27294 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
27295 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
27296 ("*vec_load_pairv2di"): Change to ...
27297 ("*vec_load_pair<mode>"): ... this one.
27299 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27301 * config/s390/constraints.md: Add comments.
27302 (jKK): Reject element sizes > 8 bytes.
27303 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
27305 * config/s390/s390.md: Add the s_operand checks formerly in
27306 s390_split_ok_p to various splitters where they are still
27308 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
27309 for 128 bit vectors. Plus two splitters.
27311 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27313 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
27316 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27319 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
27320 error if the boundary argument is not constant.
27322 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27324 PR rtl-optimization/80112
27325 * loop-doloop.c (doloop_condition_get): Don't check condition
27326 if cmp isn't SET with IF_THEN_ELSE src.
27328 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27330 PR tree-optimization/80158
27331 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
27332 replacing a candidate statement, also replace it for the
27333 candidate's alternate interpretation.
27334 (replace_rhs_if_not_dup): Likewise.
27335 (replace_one_candidate): Likewise.
27337 2017-03-24 Richard Biener <rguenther@suse.de>
27339 PR tree-optimization/80167
27340 * graphite-isl-ast-to-gimple.c
27341 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
27343 (translate_isl_ast_to_gimple::get_rename): Likewise.
27345 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
27347 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
27348 handling of certain combinations of target options, including the
27349 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
27350 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
27352 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27355 * config/arm/arm.md (*load_multiple): Add reload_completed to
27356 matching condition.
27358 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27359 Richard Biener <rguenth@suse.de>
27361 PR tree-optimization/79908
27362 PR tree-optimization/80136
27363 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27364 been cast away, gimplify_and_add suffices.
27366 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
27368 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
27370 2017-03-23 Richard Biener <rguenther@suse.de>
27372 PR tree-optimization/80032
27373 * gimplify.c (gimple_push_cleanup): Forced unconditional
27374 cleanups still have to go to the conditional_cleanups
27377 2017-03-22 Jakub Jelinek <jakub@redhat.com>
27379 PR tree-optimization/80072
27380 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
27382 (next_operand_entry_id): Change type to unsigned int.
27383 (sort_by_operand_rank): Make sure to return the right return value
27384 even if unsigned fields are bigger than INT_MAX.
27385 (struct oecount): Change cnt and id type to unsigned int.
27386 (oecount_hasher::equal): Formatting fix.
27387 (oecount_cmp): Make sure to return the right return value
27388 even if unsigned fields are bigger than INT_MAX.
27389 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
27392 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
27393 TREE_READONLY on result if writing it more than once.
27396 * doc/invoke.texi (-fsanitize=thread): Document that with
27397 -fnon-call-exceptions atomics are not able to throw
27401 * tsan.c: Include tree-eh.h.
27402 (instrument_builtin_call): Call maybe_clean_eh_stmt or
27403 maybe_clean_or_replace_eh_stmt where needed.
27404 (instrument_memory_accesses): Add cfg_changed argument.
27405 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
27406 if it returned true.
27407 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
27409 PR rtl-optimization/63191
27410 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
27411 wrapper function, moved the whole old content into ...
27412 (ix86_delegitimize_address_1): ... this. New inline function.
27413 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
27414 true as last argument instead of ix86_delegitimize_address.
27416 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
27418 * config/aarch64/aarch64.c (generic_branch_cost): Copy
27419 cortexa57_branch_cost.
27421 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
27423 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
27425 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27428 * doc/md.texi (Constraints): Document wA constraint.
27429 * config/rs6000/constraints.md (wA): New.
27430 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
27431 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
27432 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
27433 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
27435 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
27438 * gimplify.c (is_oacc_declared): New function.
27439 (oacc_default_clause): Use it to set default flags for acc declared
27440 variables inside parallel regions.
27441 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
27442 declared variables.
27443 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
27444 declare attribute to any decl as necessary.
27446 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
27449 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
27450 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
27451 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
27452 (arm_arch_lpae): This.
27453 * config/arm/arm.c (arm_arch7ve): Rename into ...
27454 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
27455 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
27458 2017-03-22 Martin Liska <mliska@suse.cz>
27461 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
27462 error message instead of an ICE.
27464 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27466 * doc/extend.texi (6.11 Additional Floating Types): Revise.
27468 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
27470 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
27472 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27475 2017-03-21 Martin Sebor <msebor@redhat.com>
27477 * doc/extend.texi: Use "cannot" instead of "can't."
27478 * doc/hostconfig.texi: Same.
27479 * doc/install.texi: Same.
27480 * doc/invoke.texi: Same.
27481 * doc/loop.texi: Same.
27482 * doc/md.texi: Same.
27483 * doc/objc.texi: Same.
27484 * doc/rtl.texi: Same.
27485 * doc/tm.texi: Same.
27486 * doc/tm.texi.in: Same.
27487 * doc/trouble.texi: Same.
27489 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
27492 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
27493 (collect_checksum_attributes): Set it.
27494 (die_checksum_ordered): Use it.
27496 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27498 PR tree-optimization/79908
27499 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
27500 change: For a VA_ARG whose LHS has been cast away, use
27501 force_gimple_operand to construct the side effects.
27503 2017-03-21 David Malcolm <dmalcolm@redhat.com>
27505 PR translation/80001
27506 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
27507 more amenable to translation.
27508 (oacc_loop_auto_partitions): Likewise.
27510 2017-03-21 Marek Polacek <polacek@redhat.com>
27511 Martin Sebor <msebor@redhat.com>
27513 PR tree-optimization/80109
27514 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
27515 on INTEGRAL_TYPE_P.
27517 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27518 Segher Boessenkool <segher@kernel.crashing.org>
27521 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
27522 check reg_used_between_p between insn and one of succ or succ2
27523 depending on if succ is artificial insn not inserted into insn
27526 2017-03-21 Martin Liska <mliska@suse.cz>
27528 PR gcov-profile/80081
27529 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
27530 * doc/gcc.texi: Include gcov-dump stuff.
27531 * doc/gcov-dump.texi: New file.
27533 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
27535 PR rtl-optimization/79150
27536 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
27537 conditional jump, if the jump is the last insn of the loop.
27539 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27540 Richard Biener <rguenth@suse.de>
27542 PR tree-optimization/79908
27543 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27544 been cast away, use force_gimple_operand to construct the side
27547 2017-03-21 Martin Liska <mliska@suse.cz>
27549 PR libfortran/79956
27550 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
27553 2017-03-21 Brad Spengler <spender@grsecurity.net>
27556 * plugin.c (htab_hash_plugin): New function.
27557 (add_new_plugin): Use it and adjust.
27558 (parse_plugin_arg_opt): Adjust.
27559 (init_one_plugin): Likewise.
27561 2017-03-21 Richard Biener <rguenther@suse.de>
27563 PR tree-optimization/80032
27564 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
27565 if set force the cleanup to happen unconditionally.
27566 (gimplify_target_expr): Push inserted clobbers with force_uncond
27567 to avoid them being removed by control-dependent DCE.
27569 2017-03-21 Richard Biener <rguenther@suse.de>
27571 PR tree-optimization/80122
27572 * tree-inline.c (copy_bb): Do not expans va-arg packs or
27573 va_arg_pack_len when the inlined call stmt requires pack
27575 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
27577 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27580 * tsan.c (instrument_builtin_call): If the memory model argument
27581 is not a constant, assume it is valid.
27584 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
27587 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27589 PR rtl-optimization/79910
27590 * combine.c (can_combine_p): Do not allow combining an I0 or I1
27591 if its dest is used by an insn before I2 (other than the combined
27592 insns themselves, which are properly handled already).
27594 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27597 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
27599 * combine.c (record_used_regs): New static function.
27600 (try_combine): Handle situations where there is an additional
27601 instruction between I2 and I3 which needs to have a LOG_LINK
27605 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27607 * combine.c (try_combine): Delete redundant i1 test. Call
27608 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27610 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27613 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
27614 alternatives 13/14.
27616 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27618 PR tree-optimization/80054
27619 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
27620 the optimization if a PHI or any of its arguments is not dominated
27621 by the candidate's basis. Use gphi* rather than gimple* as
27623 (replace_profitable_candidates): Clean up a gimple* variable that
27624 should be a gphi* variable.
27626 2017-03-20 Martin Sebor <msebor@redhat.com>
27629 * doc/extend.texi (attribute constructor): Document present limitation.
27631 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
27634 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
27635 __POWER9_VECTOR__ #ifdef control, change template definition to
27636 use Power9-specific built-in function.
27637 (vec_any_eq): Likewise.
27638 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
27639 to control outcomes from this test.
27640 (vector_ae_<mode>p): For VEC_F modes, likewise.
27642 2017-03-20 Ian Lance Taylor <iant@google.com>
27644 * config/i386/i386.c (ix86_function_regparm): Save an extra
27645 register for -fsplit-stack with DECL_STATIC_CHAIN.
27647 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
27650 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
27651 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
27653 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
27655 * config/riscv/riscv.c (riscv_print_operand): Use "fence
27657 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
27660 2017-03-20 Marek Polacek <polacek@redhat.com>
27663 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
27665 2017-03-20 Richard Biener <rguenther@suse.de>
27667 PR tree-optimization/80113
27668 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
27669 allocate extra SSA name for PHI def.
27670 (add_close_phis_to_outer_loops): Likewise.
27671 (add_close_phis_to_merge_points): Likewise.
27672 (copy_loop_close_phi_args): Likewise.
27673 (copy_cond_phi_nodes): Likewise.
27675 2017-03-20 Martin Liska <mliska@suse.cz>
27677 PR middle-end/79753
27678 * tree-chkp.c (chkp_build_returned_bound): Do not build
27679 returned bounds for a LHS that's not a BOUNDED_P type.
27681 2017-03-20 Martin Liska <mliska@suse.cz>
27685 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
27686 COMPLEX_CST and VECTOR_CST.
27688 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27691 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
27692 target operand. A new splitter adds the clobber statement in case
27693 the target operand is dead anyway.
27695 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
27697 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
27698 to age-old versions of binutils and glibc.
27700 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
27702 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
27704 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
27706 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
27708 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
27710 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
27711 requirement for binutils 2.13.
27713 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27715 * combine.c (try_combine): Delete redundant i1 test. Call
27716 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27718 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
27720 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
27721 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
27723 <riscv64-*-elf>: Re-arrange section
27724 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
27725 <riscv32-*-linux>: Likewise.
27726 <riscv64-*-elf>: Likewise
27727 <riscv64-*-linux>: Likewise.
27729 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
27732 * aarch64.opt(verbose-cost-dump): Fix typo.
27734 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
27737 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
27738 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
27740 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
27742 * reload.c (find_reloads): When reloading a nonoffsettable address,
27743 use RELOAD_OTHER for it and its address reloads.
27745 PR rtl-optimization/79910
27746 * combine.c (record_used_regs): New static function.
27747 (try_combine): Handle situations where there is an additional
27748 instruction between I2 and I3 which needs to have a LOG_LINK
27751 2017-03-17 Jeff Law <law@redhat.com>
27753 PR tree-optimization/71437
27754 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
27755 conditional in the hash table first.
27756 (vrp_dom_walker::before_dom_children): Extract condition from
27757 ASSERT_EXPR. Record condition, its inverion and any implied
27758 conditions as well.
27760 2017-03-17 Marek Polacek <polacek@redhat.com>
27761 Markus Trippelsdorf <markus@trippelsdorf.de>
27763 PR tree-optimization/80079
27764 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
27767 2017-03-17 Richard Biener <rguenther@suse.de>
27769 PR middle-end/80075
27770 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
27771 Properly verify the LHS before the RHS possibly claims to be
27773 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
27776 2017-03-17 Martin Jambor <mjambor@suse.cz>
27778 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
27779 (List of -O2 options): Likewise.
27780 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
27783 2017-03-17 Tom de Vries <tom@codesourcery.com>
27785 * gcov-dump.c (print_usage): Print bug_report_url.
27787 2017-03-17 Richard Biener <rguenther@suse.de>
27789 PR middle-end/80050
27790 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
27791 (parser::peek): Likewise.
27793 2017-03-17 Richard Biener <rguenther@suse.de>
27795 PR tree-optimization/80048
27796 * sese.c (free_sese_info): Properly release rename_map and
27797 copied_bb_map elements.
27799 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
27801 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
27802 Add linked-list forward and backlinks. Insert on
27803 construction, remove on destruction.
27804 (class pass_store_merging): Add m_stores_head field.
27805 (pass_store_merging::terminate_and_process_all_chains):
27806 Iterate over m_stores_head list.
27807 (pass_store_merging::terminate_all_aliasing_chains):
27809 (pass_store_merging::execute): Check for debug stmts first.
27810 Push new chains onto the m_stores_head stack.
27812 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
27815 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
27816 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
27817 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
27819 2017-03-16 Jeff Law <law@redhat.com>
27821 PR tree-optimization/71437
27822 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
27823 member function. Implementation moved into after_dom_children
27824 member function and into the threader's thread_outgoing_edges
27826 (dom_opt_dom_walker::after_dom_children): Simplify by moving
27827 some code into new thread_outgoing_edges.
27828 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
27829 definition. Simplify marker handling (do it here). Assume we always
27830 have the available expression and the const/copies tables.
27831 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
27833 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
27834 * tree-vrp.c (equiv_stack): No longer file scoped.
27835 (vrp_dom_walker): New class.
27836 (vrp_dom_walker::before_dom_children): New member function.
27837 (vrp_dom_walker::after_dom_children): Likewise.
27838 (identify_jump_threads): Setup domwalker. Use it rather than
27839 walking edges in a random order by hand. Simplify setup/finalization.
27840 (finalize_jump_threads): Remove.
27841 (vrp_finalize): Do not call identify_jump_threads here.
27842 (execute_vrp): Do it here instead and call thread_through_all_blocks
27845 PR tree-optimization/71437
27846 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
27848 (simplify_stmt_for_jump_threading): Add basic_block argument. All
27850 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
27851 (dom_opt_dom_walker::thread_across_edge): Remove
27852 handle_dominating_asserts argument. All callers changed.
27853 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
27854 changes. Remove calls to lhs_of_dominating_assert. Other
27855 uses of handle_dominating_asserts turn into unconditional code
27856 (simplify_control_stmt_condition_1): Likewise.
27857 (simplify_control_stmt_condition): Likewise.
27858 (thread_through_normal_block, thread_across_edge): Likewise.
27859 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
27860 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
27861 object if it is not an SSA_NAME.
27862 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
27863 before calling into the VRP specific simplifiers.
27864 (identify_jump_threads): Remove handle_dominating_asserts
27867 2017-03-16 Jakub Jelinek <jakub@redhat.com>
27870 * tree-diagnostic.c (default_tree_printer): No longer static.
27871 * tree-diagnostic.h (default_tree_printer): New prototype.
27873 2017-03-16 Tamar Christina <tamar.christina@arm.com>
27875 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
27876 Change ins into fmov.
27878 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27880 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
27881 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
27882 Use h_con constraint for operand 1.
27883 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
27884 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
27886 2017-03-15 Jeff Law <law@redhat.com>
27888 PR tree-optimization/71437
27889 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
27890 (record_temporary_equivalences): Use it.
27892 PR tree-optimization/71437
27893 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
27894 tree-ssa-scopedtables.
27895 (lookup_avail_expr, build_and_record_new_cond): Likewise.
27896 (record_conditions, record_cond, vuse_eq): Likewise.
27897 (record_edge_info): Adjust to API tweak of record_conditions.
27898 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
27899 (record_temporary_equivalences, optimize_stmt): Likewise.
27900 (eliminate_redundant_computations): Likewise.
27901 (record_equivalences_from_stmt): Likewise.
27902 * tree-ssa-scopedtables.c: Include options.h and params.h.
27903 (vuse_eq): New function, moved from tree-ssa-dom.c
27904 (build_and_record_new_cond): Likewise.
27905 (record_conditions): Likewise. Accept vector of conditions rather
27906 than edge_equivalence structure for first argument.
27907 for the first argument.
27908 (avail_exprs_stack::lookup_avail_expr): New member function, moved
27909 from tree-ssa-dom.c.
27910 (avail_exprs_stack::record_cond): Likewise.
27911 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
27912 from tree-ssa-dom.c.
27913 (avail_exprs_stack): Add new member functions lookup_avail_expr
27915 (record_conditions): Declare.
27917 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
27920 * lra-constraints.c (process_alt_operands): Increase reject for
27921 reloading an input/output operand.
27923 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
27926 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
27927 insns to convert from signed/unsigned char/short to IEEE 128-bit
27929 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
27931 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
27934 * config/i386/i386.c (ix86_vector_duplicate_value): Create
27935 subreg of inner mode for values already in registers.
27937 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
27939 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
27940 iteration reg is used after the loop.
27942 2017-03-14 Martin Sebor <msebor@redhat.com>
27944 PR tree-optimization/79800
27945 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
27946 precision in negative-positive range.
27947 (format_floating): Call non-const overload with adjusted precision.
27949 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
27952 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
27955 2017-03-14 Martin Sebor <msebor@redhat.com>
27957 PR middle-end/80020
27958 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
27959 * builtins.def (aligned_alloc): Use it.
27962 * Makefile.in (GTFILES): Add calls.c.
27963 * calls.c: Include "gt-calls.h".
27965 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
27967 PR rtl-optimization/79728
27968 * regs.h (struct target_regs): New field
27969 x_contains_allocatable_regs_of_mode.
27970 (contains_allocatable_regs_of_mode): New macro.
27971 * reginfo.c (init_reg_sets_1): Initialize it, and change
27972 contains_reg_of_mode so it includes global regs as well.
27973 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
27974 rather than contains_regs_of_mode.
27976 2017-03-14 Martin Liska <mliska@suse.cz>
27978 * doc/invoke.texi: Document options that can't be combined with
27979 -fcheck-pointer-bounds.
27981 2017-03-14 Martin Liska <mliska@suse.cz>
27983 PR middle-end/79831
27984 * doc/invoke.texi (-Wchkp): Document the option.
27986 2017-03-14 Martin Liska <mliska@suse.cz>
27988 * Makefile.in: Install gcov-dump.
27990 2017-03-14 Martin Liska <mliska@suse.cz>
27992 * multiple_target.c (expand_target_clones): Bail out for
27993 an invalid attribute.
27995 2017-03-14 Richard Biener <rguenther@suse.de>
27997 * alias.c (struct alias_set_entry): Pack properly.
27998 * cfgloop.h (struct loop): Likewise.
27999 * cse.c (struct set): Likewise.
28000 * ipa-utils.c (struct searchc_env): Likewise.
28001 * loop-invariant.c (struct invariant): Likewise.
28002 * lra-remat.c (struct cand): Likewise.
28003 * recog.c (struct change_t): Likewise.
28004 * rtl.h (struct address_info): Likewise.
28005 * symbol-summary.h (function_summary): Likewise.
28006 * tree-loop-distribution.c (struct partition): Likewise.
28007 * tree-object-size.c (struct object_size_info): Likewise.
28008 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
28009 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
28010 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
28011 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
28012 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
28013 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
28014 (struct _stmt_vec_info): Likewise.
28016 2017-03-14 Martin Liska <mliska@suse.cz>
28019 * multiple_target.c (create_dispatcher_calls): Check that
28020 a target can create a function dispatcher.
28022 2017-03-14 Martin Liska <mliska@suse.cz>
28025 * multiple_target.c (expand_target_clones): Drop local.local
28026 flag for default implementation.
28028 2017-03-14 Richard Biener <rguenther@suse.de>
28030 PR tree-optimization/80030
28031 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
28033 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
28035 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
28036 gcc_fallthrough() instead of __attribute__((fallthrough));
28038 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28040 * doc/gcc.texi: Remove "up" link to (DIR).
28041 * doc/gccint.texi: Ditto.
28043 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28045 * doc/install.texi (Specific) <avr>: Remove reference to
28048 2017-03-13 Jeff Law <law@redhat.com>
28050 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
28051 attribute rather than comments.
28053 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
28054 match_scratch operand is highest.
28056 2017-03-13 Martin Liska <mliska@suse.cz>
28058 PR middle-end/78339
28059 * ipa-pure-const.c (warn_function_noreturn): If the declarations
28060 is a CHKP clone, use original declaration.
28062 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28064 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
28065 (arc_conditional_register_usage): Use a different allocation order
28066 when optimizing for size.
28067 * common/config/arc/arc-common.c (arc_option_optimization_table):
28068 Section anchors default on when optimizing for size.
28070 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28072 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
28074 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28076 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
28077 * config/arc/arc.md (cpu_facility): Add cd variant.
28078 (*movqi_insn): Add code density variant.
28079 (*movhi_insn): Likewise.
28080 (*movqi_insn): Likewise.
28081 (*addsi3_mixed): Likewise.
28082 (subsi3_insn): Likewise.
28084 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28086 * config/arc/arc.md (movsi_cond_exec): Update constraint.
28088 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28090 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
28091 expressions with MINUS and UNARY ops.
28093 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28096 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
28098 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
28099 between vec_select and vector argument.
28100 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
28101 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
28102 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
28103 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
28104 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
28105 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
28107 2017-03-13 Richard Biener <rguenther@suse.de>
28110 * params.def (vect-max-peeling-for-alignment): Fix typo.
28112 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28114 * doc/install.texi (Specific) <mips-*-*>: Remove description of
28115 issue that only occurred with binutils below 2.18.
28117 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28119 * doc/install.texi (Specific) <cris-axis-elf>: No longer
28120 refer to binutils 2.11/2.12 minimum.
28122 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28124 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
28125 ftp.kernel.org and simplify binutils requirement.
28127 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
28129 * doc/invoke.texi (Warning Options): Fix spelling of link-time
28131 (Optimize Options): Ditto. Also remove redundancy.
28133 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28135 PR translation/79848
28136 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
28138 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
28139 to G_ to avoid double translation.
28141 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28143 PR translation/79923
28144 * auto-profile.c (get_combined_location): Convert leading
28145 character of diagnostics to lower case and remove trailing period.
28146 (read_profile): Likewise for various diagnostics.
28147 * config/arm/arm.c (arm_option_override): Remove trailing period
28148 from various diagnostics.
28149 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
28150 (msp430_expand_delay_cycles): Likewise.
28152 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28155 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
28156 full command-line argument, rather than just "str".
28157 (aarch64_validate_march): Likewise.
28158 (aarch64_validate_mtune): Likewise.
28160 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
28162 PR rtl-optimization/78911
28163 * lra-assigns.c (must_not_spill_p): New function.
28164 (spill_for): Use it.
28166 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28168 PR tree-optimization/79981
28169 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
28170 ATOMIC_COMPARE_EXCHANGE ifn result.
28171 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
28172 IFN_ATOMIC_COMPARE_EXCHANGE.
28174 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28177 * opts.c (parse_sanitizer_options): Add missing question mark to
28178 "did you mean" message.
28180 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28182 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
28184 (VMULEUH_UNS): Likewise.
28185 (VMULOUB_UNS): Likewise.
28186 (VMULOUH_UNS): Likewise.
28187 * config/rs6000/rs6000.c (builtin_function_type): Remove
28188 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
28190 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28193 * read-rtl-function.c (function_reader::read_rtx_operand): Update
28194 x with result of extra_parsing_for_operand_code_0.
28195 (function_reader::extra_parsing_for_operand_code_0): Convert
28196 return type from void to rtx, returning x. When reading
28197 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
28198 larger size containing struct block_symbol.
28200 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
28202 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
28203 -mfloat128-hardware without -m64.
28205 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
28208 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
28209 entries to the case statement that marks unsigned arguments to
28210 overloaded functions.
28212 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
28214 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
28215 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
28217 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
28220 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
28221 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
28223 2017-03-10 Martin Liska <mliska@suse.cz>
28227 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
28228 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
28231 2017-03-10 Olivier Hainque <hainque@adacore.com>
28233 * tree-switch-conversion (array_value_type): Start by resetting
28234 candidate type to it's main variant.
28236 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28238 PR rtl-optimization/79909
28239 * combine.c (try_combine): Use simplify_replace_rtx on individual
28240 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
28241 of the whole CALL_INSN_FUNCTION_USAGE.
28243 PR tree-optimization/79972
28244 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
28245 get_range_info on SSA_NAMEs. Formatting fixes.
28247 2017-03-10 Richard Biener <rguenther@suse.de>
28248 Jakub Jelinek <jakub@redhat.com>
28250 PR tree-optimization/77975
28251 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
28252 edge to be constant.
28253 (get_val_for): For constant x return it. Formatting fix.
28254 (loop_niter_by_eval): Avoid pointless looping if the next iteration
28255 would use the same bases as the current one.
28257 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28259 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
28260 instead of vec_select for V1TImode.
28261 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
28263 (VSX_LE_128): Add V1TI to this mode iterator.
28264 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
28265 (*vsx_le_perm_store_<mode>): Likewise.
28266 (pre-reload splitter for VSX stores): Likewise.
28267 (post-reload splitter for VSX stores): Likewise.
28268 (*vsx_xxpermdi2_le_<mode>): Likewise.
28269 (*vsx_lxvd2x2_le_<mode>): Likewise.
28270 (*vsx_stxvd2x2_le_<mode>): Likewise.
28272 2017-03-09 Michael Eager <eager@eagercon.com>
28274 Correct failures with --enable-checking=yes,rtl.
28276 * config/microblaze/microblaze.c (microblaze_expand_shift):
28277 Replace GET_CODE test with CONST_INT_P and INTVAL test with
28278 test for const0_rtx.
28279 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
28280 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
28282 2017-03-09 Richard Biener <rguenther@suse.de>
28284 PR tree-optimization/79977
28285 * graphite-scop-detection.c (scop_detection::merge_sese):
28286 Handle the case of extra exits to blocks dominating the entry.
28288 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
28290 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28293 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
28295 PR rtl-optimization/79949
28296 * lra-constraints.c (process_alt_operands): Check memory when
28297 trying to predict a cycle. Print about the overall increase.
28299 2017-03-09 Richard Biener <rguenther@suse.de>
28301 PR middle-end/79971
28302 * gimple-expr.c (useless_type_conversion_p): Preserve
28303 TYPE_SATURATING for fixed-point types.
28305 2017-03-09 Richard Biener <rguenther@suse.de>
28308 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
28309 alignment of BLKmode params.
28311 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28314 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
28315 (VALL_NO_V2Q): Likewise.
28317 * config/aarch64/aarch64-simd.md
28318 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
28320 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
28321 VALL_NO_V2Q mode iterator.
28322 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
28324 2017-03-09 Martin Liska <mliska@suse.cz>
28326 PR tree-optimization/79631
28327 * tree-chkp-opt.c (chkp_is_constant_addr): Call
28328 tree_int_cst_sign_bit just for INTEGER constants.
28330 2017-03-09 Martin Liska <mliska@suse.cz>
28334 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
28337 2017-03-09 Marek Polacek <polacek@redhat.com>
28340 * tree.c (inchash::add_expr): Handle TREE_VEC.
28342 2017-03-09 Martin Liska <mliska@suse.cz>
28345 (chkp_narrow_size_and_offset): New function.
28346 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
28347 (void chkp_parse_bit_field_ref): New function.
28348 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
28349 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
28351 2017-03-09 Martin Liska <mliska@suse.cz>
28354 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
28355 (chkp_find_bounds_1): Remove gcc_unreachable.
28357 2017-03-09 Jakub Jelinek <jakub@redhat.com>
28360 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
28361 BUILT_IN_SYNC*, determine the access type from the size suffix and
28362 always build a MEM_REF with that type. Handle forgotten
28363 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
28366 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
28367 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
28368 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
28369 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
28370 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
28371 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
28372 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
28373 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
28374 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
28375 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
28376 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
28377 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
28378 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
28379 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
28380 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
28381 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
28382 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
28383 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
28384 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
28385 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
28386 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
28387 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
28388 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
28389 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
28390 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
28391 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
28392 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
28393 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
28394 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
28395 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
28396 definitions outside of __OPTIMIZE__ guarded section.
28399 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
28400 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
28401 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
28402 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
28405 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28407 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
28408 ("vfenez<mode>"): Add missing constraints.
28410 2017-03-08 Martin Sebor <msebor@redhat.com>
28413 * config/nds32/nds32.c (nds32_option_override):
28414 Fix misspelled diagnostic.
28416 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28419 * gimplify.c (gimplify_omp_for): Replace index var in outer
28420 taskloop statement with an artificial variable and add
28421 OMP_CLAUSE_PRIVATE clause for it.
28423 2017-03-08 Richard Biener <rguenther@suse.de>
28425 PR tree-optimization/79955
28426 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
28427 for accesses that are completely outside of the variable.
28429 2017-03-08 Andrew Haley <aph@redhat.com>
28431 PR tree-optimization/79943
28432 * tree-ssa-loop-split.c (compute_new_first_bound): When
28433 calculating the new upper bound, (END-BEG) should be added, not
28436 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28438 * config/avr/avr.md (setmemhi): Make sure match_dup
28439 operand number comes before match_scratch.
28441 2017-03-08 Richard Biener <rguenther@suse.de>
28443 PR tree-optimization/79920
28444 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
28445 with ncopies == 1 to ...
28446 (vect_transform_slp_perm_load): ... here. Properly compute
28447 all element loads by iterating VF times over the group. Do
28448 not handle ncopies (computed in a broken way) in
28449 vect_create_mask_and_perm.
28451 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28454 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
28455 is a uniform vector, use uniform_vector_p return value instead of
28456 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
28458 2017-03-07 Marek Polacek <polacek@redhat.com>
28460 PR middle-end/79809
28461 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
28462 (alloca_call_type): Likewise.
28464 2017-03-07 Martin Liska <mliska@suse.cz>
28466 * gcov.c (process_args): Put comment to correct location.
28468 2017-03-07 Martin Liska <mliska@suse.cz>
28470 PR middle-end/68270
28471 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
28472 Use array_at_struct_end_p instead of DECL_CHAIN (field).
28473 (chkp_narrow_bounds_for_field): Likewise.
28474 (chkp_parse_array_and_component_ref): Pass one more argument to
28477 2017-03-07 Richard Biener <rguenther@suse.de>
28479 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
28482 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
28484 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
28485 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
28487 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28490 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
28491 to end of description.
28492 (PARAM_MAX_STORES_TO_MERGE): Likewise.
28494 2017-03-07 Jakub Jelinek <jakub@redhat.com>
28496 PR rtl-optimization/79901
28497 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
28499 (*avx512f_<code><mode>3<mask_name>): ... this.
28500 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
28501 iterator instead of VI8_AVX2_AVX512BW.
28503 PR rtl-optimization/79901
28504 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
28505 min/max expander, expand it using expand_vec_cond_expr.
28508 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
28511 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28514 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
28515 to void * for PCH reasons.
28516 * dwarf2out.c (output_loc_operands, output_die): Cast
28517 v.val_vec.array to unsigned char *.
28519 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
28522 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
28525 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
28527 PR rtl-optimization/79571
28528 * lra-constraints.c (process_alt_operands): Calculate static
28529 reject and subtract it from overall when only addresses will be
28532 2017-03-06 Julia Koval <julia.koval@intel.com>
28535 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
28536 incoming stack boundary to 128 for 64-bit targets.
28538 2017-03-06 Richard Biener <rguenther@suse.de>
28540 PR tree-optimization/79894
28541 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
28542 to NULL after folding it.
28544 2017-03-06 Richard Biener <rguenther@suse.de>
28546 PR tree-optimization/79824
28547 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
28548 check disabling peeling for gaps.
28550 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
28552 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28553 attributes): Document gettimeofday.
28555 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28557 * config/s390/s390.c (s390_option_override_internal): Set
28558 PARAM_MIN_VECT_LOOP_BOUND
28560 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28562 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
28563 * config/s390/s390.md: Likewise.
28565 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28568 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
28569 (<avx2_avx512>_perm<mode>): Rename to ...
28570 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
28572 (<avx512>_perm<mode>_mask): Rename to ...
28573 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
28575 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
28576 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
28577 instead of VI8F_256_512.
28578 (avx512f_perm<mode>): New define_expand.
28579 (avx512f_perm<mode>_mask): Likewise.
28580 (avx512f_perm<mode>_1<mask_name>): New define_insn.
28581 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
28583 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28585 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
28586 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
28588 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
28590 2017-03-06 Martin Liska <mliska@suse.cz>
28593 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
28594 when having a SSA NAME w/o VAR_DECL assigned to it.
28596 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28598 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
28599 msa_dpsub_<su>_d): Fix MODE for vec_select.
28601 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28603 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
28605 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
28607 2017-03-06 Richard Biener <rguenther@suse.de>
28609 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
28610 * plugin.c (register_plugin_info): Likewise.
28611 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
28613 2017-03-05 Jakub Jelinek <jakub@redhat.com>
28615 * config/i386/sse.md (sse_storehps, sse_storelps,
28616 avx_<castmode><avxsizesuffix>_<castmode>,
28617 avx512f_<castmode><avxsizesuffix>_<castmode>,
28618 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
28619 in condition that at least one operand is not a MEM.
28621 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28623 PR middle-end/79805
28624 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
28625 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
28627 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
28628 gimple_call_nothrow_p flag based on whether original builtin can throw.
28629 If it can, emit following stmts on the fallthrough edge.
28630 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
28631 don't create new bb if inserting just debug stmts on the edge, try to
28632 insert them on the fallthru bb or just reset debug stmts.
28634 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
28637 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
28638 restore recog_data (including the operand rtxes inside it) around
28639 the call to get_insn_template.
28641 2017-03-03 Martin Sebor <msebor@redhat.com>
28643 PR tree-optimization/79699
28644 * context.c (context::~context): Free MPFR caches to avoid
28645 a memory leak on program exit.
28647 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28649 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
28650 Use wide_int::ulow () instead of .elt (0).
28652 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
28654 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
28655 (*pushxf): Limit oF constraint to 32bit targets and add oC
28656 constraint for 64bit targets.
28657 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
28658 (*pushdf): Change rmF constraint to rmC.
28660 2017-03-03 Martin Liska <mliska@suse.cz>
28662 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
28663 Remove unused variable.
28665 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28668 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
28669 is a memory operand, increase num_memory.
28670 (ix86_expand_args_builtin): Likewise.
28672 2017-03-03 Jan Hubicka <jh@suse.cz>
28675 * ipa-devirt.c (maybe_record_node): Properly handle
28676 __cxa_pure_virtual visibility.
28678 2017-03-03 Martin Liska <mliska@suse.cz>
28680 PR tree-optimization/79803
28681 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
28683 (pass_loop_prefetch::execute): Disabled optimization if an
28684 assumption about L1 cache size is not met.
28686 2017-03-03 Martin Liska <mliska@suse.cz>
28688 PR rtl-optimization/79574
28689 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
28690 (hash_scan_set): Likewise.
28691 (dump_hash_table): Likewise.
28692 (hoist_code): Likewise.
28694 2017-03-03 Richard Biener <rguenther@suse.de>
28696 * fixed-value.c (fixed_from_string): Restore use of elt (1)
28697 in place of uhigh ().
28698 (fixed_convert_from_real): Likewise.
28700 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
28703 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
28705 2017-03-03 Richard Biener <rguenther@suse.de>
28707 PR middle-end/79818
28708 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
28709 TYPE_OVERFLOW_UNDEFINED check.
28711 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28713 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
28715 (vector_ae_<mode>_p): Likewise.
28716 (vector_nez_<mode>_p): Likewise.
28717 (vector_ne_v2di_p): Likewise.
28718 (vector_ae_v2di_p): Likewise.
28719 (vector_ne_<mode>_p): Likewise.
28720 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
28722 (vsx_tsqrt<mode>2_fe): Likewise.
28724 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
28727 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
28729 2017-03-02 Jakub Jelinek <jakub@redhat.com>
28731 PR rtl-optimization/79780
28732 * cprop.c (one_cprop_pass): When second and further conditional trap
28733 in a single basic block is turned into an unconditional trap, turn it
28734 into a deleted note to avoid RTL verification failures.
28736 2017-03-02 Richard Biener <rguenther@suse.de>
28738 * fold-const.c (const_binop): Use ulow () instead of elt (0).
28740 2017-03-02 Richard Biener <rguenther@suse.de>
28742 PR tree-optimization/79345
28744 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
28745 param and abort the walk, returning -1 if it is hit.
28746 (walk_aliased_vdefs): Take a limit param and pass it on.
28747 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
28748 defaulting to 0 and return a signed int.
28749 * tree-ssa-uninit.c (struct check_defs_data): New struct.
28750 (check_defs): New helper.
28751 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
28752 about uninitialized memory.
28753 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
28754 bogus uninitialized warning.
28755 (fixed_convert_from_real): Likewise.
28757 2017-03-02 Bin Cheng <bin.cheng@arm.com>
28759 PR tree-optimization/66768
28760 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
28761 iv_use if base object can't be determined.
28763 2017-03-02 Jakub Jelinek <jakub@redhat.com>
28765 PR tree-optimization/79345
28766 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
28767 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
28768 (get_pattern_stats): Initialize it.
28769 * genemit.c (gen_expand): Verify match_scratch numbers come after
28770 match_operand/match_dup numbers.
28771 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
28772 match_scratch numbers.
28773 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
28775 * config/s390/s390.md (trunctdsd2): Likewise.
28777 2017-03-02 Richard Biener <rguenther@suse.de>
28779 * wide-int.h (wide_int_storage::operator=): Implement in terms
28782 2017-03-02 Richard Biener <rguenther@suse.de>
28784 PR tree-optimization/79777
28785 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
28786 the to insert expression to sth existing.
28788 2017-03-01 Martin Sebor <msebor@redhat.com>
28790 PR middle-end/79692
28791 * gimple-ssa-sprintf.c
28792 (directive::known_width_and_precision): New function.
28793 (format_integer): Use it.
28794 (get_mpfr_format_length): Consider the full range of precision
28795 when computing %g output with the # flag. Set the likely byte
28796 count to 3 rather than 1 when precision is indeterminate.
28797 (format_floating): Correct the lower bound of precision.
28799 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28801 * doc/invoke.texi: Document default code model for 64-bit Linux.
28803 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28806 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
28807 udiv rather than div since input pattern is unsigned.
28809 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
28811 * config/i386/i386.c (print_reg): Warn for values of
28812 unsupported size in integer register.
28814 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
28817 * config/rs6000/predicates.md (current_file_function_operand): Do
28818 not allow self calls to be local if the function is replaceable.
28820 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
28823 * config/rs6000/altivec.h (vec_ctz and others): Change the
28824 preprocessor macro that controls conditional compilation from
28825 _ARCH_PWR9 to __POWER9_VECTOR__.
28826 (vec_all_ne): Change parameterization of __altivec_scalar_pred
28827 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
28828 control (instead of _ARCH_PWR9 control) so that template
28829 definition uses power9-specific function.
28830 (vec_any_eq): Likewise.
28831 (vec_all_ne): Change macro definition to use a power9-specific
28832 expansion under #ifdef __POWER9_VECTOR__ control (instead of
28833 _ARCH_PWR9 control).
28834 (vec_any_eq) Likewise.
28835 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
28836 expansion for CMPNEF to remove support for xvcmpnesp instruction.
28837 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
28838 support for xvcmpnedp instruction.
28839 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
28840 macro expansion so that Power9 implementation of vec_all_ne does
28841 not use the AltiVec predicate framework.
28842 (VCMPNEH_P): Likewise.
28843 (VCMPNEW_P): Likewise.
28844 (VCMPNED_P): Likewise.
28845 (VCMPNEFP_P): Likewise.
28846 (VCMPNEDP_P): Likewise.
28847 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
28848 implementation of vec_any_eq to not use AltiVec predicate
28850 (VCMPAEH_P): Likewise.
28851 (VCMPAEW_P): Likewise.
28852 (VCMPAED_P): Likewise.
28853 (VCMPAEFP_P): Likewise.
28854 (VCMPAEDP_P): Likewise.
28855 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
28856 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
28857 not use the AltiVec predicate framework.
28858 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
28859 of vec_any_eq to not use AltiVec predicate framework.
28860 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
28861 support for predefined __POWER9_VECTOR__ macro to indicate that
28862 Power9 instruction selection is enabled.
28863 (altivec_overloaded_builtins): Remove extraneous
28864 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
28865 function argument types RS6000_BTI_bool_V16QI and
28866 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
28867 entry for overloaded function argument types RS6000_BTI_bool_V4SI
28868 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
28869 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
28870 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
28871 Power9 for implementations of vec_cmpne. Change the signature for
28872 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
28873 (representing vec_all_ne) to remove the previously described first
28874 argument of type RS6000_BTI_INTSI, as this was an artifact of
28875 reliance on the AltiVec predicate framework, which is no longer
28876 used in the implementation of these functions. Add
28877 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
28878 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
28879 since, unlike the AltiVec predicate framework implementation, we
28880 do not share function descriptors between vec_alle and vec_anyeq.
28881 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
28882 set of modes that receive special treatment even when
28883 TARGET_P9_VECTOR is true. The special treatment emits code that
28884 does not depend on Power9 instructions.
28885 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
28886 define_expand to not rely on AltiVec predicate framework.
28887 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
28889 (vector_ne_v2di_p): Change this define_expand to not rely on
28890 AltiVec predicate framework.
28891 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
28893 (vector_ne_<mode>_p): Change this define_expand to not rely on
28894 AltiVec predicate framework.
28895 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
28897 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
28898 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
28899 define_insn pattern.
28900 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
28901 define_insn pattern because the xvcmpne<VSs>. instruction is not
28903 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
28904 instruction is not supported.
28906 2017-03-01 Jakub Jelinek <jakub@redhat.com>
28908 * config/nvptx/nvptx.c: Include intl.h.
28910 2017-03-01 Martin Jambor <mjambor@suse.cz>
28913 * ipa-prop.h (ipa_bits): Removed field known.
28914 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
28915 to pointers. Adjusted their comments to warn about their sharing.
28916 (ipcp_transformation_summary): Change bits to a vector of pointers.
28917 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
28918 (ipa_get_ipa_bits_for_value): Declare.
28919 * tree-vrp.h (value_range): Mark as GTY((for_user)).
28920 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
28921 (ipa_bits_hash_table): Likewise.
28922 (ipa_vr_ggc_hash_traits): Likewise.
28923 (ipa_vr_hash_table): Likewise.
28924 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
28925 being pointers and vr_known being removed.
28926 (ipa_set_jf_unknown): Likewise.
28927 (ipa_get_ipa_bits_for_value): New function.
28928 (ipa_set_jfunc_bits): Likewise.
28929 (ipa_get_value_range): New overloaded functions.
28930 (ipa_set_jfunc_vr): Likewise.
28931 (ipa_compute_jump_functions_for_edge): Use the above functions to
28932 construct bits and vr parts of jump functions.
28933 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
28934 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
28936 (ipcp_grow_transformations_if_necessary): Also allocate
28937 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
28939 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
28940 them. Fix too long lines.
28941 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
28942 vr_known being removed.
28943 (ipa_read_jump_function): Use new setter functions to construct bits
28944 and vr parts of jump functions or set them to NULL.
28945 (write_ipcp_transformation_info): Adjust for bits being pointers.
28946 (read_ipcp_transformation_info): Likewise.
28947 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
28949 Include gt-ipa-prop.h.
28950 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
28952 (ipcp_store_bits_results): Likewise.
28953 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
28954 Do not write to existing jump functions but use a temporary instead.
28956 2017-03-01 Jakub Jelinek <jakub@redhat.com>
28959 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
28960 attempt to use its first operand as BIT_FIELD_REF base.
28962 2017-03-01 Richard Biener <rguenther@suse.de>
28964 PR middle-end/79721
28965 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
28966 interpolating formula in wrapping arithmetic.
28967 (chrec_apply): Convert chrec_evaluate return value to wanted type.
28969 2017-03-01 Jakub Jelinek <jakub@redhat.com>
28971 PR tree-optimization/79734
28972 * tree-vect-generic.c (expand_vector_condition): Optimize
28973 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
28974 Handle VEC_COND_EXPR where comparison has different inner width from
28975 type's inner width.
28977 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
28979 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
28980 markup, and similar issues. Remove @opindex entries for things
28981 that aren't options. Add missing -mmpy-option entries.
28983 2017-02-28 Jakub Jelinek <jakub@redhat.com>
28985 PR tree-optimization/79737
28986 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
28987 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
28988 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
28989 instead of byte_size. Formatting fix.
28990 (shift_bytes_in_array_right): Formatting fix.
28992 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
28995 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
28996 condition on optimize for the leaf function test.
28998 2017-02-28 Martin Liska <mliska@suse.cz>
29001 * read-rtl-function.c (function_reader::handle_unknown_directive):
29002 Bail out when one uses -flto.
29004 2017-02-28 Martin Liska <mliska@suse.cz>
29006 * common.opt: Replace space with tabular for options of <number>
29008 * config/i386/i386.opt: Show <number> value for
29009 -mlarge-data-threshold.
29010 * opts.c (print_filtered_help): Do not display number in hexadecimal
29013 2017-02-28 Martin Liska <mliska@suse.cz>
29015 * common.opt: Fix --help=option -Q for options which are of
29018 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
29020 * config/i386/i386.c (print_reg): Error out for values
29021 of 8-bit size in invalid integer register.
29023 2017-02-28 Martin Sebor <msebor@redhat.com>
29025 PR tree-optimization/79691
29026 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
29028 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29031 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
29032 gcc_unreachable with output_operand_lossage.
29034 2017-02-28 Richard Biener <rguenther@suse.de>
29036 PR tree-optimization/79740
29037 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
29039 (visit_nary_op): Insert the nary into the hashtable if we
29040 pattern-matched sth.
29041 * tree-ssa-pre.c (eliminate_insert): Robustify.
29043 2017-02-28 Richard Biener <rguenther@suse.de>
29045 PR middle-end/79731
29046 * fold-const.c (decode_field_reference): Reject out-of-bound
29049 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29051 * config/i386/i386.c: Include intl.h.
29052 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
29053 instead of just cond ? "..." : "...".
29054 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
29055 * coverage.c (read_counts_file): Likewise.
29056 * omp-offload.c: Include intl.h.
29057 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
29058 of just cond ? "..." : "...".
29059 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
29060 of just cond ? "..." : "...".
29062 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
29065 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
29067 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
29068 'tune for' CPU name.
29069 * config/arm/arm-cpu-data.h: Regenerated.
29071 2017-02-28 Richard Biener <rguenther@suse.de>
29073 PR tree-optimization/79732
29074 * tree-inline.c (expand_call_inline): Do not shadow var.
29076 2017-02-28 Richard Biener <rguenther@suse.de>
29078 PR tree-optimization/79723
29079 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
29080 address-space properly.
29082 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
29084 * doc/optinfo.texi (Optimization groups): Fix option used for
29086 * doc/invoke.texi (-fopt-info): Document "omp".
29087 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
29088 (OPTGROUP_ALL): Add OPTGROUP_OMP.
29089 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
29090 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
29091 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
29093 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
29095 * dumpfile.c (optgroup_options): Instead of "openmp", associate
29096 OPTGROUP_OMP with "omp".
29098 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
29101 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
29102 for arithmetic shift of unsigned V2DI.
29104 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
29106 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
29107 arc/linux.h headers.
29108 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
29109 (LINK_SPEC): Likewise.
29110 (ARC_TLS_EXTRA_START_SPEC): Likewise.
29111 (EXTRA_SPECS): Likewise.
29112 (STARTFILE_SPEC): Likewise.
29113 (ENDFILE_SPEC): Likewise.
29114 (LIB_SPEC): Likewise.
29115 (TARGET_SDATA_DEFAULT): Likewise.
29116 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
29117 (MULTILIB_DEFAULTS): Likewise.
29118 (DWARF2_UNWIND_INFO): Likewise.
29119 * config/arc/big.h: New file.
29120 * config/arc/elf.h: Likewise.
29121 * config/arc/linux.h: Likewise.
29122 * config/arc/t-uClibc: Remove.
29124 2017-02-27 Bin Cheng <bin.cheng@arm.com>
29126 PR tree-optimization/77536
29127 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
29128 (tree_transform_and_unroll_loop): Use above function to compute the
29129 estimated niter of unrolled loop and use it when scaling profile.
29130 Also use count info rather than frequency if it's non-zero.
29131 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
29132 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
29133 (vect_transform_loop): Call above function.
29135 2017-02-27 Richard Biener <rguenther@suse.de>
29137 PR tree-optimization/45397
29138 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
29139 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
29140 (visit_nary_op): Add pattern matching for CSEing sign-changed
29141 or truncated operations with wider ones.
29143 2017-02-27 Richard Biener <rguenther@suse.de>
29145 PR tree-optimization/79690
29146 * tree-vect-stmts.c (vectorizable_store): Use vector type
29147 built from the DR with address-space.
29149 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
29151 * doc/invoke.texi (Optimize Options): Refine the description
29152 of asan-use-after-return.
29154 2017-02-25 Alan Modra <amodra@gmail.com>
29156 PR rtl-optimization/79584
29157 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
29158 base, not ad->base_term, the reg within base. Remove assertion
29159 that ad->base == ad->base_term. Replace gen_int_mode using
29160 bogus mode with const0_rtx.
29162 2017-02-25 Jakub Jelinek <jakub@redhat.com>
29164 PR middle-end/79396
29165 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
29166 FMA_EXPR like tcc_binary or tcc_unary.
29168 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
29171 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
29173 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
29174 (output_loc_operands): Handle DW_OP_call_ref and
29175 DW_OP_GNU_variable_value.
29176 (struct variable_value_struct): New type.
29177 (struct variable_value_hasher): Likewise.
29178 (variable_value_hash): New variable.
29179 (string_types): Remove.
29180 (copy_loc_descr): New function.
29181 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
29182 (prepend_loc_descr_to_each): New function.
29183 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
29184 instead of add_loc_descr_to_each if the first argument is single
29185 location list and the second has multiple.
29186 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
29187 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
29188 when looking for variable value which doesn't have other location info.
29189 (loc_list_from_tree): Formatting fix.
29190 (gen_array_type_die): Simplify DW_AT_string_length handling.
29191 (adjust_string_types): Remove.
29192 (gen_subprogram_die): Don't call adjust_string_types nor test/set
29193 string_types. Call resolve_variable_values.
29194 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
29195 (resolve_addr_in_expr): Likewise. Add A argument.
29196 (copy_deref_exprloc): Remove deref argument. Adjust for the
29197 original expression being DW_OP_GNU_variable_value with optionally
29198 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
29199 optionally after it.
29200 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
29201 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
29202 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
29203 (variable_value_hasher::hash, variable_value_hasher::equal): New
29205 (resolve_variable_value_in_expr, resolve_variable_value,
29206 resolve_variable_values, note_variable_value_in_expr,
29207 note_variable_value): New functions.
29208 (dwarf2out_early_finish): Call note_variable_value on all toplevel
29211 2017-02-24 Jakub Jelinek <jakub@redhat.com>
29214 * opts.h (handle_generated_option): Add GENERATED_P argument.
29215 * opts-common.c (handle_option): Adjust function comment.
29216 (handle_generated_option): Add GENERATED_P argument, pass it to
29218 (control_warning_option): Pass false to handle_generated_option
29220 * opts.c (maybe_default_option): Pass true to handle_generated_option
29222 * optc-gen.awk: Likewise.
29224 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29226 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
29227 a REG, look at the REG it is a SUBREG of.
29228 (splitter for cmpeqsi_t): Ditto.
29230 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29232 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
29233 the special USEs with the pattern of the insn, not the insn itself.
29235 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
29238 * doc/invoke.texi: Document -mload-store-pairs.
29240 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29241 Sandra Loosemore <sandra@codesourcery.com>
29243 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
29244 argument isn't a CONST_INT.
29245 (nios2_alternate_compare_const): Assert op is a CONST_INT.
29246 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
29247 (nios2_validate_compare): Bypass alternate compare logic if *op2
29248 is not a CONST_INT.
29249 (ldstwm_operation_p): Return false if first_base is not a REG or
29250 if first_offset is not a CONST_INT.
29252 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29254 * config/cris/cris.md: Use correct operand in a define_peephole2.
29256 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29258 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
29260 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29262 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
29263 this_insn if it is an INSN or JUMP_INSN.
29264 (force_offsettable): Look at base, not at addr.
29265 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
29266 on things that aren't necessarily CONST_INTs.
29268 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
29270 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
29271 -mfpmath=sse is the default also for x86-32 targets with SSE2
29272 instruction set when @option{-ffast-math} is enabled
29274 2017-02-24 Jeff Law <law@redhat.com>
29276 PR rtl-optimizatoin/79286
29277 * ira.c (update_equiv_regs): Drop may_trap_p exception to
29280 2017-02-24 Richard Biener <rguenther@suse.de>
29282 PR tree-optimization/79389
29283 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
29286 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
29288 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
29289 function comment to reflect reality.
29290 (loop_exits_before_overflow): Fix typo in function description.
29292 2017-02-24 Richard Biener <rguenther@suse.de>
29294 PR tree-optimization/79389
29295 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
29296 properly that a threading opportunity exists. Detect conditional
29297 copy/constant propagation opportunities.
29299 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
29301 * config/visium/visium.md (type): Add trap.
29302 (b): New mode attribute.
29303 (*btst): Rename into...
29304 (*btst<mode>): ...this and adjust.
29305 (*cbranchsi4_btst_insn): Rename into...
29306 (*cbranch<mode>4_btst_insn): ...this and adjust.
29307 (trap): New define_insn.
29309 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29311 PR tree-optimization/79389
29312 * ifcvt.c (struct noce_if_info): Add rev_cond field.
29313 (noce_reversed_cond_code): New function.
29314 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
29315 reversed_comparison_code. Formatting fix.
29316 (noce_try_store_flag): Test rev_cond != NULL in addition to
29317 reversed_comparison_code.
29318 (noce_try_store_flag_constants): Likewise.
29319 (noce_try_store_flag_mask): Likewise.
29320 (noce_try_addcc): Use rev_cond if non-NULL instead of
29321 reversed_comparison_code.
29322 (noce_try_cmove_arith): Likewise. Formatting fixes.
29323 (noce_try_minmax, noce_try_abs): Clear rev_cond.
29324 (noce_find_if_block): Initialize rev_cond.
29325 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
29326 instead of false as last argument never attempt to reverse it
29329 2017-02-23 Bin Cheng <bin.cheng@arm.com>
29331 PR tree-optimization/79663
29332 * tree-predcom.c (combine_chains): Process refs in reverse order
29333 only for ZERO length chains, and add explaining comment.
29335 2017-02-23 Jeff Law <law@redhat.com>
29337 PR tree-optimization/79578
29338 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
29339 in call to operand_equal_p.
29341 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
29344 * config/i386/cpuid.h: Fix another undefined behavior.
29346 2017-02-23 Richard Biener <rguenther@suse.de>
29348 PR tree-optimization/79683
29349 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
29350 vector types for data-refs.
29352 2017-02-23 Martin Liska <mliska@suse.cz>
29354 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
29356 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29358 PR middle-end/79665
29359 * internal-fn.c (get_range_pos_neg): Moved to ...
29360 * tree.c (get_range_pos_neg): ... here. No longer static.
29361 * tree.h (get_range_pos_neg): New prototype.
29362 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
29363 are known to be in between 0 and signed maximum inclusive, try to
29364 expand both unsigned and signed divmod and use the cheaper one from
29367 2017-02-22 Jeff Law <law@redhat.com>
29369 PR tree-optimization/79578
29370 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
29371 to compare base operands.
29373 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
29376 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
29377 gpc_reg_operand instead of fpr_reg_operand.
29379 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
29381 * config/mips/mips.c (mips_return_in_memory): Force FP
29382 vector types to be returned in memory for o32 ABI.
29384 2017-02-22 Jakub Jelinek <jakub@redhat.com>
29386 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
29387 instead of DW_TAG_member for static data member declarations and don't
29388 set no_linkage_name for static inline data members.
29389 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
29392 2017-02-22 Martin Liska <mliska@suse.cz>
29394 * doc/invoke.texi: Replace inequality signs with square brackets
29397 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
29400 * lra-constraints.c (simplify_operand_subreg): Handle
29401 WORD_REGISTER_OPERATIONS targets.
29403 2017-02-22 Jakub Jelinek <jakub@redhat.com>
29406 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
29407 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
29408 elimination by swapping fld*.
29410 2017-02-22 Richard Biener <rguenther@suse.de>
29412 PR tree-optimization/79673
29413 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
29414 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
29415 irrelevant address-space qualifiers and avoiding a
29416 ADDR_SPACE_CONVERT_EXPR from fold_convert.
29418 2017-02-22 Richard Biener <rguenther@suse.de>
29420 PR tree-optimization/79666
29421 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
29422 to not symbolically negate if that may introduce undefined
29425 2017-02-22 Martin Liska <mliska@suse.cz>
29428 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
29429 * data-streamer-out.c (streamer_write_gcov_count_stream):
29431 * value-prof.c (stream_out_histogram_value): Make assert more
29432 precise based on type of counter.
29434 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
29437 * config/i386/i386.md (standard_x87sse_constant_load splitter):
29438 Use nonimmediate_operand instead of memory_operand for operand 1.
29439 (float-extend standard_x87sse_constant_load splitter): Ditto.
29441 2017-02-21 Jeff Law <law@redhat.com>
29443 PR tree-optimization/79621
29444 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
29445 blocks with edges to themselves.
29447 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29450 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
29451 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
29452 Use gimple_call_builtin_p.
29455 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
29456 on temporarily removed DEBUG_INSNs.
29458 PR tree-optimization/79649
29459 * tree-loop-distribution.c (classify_partition): Give up on
29460 non-generic address space loads/stores.
29462 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
29464 * doc/loop.texi (Loop manipulation): Remove nonexistent
29465 tree_ssa_loop_version from the documentation.
29466 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
29468 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29471 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
29472 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
29473 * config/rs6000/rs6000.c: Include except.h.
29474 (rs6000_expand_split_stack_prologue): Call
29475 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
29477 2017-02-21 Martin Jambor <mjambor@suse.cz>
29480 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
29481 have been analyzed.
29483 2017-02-21 Martin Jambor <mjambor@suse.cz>
29485 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
29486 for backward compatibility only.
29487 * doc/invoke.texi (Option Summary): Remove all references to
29488 -fipa-cp-alignment.
29490 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
29494 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29496 * lra-constraints.c (curr_insn_transform): Handle
29497 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29499 2017-02-21 Martin Liska <mliska@suse.cz>
29501 * config/i386/i386.opt: Replace -masm-dialect with -masm.
29503 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
29505 PR translation/79638
29506 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
29508 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
29511 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
29512 (arm_function_ok_for_sibcall): Return false for an indirect call by
29513 descriptor if all the argument registers are used.
29514 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
29515 alignment of the function.
29517 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29519 PR tree-optimization/61441
29520 * simplify-rtx.c (simplify_const_unary_operation): For
29521 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
29522 the sNaN unmodified.
29524 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
29526 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29527 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29528 instead of SYSTEM_HEADER_DIR.
29530 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
29531 Martin Liška <mliska@suse.cz>
29533 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
29534 Fix typos and grammar, use active voice, and clarify.
29536 2017-02-20 Marek Polacek <polacek@redhat.com>
29538 PR middle-end/79537
29539 * gimplify.c (gimplify_expr): Handle unused *&&L;.
29542 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
29544 2017-02-20 Jakub Jelinek <jakub@redhat.com>
29547 * config/i386/i386.c (ix86_expand_builtin): Handle
29548 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
29549 ix86_builtins_isa[fcode].isa as a requirement of those
29550 flags and any other flag in the bitmask.
29551 (ix86_init_mmx_sse_builtins): Use 0 instead of
29552 ~OPTION_MASK_ISA_64BIT as mask.
29553 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
29554 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
29555 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
29556 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
29558 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29561 * lra-constraints.c (split_reg): Check requested split mode
29562 is supported by the register.
29564 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29566 * lra-constraints.c (simplify_operand_subreg): Remove early
29569 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29572 * lra-constraints.c (curr_insn_transform): Tighten condition
29573 for converting SUBREG reloads from OP_OUT to OP_INOUT.
29575 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29578 * lra-constraints.c (curr_insn_transform): Handle
29579 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29581 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
29584 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
29586 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
29588 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
29591 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
29594 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29596 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
29599 2017-02-18 Jakub Jelinek <jakub@redhat.com>
29601 * final.c (last_columnnum, override_columnnum): New variables.
29602 (final_start_function): Set last_columnnum, pass it to begin_prologue
29603 hook and pass 0 to dwarf2out_begin_prologue.
29604 (final_scan_insn): Update override_columnnum. Pass last_columnnum
29605 to source_line debug hook.
29606 (notice_source_line): Compute last_columnnum and for debug_column_info
29607 return true on column changes.
29608 * debug.h (struct gcc_debug_hooks): Add column argument to
29609 source_line and begin_prologue hooks.
29610 (debug_nothing_int_charstar_int_bool): Remove prototype.
29611 (debug_nothing_int_int_charstar,
29612 debug_nothing_int_int_charstar_int_bool): New prototypes.
29613 (dwarf2out_begin_prologue): Add column argument.
29614 * debug.c (do_nothing_debug_hooks): Adjust source_line and
29615 begin_prologue hooks.
29616 (debug_nothing_int_charstar_int_bool): Remove.
29617 (debug_nothing_int_int_charstar,
29618 debug_nothing_int_int_charstar_int_bool): New functions.
29619 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
29620 through to dwarf2out_source_line.
29621 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
29622 (dwarf2out_source_line): Add column argument, emit it if requested.
29623 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
29625 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29626 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29627 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
29628 through to dwarf2out_begin_prologue.
29629 (vmsdbgout_source_line): Add column argument, pass it through to
29630 dwarf2out_source_line.
29631 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
29632 dbxout_source_line caller.
29633 (dbxout_source_line): Add column argument.
29635 * common.opt (gno-column-info, gcolumn-info): New options.
29636 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
29637 (check_die): Also test for multiple DW_AT_decl_column attributes.
29638 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
29639 DW_AT_decl_column if requested.
29640 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
29642 (gen_variable_die): Likewise.
29643 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
29644 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
29647 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
29648 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
29649 (ix86_handle_option): Handle OPT_m3dnowa.
29650 * doc/invoke.texi (-m3dnowa): Document.
29651 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
29652 -m3dnowa instead of -m3dnow -march=athlon.
29655 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
29656 instead of gcc_assert for K, r and R code checks. Formatting fixes.
29658 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29661 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
29662 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
29663 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
29664 generator for vsx_xxpermdi_<mode>_be.
29665 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
29666 force big-endian semantics.
29667 (vsx_xxpermdi_<mode>_be): New define_expand with same
29668 implementation as previous version of vsx_xxpermdi_<mode>.
29670 2017-02-17 Jakub Jelinek <jakub@redhat.com>
29672 PR tree-optimization/79327
29673 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
29674 variable, its initialization and use.
29676 2017-02-17 Julia Koval <julia.koval@intel.com>
29678 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
29679 (OPTION_MASK_ISA_PKU_UNSET): New.
29680 (ix86_handle_option): Handle -mrdpid.
29681 * config/i386/cpuid.h (bit_RDPID): New.
29682 * config/i386/driver-i386.c (host_detect_local_cpu):
29683 Detect RDPID feature.
29684 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
29685 * config/i386/i386-c.c (ix86_target_macros_internal):
29687 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
29688 (ix86_valid_target_attribute_inner_p): Add "rdpid".
29689 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
29690 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
29691 * config/i386/i386.md (define_insn "rdpid"): New.
29692 * config/i386/i386.opt Add -mrdpid.
29693 * config/i386/immintrin.h (_rdpid_u32): New.
29695 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
29697 PR rtl-optimization/79541
29698 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
29699 instead of transforming it into USE.
29701 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
29703 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
29704 If HONOR_SNANS (SFmode) force the input to a register.
29705 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
29706 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
29707 an frsp or similar insn.
29709 2017-02-17 Martin Liska <mliska@suse.cz>
29711 PR rtl-optimization/79577
29712 * params.def (selsched-max-sched-times): Increase minimum to 1.
29714 2017-02-17 Martin Liska <mliska@suse.cz>
29716 PR rtl-optimization/79574
29717 * gcse.c (want_to_gcse_p): Prevent integer overflow.
29719 2017-02-17 Martin Liska <mliska@suse.cz>
29721 PR tree-optimization/79529
29722 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
29723 ssa_defined_default_def_p to handle cases which are implicitly
29725 * tree-ssa.c (ssa_defined_default_def_p): New function.
29726 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
29727 which are implicitly defined.
29728 * tree-ssa.h (ssa_defined_default_def_p): Declare.
29730 2017-02-17 Richard Biener <rguenther@suse.de>
29732 PR middle-end/79576
29733 * params.def (max-ssa-name-query-depth): Limit to 10.
29735 2017-02-17 Richard Biener <rguenther@suse.de>
29737 PR tree-optimization/79552
29738 * tree-ssa-structalias.c (visit_loadstore): Properly verify
29741 2017-02-17 Richard Biener <rguenther@suse.de>
29744 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
29746 2017-02-17 Marek Polacek <polacek@redhat.com>
29748 PR middle-end/79536
29749 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
29750 (fold_negate_expr): New wrapper.
29752 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
29754 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
29755 Correct terminology and de-emphasize pre-standard behavior.
29757 2017-02-16 Alan Modra <amodra@gmail.com>
29759 PR rtl-optimization/79286
29760 * ira.c (def_dominates_uses): New function.
29761 (update_equiv_regs): Don't create an equivalence for insns that
29762 may trap where the register def does not dominate the use.
29764 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
29766 PR rtl-optimization/78127
29767 * lra.c (lra): Call lra_eliminate before finish the loop after
29770 2017-02-16 Richard Biener <rguenther@suse.de>
29772 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
29774 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
29775 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
29776 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
29777 (isl_val_int_from_wi): New function.
29778 (extract_affine_gmp): Rename to ...
29779 (extract_affine_wi): ... this, take a widest_int.
29780 (extract_affine_int): Just wrap extract_affine_wi.
29781 (add_param_constraints): Use isl_val_int_from_wi.
29782 (add_loop_constraints): Likewise, and extract_affine_wi.
29784 2017-02-15 Jeff Law <law@redhat.com>
29786 PR middle-end/79521
29787 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
29788 ira_init_register_move_cost_if_necessary.
29790 2017-02-15 Martin Sebor <msebor@redhat.com>
29792 PR middle-end/32003
29793 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
29794 removed in a prior commit.
29796 2017-02-15 Bin Cheng <bin.cheng@arm.com>
29798 PR tree-optimization/79347
29799 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
29800 counters during peeling.
29802 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
29804 * Makefile.in (site.exp): Remove "set ISLVER".
29806 2017-02-15 Jakub Jelinek <jakub@redhat.com>
29809 * real.c (real_from_integer): Call real_convert even for decimal.
29811 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
29814 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
29816 2017-02-14 Andrew Pinski <apinski@cavium.com>
29818 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
29819 cores and change the partno/implementer to be correct.
29820 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
29821 the 'B" as the implementer.
29822 * config/aarch64/aarch64-tune.md: Regenerate.
29824 2017-02-14 Carl Love <cel@us.ibm.com>
29826 * config/rs6000/rs6000.c: Add case statement entry to make the
29827 xvcvuxdsp built-in argument unsigned.
29828 * config/rs6000/vsx.md: Fix the source and return operand types so they
29829 match the instruction definitions from the ISA document. Fix typo
29830 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
29833 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
29836 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
29837 member early_clobber_alts.
29838 * lra-lives.c (reg_early_clobber_p): New.
29839 (process_bb_lives): Use it.
29840 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
29841 (debug_operand_data): Initialize early_clobber_alts.
29842 (setup_operand_alternative): Set up early_clobber_alts.
29843 (collect_non_operand_hard_regs): Ditto. Pass early clobber
29844 alternatives to new_insn_reg.
29845 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
29847 (lra_update_insn_regno_info): Pass the new arg.
29849 2017-02-14 Jakub Jelinek <jakub@redhat.com>
29851 PR middle-end/79505
29852 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
29853 (new_oacc_loop_raw): Don't clear already cleared fields.
29856 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
29857 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
29858 _mm512_prefetch_i64gather_ps): New inline functions and macros.
29860 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
29863 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
29865 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
29868 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
29869 the extra instruction to the right place to store 128-bit constant
29872 2017-02-14 Martin Sebor <msebor@redhat.com>
29874 PR middle-end/79448
29875 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
29876 warning for strings of unknown length.
29878 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
29880 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
29882 2017-02-14 Jeff Law <law@redhat.com>
29885 * ira-costs.c (scan_one_insn): Initialize register move costs
29886 for pseudos seen in USE/CLOBBER insns.
29888 PR tree-optimization/79095
29889 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
29890 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
29891 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
29892 if the operands are known to be not equal, then the resulting range
29894 (intersect_ranges): If the new range is ~[0,0] and the old range is
29895 wide, then prefer ~[0,0].
29896 * tree-vrp.c (overflow_comparison_p_1): New function.
29897 (overflow_comparison_p): New function.
29898 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
29899 if NAME is used in an overflow test.
29900 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
29901 overflow check that can be expressed as an equality test, then adjust
29902 ops to be that equality test.
29904 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29906 * config/s390/s390-builtin-types.def: Remove flags argument.
29907 * config/s390/s390.c (s390_init_builtins): Likewise.
29909 2017-02-14 Martin Liska <mliska@suse.cz>
29911 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
29912 vector. Fix trailing white spaces.
29914 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
29916 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
29919 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29921 PR rtl-optimization/68664
29922 * config/arm/arm.c (arm_sched_can_speculate_insn):
29923 New function. Declare prototype.
29924 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
29926 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29928 PR rtl-optimization/68664
29929 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
29931 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
29933 2017-02-14 Amit Pawar <amit.pawar@amd.com>
29935 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
29936 max skip bytes for function, loop and jump.
29938 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29940 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
29941 ABS_EXPR for gimple dump.
29943 2017-02-14 Jakub Jelinek <jakub@redhat.com>
29946 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
29948 PR tree-optimization/79408
29949 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
29950 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
29951 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
29952 also if rhs1 is INTEGER_CST.
29954 2017-02-14 Richard Biener <rguenther@suse.de>
29956 PR middle-end/79432
29957 * tree-into-ssa.c (insert_phi_nodes): When the function can
29958 have abnormal edges rewrite SSA names with broken use-def
29959 dominance out of SSA and register them for PHI insertion.
29961 2017-02-13 Martin Sebor <msebor@redhat.com>
29963 PR middle-end/79496
29964 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
29965 clearing info.nowrite flag when snprintf size argument is a range.
29967 2017-02-13 Jakub Jelinek <jakub@redhat.com>
29969 * cprop.c (cprop_jump): Add missing space in string literal.
29970 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
29971 (get_constraint_for_component_ref): Likewise.
29972 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
29973 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
29974 * lra-constraints.c (process_alt_operands): Likewise.
29975 * ipa-inline.c (inline_small_functions): Likewise.
29976 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
29977 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
29978 * trans-mem.c (diagnose_tm_1_op): Likewise.
29979 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
29980 (grid_parallel_clauses_gridifiable): Likewise.
29982 * config/nvptx/mkoffload.c (process): Add space in between
29985 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
29986 "MOD4_SSE_REGS" and "ALL_REGS".
29988 * spellcheck.c (test_data): Add , in between "foo" and "food".
29990 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29993 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
29994 boundary crossing check and subsequent code generation agree.
29996 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29998 * config/aarch64/aarch64.c (has_memory_op): Delete.
29999 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
30002 2017-02-13 Jakub Jelinek <jakub@redhat.com>
30004 PR rtl-optimization/79388
30005 PR rtl-optimization/79450
30006 * combine.c (distribute_notes): When removing TEM_INSN for which
30007 corresponding dest has last value recorded, invalidate that last
30010 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30012 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
30013 of explicit '@'. Add missing assembly comment marker on branch costs
30016 2017-02-13 Nathan Sidwell <nathan@acm.org>
30018 * gengtype-lex.l (<in_struct>): Add '/'.
30020 2017-02-13 Martin Liska <mliska@suse.cz>
30023 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
30025 2017-02-13 Richard Biener <rguenther@suse.de>
30027 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
30029 * configure: Re-generate.
30030 * config.in: Likewise.
30031 * graphite-dependences.c: Simplify as if
30032 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
30033 * graphite-isl-ast-to-gimple.c: Likewise.
30034 * graphite-optimize-isl.c: Likewise.
30035 * graphite-poly.c: Likewise.
30036 * graphite-sese-to-poly.c: Likewise.
30037 * graphite.h: Likewise.
30038 * toplev.c: Include isl/version.h and use isl_version () for
30039 printing the ISL version.
30040 * doc/install.texi: Update ISL requirement.
30042 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30044 * doc/standards.texi (Standards): Update reference to
30047 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30049 * doc/extend.texi (Named Address Spaces): sourceware.org now
30051 * doc/install.texi (Binaries): Ditto.
30054 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30056 * doc/cpp.texi: Replace "stringify"/"stringification" with C
30057 standard terminology "stringize"/"stringizing" throughout.
30058 * doc/cppinternals.texi: Likewise.
30060 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30062 * doc/extend.texi: Fix some spelling mistakes and typos.
30063 * doc/invoke.texi: Likewise.
30065 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30068 * params.def (inline-min-speedup) Change from 10 to 8.
30070 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30072 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
30075 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30078 * ipa-inline-analysis.c (get_minimal_bb): New function.
30079 (record_modified): Use it.
30080 (remap_edge_change_prob): Handle also ancestor functions.
30082 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
30084 * doc/contrib.texi (Contributors): Remove broken link into
30085 the Mauve CVS repository.
30087 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30089 PR middle-end/79454
30090 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
30091 result computation whenever lhs doesn't have vector mode, not
30092 just when it has BLKmode.
30094 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30096 * doc/makefile.texi (profiledbootstrap): Refer to the
30097 installation instructions only in textual form.
30099 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30102 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
30104 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30106 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
30107 (Specific): Update mingw-w64 reference.
30109 (Specific): Remove broken link to Renesas RX processor.
30111 2017-02-10 Richard Biener <rguenther@suse.de>
30113 * toplev.c (process_options): Do not mention obsolete graphite
30114 options when printing sorry message about missing graphite support.
30115 Mention -floop-nest-optimize.
30117 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
30119 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
30120 (vtst_p16): Likewise.
30121 (vtstq_p8): Likewise.
30122 (vtstq_p16): Likewise.
30124 (vtstq_p64): Likewise.
30125 * config/arm/arm_neon.h (vgetq_lane_p64): New.
30126 (vset_lane_p64): New.
30127 (vsetq_lane_p64): New.
30129 2017-02-10 Jakub Jelinek <jakub@redhat.com>
30131 PR tree-optimization/79411
30132 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
30133 stmt operands are SSA_NAMEs used in abnormal phis.
30134 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
30137 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30140 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
30143 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30145 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
30147 2017-02-09 Jakub Jelinek <jakub@redhat.com>
30149 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
30153 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
30154 not arbitrary TREE_CONSTANT.
30157 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
30158 "omp declare target link" attribute unless is_global_var.
30159 * omp-offload.c (find_link_var_op): Likewise.
30161 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
30162 Chung-Lin Tang <cltang@codesourcery.com>
30164 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
30166 (gimplify_adjust_omp_clauses): Don't delete TILE.
30167 (gimplify_omp_for): Deal with TILE.
30168 * internal-fn.c (expand_GOACC_TILE): New function.
30169 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
30171 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
30172 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
30174 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
30175 avoid DIV for outermost collapse var.
30176 (expand_oacc_for): Insert tile element loop as needed. Adjust.
30177 Remove out of date comments, fix whitespace.
30178 * omp-general.c (omp_extract_for_data): Deal with tiling.
30179 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
30180 adjust OLF_DIM_BASE value.
30181 (struct omp_for_data): Add tiling field.
30182 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
30183 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
30184 for auto loops. Remove default auto determining, moved to
30185 oacc_loop_fixed_partitions.
30186 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
30187 stmts, add e_mask field.
30188 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
30189 (oacc_thread_numbers): Use oacc_dim_call.
30190 (oacc_xform_tile): New.
30191 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
30192 (finish_oacc_loop): Adjust for ifns vector.
30193 (oacc_loop_discover_walk): Append loop abstraction sites to list,
30194 add case for GOACC_TILE fns.
30195 (oacc_loop_xform_loop): Delete.
30196 (oacc_loop_process): Iterate over call list directly, and add
30197 handling for GOACC_TILE fns.
30198 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
30200 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
30201 vector partitioning to outer loops. Assign 2 partitions to loops
30202 when available. Add TILE handling.
30203 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
30204 (execite_oacc_device_lower): Process GOACC_TILE fns,
30205 ignore unknown specs.
30206 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
30207 * tree.c (omp_clause_num_ops): Adjust TILE ops.
30208 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
30210 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
30212 * configure.ac (ACX_BUGURL): Update.
30213 * configure: Regenerate.
30215 2017-02-09 Richard Biener <rguenther@suse.de>
30217 PR tree-optimization/69823
30218 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
30219 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
30221 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30223 * config/arc/arc-c.def: Add __NPS400__ definition.
30224 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
30225 (TARGET_NPS400): Define.
30227 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30229 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
30231 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
30232 pointer, arch_info.
30233 (arc_cpu_types): Fill the arch_info field with a pointer into the
30234 arc_arch_types table.
30235 (arc_selected_cpu): Declare.
30236 * config/arc/arc.c (arc_selected_cpu): Make global.
30237 (arc_selected_arch): Delete.
30238 (arc_base_cpu): Delete.
30239 (arc_override_options): Remove references to deleted variables,
30240 update access to arch information.
30241 (ARC_OPT): Update access to arch information.
30242 (ARC_OPTX): Likewise.
30243 * config/arc/arc.h (arc_base_cpu): Remove declaration.
30244 (TARGET_ARC600): Update access to arch information.
30245 (TARGET_ARC601): Likewise.
30246 (TARGET_ARC700): Likewise.
30247 (TARGET_EM): Likewise.
30248 (TARGET_HS): Likewise.
30249 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
30252 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
30255 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
30256 condition/operands for integer GE/LE/GEU/LEU operations.
30258 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
30260 PR translation/79397
30261 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
30264 2017-02-08 Martin Jambor <mjambor@suse.cz>
30267 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
30268 whether allocation happened.
30269 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
30270 nothing was allocated.
30272 2017-02-08 Jakub Jelinek <jakub@redhat.com>
30274 PR tree-optimization/79408
30275 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
30276 constant, but SSA_NAME with a known integer range, use the minimum
30277 of that range instead of op1 to determine if modulo can be replaced
30278 with its first operand.
30280 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30282 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
30284 2017-02-08 Richard Biener <rguenther@suse.de>
30286 PR tree-optimization/71824
30287 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
30288 Check all loops contained in the merged region.
30290 2017-02-07 Andrew Pinski <apinski@cavium.com>
30292 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
30294 2017-02-07 Andrew Pinski <apinski@cavium.com>
30296 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
30297 (thunderxt88): Likewise.
30298 (thunderxt81): Disable LSE and change v8.1 to v8.
30299 (thunderxt83): Likewise.
30301 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30302 Richard Biener <rguenther@suse.de>
30304 PR middle-end/79399
30305 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
30306 type from int to size_t.
30307 * ira-costs.c (struct_costs_size): Change type from int to size_t.
30309 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30311 PR rtl-optimization/79386
30312 * cprop.c (bypass_conditional_jumps): Initialize
30313 bypass_last_basic_block already before splitting bbs after
30314 unconditional traps...
30315 (bypass_conditional_jumps): ... rather than here.
30318 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
30319 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
30320 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
30321 fix -masm=intel patterns.
30323 2017-02-07 Richard Biener <rguenther@suse.de>
30325 PR tree-optimization/79256
30326 PR middle-end/79278
30327 * builtins.c (get_object_alignment_2): Use min_align_of_type
30328 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
30329 and ADJUST_FIELD_ALIGN.
30331 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
30333 * doc/tm.texi: Regenerate.
30334 * stor-layout.c (layout_decl): Adjust.
30335 (update_alignment_for_field): Likewise.
30336 (place_field): Likewise.
30337 (min_align_of_type): Likewise.
30338 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
30339 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
30340 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
30341 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
30342 * config/frv/frv.c (frv_adjust_field_align): Likewise.
30343 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
30344 * config/i386/i386.c (x86_field_alignment): Likewise.
30345 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
30346 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
30347 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
30348 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
30349 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
30350 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
30354 2017-01-30 Richard Biener <rguenther@suse.de>
30356 PR tree-optimization/79256
30357 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
30358 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
30361 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
30363 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
30364 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
30365 builtins to SImode and emit a zero-extend, if necessary.
30367 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30369 * docs/invoke.texi (RISC-V Options): Alphabetize.
30371 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30373 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
30376 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30378 * config/riscv/riscv.c: New file.
30379 * common/config/riscv/riscv-common.c: Likewise.
30380 * config.gcc: Likewise.
30381 * config/riscv/constraints.md: Likewise.
30382 * config/riscv/elf.h: Likewise.
30383 * config/riscv/generic.md: Likewise.
30384 * config/riscv/linux.h: Likewise.
30385 * config/riscv/multilib-generator: Likewise.
30386 * config/riscv/peephole.md: Likewise.
30387 * config/riscv/pic.md: Likewise.
30388 * config/riscv/predicates.md: Likewise.
30389 * config/riscv/riscv-builtins.c: Likewise.
30390 * config/riscv/riscv-c.c: Likewise.
30391 * config/riscv/riscv-ftypes.def: Likewise.
30392 * config/riscv/riscv-modes.def: Likewise.
30393 * config/riscv/riscv-opts.h: Likewise.
30394 * config/riscv/riscv-protos.h: Likewise.
30395 * config/riscv/riscv.h: Likewise.
30396 * config/riscv/riscv.md: Likewise.
30397 * config/riscv/riscv.opt: Likewise.
30398 * config/riscv/sync.md: Likewise.
30399 * config/riscv/t-elf-multilib: Likewise.
30400 * config/riscv/t-linux: Likewise.
30401 * config/riscv/t-linux-multilib: Likewise.
30402 * config/riscv/t-riscv: Likewise.
30403 * configure.ac: Likewise.
30404 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
30405 Waterman as RISC-V maintainers.
30406 * doc/install.texi: Add RISC-V entries.
30407 * doc/invoke.texi: Add RISC-V options section.
30408 * doc/md.texi: Add RISC-V constraints section.
30409 * configure: Regenerated.
30411 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
30414 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
30415 false values to be constant vectors with all 0 or all 1 bits set.
30416 (vcondu<mode><mode>): Likewise.
30417 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
30419 (fpmask_comparison_operator): Update comment.
30420 (vecint_comparison_operator): New predicate.
30421 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
30422 vector conditionals when the true and false values are constant
30423 vectors with all 0 bits or all 1 bits set.
30425 2017-02-06 Martin Sebor <msebor@redhat.com>
30427 PR tree-optimization/79376
30428 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
30430 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
30432 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
30433 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
30434 to simplify split condition.
30436 2017-02-06 Jakub Jelinek <jakub@redhat.com>
30438 * omp-expand.c (oxpand_omp_atomic_fetch_op,
30439 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
30442 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
30444 PR rtl-optimization/68664
30445 * target.def (can_speculate_insn): New hook.
30446 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
30447 * doc/tm.texi: Regenerate.
30448 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
30449 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
30450 (rs6000_sched_can_speculate_insn): New function.
30452 2017-02-06 Jakub Jelinek <jakub@redhat.com>
30454 PR tree-optimization/79284
30455 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
30456 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
30457 vectorizable_mask_load_store, vectorizable_operation,
30458 vect_is_simple_cond, get_same_sized_vectype): Use it instead
30459 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
30460 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
30461 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
30462 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
30463 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
30464 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
30465 is_gimple_assign (stmt). Replace another such test with
30466 is_gimple_assign (stmt).
30468 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
30471 * config/avr/avr.c (rtl-iter.h): Include it.
30472 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
30473 (avr_legitimate_combined_insn): ...and implementation.
30475 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
30477 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
30478 * config/s390/s390.c (s390_const_operand_ok)
30479 (s390_canonicalize_comparison, s390_extract_part)
30480 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
30481 (s390_contiguous_bitmask_p, s390_rtx_costs)
30482 (legitimize_pic_address): Likewise.
30483 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
30484 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
30485 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
30486 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
30487 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
30489 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
30491 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
30492 REGNO($0) == REGNO($1).
30494 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30496 * config/s390/linux.h(SIZE_TYPE): Add comment.
30498 2017-02-06 Julian Brown <julian@codesourcery.com>
30499 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30500 Virendra Pathak <virendra.pathak@broadcom.com>
30502 * config/aarch64/aarch64-cores.def: Change the scheduler
30504 * config/aarch64/aarch64.md: Include thunderx2t99.md.
30505 * config/aarch64/thunderx2t99.md: New file.
30507 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30509 * doc/standards.texi (Go Language): Update link to language
30512 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30514 * tree-eh.c (lower_resx): Sanitize profile.
30515 (cleanup_empty_eh_move_lp): Likewise.
30517 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30520 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
30522 * cfgloopmanip.h (loop_version): Update prototype.
30523 * modulo-sched.c (sms_schedule): Update call of loop_version.
30524 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
30525 * tree-parloops.c (gen_parallel_loop): Likewise.
30526 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
30527 * tree-ssa-loop-split.c (split_loop): Likewise.
30528 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
30529 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
30531 2017-02-05 Martin Liska <mliska@suse.cz>
30534 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
30536 (print_operand_address): Initialize a struct to zero.
30538 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30540 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
30541 garbage collector only in textual form.
30543 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30545 * doc/extend.texi (x86 specific memory model extensions for
30546 transactional memory): Simplify a phrase.
30548 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
30551 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
30552 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
30553 (atomic_storedi_1): Likewise.
30555 2017-02-04 Jakub Jelinek <jakub@redhat.com>
30557 PR tree-optimization/79338
30558 * tree-parloops.c (gather_scalar_reductions): Don't call
30559 vect_analyze_loop_form for loop->inner before destroying loop's
30562 2017-02-03 Martin Sebor <msebor@redhat.com>
30564 PR tree-optimization/79327
30565 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
30566 when precision has resulted in leading zeros.
30567 (format_integer): Adjust the likely counter to assume an unknown
30568 argument that may be zero is non-zero.
30570 2017-02-03 Jason Merrill <jason@redhat.com>
30573 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
30574 avoid copying non-taken branch.
30576 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30578 PR tree-optimization/79340
30579 * tree-vect-loop.c (vectorizable_reduction): Release
30580 vec_defs elements after safe_splicing them into other vectors.
30583 PR tree-optimization/79327
30584 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
30585 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
30587 (format_integer): Use wide_int_to_tree instead of build_int_cst
30588 + to_?hwi. If argmin is NULL, just set argmin and argmax to
30589 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
30590 of shortest and longest sequence.
30592 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
30594 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
30595 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
30597 2017-02-03 Walter Lee <walt@tilera.com>
30600 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
30601 after initial stackframe link reg save.
30602 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
30604 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30607 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
30608 wu for stxssp alternative.
30610 2017-02-03 Martin Sebor <msebor@redhat.com>
30612 PR tree-optimization/79352
30613 * gimple-fold.c (get_range_strlen): Add argument.
30614 (get_range_strlen): Change return type to bool.
30615 (get_maxval_strlen): Pass in a dummy argument.
30616 * gimple-fold.h (get_range_strlen): Change return type to bool.
30617 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
30618 * tree.h (array_at_struct_end_p): Add argument.
30619 * tree.c (array_at_struct_end_p): Handle it.
30621 2017-02-03 Martin Liska <mliska@suse.cz>
30624 * multiple_target.c (create_dispatcher_calls): Redirect edge
30625 from a caller of a dispatcher.
30626 (expand_target_clones): Make the clones local.
30627 (ipa_target_clone): Do both target clones and resolvers.
30628 (ipa_dispatcher_calls): Remove the pass.
30629 (pass_dispatcher_calls::gate): Likewise.
30630 (make_pass_dispatcher_calls): Likewise.
30631 * passes.def (pass_target_clone): Put as very first IPA early
30634 2017-02-03 Martin Liska <mliska@suse.cz>
30636 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
30637 in case of a function with ifunc attribute.
30639 2017-02-03 Martin Liska <mliska@suse.cz>
30641 * cgraph.c (cgraph_node::dump): Dump function version info.
30642 * symtab.c (symtab_node::dump_base): Add missing new line.
30644 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
30646 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
30647 (ifcombine_ifandif): Use it.
30649 2017-02-03 Martin Liska <mliska@suse.cz>
30651 * doc/invoke.texi: Document default value for
30652 use-after-scope-direct-emission-threshold.
30654 2017-02-03 Martin Liska <mliska@suse.cz>
30656 PR tree-optimization/79339
30657 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
30658 (format_floating): Likewise.
30660 2017-02-03 Martin Liska <mliska@suse.cz>
30663 * ipa-prop.c (ipa_node_params_t::insert): Remove current
30665 (ipa_node_params_t::remove): Likewise.
30666 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
30667 initialization from removed ipa_node_params_t::insert.
30668 (ipa_node_params::~ipa_node_params): Move from removed
30669 ipa_node_params_t::release.
30670 * symbol-summary.h (symbol_summary::m_released): New member.
30671 Do not release a summary twice. Do not allow to call finalizer
30672 for types of a summary that live in GGC memory.
30674 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30676 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
30679 2017-02-02 Martin Sebor <msebor@redhat.com>
30681 PR middle-end/79275
30682 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
30683 (format_string): Tighten up the range of output for non-constant
30684 strings and correct the expected range for wide non-constant strings.
30686 2017-02-02 Martin Sebor <msebor@redhat.com>
30688 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
30690 PR middle-end/32003
30691 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
30693 (-fdump-tree-@var): Add to index and document how to come up
30694 with pass-specific option and dump file names.
30695 (-fdump-passes): Clarify where to look for output.
30697 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
30699 PR middle-end/77445
30700 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
30701 statistics of the analyzed path; allow threading for speed when
30702 any of BBs along the path are optimized for speed.
30704 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
30706 PR middle-end/78468
30707 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
30708 settings of the virtual registers.
30711 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
30713 * explow.c (get_dynamic_stack_size): Take known alignment of stack
30714 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
30717 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30719 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
30720 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
30722 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30724 * config/s390/s390.md: Add missing comments with the expanded
30726 * config/s390/vector.md: Likewise.
30727 * config/s390/vx-builtins.md: Likewise.
30729 2017-02-02 Jakub Jelinek <jakub@redhat.com>
30732 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
30733 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
30734 conditions on a single line.
30736 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30738 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
30739 __S390_VX__ to __VX__.
30741 2017-02-01 Andrew Pinski <apinski@cavium.com>
30743 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
30744 stmt_info to record_stmt_cost.
30745 (vect_get_known_peeling_cost): Pass stmt_info if known to
30747 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
30748 cpu_vector_cost field into
30749 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
30750 field into vec_int_stmt_cost and vec_fp_stmt_cost.
30751 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
30752 splitting of scalar_stmt_cost and vec_stmt_cost.
30753 (thunderx_vector_cost): Likewise.
30754 (cortexa57_vector_cost): LIkewise.
30755 (exynosm1_vector_cost): Likewise.
30756 (xgene1_vector_cost): Likewise.
30757 (thunderx2t99_vector_cost): Improve after the splitting of the two
30759 (aarch64_builtin_vectorization_cost): Update for the splitting of
30760 scalar_stmt_cost and vec_stmt_cost.
30762 2017-02-01 Torvald Riegel <triegel@redhat.com>
30763 Richard Henderson <rth@redhat.com>
30765 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
30766 conditional on existance of a fast atomic load.
30767 * optabs-query.c (can_atomic_load_p): New function.
30768 * optabs-query.h (can_atomic_load_p): Declare it.
30769 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
30770 no fast atomic load is available for the particular size of access.
30771 (expand_atomic_compare_and_swap): Likewise.
30772 (expand_atomic_load): Likewise.
30773 (expand_atomic_store): Likewise.
30774 (expand_atomic_fetch_op): Likewise.
30775 * testsuite/lib/target-supports.exp
30776 (check_effective_target_sync_int_128): Remove x86 because it provides
30777 no fast atomic load.
30778 (check_effective_target_sync_int_128_runtime): Likewise.
30780 2017-02-01 Richard Biener <rguenther@suse.de>
30782 * graphite.c: Include tree-vectorizer.h for find_loop_location.
30783 (graphite_transform_loops): Provide opt-info for optimized nests.
30784 * tree-parloop.c (parallelize_loops): Provide opt-info for
30785 parallelized loops.
30787 2017-02-01 Richard Biener <rguenther@suse.de>
30789 PR middle-end/79315
30790 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
30791 was not set before.
30793 2017-02-01 Richard Biener <rguenther@suse.de>
30795 PR tree-optimization/71824
30796 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
30797 Verify the loops are valid in the merged SESE region.
30798 (scop_detection::can_represent_loop_1): Check analyzing the
30799 evolution of the number of iterations in the region succeeds.
30801 2017-01-31 Ian Lance Taylor <iant@golang.org>
30803 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
30804 REG_ARGS_SIZE note to 32-bit push insns and call insn.
30806 2017-01-31 David Malcolm <dmalcolm@redhat.com>
30808 PR preprocessor/79210
30809 * input.c (get_substring_ranges_for_loc): Replace line_width
30810 assertion with error-handling.
30812 2017-01-31 Richard Biener <rguenther@suse.de>
30814 PR tree-optimization/77318
30815 * graphite-sese-to-poly.c (extract_affine): Fix assert.
30816 (create_pw_aff_from_tree): Take loop parameter.
30817 (add_condition_to_pbb): Pass loop of the condition to
30818 create_pw_aff_from_tree.
30820 2017-01-31 Jakub Jelinek <jakub@redhat.com>
30822 * config/s390/s390.c (s390_asan_shadow_offset): New function.
30823 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
30825 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
30829 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
30831 (convert_int_to_float128): Likewise.
30832 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
30833 (convert_int_to_float128): Likewise.
30834 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
30835 (UNSPEC_IEEE128_CONVERT): Likewise.
30836 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
30837 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
30838 Use local variables for IBM extended format.
30839 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
30840 (fix_trunc<mode>si2_fprs): Likewise.
30841 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
30842 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
30843 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
30844 to know that we can now have integers of all sizes in vector
30846 (fix<uns>_<mode>di2_hw): Likewise.
30847 (float<uns>_<mode>si2_hw): Likewise.
30848 (fix_<mode>si2_hw): Likewise.
30849 (fixuns_<mode>si2_hw): Likewise.
30850 (float<uns>_<mode>di2_hw): Likewise.
30851 (float_<mode>di2_hw): Likewise.
30852 (float_<mode>si2_hw): Likewise.
30853 (floatuns_<mode>di2_hw): Likewise.
30854 (floatuns_<mode>si2_hw): Likewise.
30855 (xscvqp<su>wz_<mode>): Delete, no longer used.
30856 (xscvqp<su>dz_<mode>): Likewise.
30857 (xscv<su>dqp_<mode>): Likewise.
30858 (ieee128_mfvsrd_64bit): Likewise.
30859 (ieee128_mfvsrd_32bit): Likewise.
30860 (ieee128_mfvsrwz): Likewise.
30861 (ieee128_mtvsrw): Likewise.
30862 (ieee128_mtvsrd_64bit): Likewise.
30863 (ieee128_mtvsrd_32bit): Likewise.
30865 2017-01-31 Martin Liska <mliska@suse.cz>
30868 * ipa-prop.c (ipa_free_all_node_params): Call release method
30869 instead of ~sumbol_summary to not to trigger double times
30872 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
30874 PR tree-optimization/71691
30875 * bitmap.h (class auto_bitmap): New.
30876 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
30877 is_maybe_undefined instead of ssa_undefined_value_p.
30879 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30881 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
30882 __S390_ARCH_LEVEL__ to __ARCH__.
30884 2017-01-31 Jakub Jelinek <jakub@redhat.com>
30886 PR tree-optimization/79267
30887 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
30888 if should_remove_lhs_p is true.
30890 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
30893 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
30894 (add_alignment_attribute): New.
30895 (base_type_die): Add alignment attribute.
30896 (subrange_type_die): Likewise.
30897 (modified_type_die): Likewise.
30898 (gen_array_type_die): Likewise.
30899 (gen_descr_array_type_die: Likewise.
30900 (gen_enumeration_type_die): Likewise.
30901 (gen_subprogram_die): Likewise.
30902 (gen_variable_die): Likewise.
30903 (gen_field_die): Likewise.
30904 (gen_ptr_to_mbr_type_die): Likewise.
30905 (gen_struct_or_union_type_die): Likewise.
30906 (gen_subroutine_type_die): Likewise.
30907 (gen_typedef_die): Likewise.
30908 (base_type_cmp): Compare alignment attribute.
30910 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30913 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
30914 (setb_unsigned) New pattern for setb with CCUNS.
30915 * config/rs6000/rs6000.c (expand_block_compare): Use a different
30916 subfc./subfe sequence to avoid overflow problems. Generate a
30917 shorter sequence with cmpld/setb for power9.
30918 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
30919 for generating subfc. instruction.
30920 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
30921 now uses this instruction.
30923 2017-01-30 Ian Lance Taylor <iant@google.com>
30926 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
30927 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
30929 2017-01-30 Martin Sebor <msebor@redhat.com>
30931 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
30932 Move constant to the right of a relational operator.
30933 (get_mpfr_format_length, format_character, format_string): Ditto.
30934 (should_warn_p, maybe_warn): Same.
30936 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
30938 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
30941 * asan.c (get_translation_unit_decl): Remove function.
30942 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
30944 2017-01-30 Martin Liska <mliska@suse.cz>
30946 PR gcov-profile/79259
30947 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
30948 -fprofile-generate.
30950 2017-01-30 Martin Liska <mliska@suse.cz>
30953 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
30954 Initialize variables with NULL value.
30956 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
30959 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
30961 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
30963 2017-01-30 Richard Biener <rguenther@suse.de>
30965 PR tree-optimization/79276
30966 * tree-vrp.c (process_assert_insertions): Properly adjust common
30967 when removing a duplicate.
30969 2017-01-30 Richard Biener <rguenther@suse.de>
30971 PR tree-optimization/79256
30972 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
30973 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
30975 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
30977 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
30980 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
30981 ("*r<noxa>sbg_<mode>_sll_bitmask")
30982 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
30983 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
30984 Use contiguous_bitmask_nowrap_operand.
30986 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30989 * config/rs6000/altivec.h (vec_xl): Revise #define.
30990 (vec_xst): Likewise.
30992 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
30994 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
30996 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
30998 PR rtl-optimization/79194
30999 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
31000 traps before call to bypass_conditional_jumps.
31002 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31004 PR tree-optimization/71374
31005 * lra-constraints.c (check_conflict_input_operands): New.
31006 (match_reload): Use it.
31008 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31011 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
31012 account to calculate conflict_set.
31014 2017-01-27 Bin Cheng <bin.cheng@arm.com>
31016 PR rtl-optimization/78559
31017 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
31018 other_insn in combine.
31020 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31022 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
31023 uint16_type_node for BT_UINT16.
31025 2017-01-27 David Malcolm <dmalcolm@redhat.com>
31027 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
31028 "RTL Tests" to menu.
31029 (GIMPLE Tests): New node.
31030 (RTL Tests): New node.
31032 2017-01-27 Richard Biener <rguenther@suse.de>
31034 PR tree-optimization/79245
31035 * tree-loop-distribution.c (distribute_loop): Apply cost
31036 modeling also to detected patterns.
31038 2017-01-27 Richard Biener <rguenther@suse.de>
31040 PR tree-optimization/71433
31041 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
31042 (compare_assert_loc): New function.
31043 (process_assert_insertions): Sort and optimize assert locations
31044 to remove duplicates and push down identical assertions on
31045 edges to their destination block.
31047 2017-01-27 Richard Biener <rguenther@suse.de>
31049 PR tree-optimization/79244
31050 * tree-vrp.c (remove_range_assertions): Forcefully propagate
31051 out SSA names even if abnormal.
31053 2017-01-27 Jakub Jelinek <jakub@redhat.com>
31055 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
31056 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
31057 instead of MPFR_RNDN.
31059 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
31062 * arm.c (arm_option_override): Don't call build_target_option_node
31063 until after doing all option overrides.
31064 (arm_valid_target_attribute_tree): Likewise.
31066 2017-01-27 Martin Liska <mliska@suse.cz>
31068 * doc/invoke.texi (-fprofile-arcs): Document profiling support
31069 for {cd}tors and C++ {cd}tors.
31071 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
31073 * config/s390/s390.md ("*setmem_long_and")
31074 ("*setmem_long_and_31z"): Use zero_extend instead of and.
31076 2017-01-26 Martin Sebor <msebor@redhat.com>
31078 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
31081 2017-01-26 Martin Sebor <msebor@redhat.com>
31083 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
31084 HAVE_DFmode before using XFmode or DFmode.
31085 (parse_directive): Avoid using the z length modifier to avoid
31086 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
31088 PR middle-end/78703
31089 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
31090 to accept adjustment as an array.
31091 (get_int_range): New function.
31092 (struct directive): Make width and prec arrays.
31093 (directive::set_width, directive::set_precision): Call get_int_range.
31094 (format_integer, format_floating): Handle width and precision ranges.
31095 (format_string, parse_directive): Same.
31097 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31100 * dwarf2out.c (generate_skeleton_bottom_up): For children with
31101 comdat_type_p set, just clone them, but keep the children in the
31105 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
31106 which have direct callers with -fvar-tracking-assignments enabled
31108 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
31109 inside of type units.
31111 2017-01-26 Martin Sebor <msebor@redhat.com>
31113 PR middle-end/78703
31114 * gimple-ssa-sprintf.c (struct result_range): Add likely and
31116 (struct format_result): Replace number_chars, number_chars_min,
31117 and number_chars_max with a single member of struct result_range.
31119 (format_result::operator+=): Adjust.
31120 (struct fmtresult): Remove bounded. Handle likely and unlikely
31122 (fmtresult::adjust_for_width_or_precision): New function.
31123 (fmtresult:type_max_digits): New function.
31124 (bytes_remaining): Handle likely and unlikely counters.
31125 (min_bytes_remaining): Remove.
31126 (format_percent): Simplify.
31127 (format_integer, format_floating): Set likely and unlikely counters.
31128 (get_string_length, format_character, format_string): Same.
31129 (format_plain, should_warn_p): New function.
31130 (maybe_warn): Call should_warn_p. Update diagnostic messages
31131 and handle those for all directives, including plain strings.
31132 (format_directive): Handle likely and unlikely counters.
31133 Remove unnecessary quoting from diagnostics. Add an informational
31135 (add_bytes): Remove.
31136 (pass_sprintf_length::compute_format_length): Simplify.
31137 (try_substitute_return_value): Handle likely and unlikely counters.
31139 2017-01-26 Carl Love <cel@us.ibm.com>
31141 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
31142 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
31144 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
31147 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
31148 endianess for subregs into account.
31149 * lra-constraints.c (lra_constraints): Do risky transformations
31150 always on the first iteration.
31151 * lra-lives.c (check_pseudos_live_through_calls): Add arg
31152 last_call_used_reg_set.
31153 (process_bb_lives): Define and use last_call_used_reg_set.
31154 * lra.c (lra): Always continue after lra_constraints on the first
31157 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
31159 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
31161 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
31163 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31165 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
31166 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
31167 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
31168 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
31169 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
31170 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
31171 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
31172 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
31173 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
31175 2017-01-26 Marek Polacek <polacek@redhat.com>
31178 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
31179 for the third operand.
31181 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31183 PR middle-end/79236
31184 * omp-low.c (struct omp_context): Add simt_stmt field.
31185 (scan_omp_for): Return omp_context *.
31186 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
31187 context to the _simt_ SIMD stmt.
31188 (lower_omp_for): For combined SIMD with sibling _simt_
31189 SIMD, make sure to use the same decls in _looptemp_
31190 clauses as in the sibling.
31192 2017-01-26 David Sherwood <david.sherwood@arm.com>
31194 PR middle-end/79212
31195 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
31198 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31201 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
31202 emit fld b; fld a; if possible.
31204 * brig-builtins.def: Update copyright years.
31205 * config/arm/arm_acle_builtins.def: Update copyright years.
31207 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
31210 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
31211 constraint instead of o for the stxsd instruction.
31213 2017-01-25 Carl Love <cel@us.ibm.com>
31215 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
31216 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
31218 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
31220 * doc/invoke.texi (C++ Dialect Options): Fix typo.
31222 2017-01-25 Richard Biener <rguenther@suse.de>
31224 PR tree-optimization/69264
31225 * target.def (vector_alignment_reachable): Improve documentation.
31226 * doc/tm.texi: Regenerate.
31227 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
31229 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
31230 earlier changes with respect to TYPE_USER_ALIGN.
31231 (vector_alignment_reachable_p): Likewise. Improve dumping.
31233 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31236 * config/arm/arm.md (xordi3): Force constant operand into a register
31239 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31241 * doc/invoke.texi (-fstore-merging): Correct default optimization
31242 levels at which it is enabled.
31243 (-O): Move -fstore-merging from list to...
31246 2017-01-25 Richard Biener <rguenther@suse.de>
31249 * omp-expand.c: Include debug.h.
31250 (expand_omp_taskreg): Make sure to generate early debug before
31251 outlining anything from a function.
31252 (expand_omp_target): Likewise.
31253 (grid_expand_target_grid_body): Likewise.
31255 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
31258 * asan.c (get_translation_unit_decl): New function.
31259 (asan_add_global): Extract modules file name from globals
31260 TRANSLATION_UNIT_DECL name.
31262 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
31265 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
31266 for long calls with APCS frame and VFP.
31268 2017-01-24 David Malcolm <dmalcolm@redhat.com>
31270 * cfg.c (original_copy_tables_initialized_p): New function.
31271 * cfg.h (original_copy_tables_initialized_p): New decl.
31272 * cfgrtl.c (relink_block_chain): Guard the call to
31273 free_original_copy_tables with a call to
31274 original_copy_tables_initialized_p.
31275 * cgraph.h (symtab_node::native_rtl_p): New decl.
31276 * cgraphunit.c (symtab_node::native_rtl_p): New function.
31277 (symtab_node::needed_p): Don't assert for early assembly output
31278 for __RTL functions.
31279 (cgraph_node::finalize_function): Set "force_output" for __RTL
31281 (cgraph_node::analyze): Bail out early for __RTL functions.
31282 (analyze_functions): Update assertion to support __RTL functions.
31283 (cgraph_node::expand): Bail out early for __RTL functions.
31284 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
31286 * function.h (struct function): Update comment for field
31288 * gimple-expr.c: Include "tree-pass.h".
31289 (gimple_has_body_p): Return false for __RTL functions.
31290 * Makefile.in (OBJS): Add run-rtl-passes.o.
31291 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
31293 (gcc::pass_manager::get_clean_slate): New accessor.
31294 * passes.c: Include "insn-addr.h".
31295 (should_skip_pass_p): Add logging. Update logic for running
31296 "expand" to be compatible with both __GIMPLE and __RTL. Guard
31297 property-provider override so it is only done for gimple passes.
31299 (skip_pass): New function.
31300 (execute_one_pass): Call skip_pass when skipping passes.
31301 * read-md.c (md_reader::read_char): Support filtering
31302 the input to a subset of line numbers.
31303 (md_reader::md_reader): Initialize fields
31304 m_first_line and m_last_line.
31305 (md_reader::read_file_fragment): New function.
31306 * read-md.h (md_reader::read_file_fragment): New decl.
31307 (md_reader::m_first_line): New field.
31308 (md_reader::m_last_line): New field.
31309 * read-rtl-function.c (function_reader::create_function): Only
31310 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
31311 curr_properties. Set DECL_INITIAL to a dummy block.
31312 (read_rtl_function_body_from_file_range): New function.
31313 * read-rtl-function.h (read_rtl_function_body_from_file_range):
31315 * run-rtl-passes.c: New file.
31316 * run-rtl-passes.h: New file.
31318 2017-01-24 Jeff Law <law@redhat.com>
31320 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
31323 2017-01-24 Bin Cheng <bin.cheng@arm.com>
31325 PR tree-optimization/79159
31326 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
31327 (record_nonwrapping_iv): Improve boundary using above function if no
31328 value range information.
31330 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
31331 Martin Jambor <mjambor@suse.cz>
31333 * brig-builtins.def: New file.
31334 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
31335 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
31336 (DEF_HSAIL_SAT_BUILTIN): Likewise.
31337 (DEF_HSAIL_INTR_BUILTIN): Likewise.
31338 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
31339 * builtin-types.def (BT_INT8): New.
31340 (BT_INT16): Likewise.
31341 (BT_UINT8): Likewise.
31342 (BT_UINT16): Likewise.
31343 (BT_FN_ULONG): Likewise.
31344 (BT_FN_UINT_INT): Likewise.
31345 (BT_FN_UINT_ULONG): Likewise.
31346 (BT_FN_UINT_LONG): Likewise.
31347 (BT_FN_UINT_PTR): Likewise.
31348 (BT_FN_ULONG_PTR): Likewise.
31349 (BT_FN_INT8_FLOAT): Likewise.
31350 (BT_FN_INT16_FLOAT): Likewise.
31351 (BT_FN_UINT32_FLOAT): Likewise.
31352 (BT_FN_UINT16_FLOAT): Likewise.
31353 (BT_FN_UINT8_FLOAT): Likewise.
31354 (BT_FN_UINT64_FLOAT): Likewise.
31355 (BT_FN_UINT16_UINT32): Likewise.
31356 (BT_FN_UINT32_UINT16): Likewise.
31357 (BT_FN_UINT16_UINT16_UINT16): Likewise.
31358 (BT_FN_INT_PTR_INT): Likewise.
31359 (BT_FN_UINT_PTR_UINT): Likewise.
31360 (BT_FN_LONG_PTR_LONG): Likewise.
31361 (BT_FN_ULONG_PTR_ULONG): Likewise.
31362 (BT_FN_VOID_UINT64_UINT64): Likewise.
31363 (BT_FN_UINT8_UINT8_UINT8): Likewise.
31364 (BT_FN_INT8_INT8_INT8): Likewise.
31365 (BT_FN_INT16_INT16_INT16): Likewise.
31366 (BT_FN_INT_INT_INT): Likewise.
31367 (BT_FN_UINT_FLOAT_UINT): Likewise.
31368 (BT_FN_FLOAT_UINT_UINT): Likewise.
31369 (BT_FN_ULONG_UINT_UINT): Likewise.
31370 (BT_FN_ULONG_UINT_PTR): Likewise.
31371 (BT_FN_ULONG_ULONG_ULONG): Likewise.
31372 (BT_FN_UINT_UINT_UINT): Likewise.
31373 (BT_FN_VOID_UINT_PTR): Likewise.
31374 (BT_FN_UINT_UINT_PTR: Likewise.
31375 (BT_FN_UINT32_UINT64_PTR): Likewise.
31376 (BT_FN_INT_INT_UINT_UINT): Likewise.
31377 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
31378 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
31379 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
31380 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
31381 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
31382 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
31383 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
31384 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
31385 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
31386 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
31387 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
31388 * doc/frontends.texi: List BRIG FE.
31389 * doc/install.texi (Testing): Add BRIG tesring requirements.
31390 * doc/invoke.texi (Overall Options): Mention BRIG.
31391 * doc/standards.texi (Standards): Doucment BRIG HSA version.
31393 2017-01-24 Richard Biener <rguenther@suse.de>
31395 PR translation/79208
31396 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
31398 2017-01-24 Martin Jambor <mjambor@suse.cz>
31401 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
31402 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
31403 and known_contexts.
31405 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
31407 PR middle-end/79123
31408 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
31409 casts from signed to unsigned really don't have a range.
31411 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
31413 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
31414 GMP_RNDx for compatiblity.
31416 2017-01-24 Martin Liska <mliska@suse.cz>
31419 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
31420 that would prevent us to call alloca with -1 as argument.
31422 2017-01-24 Jakub Jelinek <jakub@redhat.com>
31424 * dwarf2out.c (output_compilation_unit_header, output_file_names):
31425 Avoid -Wformat-security warning.
31427 2017-01-23 Andrew Pinski <apinski@cavium.com>
31429 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
31432 2017-01-23 Martin Sebor <msebor@redhat.com>
31434 PR middle-end/78703
31435 * gimple-ssa-sprintf.c (warn_level): New global.
31436 (format_integer): Use it here and throughout the rest of the file.
31437 Use the same switch to compute sign as base.
31438 (maybe_warn): New function.
31439 (format_directive): Factor out warnings into maybe_warn.
31440 Add debugging output. Use warn_level.
31441 (add_bytes): Use warn_level.
31442 (pass_sprintf_length::compute_format_length): Add debugging output.
31443 (try_substitute_return_value): Same.
31444 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
31446 PR middle-end/78703
31447 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
31448 (struct fmtresult, format_integer, format_floating): Adjust.
31449 (fmtresult::fmtresult): Set max correctly in two argument ctor.
31450 (get_string_length, format_string,format_directive): Same.
31451 (pass_sprintf_length::compute_format_length): Same.
31452 (try_substitute_return_value): Simplify slightly.
31454 PR middle-end/78703
31455 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
31456 (fmtresult::operator+=): Outlined.
31457 (struct fmtresult): Add ctors.
31458 (struct conversion_spec): Rename...
31459 (struct directive): ...to this. Add and remove data members.
31460 (directive::set_width, directive::set_precision): New functions.
31461 (format_percent): Use fmtresult ctor.
31462 (get_width_and_precision): Remove.
31463 (format_integer): Make naming changes. Avoid computing width and
31465 (format_floating): Same. Adjust indentation.
31466 (format_character, format_none): New functions.
31467 (format_string): Moved character handling to format_character.
31468 (format_directive): Remove arguments, change return type.
31469 (parse_directive): New function.
31470 (pass_sprintf_length::compute_format_length): Move directive
31471 parsing to parse_directive.
31473 2017-01-23 Jakub Jelinek <jakub@redhat.com>
31475 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
31476 (assign_assembler_name_if_needed): ... this.
31477 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
31478 (assign_assembler_name_if_needed): ... this.
31479 (free_lang_data_in_cgraph): Adjust callers.
31480 * cgraphunit.c (cgraph_node::analyze): Likewise.
31481 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
31484 2017-01-23 Richard Biener <rguenther@suse.de>
31486 PR tree-optimization/79088
31487 PR tree-optimization/79188
31488 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
31489 resetting loop bounds after last path deletion. Reset loop
31490 bounds of the target loop, make code match the comments.
31491 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
31492 Make sure loops need no fixups.
31494 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
31496 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
31497 exponent support with double type for first argument.
31498 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
31499 type returned by __builtin_vec_extract_sig,
31500 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
31501 functions from "vector int" to "vector unsigned int" or from
31502 "vector long long int" to "vector unsigned long long int".
31503 Changed type returned by __builtin_vec_extract_exp,
31504 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
31505 functions from "vector int" to "vector unsigned int" or from
31506 "vector long long int" to "vector unsigned long long int".
31507 Changed return type of __builtin_vec_test_data_class,
31508 __builtin_vec_test_data_class_sp, and
31509 __builtin_vec_test_data_class_dp from "vector int" to
31510 "vector bool int" or from "vector long long int" to "vector bool
31511 long long int" and changed second argument type from "unsigned
31512 int" to "int". Added new overloaded function forms "vector float
31513 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
31514 "vector float __builtin_vec_insert_exp_sp (vector float, vector
31515 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
31516 double, vector unsigned long long int)" and "vector double
31517 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
31518 long int)". Changed return type of
31519 __builtin_scalar_test_data_class and
31520 __builtin_scalar_test_data_class_sp and
31521 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
31522 int" and changed second argument from "unsigned int" to "int".
31523 Changed type returned by __builtin_scalar_test_neg,
31524 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
31525 from "int" to "bool int". Added new overloaded function form
31526 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
31527 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
31528 exponent double-precision with floating point first argument.
31529 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
31530 documentation of scalar_test_data_class, scalar_test_neg,
31531 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
31532 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
31533 vec_test_data_class built-in functions to reflect refinements in
31534 their type signatures.
31536 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
31538 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
31540 (aarch64_elf_asm_destructor): Likewise.
31542 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
31544 PR rtl-optimization/78634
31545 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
31546 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
31547 * ifcvt.c (noce_try_cmove): Add missing cost check.
31549 PR rtl-optimization/71724
31550 * combine.c (if_then_else_cond): Look for situations where it is
31551 beneficial to undo the work of one of the recursive calls.
31553 2017-01-23 Bin Cheng <bin.cheng@arm.com>
31555 PR tree-optimization/70754
31556 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
31557 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
31558 combined stmt before it if not NULL.
31559 (combine_chains): Process refs reversely and compute dominance point
31562 2017-01-23 Martin Liska <mliska@suse.cz>
31564 PR tree-optimization/79196
31565 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
31566 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
31568 (strlen_optimize_stmt): Call the renamed function.
31570 2017-01-23 Michael Matz <matz@suse.de>
31572 PR tree-optimization/78384
31573 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
31575 2017-01-23 Richard Biener <rguenther@suse.de>
31577 PR tree-optimization/79186
31578 * tree-vrp.c (register_new_assert_for): Make sure we've seen
31579 both incoming edges before moving an assert.
31581 2017-01-23 Martin Jambor <mjambor@suse.cz>
31583 * ipa-prop.c (load_from_param_1): Removed.
31584 (load_from_unmodified_param): Bits from load_from_param_1 put back
31586 (load_from_param): Removed.
31587 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
31588 with stmt. Reverted back to use of load_from_unmodified_param.
31590 2017-01-23 Martin Jambor <mjambor@suse.cz>
31593 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
31594 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
31595 field a pointer to garbage collected vector, mark lattices and
31596 ipcp_orig_node with GTY((skip)).
31597 (ipa_get_param_count): Adjust to descriptors being a pointer.
31598 (ipa_get_param): Likewise.
31599 (ipa_get_type): Likewise.
31600 (ipa_get_param_move_cost): Likewise.
31601 (ipa_set_param_used): Likewise.
31602 (ipa_get_controlled_uses): Likewise.
31603 (ipa_set_controlled_uses): Likewise.
31604 (ipa_is_param_used): Likewise.
31605 (ipa_node_params_t): Move into garbage collector. New methods insert
31607 (ipa_node_params_sum): Annotate wth GTY(()).
31608 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
31610 (ipa_load_from_parm_agg): Adjust declaration.
31611 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
31612 * ipa-profile.c (ipa_profile): Likewise.
31613 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
31614 (ipa_populate_param_decls): Make descriptors parameter garbage
31616 (ipa_dump_param): Adjust to descriptors being a pointer.
31617 (ipa_alloc_node_params): Likewise.
31618 (ipa_initialize_node_params): Likewise.
31619 (load_from_param_1): Make descriptors parameter garbage collected.
31620 (load_from_unmodified_param): Likewise.
31621 (load_from_param): Likewise.
31622 (ipa_load_from_parm_agg): Likewise.
31623 (ipa_node_params::~ipa_node_params): Removed.
31624 (ipa_free_all_node_params): Remove call to delete operator.
31625 (ipa_node_params_t::insert): New.
31626 (ipa_node_params_t::remove): Likewise.
31627 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
31628 copy known_csts and known_contexts vectors.
31629 (ipa_read_node_info): Adjust to descriptors being a pointer.
31630 (ipcp_modif_dom_walker): Make m_descriptors field garbage
31632 (ipcp_transform_function): Make descriptors variable garbage
31635 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
31637 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
31638 * config/i386/avx512dqintrin.h: Ditto.
31639 * config/i386/avx512fintrin.h: Ditto.
31640 * config/i386/i386.c: Handle new builtins.
31641 * config/i386/i386-builtin.def: Add new builtins.
31642 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
31643 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
31645 2017-01-23 Jakub Jelinek <jakub@redhat.com>
31646 Martin Liska <mliska@suse.cz>
31648 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
31649 * asan.c (asan_expand_poison_ifn): Support stores and use
31650 appropriate ASAN report function.
31651 * internal-fn.c (expand_ASAN_POISON_USE): New function.
31652 * internal-fn.def (ASAN_POISON_USE): Declare.
31653 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
31654 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
31655 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
31656 ASAN_POISON calls w/o LHS.
31657 * tree-ssa.c (execute_update_addresses_taken): Create clobber
31658 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
31659 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
31660 * gimplify.c (asan_poison_variables): Add attribute
31661 use_after_scope_memory to variables that really needs to live
31663 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
31664 having the attribute.
31666 2017-01-23 Martin Liska <mliska@suse.cz>
31668 * asan.c (create_asan_shadow_var): New function.
31669 (asan_expand_poison_ifn): Likewise.
31670 * asan.h (asan_expand_poison_ifn): New declaration.
31671 * internal-fn.c (expand_ASAN_POISON): Likewise.
31672 * internal-fn.def (ASAN_POISON): New builtin.
31673 * sanopt.c (pass_sanopt::execute): Expand
31674 asan_expand_poison_ifn.
31675 * tree-inline.c (copy_decl_for_dup_finish): Make function
31677 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
31678 * tree-ssa.c (is_asan_mark_p): New function.
31679 (execute_update_addresses_taken): Rewrite local variables
31680 (identified just by use-after-scope as addressable) into SSA.
31682 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
31684 * doc/install.texi (Specific): opensource.apple.com uses https
31685 now. Remove trailing slash.
31687 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
31689 * README.Portability: Remove note on an Irix compatibility issue.
31691 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
31693 * gcov.c (INCLUDE_ALGORITHM): Define.
31694 (INCLUDE_VECTOR): Define.
31695 No longer include <vector> and <algorithm> directly.
31697 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
31699 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
31701 * doc/invoke.texi (Code Gen Options): Ditto.
31703 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
31706 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
31708 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
31710 rtl-optimization/79125
31711 * cprop.c (local_cprop_pass): Handle cases where we make an
31712 unconditional trap.
31714 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
31718 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
31719 read from, for big endian.
31721 2017-01-20 Jiong Wang <jiong.wang@arm.com>
31723 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
31724 register pauth builtins for LP64 only.
31726 2017-01-20 Marek Polacek <polacek@redhat.com>
31729 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
31732 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
31734 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
31736 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
31737 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
31738 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
31740 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31743 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
31744 in big-endian mode when they are not a single duplicated value.
31746 2017-01-20 Richard Biener <rguenther@suse.de>
31748 * BASE-VER: Bump to 7.0.1.
31750 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
31752 * omp-low.c (omplow_simd_context): New struct. Use it...
31753 (lower_rec_simd_input_clauses): ...here and...
31754 (lower_rec_input_clauses): ...here to hold common data. Adjust all
31755 references to idx, lane, max_vf, is_simt.
31757 2017-01-20 Graham Markall <graham.markall@embecosm.com>
31759 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
31762 2017-01-20 Martin Jambor <mjambor@suse.cz>
31764 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
31765 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
31767 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
31768 (GTFILES): Rename hsa.c to hsa-common.c.
31769 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
31770 * hsa-dump.c: Likewise.
31771 * hsa-gen.c: Likewise.
31772 * hsa-regalloc.c: Likewise.
31773 * ipa-hsa.c: Likewise.
31774 * omp-expand.c: Likewise.
31775 * omp-low.c: Likewise.
31776 * toplev.c: Likewise.
31778 2017-01-20 Marek Polacek <polacek@redhat.com>
31781 * doc/invoke.texi: Document -Wduplicated-branches.
31782 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
31783 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
31784 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
31785 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
31786 return 0 only when not OEP_LEXICOGRAPHIC.
31787 (fold_build_cleanup_point_expr): Use the expression
31788 location when building CLEANUP_POINT_EXPR.
31789 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
31790 * tree.c (add_expr): Handle error_mark_node.
31792 2017-01-20 Martin Liska <mliska@suse.cz>
31795 * tree-profile.c (init_ic_make_global_vars): Do not call
31797 (gimple_init_gcov_profiler): Likewise.
31799 2017-01-20 Martin Liska <mliska@suse.cz>
31802 * cgraph.h (maybe_create_reference): Remove argument and
31804 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
31806 * ipa-cp.c (create_specialized_node): Likewise.
31807 * symtab.c (symtab_node::maybe_create_reference): Handle
31808 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
31810 2017-01-20 Martin Liska <mliska@suse.cz>
31812 * read-rtl-function.c (function_reader::create_function): Use
31813 build_decl instread of build_decl_stat.
31815 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
31817 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
31818 * config/i386/avx512dqintrin.h: Ditto.
31819 * config/i386/avx512fintrin.h: Ditto.
31820 * config/i386/i386-builtin-types.def: Add new types.
31821 * config/i386/i386.c: Handle new types.
31822 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
31823 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
31824 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
31825 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
31826 (__builtin_ia32_kshiftridi): New.
31827 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
31829 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
31833 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
31834 define to rs6000_init_stack_protect_guard.
31835 (rs6000_init_stack_protect_guard): New function.
31837 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
31838 Yunqiang Su <yunqiang.su@imgtec.com>
31840 * config.gcc (supported_defaults): Add madd4.
31841 (with_madd4): Add validation.
31842 (all_defaults): Add madd4.
31843 * config/mips/mips.opt (mmadd4): New option.
31844 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
31846 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
31848 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
31849 (ISA_HAS_FUSED_MADD4): Likewise.
31850 * doc/invoke.texi (-mmadd4): Document the new option.
31851 * doc/install.texi (--with-madd4): Document the new option.
31853 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31855 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
31856 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
31857 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
31858 (aarch64_init_pauth_hint_builtins): New.
31859 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
31860 (aarch64_expand_builtin): Expand new builtins.
31862 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31864 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
31865 * combine-stack-adj.c (no_unhandled_cfa): Handle
31866 REG_CFA_TOGGLE_RA_MANGLE.
31867 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
31868 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
31869 info for return address signing.
31870 (aarch64_expand_epilogue): Likewise.
31872 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31874 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
31875 * config/aarch64/aarch64-protos.h
31876 (aarch64_return_address_signing_enabled): New declaration.
31877 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
31879 (aarch64_expand_prologue): Sign return address before it's pushed onto
31881 (aarch64_expand_epilogue): Authenticate return address fetched from
31883 (aarch64_override_options): Sanity check for ILP32 and ISA level.
31884 (aarch64_attributes): New function attributes for "sign-return-address".
31885 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
31886 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
31887 ("*do_return"): Generate combined instructions according to key index.
31888 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
31889 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
31891 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
31892 * config/aarch64/aarch64.opt (msign-return-address=): New.
31893 * doc/extend.texi (AArch64 Function Attributes): Documents
31894 "sign-return-address=".
31895 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
31897 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
31899 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
31900 overall option summary.
31902 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31904 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
31905 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
31906 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
31907 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
31909 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
31911 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
31912 -mpower9-minmax by default for -mcpu=power9.
31913 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
31914 128-bit floating point.
31916 2017-01-20 Alan Modra <amodra@gmail.com>
31918 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
31919 optimizing for size.
31921 2017-01-20 Alan Modra <amodra@gmail.com>
31924 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
31925 for strcmp and strncmp from corresponding builtin decl.
31927 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
31929 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
31930 instead of i386/rtems-64.h.
31931 * config/i386/rtems-64.h: Remove.
31933 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
31937 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
31939 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
31941 2017-01-19 Tamar Christina <tamar.christina@arm.com>
31943 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
31944 Change int to HOST_WIDE_INT.
31945 * config/aarch64/aarch64-protos.h
31946 (aarch64_simd_gen_const_vector_dup): Likewise.
31947 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
31949 2017-01-19 David Malcolm <dmalcolm@redhat.com>
31951 * langhooks-def.h (lhd_type_for_size): New decl.
31952 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
31953 * langhooks.c (lhd_type_for_size): New function, taken from
31956 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
31958 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
31959 define_bypass for CR latency.
31960 (power9-cracked-alu): Update bypass latency and remove power9-branch.
31961 (power9-alu2): Add define_bypass for CR latency.
31963 (power9-mul): Update insn latency.
31964 (power9-mul-compare): Update insn latency, bypass latency and remove
31967 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31969 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
31971 * config/aarch64/aarch64.md
31972 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
31973 aarch64_nopcrelative_literal_loads.
31974 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
31976 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
31978 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
31979 TARGET_LOONGSON_3A.
31980 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
31982 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
31985 * config.gcc (supported_defaults): Add lxc1-sxc1.
31986 (with_lxc1_sxc1): Add validation.
31987 (all_defaults): Add lxc1-sxc1.
31988 * config/mips/mips.opt (mlxc1-sxc1): New option.
31989 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
31991 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
31992 __mips_no_lxc1_sxc1.
31993 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
31994 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
31995 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
31997 2017-01-19 Richard Biener <rguenther@suse.de>
31999 PR tree-optimization/72488
32000 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
32001 sure to restore SSA info.
32002 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
32004 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
32006 PR rtl-optimization/79121
32007 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
32008 of the inner type when shifting an extended value.
32010 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
32013 * symtab.c (symtab_node::equal_address_to): Fix comparing of
32014 interposable aliases.
32016 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
32019 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
32020 Use the evmergelohi instruction.
32021 (mov_si<mode>_e500_subreg4_2_le): Likewise.
32022 (mov_sitf_e500_subreg8_2_be): Likewise.
32023 (mov_sitf_e500_subreg12_2_le): Likewise.
32024 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
32025 (mov_si<mode>_e500_subreg4_2_be): Likewise.
32026 (mov_sitf_e500_subreg8_2_le): Likewise.
32027 (mov_sitf_e500_subreg12_2_be): Likewise.
32029 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32031 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
32032 attribute from vecsimple to vecperm.
32033 (altivec_vbpermq2): Likewise.
32035 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32038 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
32040 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32041 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
32042 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
32043 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
32044 case where N arg is SIZE_MAX.
32045 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
32046 (cmpstrsi): Add pattern.
32048 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
32050 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
32051 __builtin_vec_revb builtins.
32052 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
32053 built-in functions to support generation of the ISA 3.0 XXBR<x>
32054 vector byte reverse instructions.
32055 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
32056 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
32057 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
32058 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
32059 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
32060 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
32061 (P9V_BUILTIN_VEC_REVB): Likewise.
32062 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
32063 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
32064 (p9_xxbrq_v16qi): Likewise.
32065 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
32066 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
32067 (p9_xxbrh_v8hi): Likewise.
32068 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
32069 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
32070 vec_revb built-in functions.
32072 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
32074 PR rtl-optimization/78952
32075 * config/i386/i386.md (any_extract): New code iterator.
32076 (*insvqi_2): Use any_extract for source operand.
32077 (*insvqi_3): Use any_shiftrt for source operand.
32079 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
32081 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
32083 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
32085 2017-01-18 Matthias Klose <doko@ubuntu.com>
32087 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
32089 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32091 * config/rs6000/altivec.h (vec_bperm): Change #define.
32092 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
32093 (altivec_vbpermq2): New define_insn.
32094 (altivec_vbpermd): Likewise.
32095 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
32096 function interface.
32097 (VBPERMD): Likewise.
32098 (VBPERM): New polymorphic function interface.
32099 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
32100 Add entries for P9V_BUILTIN_VEC_VBPERM.
32101 * doc/extend.texi: Add interfaces for vec_bperm.
32103 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32105 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
32106 first letter of error messages.
32107 (s390_resolve_overloaded_builtin): Likewise.
32108 * config/s390/s390.c (s390_expand_builtin): Likewise.
32109 (s390_invalid_arg_for_unprototyped_fn): Likewise.
32110 (s390_valid_target_attribute_inner_p): Likewise.
32111 * config/s390/s390.md ("tabort"): Likewise.
32113 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
32115 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
32116 (ISA_AVOID_DIV_HILO): New macro.
32117 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
32118 (ISA_HAS_DDIV): Likewise.
32120 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32122 * doc/invoke.texi (fabi-version): Correct number of occurrences.
32124 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32126 * doc/invoke.texi (fabi-version): Spelling fix.
32128 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32131 * doc/invoke.texi (fabi-version): Mention mangling fix for
32134 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32137 * doc/invoke.texi (fabi-version): Document discriminator mangling.
32139 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
32142 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
32143 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
32145 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
32147 (stack_protect_test): Ditto.
32148 * config/rs6000/rs6000.opt (mstack-protector-guard=,
32149 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
32151 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
32152 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
32153 -mstack-protector-guard-offset=.
32154 (RS/6000 and PowerPC Options): Ditto.
32156 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32158 * config/i386/i386.h (MASK_CLASS_P): New define.
32159 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
32160 there are no registers from different register sets also when
32161 mask registers are used. Update function comment.
32162 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
32163 to (*k/*r) and (*k/*km) alternatives.
32165 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
32167 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
32168 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
32169 (EH_RETURN_HANDLER_RTX): New define.
32170 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
32171 Force frame pointer in EH return functions.
32172 (aarch64_expand_epilogue): Add barrier for eh_return.
32173 (aarch64_final_eh_return_addr): Remove.
32174 (aarch64_eh_return_handler_rtx): New function.
32175 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
32177 (aarch64_eh_return_handler_rtx): New prototype.
32179 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32181 * config/rs6000/altivec.h (vec_rlmi): New #define.
32182 (vec_vrlnm): Likewise.
32183 (vec_rlnm): Likewise.
32184 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
32185 (UNSPEC_VRLNM): Likewise.
32186 (VIlong): New mode iterator.
32187 (altivec_vrl<VI_char>mi): New define_insn.
32188 (altivec_vrl<VI_char>nm): Likewise.
32189 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
32191 (VRLDNM): Likewise.
32192 (RLNM): New polymorphic function entry.
32193 (VRLWMI): New monomorphic function entry.
32194 (VRLDMI): Likewise.
32195 (RLMI): New polymorphic function entry.
32196 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
32197 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
32198 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
32201 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32204 * dwarf2out.c (field_byte_offset): Restore the
32205 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
32206 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
32207 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
32210 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32213 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
32215 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32218 * dwarf2out.c (add_data_member_location_attribute): For constant
32219 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
32220 instead of DW_AT_data_member_location, DW_AT_bit_offset and
32221 DW_AT_byte_size attributes.
32223 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32225 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
32226 after forcing to constant memory when the code model is medium.
32228 2017-01-17 Julia Koval <julia.koval@intel.com>
32231 * config/i386/avx512fintrin.h
32232 (_mm512_i32gather_ps): Change __addr type to void const*.
32233 (_mm512_mask_i32gather_ps): Ditto.
32234 (_mm512_i32gather_pd): Ditto.
32235 (_mm512_mask_i32gather_pd): Ditto.
32236 (_mm512_i64gather_ps): Ditto.
32237 (_mm512_mask_i64gather_ps): Ditto.
32238 (_mm512_i64gather_pd): Ditto.
32239 (_mm512_mask_i64gather_pd): Ditto.
32240 (_mm512_i32gather_epi32): Ditto.
32241 (_mm512_mask_i32gather_epi32): Ditto.
32242 (_mm512_i32gather_epi64): Ditto.
32243 (_mm512_mask_i32gather_epi64): Ditto.
32244 (_mm512_i64gather_epi32): Ditto.
32245 (_mm512_mask_i64gather_epi32): Ditto.
32246 (_mm512_i64gather_epi64): Ditto.
32247 (_mm512_mask_i64gather_epi64): Ditto.
32248 (_mm512_i32scatter_ps): Change __addr type to void*.
32249 (_mm512_mask_i32scatter_ps): Ditto.
32250 (_mm512_i32scatter_pd): Ditto.
32251 (_mm512_mask_i32scatter_pd): Ditto.
32252 (_mm512_i64scatter_ps): Ditto.
32253 (_mm512_mask_i64scatter_ps): Ditto.
32254 (_mm512_i64scatter_pd): Ditto.
32255 (_mm512_mask_i64scatter_pd): Ditto.
32256 (_mm512_i32scatter_epi32): Ditto.
32257 (_mm512_mask_i32scatter_epi32): Ditto.
32258 (_mm512_i32scatter_epi64): Ditto.
32259 (_mm512_mask_i32scatter_epi64): Ditto.
32260 (_mm512_i64scatter_epi32): Ditto.
32261 (_mm512_mask_i64scatter_epi32): Ditto.
32262 (_mm512_i64scatter_epi64): Ditto.
32263 (_mm512_mask_i64scatter_epi64): Ditto.
32264 * config/i386/avx512pfintrin.h
32265 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
32266 (_mm512_mask_prefetch_i32gather_ps): Ditto.
32267 (_mm512_mask_prefetch_i64gather_pd): Ditto.
32268 (_mm512_mask_prefetch_i64gather_ps): Ditto.
32269 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
32270 (_mm512_prefetch_i32scatter_ps): Ditto.
32271 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
32272 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
32273 (_mm512_prefetch_i64scatter_pd): Ditto.
32274 (_mm512_prefetch_i64scatter_ps): Ditto.
32275 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
32276 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
32277 * config/i386/avx512vlintrin.h
32278 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
32279 (_mm_mmask_i32gather_ps): Ditto.
32280 (_mm256_mmask_i32gather_pd): Ditto.
32281 (_mm_mmask_i32gather_pd): Ditto.
32282 (_mm256_mmask_i64gather_ps): Ditto.
32283 (_mm_mmask_i64gather_ps): Ditto.
32284 (_mm256_mmask_i64gather_pd): Ditto.
32285 (_mm_mmask_i64gather_pd): Ditto.
32286 (_mm256_mmask_i32gather_epi32): Ditto.
32287 (_mm_mmask_i32gather_epi32): Ditto.
32288 (_mm256_mmask_i32gather_epi64): Ditto.
32289 (_mm_mmask_i32gather_epi64): Ditto.
32290 (_mm256_mmask_i64gather_epi32): Ditto.
32291 (_mm_mmask_i64gather_epi32): Ditto.
32292 (_mm256_mmask_i64gather_epi64): Ditto.
32293 (_mm_mmask_i64gather_epi64): Ditto.
32294 (_mm256_i32scatter_ps): Change __addr type to void*.
32295 (_mm256_mask_i32scatter_ps): Ditto.
32296 (_mm_i32scatter_ps): Ditto.
32297 (_mm_mask_i32scatter_ps): Ditto.
32298 (_mm256_i32scatter_pd): Ditto.
32299 (_mm256_mask_i32scatter_pd): Ditto.
32300 (_mm_i32scatter_pd): Ditto.
32301 (_mm_mask_i32scatter_pd): Ditto.
32302 (_mm256_i64scatter_ps): Ditto.
32303 (_mm256_mask_i64scatter_ps): Ditto.
32304 (_mm_i64scatter_ps): Ditto.
32305 (_mm_mask_i64scatter_ps): Ditto.
32306 (_mm256_i64scatter_pd): Ditto.
32307 (_mm256_mask_i64scatter_pd): Ditto.
32308 (_mm_i64scatter_pd): Ditto.
32309 (_mm_mask_i64scatter_pd): Ditto.
32310 (_mm256_i32scatter_epi32): Ditto.
32311 (_mm256_mask_i32scatter_epi32): Ditto.
32312 (_mm_i32scatter_epi32): Ditto.
32313 (_mm_mask_i32scatter_epi32): Ditto.
32314 (_mm256_i32scatter_epi64): Ditto.
32315 (_mm256_mask_i32scatter_epi64): Ditto.
32316 (_mm_i32scatter_epi64): Ditto.
32317 (_mm_mask_i32scatter_epi64): Ditto.
32318 (_mm256_i64scatter_epi32): Ditto.
32319 (_mm256_mask_i64scatter_epi32): Ditto.
32320 (_mm_i64scatter_epi32): Ditto.
32321 (_mm_mask_i64scatter_epi32): Ditto.
32322 (_mm256_i64scatter_epi64): Ditto.
32323 (_mm256_mask_i64scatter_epi64): Ditto.
32324 (_mm_i64scatter_epi64): Ditto.
32325 (_mm_mask_i64scatter_epi64): Ditto.
32326 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
32327 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
32328 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
32329 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
32330 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
32331 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
32332 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
32333 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
32334 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
32335 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
32336 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
32337 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
32338 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
32339 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
32340 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
32341 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
32342 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
32343 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
32344 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
32345 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
32346 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
32347 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
32348 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
32349 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
32350 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
32351 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
32352 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
32353 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
32354 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
32355 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
32356 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
32357 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
32358 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
32359 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
32360 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
32361 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
32362 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
32363 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
32364 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
32365 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
32366 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
32367 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
32368 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
32369 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
32370 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
32371 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
32372 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
32373 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
32374 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
32375 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
32376 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
32377 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
32378 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
32379 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
32380 definitions accordingly.
32382 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
32383 Kuan-Lin Chen <kuanlinchentw@gmail.com>
32386 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
32389 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
32392 * ira-conflicts.c (ira_build_conflicts): Update total conflict
32393 hard regs for inner regno.
32395 2017-01-17 Martin Liska <mliska@suse.cz>
32398 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
32399 assumption and add comment.
32401 2017-01-17 Nathan Sidwell <nathan@acm.org>
32403 * ipa-visibility.c (localize_node): New function, broken out of ...
32404 (function_and_variable_visibility): ... here. Call it.
32406 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
32408 PR middle-end/77445
32409 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
32410 correctly set frequency of oudgoing edge.
32411 (duplicate_thread_path): Fix profile updating.
32413 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32416 * configure.ac: Add GCC_BASE_VER.
32417 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
32418 version from BASE-VER file.
32419 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
32420 (gcc.o): Depend on $(BASEVER).
32421 * common.opt (dumpfullversion): New option.
32422 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
32423 * doc/invoke.texi: Document -dumpfullversion.
32424 * doc/install.texi: Document --with-gcc-major-version-only.
32425 * configure: Regenerated.
32427 2017-01-17 Richard Biener <rguenther@suse.de>
32429 PR tree-optimization/71433
32430 * tree-vrp.c (register_new_assert_for): Merge same asserts
32431 on all incoming edges.
32432 (process_assert_insertions_for): Handle insertions at the
32435 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
32437 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
32438 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
32440 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
32443 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
32446 2017-01-17 Alan Modra <amodra@gmail.com>
32449 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
32450 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
32451 symbolic stack limit when pic.
32453 2017-01-16 Martin Sebor <msebor@redhat.com>
32455 PR tree-optimization/78608
32456 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
32458 2017-01-16 Jeff Law <law@redhat.com>
32461 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
32462 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
32463 for several include directories that may be relative to sysroot.
32464 * config/i386/x-mingw32 (gplus_includedir): Define.
32465 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
32466 (native_system_includedir): Likewise.
32467 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
32468 override if TARGET_SYSTEM_ROOT is defined.
32469 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
32471 PR tree-optimization/79090
32472 PR tree-optimization/33562
32473 PR tree-optimization/61912
32474 PR tree-optimization/77485
32475 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
32476 and computed trims into the dump file.
32478 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32480 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
32482 2017-01-16 Jakub Jelinek <jakub@redhat.com>
32485 * gimplify.c (gimplify_init_constructor): If want_value and
32486 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
32490 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
32491 sequence. Formatting fixes.
32492 (doloop_optimize): Formatting fixes.
32495 * gcc.c (debug_level_greater_than_spec_func): New function.
32496 (static_spec_functions): Add debug-level-gt spec function.
32497 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
32499 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32500 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32501 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
32502 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
32503 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
32504 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
32506 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
32508 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
32509 QImode fixups to general and mask registers only.
32511 2017-01-16 Carl Love <cel@us.ibm.com>
32513 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
32514 for built-in functions
32515 vector signed char vec_nabs (vector signed char)
32516 vector signed short vec_nabs (vector signed short)
32517 vector signed int vec_nabs (vector signed int)
32518 vector signed long long vec_nabs (vector signed long long)
32519 vector float vec_nabs (vector float)
32520 vector double vec_nabs (vector double)
32521 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
32523 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
32524 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
32525 * doc/extend.texi: Update the documentation file for the new built-in
32528 2017-01-16 Martin Sebor <msebor@redhat.com>
32530 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
32533 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32535 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
32536 UNSPEC_VSX__XXSPLTD to require special splat handling.
32538 2017-01-16 David Malcolm <dmalcolm@redhat.com>
32541 * system.h: Poison strndup.
32543 2017-01-16 Alan Modra <amodra@gmail.com>
32546 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
32549 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
32551 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
32553 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
32555 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
32556 call recog here. Assert that INSN_CODE (insn) is non-negative.
32558 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32561 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
32563 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
32564 in the currently scheduled RTL fragment.
32566 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32568 PR rtl-optimization/78751
32569 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
32572 2017-01-14 Jeff Law <law@redhat.com>
32574 PR tree-optimization/79090
32575 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
32576 variable length stores.
32577 (compute_trims): Delete dead assignment to *trim_tail.
32578 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
32581 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
32583 PR rtl-optimization/78626
32584 PR rtl-optimization/78727
32585 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
32586 of a block, and split such blocks after everything else is finished.
32588 2017-01-14 Alan Modra <amodra@gmail.com>
32591 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
32592 target legitimate_combined_insn.
32593 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
32594 (rs6000_legitimate_combined_insn): New function.
32595 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
32597 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
32598 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
32599 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
32601 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
32603 * doc/frontends.texi (G++ and GCC): Remove references to Java.
32605 2017-01-13 Jeff Law <law@redhat.com>
32607 PR tree-optimization/33562
32608 PR tree-optimization/61912
32609 PR tree-optimization/77485
32610 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
32612 (delete_dead_assignment): Likewise.
32613 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
32614 statement to delete_dead_call and delete_dead_assignment.
32616 2017-01-13 David Malcolm <dmalcolm@redhat.com>
32619 * substring-locations.c (format_warning_va): Strengthen case 1 so
32620 that both endpoints of the substring must be within the format
32621 range for just the substring to be printed.
32623 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
32625 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
32626 * config/i386/i386.c (ix86_target_string): Add missing options
32627 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
32628 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
32629 flags_other and ix86_target_other to flags2_other. Display unknown
32631 (ix86_valid_target_attribute_inner_p): Add missing options and
32632 reorder options by implied ISAs, as in ix86_target_string.
32634 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
32636 * hash-table.h (hash_table::too_empty_p): New function.
32637 (hash_table::expand): Use it.
32638 (hash_table::traverse): Likewise.
32639 (hash_table::empty_slot): Use sizeof (value_type) instead of
32640 sizeof (PTR) to convert bytes to elements. Shrink the table
32641 if the current size is excessive for the current number of
32644 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
32646 * ira-costs.c (record_reg_classes): Break from the inner loop
32647 early once alt_fail is known to be true. Update outer loop
32648 handling accordingly.
32650 2017-01-13 Jeff Law <law@redhat.com>
32652 * tree-ssa-dse.c (decrement_count): New function.
32653 (increment_start_addr, maybe_trim_memstar_call): Likewise.
32654 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
32655 when we know the partially dead statement is a mem* function.
32657 PR tree-optimization/61912
32658 PR tree-optimization/77485
32659 * tree-ssa-dse.c: Include expr.h.
32660 (maybe_trim_constructor_store): New function.
32661 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
32663 PR tree-optimization/33562
32664 PR tree-optimization/61912
32665 PR tree-optimization/77485
32666 * doc/invoke.texi: Document new dse-max-object-size param.
32667 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
32668 * tree-ssa-dse.c: Include params.h.
32669 (dse_store_status): New enum.
32670 (initialize_ao_ref_for_dse): New, partially extracted from
32672 (valid_ao_ref_for_dse, normalize_ref): New.
32673 (setup_live_bytes_from_ref, compute_trims): Likewise.
32674 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
32675 (maybe_trim_partially_dead_store): Likewise.
32676 (maybe_trim_complex_store): Likewise.
32677 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
32678 Track what bytes live from the original store. Return tri-state
32679 for dead, partially dead or live.
32680 (dse_dom_walker): Add constructor, destructor and new private members.
32681 (delete_dead_call, delete_dead_assignment): New extracted from
32683 (dse_optimize_stmt): Make a member of dse_dom_walker.
32684 Use initialize_ao_ref_for_dse.
32686 PR tree-optimization/33562
32687 PR tree-optimization/61912
32688 PR tree-optimization/77485
32689 * sbitmap.h (bitmap_count_bits): Prototype.
32690 (bitmap_clear_range, bitmap_set_range): Likewise.
32691 * sbitmap.c (bitmap_clear_range): New function.
32692 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
32694 2017-01-13 Martin Liska <mliska@suse.cz>
32697 * function.c (set_cfun): Add new argument force.
32698 * function.h (set_cfun): Likewise.
32699 * ipa-inline-transform.c (inline_call): Use the function when
32700 strict alising from is dropped for function we inline to.
32702 2017-01-13 Richard Biener <rguenther@suse.de>
32704 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
32705 for dumping GIMPLE INTEGER_CSTs.
32707 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
32709 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
32710 to 201112L since C++17.
32712 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
32715 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
32716 if -fsanitize=kernel-address is present.
32718 2017-01-13 Richard Biener <rguenther@suse.de>
32720 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
32721 as _Literal ( type ) number in case usual suffixes do not
32722 preserve all information.
32724 2017-01-13 Richard Biener <rguenther@suse.de>
32726 PR tree-optimization/77283
32727 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
32728 and ssa-iterators.h.
32729 (is_feasible_trace): Implement a cost model based on joiner
32732 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
32735 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
32736 char or short to __float128/_Float128 directly.
32738 2017-01-12 Martin Sebor <msebor@redhat.com>
32740 to -Wformat-overflow.
32741 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
32742 (min_bytes_remaining): Same.
32743 (get_string_length): Same.
32744 (format_string): Same.
32745 (format_directive): Same.
32747 (pass_sprintf_length::handle_gimple_call): Same.
32749 2017-01-12 Jakub Jelinek <jakub@redhat.com>
32751 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
32752 info.nowrite calls with no lhs that can't throw. Return bool
32753 whether gsi_remove has been called or not.
32754 (pass_sprintf_length::handle_gimple_call): Return bool whether
32755 try_substitute_return_value called gsi_remove. Formatting fix.
32756 (pass_sprintf_length::execute): Don't use gsi_remove if
32757 handle_gimple_call returned true.
32760 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
32761 be removed due to side-effects, don't remove following barrier nor
32762 turn the successor edge into fallthru edge.
32764 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32767 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
32768 element-reversing loads and stores as not swappable.
32770 2017-01-12 Nathan Sidwell <nathan@acm.org>
32771 Nicolai Stange <nicstange@gmail.com>
32773 * combine.c (try_combine): Don't ignore result of overlap checking
32774 loop. Combine overlap & asm check into single loop.
32776 2017-01-12 Richard Biener <rguenther@suse.de>
32778 * tree-pretty-print.c (dump_generic_node): Provide -gimple
32779 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
32781 2017-01-12 Richard Biener <rguenther@suse.de>
32783 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
32784 and TS_TARGET_OPTION directly derive from TS_BASE.
32785 * tree-core.h (tree_optimization_option): Derive from tree_base.
32786 (tree_target_option): Likewise.
32788 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
32790 * config/i386/i386.c (memory_address_length): Increase len
32791 only when rip_relative_addr_p returns false.
32793 2017-01-11 Julia Koval <julia.koval@intel.com>
32795 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
32796 (OPTION_MASK_ISA_SGX_SET): New.
32797 (ix86_handle_option): Handle OPT_msgx.
32798 * config.gcc: Added sgxintrin.h.
32799 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
32800 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
32801 * config/i386/i386.c (ix86_target_string): Add -msgx.
32803 (ix86_option_override_internal): Handle new options.
32804 (ix86_valid_target_attribute_inner_p): Add sgx.
32805 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
32806 * config/i386/i386.opt: Add msgx.
32807 * config/i386/sgxintrin.h: New file.
32808 * config/i386/x86intrin.h: Add sgxintrin.h.
32810 2017-01-11 Jakub Jelinek <jakub@redhat.com>
32813 * fold-const.c (maybe_nonzero_address): Return 1 for function
32815 (tree_single_nonzero_warnv_p): Don't handle function local objects
32819 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
32822 2017-01-11 David Malcolm <dmalcolm@redhat.com>
32825 * opts.c: Include "spellcheck.h"
32826 (struct string_fragment): New struct.
32827 (struct edit_distance_traits<const string_fragment &>): New
32829 (get_closest_sanitizer_option): New function.
32830 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
32832 2017-01-11 Jakub Jelinek <jakub@redhat.com>
32834 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
32836 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
32837 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
32838 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
32839 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
32840 for initial die_offset if dwarf_split_debug_info.
32841 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
32842 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
32844 (output_skeleton_debug_sections): Formatting fix. Use
32845 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
32846 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
32848 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
32850 * config/arm/cortex-a53.md: Add bypasses for
32851 cortex_a53_r2f_cvt.
32852 (cortex_a53_r2f): Only use for transfers.
32853 (cortex_a53_f2r): Likewise.
32854 (cortex_a53_r2f_cvt): Add reservation for conversions.
32855 (cortex_a53_f2r_cvt): Likewise.
32857 2017-01-11 Tamar Christina <tamar.christina@arm.com>
32859 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
32860 to all inlined functions, change static to extern.
32862 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
32865 * config/arm/arm.c (legitimize_pic_address): Handle reference to
32867 (arm_assemble_integer): Likewise.
32869 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
32871 * config.gcc: Use new awk script to check CPU, FPU and architecture
32872 parameters for --with-... options.
32873 * config/arm/parsecpu.awk: New file
32874 * config/arm/arm-cpus.in: New file.
32875 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
32877 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
32879 * config/arm/t-arm: Update dependency rules.
32880 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
32881 of processing .def files.
32882 * config/arm/genopt.sh: Deleted.
32883 * config/arm/gentune.sh: Deleted.
32884 * config/arm/arm-cores.def: Deleted.
32885 * config/arm/arm-arches.def: Deleted.
32886 * config/arm/arm-fpus.def: Deleted.
32887 * config/arm/arm-tune.md: Regenerated.
32888 * config/arm/arm-tables.opt: Regenerated.
32889 * config/arm/arm-cpu.h: New generated file.
32890 * config/arm/arm-cpu-data.h: New generated file.
32891 * config/arm/arm-cpu-cdata.h: New generated file.
32893 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
32896 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
32898 (input_varpool_node): Unpack dynamically_initialized bit.
32900 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
32902 PR rtl-optimization/79032
32903 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
32904 the alignment of the adjusted memory reference against that of MODE,
32905 instead of the alignment of the original memory reference.
32907 2017-01-11 Martin Jambor <mjambor@suse.cz>
32909 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
32911 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
32912 decorated functions.
32914 2017-01-11 Richard Biener <rguenther@suse.de>
32916 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
32917 set range/nonnull info for PHI results. Do not set it on
32918 stmts marked for removal.
32920 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
32922 * expr.c (store_field): In the bitfield case, fetch the return value
32923 from the registers before applying a single big-endian adjustment.
32924 Always do a final load for a BLKmode value not larger than a word.
32926 2017-01-10 David Malcolm <dmalcolm@redhat.com>
32929 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
32930 that we correctly handle column numbers greater than
32931 LINE_MAP_MAX_COLUMN_NUMBER.
32933 2017-01-10 Martin Sebor <msebor@redhat.com>
32935 PR middle-end/78245
32936 * gimple-ssa-sprintf.c (get_destination_size): Call
32937 {init,fini}object_sizes.
32938 * tree-object-size.c (addr_object_size): Adjust.
32939 (pass_through_call): Adjust.
32940 (pass_object_sizes::execute): Adjust.
32941 * tree-object-size.h (fini_object_sizes): Declare.
32943 2017-01-10 Martin Sebor <msebor@redhat.com>
32945 PR tree-optimization/78775
32946 * builtins.c (get_size_range): Move...
32947 * calls.c: ...to here.
32948 (alloc_max_size): Accept zero argument.
32949 (operand_signed_p): Remove.
32950 (maybe_warn_alloc_args_overflow): Call get_size_range.
32951 * calls.h (get_size_range): Declare.
32953 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
32955 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
32956 from TI's devices.csv file as of September 2016.
32957 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
32959 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
32961 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
32962 * doc/invoke.texi: Likewise.
32963 * doc/md.texi: Likewise.
32964 * doc/objc.texi: Likewise.
32966 2017-01-10 Joshua Conner <joshconner@google.com>
32968 * config/arm/fuchsia-elf.h: New file.
32969 * config/fuchsia.h: New file.
32970 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
32971 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
32973 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
32975 2016-01-10 Richard Biener <rguenther@suse.de>
32977 PR tree-optimization/79034
32978 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
32979 Propagate out degenerate PHIs in the joiner.
32981 2017-01-10 Martin Liska <mliska@suse.cz>
32983 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
32984 (sort_congruence_classes_by_decl_uid): Likewise.
32985 (sort_congruence_class_groups_by_decl_uid): Likewise.
32986 (sem_item_optimizer::merge_classes): Sort class, groups in these
32987 classes and members in the groups by DECL_UID of declarations.
32988 This would make merge operations stable.
32990 2017-01-10 Martin Liska <mliska@suse.cz>
32992 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
32993 usage of m_classes_vec.
32994 (sem_item_optimizer::~sem_item_optimizer): Likewise.
32995 (sem_item_optimizer::get_group_by_hash): Likewise.
32996 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
32997 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
32998 (sem_item_optimizer::verify_classes): Likewise.
32999 (sem_item_optimizer::process_cong_reduction): Likewise.
33000 (sem_item_optimizer::dump_cong_classes): Likewise.
33001 (sem_item_optimizer::merge_classes): Likewise.
33002 * ipa-icf.h (congruence_class_hash): Rename from
33003 congruence_class_group_hash. Remove declaration of m_classes_vec.
33005 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
33007 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
33008 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
33009 * config.gcc: Add avx512vpopcntdqintrin.h.
33010 * config/i386/avx512vpopcntdqintrin.h: New.
33011 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
33012 * config/i386/i386-builtin-types.def: Add new types.
33013 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
33014 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
33015 __builtin_ia32_vpopcountq_v8di_mask): New.
33016 * config/i386/i386-c.c (ix86_target_macros_internal): Define
33017 __AVX512VPOPCNTDQ__.
33018 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
33019 (PTA_AVX512VPOPCNTDQ): Define.
33020 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
33021 TARGET_AVX512VPOPCNTDQ_P): Define.
33022 * config/i386/i386.opt: Add mavx512vpopcntdq.
33023 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
33024 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
33026 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33028 PR middle-end/77484
33029 * predict.def (PRED_CALL): Set to 67.
33031 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
33033 * expr.c (store_field): In the bitfield case, if the value comes from
33034 a function call and is of an aggregate type returned in registers, do
33035 not modify the field mode; extract the value in all cases if the mode
33036 is BLKmode and the size is not larger than a word.
33038 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
33041 * config/i386/cpuid.h: Fix undefined behavior.
33043 2017-01-04 Jeff Law <law@redhat.com>
33045 PR tree-optimization/79007
33046 PR tree-optimization/67955
33047 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
33048 conservative for pt.null when flag_non_call_exceptions is on.
33050 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33052 PR translation/79019
33053 PR translation/79020
33054 * params.def (PARAM_INLINE_MIN_SPEEDUP,
33055 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
33056 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
33058 * config/avr/avr.opt (maccumulate-args): Likewise.
33059 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
33060 * common.opt (freport-bug): Likewise.
33061 * cif-code.def (CIF_FINAL_ERROR): Likewise.
33062 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
33063 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
33064 translatable string.
33065 * config/i386/i386.c (function_value_32): Likewise.
33066 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
33067 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
33069 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
33070 * common/config/msp430/msp430-common.c (msp430_handle_option):
33072 * symtab.c (symtab_node::verify_base): Likewise.
33073 * opts.c (set_debug_level): Likewise.
33074 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
33075 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
33076 missing whitespace to translatable strings.
33077 * config/avr/avr.md (bswapsi2): Fix typo in comment.
33078 * config/sh/superh.h: Likewise.
33079 * config/i386/xopintrin.h: Likewise.
33080 * config/i386/znver1.md: Likewise.
33081 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
33082 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
33083 * double-int.h (struct double_int): Likewise.
33084 * double-int.c (div_and_round_double): Likewise.
33085 * wide-int.cc: Likewise.
33086 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
33087 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
33088 * cfgcleanup.c (crossjumps_occured): Renamed to ...
33089 (crossjumps_occurred): ... this.
33090 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
33093 PR tree-optimization/78899
33094 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
33095 returning bool return struct loop *, NULL for failure and the new
33097 (versionable_outer_loop_p): Don't version outer loop if it has
33098 dont_vectorized bit set.
33099 (tree_if_conversion): When versioning outer loop, ensure
33100 tree_if_conversion is performed also on the inner loop of the
33101 non-vectorizable outer loop copy.
33102 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
33103 LOOP_VECTORIZED in inner loop of the scalar outer loop and
33104 prevent vectorization of it.
33105 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
33106 the outer loop vectorization of the non-scalar version is attempted
33107 before vectorization of the inner loop in scalar version. If
33108 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
33109 vectorization of its inner loop.
33110 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
33111 has 2 inner loops, rename also on edges from bb whose single pred
33112 is outer_loop->header. Fix typo in function comment.
33114 2017-01-09 Martin Sebor <msebor@redhat.com>
33117 * asan.c (asan_emit_stack_protection): Increase local buffer size
33118 to avoid snprintf truncation warning.
33120 2017-01-09 Andrew Pinski <apinski@cavium.com>
33122 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
33123 to reference thunderx2t99 for the tuning structure
33124 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
33126 (thunderx2t99_extra_costs): This.
33127 * config/aarch64/aarch64-tune.md: Regenerate.
33128 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
33129 (vulcan_addrcost_table): This.
33130 (vulcan_regmove_cost): Rename to ...
33131 (thunderx2t99_regmove_cost): This.
33132 (vulcan_vector_cost): Rename to ...
33133 (thunderx2t99_vector_cost): this.
33134 (vulcan_branch_cost): Rename to ...
33135 (thunderx2t99_branch_cost): This.
33136 (vulcan_tunings): Rename to ...
33137 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
33138 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
33140 2017-01-09 Martin Jambor <mjambor@suse.cz>
33144 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
33145 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
33146 (propagate_vr_accross_jump_function): Use the above function for all
33147 value range computations for pass-through jump functions and type
33148 converasion from explicit value range values.
33149 (ipcp_propagate_stage): Do not attempt to deduce types of formal
33150 parameters from TYPE_ARG_TYPES.
33151 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
33152 (ipa_write_node_info): Stream type of the actual argument.
33153 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
33155 2017-01-09 Martin Liska <mliska@suse.cz>
33158 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
33159 (lookup_compiler): Do not show error message with have_E.
33161 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33163 PR tree-optimization/78938
33164 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
33165 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
33166 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
33167 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
33170 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33172 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
33175 2017-01-09 Richard Biener <rguenther@suse.de>
33177 PR tree-optimization/78997
33178 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
33179 name condition properly.
33181 2017-01-09 Richard Biener <rguenther@suse.de>
33184 * dwarf2out.c (is_cxx): New overload with context.
33185 (is_naming_typedef_decl): Use it.
33187 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33189 * invoke.texi (Option Summary): Correct spacing in option lists
33190 and add line breaks to fix over-long lines.
33192 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33194 PR middle-end/17660
33196 * extend.texi (Common Variable Attributes): Add xref to GCC
33197 Internals manual to explain mode attribute keywords.
33199 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33202 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
33203 and Preprocessor Options.
33204 (Options for Linking): Document -pthread here....
33205 (RS/6000 and PowerPC Options): ...not here.
33206 (Solaris 2 Options): ...or here.
33207 * doc/cppopts.texi: Document -pthread.
33209 2017-01-08 Martin Sebor <msebor@redhat.com>
33211 PR middle-end/77708
33212 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
33213 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
33214 New member functions.
33215 (format_directive): Used them.
33217 (pass_sprintf_length::handle_gimple_call): Same.
33218 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
33219 to avoid truncation for any argument.
33220 (extract_affine_mul): Same.
33221 * tree.c (get_file_function_name): Same.
33223 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33225 PR middle-end/77484
33226 * predict.def (PRED_INDIR_CALL): Set to 86.
33228 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33230 PR preprocessor/54124
33231 * doc/cppopts.texi: Reformat -d subtable to list the full name
33232 of the options. Add cross-reference to the docs for the general
33233 compiler -d options.
33234 * doc/invoke.texi (Developer Options): Add cross-reference to the
33235 preprocessor-specific -d option documentation.
33237 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33239 PR preprocessor/13498
33240 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
33241 redudant material, and reflect new command-line options.
33242 (System Headers): Likewise.
33244 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33246 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
33247 -isystem, and -idirafter. Copy-edit.
33248 * doc/cppopts.texi: Copy-edit. Remove contradiction about
33249 default for -ftrack-macro-expansion. Delete obsolete and
33250 badly-formatted implementation details about -fdebug-cpp output.
33251 * doc/cppwarnopts.texi: Copy-edit.
33253 2017-01-07 David Malcolm <dmalcolm@redhat.com>
33256 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
33257 that the transition from a max line width >= 1<<10 to narrower
33258 lines works correctly.
33260 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
33262 * doc/options.texi (PerFunction): New.
33263 * opt-functions.awk (switch_flags): Map both Optimization and
33264 PerFunction to CL_OPTIMIZATION.
33265 * opth-gen.awk: Test for PerFunction flag along with
33267 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
33268 it only when the latter is present. Skip those that don't in
33269 the hash function generator.
33270 * common.opt (fvar-tracking): Mark as PerFunction instead of
33272 (fvar-tracking-assignments): Likewise.
33273 (fvar-tracking-assignments-toggle): Likewise.
33274 (fvar-tracking-uninit): Likewise.
33276 2017-01-07 Jakub Jelinek <jakub@redhat.com>
33278 PR translation/79018
33279 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
33282 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
33285 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
33286 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
33287 TARGET_LONG_JUMP_TABLE_OFFSETS.
33288 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
33289 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
33290 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
33291 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
33292 * config/m68k/m68k.md (tablejump expander): Likewise.
33293 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
33294 TARGET_LONG_JUMP_TABLE_OFFSETS.
33295 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
33296 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
33298 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
33299 David Holsgrove <david.holsgrove@xilinx.com>
33301 * common/config/microblaze/microblaze-common.c
33302 (TARGET_EXCEPT_UNWIND_INFO): Remove.
33303 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
33305 * config/microblaze/microblaze.c (microblaze_must_save_register)
33306 (microblaze_expand_epilogue, microblaze_return_addr): Handle
33308 (microblaze_eh_return): New function.
33309 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
33310 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
33311 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
33312 * config/microblaze/microblaze.md (eh_return): New pattern.
33314 2017-01-06 Jakub Jelinek <jakub@redhat.com>
33316 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
33317 GCC_DIAGNOSTIC_STRINGIFY): Define.
33319 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
33321 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33323 * config/arm/arm.md (<mcrr>): New.
33325 * config/arm/arm.c (arm_arch5te): New.
33326 (arm_option_override): Set arm_arch5te.
33327 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
33329 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
33330 (arm_mcrr_qualifiers): ... this. New.
33331 (MRRC_QUALIFIERS): Define to...
33332 (arm_mrrc_qualifiers): ... this. New.
33333 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
33335 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
33336 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
33337 (MRRCI, mrrc, MRRC): New.
33338 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
33339 VUNSPEC_MRRC2): New.
33341 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33343 * config/arm/arm.md (<mcr>): New.
33345 * config/arm/arm.c (arm_coproc_builtin_available): Add
33346 support for mcr, mrc, mcr2 and mrc2.
33347 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
33348 (arm_mcr_qualifiers): ... this. New.
33349 (MRC_QUALIFIERS): Define to ...
33350 (arm_mrc_qualifiers): ... this. New.
33351 (MCR_QUALIFIERS): Define to ...
33352 (arm_mcr_qualifiers): ... this. New.
33353 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
33355 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
33356 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
33357 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
33358 VUNSPEC_MRC2): New.
33360 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33362 * config/arm/arm.md (*ldc): New.
33366 * config/arm/arm.c (arm_coproc_builtin_available): Add
33367 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
33368 (arm_coproc_ldc_stc_legitimate_address): New.
33369 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
33370 'qualifier_const_pointer'.
33371 (LDC_QUALIFIERS): Define to...
33372 (arm_ldc_qualifiers): ... this. New.
33373 (STC_QUALIFIERS): Define to...
33374 (arm_stc_qualifiers): ... this. New.
33375 * config/arm/arm-protos.h
33376 (arm_coproc_ldc_stc_legitimate_address): New.
33377 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
33378 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
33379 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
33380 stc2, stcl, stc2l): New.
33381 * config/arm/constraints.md (Uz): New.
33382 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
33383 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
33384 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
33385 VUNSPEC_STC2L): New.
33387 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33389 * config/arm/arm.md (<cdp>): New.
33390 * config/arm/arm.c (neon_const_bounds): Rename this ...
33391 (arm_const_bounds): ... this.
33392 (arm_coproc_builtin_available): New.
33393 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
33394 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
33395 (CDP_QUALIFIERS): Define to...
33396 (arm_cdp_qualifiers): ... this. New.
33398 (arm_expand_builtin_args): Add case for 6 arguments.
33399 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
33400 (arm_const_bounds): ... this.
33401 (arm_coproc_builtin_available): New.
33402 * config/arm/arm_acle.h (__arm_cdp): New.
33404 * config/arm/arm_acle_builtins.def (cdp): New.
33406 * config/arm/iterators.md (CDPI,CDP,cdp): New.
33407 * config/arm/neon.md: Rename all 'neon_const_bounds' to
33408 'arm_const_bounds'.
33409 * config/arm/types.md (coproc): New.
33410 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
33411 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
33412 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
33413 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
33415 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33417 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
33418 (UBINOP_QUALIFIERS): New.
33420 (acle_builtin_data): New. Change comment.
33421 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
33422 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
33423 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
33424 arm_acle_builtins.def.
33425 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
33426 (arm_init_acle_builtins): New.
33427 (CRC32_BUILTIN): Remove.
33428 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
33429 crc32cb, crc32ch and crc32cw.
33430 (arm_init_crc32_builtins): Remove.
33431 (arm_init_builtins): Use arm_init_acle_builtins rather
33432 than arm_init_crc32_builtins.
33433 (arm_expand_acle_builtin): New.
33434 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
33435 * config/arm/arm_acle_builtins.def: New.
33437 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33439 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
33440 (arm_builtin_datum): ... this.
33441 (arm_init_neon_builtin): Rename to ...
33442 (arm_init_builtin): ... this. Add a new parameters PREFIX
33443 and USE_SIG_IN_NAME.
33444 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
33445 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
33446 'arm_builtin_datum'.
33447 (arm_init_vfp_builtins): Likewise.
33448 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
33449 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
33450 (arm_expand_neon_args): Rename to ...
33451 (arm_expand_builtin_args): ... this. Rename builtin_arg
33452 enum values and differentiate between ARG_BUILTIN_MEMORY
33453 and ARG_BUILTIN_NEON_MEMORY.
33454 (arm_expand_neon_builtin_1): Rename to ...
33455 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
33456 values, arm_expand_builtin_args and add bool parameter NEON.
33457 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
33458 (arm_expand_vfp_builtin): Likewise.
33459 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
33461 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33463 PR middle-end/77484
33464 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
33465 * predict.c (tree_estimate_probability_bb): Reverse direction of
33466 polymorphic call predictor.
33468 2017-01-06 David Malcolm <dmalcolm@redhat.com>
33470 * passes.c (execute_one_pass): Split out pass-skipping logic into...
33471 (determine_pass_name_match): ...this new function and...
33472 (should_skip_pass_p): ...this new function.
33474 2017-01-06 Nathan Sidwell <nathan@acm.org>
33476 * ipa-visibility.c (function_and_variable_visibility): Reformat
33477 comments and long lines. Remove extrneous if.
33478 * symtab.c (symtab_node::make_decl_local): Fix code format.
33479 (symtab_node::set_section_for_node): Fix comment typo.
33481 2017-01-06 Martin Liska <mliska@suse.cz>
33484 * lra-constraints.c: Rename invariant to lra_invariant.
33485 * predict.c (set_even_probabilities): Initialize e to NULL.
33487 2017-01-05 Martin Sebor <msebor@redhat.com>
33489 PR tree-optimization/78910
33490 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
33491 (format_integer): Correct off-by-one error in the handling
33492 of precision with negative numbers in signed conversions..
33494 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
33496 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
33498 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33500 PR tree-optimization/71016
33501 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
33502 factor_out_conditional_conversion. Formatting fix.
33503 (factor_out_conditional_conversion): Add cond_stmt argument.
33504 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
33505 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
33508 2017-01-05 David Malcolm <dmalcolm@redhat.com>
33510 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
33511 read-rtl-function.o, and selftest-rtl.o.
33512 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
33513 (selftest::aarch64_test_loading_full_dump): New function.
33514 (selftest::aarch64_run_selftests): New function.
33515 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
33516 selftest::aarch64_run_selftests.
33517 * config/i386/i386.c
33518 (selftest::ix86_test_loading_dump_fragment_1): New function.
33519 (selftest::ix86_test_loading_call_insn): New function.
33520 (selftest::ix86_test_loading_full_dump): New function.
33521 (selftest::ix86_test_loading_unspec): New function.
33522 (selftest::ix86_run_selftests): Call the new functions.
33523 * emit-rtl.c (maybe_set_max_label_num): New function.
33524 * emit-rtl.h (maybe_set_max_label_num): New decl.
33525 * function.c (instantiate_decls): Guard call to
33526 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
33527 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
33529 * gensupport.c (gen_reader::gen_reader): Pass "false"
33530 for new "compact" param of rtx_reader.
33531 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
33532 rather than an empty string for NULL strings.
33533 * read-md.c: Potentially include config.h rather than bconfig.h.
33534 Wrap include of errors.h with #ifdef GENERATOR_FILE.
33535 (have_error): New global, copied from errors.c.
33536 (md_reader::read_name): Rename to...
33537 (md_reader::read_name_1): ...this, adding "out_loc" param,
33538 and converting "missing name or number" to returning false, rather
33540 (md_reader::read_name): Reimplement in terms of read_name_1.
33541 (md_reader::read_name_or_nil): New function.
33542 (md_reader::read_string): Handle "(nil)" by returning NULL.
33543 (md_reader::md_reader): Add new param "compact".
33544 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
33545 (md_reader::read_file): New method.
33546 * read-md.h (md_reader::md_reader): Add new param "compact".
33547 (md_reader::read_file): New method.
33548 (md_reader::is_compact): New accessor.
33549 (md_reader::read_name): Convert return type from void to file_location.
33550 (md_reader::read_name_or_nil): New decl.
33551 (md_reader::read_name_1): New decl.
33552 (md_reader::m_compact): New field.
33553 (noop_reader::noop_reader): Pass "false" for new "compact" param
33555 (rtx_reader::rtx_reader): Add new "compact" param.
33556 (rtx_reader::read_rtx_operand): Make virtual and convert return
33557 type from void to rtx.
33558 (rtx_reader::read_until): New decl.
33559 (rtx_reader::handle_any_trailing_information): New virtual function.
33560 (rtx_reader::postprocess): New virtual function.
33561 (rtx_reader::finalize_string): New virtual function.
33562 (rtx_reader::m_in_call_function_usage): New field.
33563 (rtx_reader::m_reuse_rtx_by_id): New field.
33564 * read-rtl-function.c: New file.
33565 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
33566 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
33567 (selftest::verify_three_block_rtl_cfg): New decl.
33568 * read-rtl-function.h: New file.
33569 * read-rtl.c: Potentially include config.h rather than bconfig.h.
33570 For host, include function.h, memmodel.h, and emit-rtl.h.
33571 (one_time_initialization): New function.
33572 (struct compact_insn_name): New struct.
33573 (compact_insn_names): New array.
33574 (find_code): Handle insn codes in compact dumps.
33575 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
33576 (bind_subst_iter_and_attr): Likewise.
33577 (add_condition_to_string): Likewise.
33578 (add_condition_to_rtx): Likewise.
33579 (apply_attribute_uses): Likewise.
33580 (add_current_iterators): Likewise.
33581 (apply_iterators): Likewise.
33582 (initialize_iterators): Guard usage of apply_subst_iterator with
33583 #ifdef GENERATOR_FILE.
33584 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
33585 (md_reader::read_mapping): Likewise.
33586 (add_define_attr_for_define_subst): Likewise.
33587 (add_define_subst_attr): Likewise.
33588 (read_subst_mapping): Likewise.
33589 (check_code_iterator): Likewise.
33590 (rtx_reader::read_rtx): Likewise. Move one-time initialization
33592 (one_time_initialization): New function.
33593 (rtx_reader::read_until): New method.
33594 (read_flags): New function.
33595 (parse_reg_note_name): New function.
33596 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
33597 Handle reuse_rtx ids.
33598 Wrap iterator lookup within #ifdef GENERATOR_FILE.
33599 Add parsing support for RTL dumps, mirroring the special-cases in
33600 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
33601 values, and calling handle_any_trailing_information.
33602 (rtx_reader::read_rtx_operand): Convert return type from void
33603 to rtx, returning return_rtx. Handle case 'e'. Call
33604 finalize_string on XSTR and XTMPL fields.
33605 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
33606 "(nil)" values were omitted. Call the postprocess vfunc on the
33608 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
33609 class ctor. Initialize m_in_call_function_usage. Call
33610 one_time_initialization.
33611 * rtl-tests.c (selftest::test_uncond_jump): Call
33612 set_new_first_and_last_insn.
33613 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
33614 * selftest-rtl.c: New file.
33615 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
33616 (selftest::get_insn_by_uid): New decl.
33617 * selftest-run-tests.c (selftest::run_tests): Call
33618 read_rtl_function_c_tests.
33619 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
33620 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
33623 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
33625 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
33626 operands in a special way. Assert that pos+len <= mode precision.
33628 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33630 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
33631 3 argument Alias with unlimited for the negative form.
33632 (fno-vect-cost-model): Removed.
33634 2017-01-05 Martin Liska <mliska@suse.cz>
33636 * hsa-gen.c (gen_hsa_divmod): New function.
33637 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
33639 2017-01-05 Martin Liska <mliska@suse.cz>
33642 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
33645 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33647 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
33648 small constant length operands.
33650 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33652 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
33653 between loop iterations.
33655 2017-01-05 Martin Liska <mliska@suse.cz>
33658 * gimplify.c (gimplify_decl_expr): Compare to
33659 asan_poisoned_variables instread of checking flags.
33660 (gimplify_target_expr): Likewise.
33661 (gimplify_expr): Likewise.
33662 (gimplify_function_tree): Conditionally initialize
33663 asan_poisoned_variables.
33665 2017-01-04 Jeff Law <law@redhat.com>
33667 PR tree-optimizatin/78812
33668 * rtl.h (contains_mem_rtx_p): Prototype.
33669 * ifcvt.c (containts_mem_rtx_p): Move from here to...
33670 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
33671 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
33672 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
33673 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
33675 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33677 * input.c (assert_char_at_range): Default-initialize actual_range.
33679 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33681 * df-scan.c (df_ref_create_structure): Make regno unsigned,
33682 to match the caller.
33684 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33686 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
33687 insns after final jump in test to emit dummy move.
33689 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33691 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
33692 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
33694 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33696 * multiple_target.c (create_dispatcher_calls): Init e_next.
33697 * tree-ssa-loop-split.c (split_loop): Init border.
33698 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
33701 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
33706 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
33707 (altivec_register_operand): Do not return true if the operand
33708 contains a SUBREG mixing SImode and SFmode.
33709 (vsx_register_operand): Likewise.
33710 (vsx_reg_sfsubreg_ok): New predicate.
33711 (vfloat_operand): Do not return true if the operand contains a
33712 SUBREG mixing SImode and SFmode.
33713 (vint_operand): Likewise.
33714 (vlogical_operand): Likewise.
33715 (gpc_reg_operand): Likewise.
33716 (int_reg_operand): Likewise.
33717 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
33718 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
33719 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
33721 (rs6000_emit_move_si_sf_subreg): New helper function.
33722 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
33723 fixup SUBREGs involving SImode and SFmode.
33724 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
33725 numbers for the new peephole2 optimization.
33726 (peephole2 for SFmode unions): New peephole2 to optimize cases in
33727 the GLIBC math library that do AND/IOR/XOR operations on single
33728 precision floating point.
33729 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
33730 target macros to say whether we need to avoid SUBREGs mixing
33732 (TARGET_ALLOW_SF_SUBREG): Likewise.
33733 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
33734 (UNSPEC_SI_FROM_SF): Likewise.
33735 (iorxor): Change spacing.
33736 (and_ior_xor): New iterator for AND, IOR, and XOR.
33737 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
33738 (movdi_from_sf_zero_ext): Likewise.
33739 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
33740 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
33741 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
33742 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
33743 (fms<mode>4): Likewise.
33744 (fnma<mode>4): Likewise.
33745 (fnms<mode>4): Likewise.
33746 (nfma<mode>4): Likewise.
33747 (nfms<mode>4): Likewise.
33749 2017-01-04 Marek Polacek <polacek@redhat.com>
33752 * doc/invoke.texi: Document -Wpointer-compare.
33754 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33756 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
33759 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
33760 descriptions for -gdwarf-5 and emit them as uleb128 instead of
33763 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
33766 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
33767 documentation of the powerpc_popcntb_ok attribute.
33768 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
33769 code to issue warning messages if a requested CPU configuration is
33770 not supported by the binary (assembler and loader) toolchain.
33771 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
33772 made to define a built-in function that has been disabled.
33773 (paired_init_builtins): Add assertion to prevent ICE if attempt is
33774 made to define a built-in function that has been disabled.
33775 (altivec_init_builtins): Add comment explaining why definition
33776 of the DST built-in functions is not preceded by an assertion
33777 check. Add assertions to prevent ICE if attempts are made to
33778 define an altivec predicate or an abs* built-in function that has
33780 (htm_init_builtins): Add comment explaining why definition of the
33781 htm built-in functions is not preceded by an assertion check.
33783 2017-01-04 Jeff Law <law@redhat.com>
33785 PR tree-optimizatin/67955
33786 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
33787 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
33788 the points-to solution does not include pt_null. Use DECL_PT_UID
33791 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
33793 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
33794 Use gen_int_mode instead of gen_lopwart for const_int operands.
33796 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33798 PR tree-optimization/71563
33799 * match.pd: Simplify X << Y into X if Y is known to be 0 or
33800 out of range value - has low bits known to be zero.
33802 2017-01-04 Alan Modra <amodra@gmail.com>
33804 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
33805 * configure: Regenerate.
33806 * config.in: Regenerate.
33808 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33811 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
33812 a substring of the message, but strcmp with the whole message. Ifdef
33813 ENABLE_NLS, translate the message first using dgettext.
33815 2017-01-03 Jeff Law <law@redhat.com>
33817 PR tree-optimizatin/78856
33818 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
33819 (mark_threaded_blocks): Remove code to truncate thread paths that
33820 cross multiple loop headers. Instead invalidate the cached loop
33821 iteration information and handle case of a thread path walking
33822 into an irreducible region.
33824 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
33827 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
33828 assertions. Add support for doing the signbit if the IEEE 128-bit
33829 floating point value is in a GPR.
33830 * config/rs6000/rs6000.md (Fsignbit): Delete.
33831 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
33832 Update the length attribute if the value is in a GPR.
33833 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
33834 the sign or zero extension instruction, since the value is always 0/1.
33835 (signbit<mode>2_dm2): Delete using <Fsignbit>.
33838 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
33839 extracting SImode to a GPR register so that we can generate a
33840 store, limit the vector to be in a traditional Altivec register
33841 for the vextuwrx instruction.
33843 2017-01-03 Ian Lance Taylor <iant@google.com>
33845 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
33847 2017-01-03 Martin Sebor <msebor@redhat.com>
33849 PR tree-optimization/78696
33850 * gimple-ssa-sprintf.c (format_floating): Correct handling of
33851 precision. Use MPFR for %f for greater fidelity. Correct handling
33853 (pass_sprintf_length::compute_format_length): Set width and precision
33854 specified by asrerisk to void_node for vararg functions.
33855 (try_substitute_return_value): Adjust dump output.
33857 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
33859 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
33861 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
33863 * doc/invoke.texi (SPARC options): Document -mlra as the default.
33864 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
33865 -mlra/-mno-lra was passed to the compiler.
33867 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
33869 PR rtl-optimization/65618
33870 * emit-rtl.c (try_split): Move initialization of "before" and
33871 "after" to just before the call to emit_insn_after_setloc.
33873 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
33875 * doc/md.texi (Standard Names): Remove reference to Java frontend.
33877 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
33879 * dwarf2out.c (gen_enumeration_type_die): When
33880 -gno-strict-dwarf, add a DW_AT_encoding attribute.
33882 2017-01-03 Jakub Jelinek <jakub@redhat.com>
33884 PR tree-optimization/78965
33885 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
33886 Change first argument from const call_info & to call_info &. For %n
33887 set info.nowrite to false.
33889 PR middle-end/78901
33890 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
33891 possibly throwing calls.
33893 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
33894 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
33895 and fns handling, rather than in a separate case SSA_NAME.
33897 2017-01-02 Jeff Law <law@redhat.com>
33899 * config/darwin-driver.c (darwin_driver_init): Const-correctness
33900 fixes for first_period and second_period variables.
33902 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
33905 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
33906 (*insvqi_1): New insn pattern.
33907 (*insvqi_1_mem_rex64): Ditto.
33908 (*insvqi_2): Ditto.
33909 (*insvqi_3): Rename from *insvqi.
33911 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
33913 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
33915 * doc/cfg.texi (Edges): Remove reference to Java.
33916 (Maintaining the CFG): Ditto.
33918 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33920 PR middle-end/77674
33921 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
33922 transparent aliases.
33924 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33926 PR middle-end/77484
33927 * predict.def (PRED_CALL): Update hitrate.
33928 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
33929 * predict.c (tree_estimate_probability_bb): Split CALL predictor
33930 into direct/indirect/polymorphic variants.
33932 2017-01-01 Jakub Jelinek <jakub@redhat.com>
33934 Update copyright years.
33936 * gcc.c (process_command): Update copyright notice dates.
33937 * gcov-dump.c (print_version): Ditto.
33938 * gcov.c (print_version): Ditto.
33939 * gcov-tool.c (print_version): Ditto.
33940 * gengtype.c (create_file): Ditto.
33941 * doc/cpp.texi: Bump @copying's copyright year.
33942 * doc/cppinternals.texi: Ditto.
33943 * doc/gcc.texi: Ditto.
33944 * doc/gccint.texi: Ditto.
33945 * doc/gcov.texi: Ditto.
33946 * doc/install.texi: Ditto.
33947 * doc/invoke.texi: Ditto.
33949 Copyright (C) 2017 Free Software Foundation, Inc.
33951 Copying and distribution of this file, with or without modification,
33952 are permitted in any medium without royalty provided the copyright
33953 notice and this notice are preserved.