1 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
5 * config/aarch64/constraints.md (Uml): New constraint.
6 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
9 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
12 of two vec_duplicates into a vec_concat.
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 vec_concat.
18 * config/aarch64/constraints.md (Utq): New constraint.
19 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
22 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
25 Simplify vec_merge of vec_duplicate and const_vector.
26 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
28 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
29 mode iterator. Update predicate on operand 1 to
30 handle non-const_vec constants. Delete constraints.
31 (*aarch64_combinez_be<mode>): Likewise for operand 2.
33 2017-11-08 Jakub Jelinek <jakub@redhat.com>
35 PR tree-optimization/78821
36 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
38 (store_operand_info::store_operand_info): Initialize it to false.
39 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
40 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
41 store in the group, and if chain_info is non-NULL, to ignore altogether
43 (compatible_load_p): Fail if bit_not_p does not match.
44 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
45 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
46 (pass_store_merging::process_store): Adjust
47 terminate_all_aliasing_chains calls to pass NULL in all current spots,
48 call terminate_all_aliasing_chains newly when adding a store into
49 a chain with non-NULL chain_info.
51 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
53 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
55 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
57 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
59 (aarch64_layout_frame): Initialise emit_frame_chain.
60 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
61 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
63 2017-11-08 Martin Liska <mliska@suse.cz>
65 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
66 of gimple_call_internal_p.
68 2017-11-07 Tom de Vries <tom@codesourcery.com>
70 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
72 2017-11-07 Tom de Vries <tom@codesourcery.com>
74 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
77 2017-11-08 Martin Liska <mliska@suse.cz>
80 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
82 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
84 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
86 2017-11-07 Jakub Jelinek <jakub@redhat.com>
89 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
90 (*cmp<mode>_ccz_1): New insn with $k alternative.
93 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
94 RTX_COMM_COMPARE as commutative as well.
95 (ix86_binary_operator_ok): Formatting fix.
96 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
97 *<code><mode>3<mask_name><round_saeonly_name>,
98 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
99 *<s>mul<mode>3_highpart<mask_name>,
100 *vec_widen_umult_even_v16si<mask_name>,
101 *vec_widen_umult_even_v8si<mask_name>,
102 *vec_widen_umult_even_v4si<mask_name>,
103 *vec_widen_smult_even_v16si<mask_name>,
104 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
105 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
106 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
107 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
108 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
109 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
110 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
111 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
112 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
113 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
114 ix86_binary_operator_ok. Formatting fixes.
115 (*<plusminus_insn><mode>3<mask_name><round_name>,
116 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
119 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
121 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
122 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
123 eq and ne if TARGET_ISEL.
124 (cmp): New code_iterator.
125 (UNS, UNSU_, UNSIK): New code_attrs.
126 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
127 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
129 ("eq<mode>3"): ... this.
130 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
132 ("ne<mode>3"): ... this.
134 2017-11-07 Julia Koval <julia.koval@intel.com>
137 * common/config/i386/i386-common.c
138 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
139 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
140 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
141 * config/i386/i386.opt: Ditto.
142 * config/i386/i386.c (ix86_target_string): Ditto.
143 (ix86_option_override_internal): Ditto.
144 (ix86_init_mpx_builtins): Move MPX to args2.
145 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
146 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
147 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
148 __builtin_ia32_vgf2p8affineinvqb_v32qi,
149 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
150 __builtin_ia32_vgf2p8affineinvqb_v16qi,
151 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
153 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
156 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
157 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
158 (zero-extendsidi peephole2): Remove peephole.
160 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
163 * stor-layout.c: Include attribs.h.
164 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
165 explicit lookup of "aligned" attribute.
167 2017-11-07 Andrew Waterman <andrew@sifive.com>
169 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
170 (riscv_expand_block_move): Likewise.
171 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
173 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
174 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
175 * config/riscv/riscv.c (riscv_block_move_straight): New function.
176 (riscv_adjust_block_mem): Likewise.
177 (riscv_block_move_loop): Likewise.
178 (riscv_expand_block_move): Likewise.
179 gcc/config/riscv/riscv.md (movmemsi): New pattern.
181 2017-11-07 Michael Clark <michaeljclark@mac.com>
183 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
184 (MUSL_DYNAMIC_LINKER): Likewise.
186 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
188 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
189 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
192 2017-11-07 Richard Biener <rguenther@suse.de>
194 * match.pd: Fix build.
196 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
197 Jackson Woodruff <jackson.woodruff@arm.com>
199 PR tree-optimization/71026
200 * match.pd: Canonicalize negate in division.
202 2017-11-07 Sudakshina Das <sudi.das@arm.com>
205 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
207 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
209 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
210 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
212 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
214 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
215 non-scalar integral types.
216 * match.pd (negate_expr_p): Handle MINUS_EXPR.
217 (-(A-B), -(~A)): New transformations.
219 2017-11-07 Tom de Vries <tom@codesourcery.com>
221 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
222 semicolon after "do {} while (0)".
223 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
224 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
225 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
226 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
227 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
229 2017-11-07 Tom de Vries <tom@codesourcery.com>
231 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
232 after "do {} while (0)".
233 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
234 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
235 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
236 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
237 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
239 2017-11-07 Tom de Vries <tom@codesourcery.com>
242 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
244 (arm_rtx_costs_internal): Add missing semicolon after
245 HANDLE_NARROW_SHIFT_ARITH call.
247 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
249 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
250 disable isel if it was not set explicitly.
252 2017-11-06 James Bowman <james.bowman@ftdichip.com>
254 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
256 (add_type_attribute) likewise.
258 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
260 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
261 of struct ix86_frame.
262 (ix86_initial_elimination_offset): Likewise.
263 (ix86_expand_split_stack_prologue): Likewise.
265 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
267 * tree-vrp.h (enum value_range_type): Update stale comment.
269 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
270 Alan Hayward <alan.hayward@arm.com>
271 David Sherwood <david.sherwood@arm.com>
273 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
274 (aarch64_expand_vec_perm_const): Take the number of units too.
275 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
276 (aarch64_expand_vec_perm_const): Likewise.
277 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
278 (vec_perm<mode>): Update accordingly.
280 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
281 Alan Hayward <alan.hayward@arm.com>
282 David Sherwood <david.sherwood@arm.com>
284 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
285 Take the number of units too.
286 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
287 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
288 but check for a vector mode before rather than after the call.
289 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
290 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
291 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
292 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
293 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
294 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
295 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
296 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
297 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
298 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
299 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
300 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
301 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
302 (aarch64_sqdmull2_n<mode>): Update accordingly.
304 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
305 Alan Hayward <alan.hayward@arm.com>
306 David Sherwood <david.sherwood@arm.com>
308 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
309 the number of units too.
310 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
311 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
312 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
313 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
314 (vec_store_lanesxi<mode>): Update accordingly.
316 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
317 Alan Hayward <alan.hayward@arm.com>
318 David Sherwood <david.sherwood@arm.com>
320 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
321 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
322 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
323 of units rather than the mode.
324 * config/aarch64/iterators.md (nunits): New mode attribute.
325 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
326 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
327 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
328 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
329 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
330 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
331 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
332 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
333 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
334 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
335 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
336 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
337 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
338 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
339 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
340 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
341 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
342 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
343 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
344 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
345 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
346 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
347 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
348 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
349 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
350 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
351 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
352 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
353 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
354 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
355 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
356 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
357 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
358 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
359 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
360 (aarch64_simd_vec_setv2di): Likewise.
362 2017-11-06 Carl Love <cel@us.ibm.com>
364 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
366 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
367 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
369 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
371 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
372 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
373 Add power 8 macro expansions.
374 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
375 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
376 power 8 instructions. (VSX_XXBR): Add iterator.
378 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
380 * config/arm/arm.md (predicable_short_it): Change default to "no",
381 improve documentation, remove uses that are identical to the default.
382 (enabled_for_depr_it): Rename to enabled_for_short_it.
383 * gcc/config/arm/arm-fixed.md (predicable_short_it): Remove default uses.
384 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
385 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
386 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
387 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
389 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
392 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
393 float128 helper macros, which are no longer used after deleting
394 the old 'q' built-in functions, and moving the round to odd
395 built-in functions to being special built-in functions.
396 (BU_FLOAT128_2): Likewise.
397 (BU_FLOAT128_1_HW): Likewise.
398 (BU_FLOAT128_2_HW): Likewise.
399 (BU_FLOAT128_3_HW): Likewise.
400 (FABSQ): Delete old 'q' built-in functions.
401 (COPYSIGNQ): Likewise.
402 (SQRTF128_ODD): Move round to odd built-in functions to be
403 special built-in functions, so that we can handle
404 -mabi=ieeelongdouble.
405 (TRUNCF128_ODD): Likewise.
406 (ADDF128_ODD): Likewise.
407 (SUBF128_ODD): Likewise.
408 (MULF128_ODD): Likewise.
409 (DIVF128_ODD): Likewise.
410 (FMAF128_ODD): Likewise.
411 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
412 built-in names to 'f128'.
413 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
414 old 'q' built-in functions, as the machine independent code for
415 'f128' built-in functions handles this.
416 (rs6000_expand_builtin): Add expansion for float128 round to odd
417 functions, keying off on -mabi=ieeelongdouble of whether to use
418 the KFmode or TFmode variant.
419 (rs6000_init_builtins): Initialize the _Float128 round to odd
421 * doc/extend.texi (PowerPC Built-in Functions): Document the old
422 _Float128 'q' built-in functions are now mapped into the new
423 'f128' built-in functions.
425 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
427 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
428 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
430 2017-11-06 David Malcolm <dmalcolm@redhat.com>
433 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
434 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
435 * toplev.c: Include "ipa-fnsummary.h".
436 (toplev::finalize): Call ipa_fnsummary_c_finalize.
438 2017-11-06 Jakub Jelinek <jakub@redhat.com>
440 PR tree-optimization/82838
441 * gimple-ssa-store-merging.c
442 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
443 on a separate gimple_seq which is then appended to seq.
445 2017-11-06 Jeff Law <law@redhat.com>
448 * config/i386/i386.c (PROBE_INTERVAL): Remove.
449 (get_probe_interval): New functions.
450 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
451 (ix86_adjust_stack_and_probe): Likewise.
452 (output_adjust_stack_and_probe): Likewise.
453 (ix86_emit_probe_stack_range): Likewise.
454 (ix86_expand_prologue): Likewise.
456 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
458 PR tree-optimization/82816
459 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
460 if the modes of the two types are the same.
461 (convert_plusminus_to_widen): Likewise.
463 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
465 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
467 (usadv16qi): New define_expand.
468 (usadv8hi): New define_expand.
470 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
473 * ipa-inline-transform.c (update_noncloned_frequencies): Always
475 (inline_transform): Likewise.
476 * predict.c (counts_to_freqs): Remove useless conditional.
477 * profile-count.h (profile_count::apply_scale): Move sanity check.
478 * tree-inline.c (copy_bb): Always scale.
479 (copy_cfg_body): Likewise.
481 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
484 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
486 (*cmp_ite0): Add enabled_for_depr_it attribute.
487 (*cmp_ite1): Likewise.
489 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
491 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
494 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
496 * tree-vrp.c (vrp_int_const_binop): Return true on success and
497 return the value by pointer.
498 (extract_range_from_multiplicative_op_1): Update accordingly.
499 Return as soon as an operation fails.
501 2017-11-05 Tom de Vries <tom@codesourcery.com>
504 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
505 (DEF_SANITIZER_BUILTIN): ... here.
506 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
507 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
509 2017-11-05 Tom de Vries <tom@codesourcery.com>
512 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
514 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
515 ASM_OUTPUT_BEFORE_CASE_LABEL call.
516 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
518 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
519 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
520 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
522 2017-11-05 Tom de Vries <tom@codesourcery.com>
525 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
528 2017-11-04 Michael Clark <michaeljclark@mac.com>
530 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
531 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
533 (*addsi3_extended): Likewise.
534 (*addsi3_extended2): Likewise.
535 (<optab>si3): Likewise.
536 (<optab>di3): Likewise.
537 (<optab><mode>3): Likewise.
538 (<*optabe>si3_internal): Likewise.
539 (zero_extendqi<SUPERQI:mode>2): Likewise.
540 (*add<mode>hi3): Likewise.
541 (*xor<mode>hi3): Likewise.
542 (<optab>di3): Likewise.
543 (*<optab>si3_extend): Likewise.
544 (*sge<u>_<X:mode><GPR:mode>): Likewise.
545 (*slt<u>_<X:mode><GPR:mode>): Likewise.
546 (*sle<u>_<X:mode><GPR:mode>): Likewise.
548 2017-11-04 Andrew Waterman <andrew@sifive.com>
550 * config/riscv/riscv.c (riscv_option_override): Conditionally set
551 TARGET_STRICT_ALIGN based upon -mtune argument.
553 2017-11-04 Andrew Waterman <andrew@sifive.com>
555 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
557 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
559 config/i386/i386.c (choose_basereg): Use optional scratch
560 register and add assertion.
561 (x86_emit_outlined_ms2sysv_save): Use scratch register when
562 needed, and don't allocate stack.
563 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
564 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
565 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
567 2017-11-03 Jeff Law <law@redhat.com>
569 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
570 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
571 to probe at the start of a noreturn function.
573 2017-11-03 Jakub Jelinek <jakub@redhat.com>
575 PR tree-optimization/78821
576 * gimple-ssa-store-merging.c: Update the file comment.
577 (MAX_STORE_ALIAS_CHECKS): Define.
578 (struct store_operand_info): New type.
579 (store_operand_info::store_operand_info): New constructor.
580 (struct store_immediate_info): Add rhs_code and ops data members.
581 (store_immediate_info::store_immediate_info): Add rhscode, op0r
582 and op1r arguments to the ctor, initialize corresponding data members.
583 (struct merged_store_group): Add load_align_base and load_align
585 (merged_store_group::merged_store_group): Initialize them.
586 (merged_store_group::do_merge): Update them.
587 (merged_store_group::apply_stores): Pick the constant for
588 encode_tree_to_bitpos from one of the two operands, or skip
589 encode_tree_to_bitpos if neither operand is a constant.
590 (class pass_store_merging): Add process_store method decl. Remove
591 bool argument from terminate_all_aliasing_chains method decl.
592 (pass_store_merging::terminate_all_aliasing_chains): Remove
593 var_offset_p argument and corresponding handling.
594 (stmts_may_clobber_ref_p): New function.
595 (compatible_load_p): New function.
596 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
597 if there is overlap and rhs_code is not INTEGER_CST. For
598 non-overlapping stores terminate group if rhs is not mergeable.
599 (get_alias_type_for_stmts): Change first argument from
600 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
601 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
602 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
604 (get_location_for_stmts): Change first argument from
605 auto_vec<gimple *> & to vec<gimple *> &.
606 (struct split_store): Remove orig_stmts data member, add orig_stores.
607 (split_store::split_store): Create orig_stores rather than orig_stmts.
608 (find_constituent_stmts): Renamed to ...
609 (find_constituent_stores): ... this. Change second argument from
610 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
611 to info structures rather than the statements.
612 (split_group): Rename ALLOW_UNALIGNED argument to
613 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
614 it. Adjust find_constituent_stores caller.
615 (imm_store_chain_info::output_merged_store): Handle rhs_code other
616 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
617 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
618 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
619 (mem_valid_for_store_merging): New function.
620 (handled_load): New function.
621 (pass_store_merging::process_store): New method.
622 (pass_store_merging::execute): Use process_store method. Adjust
623 terminate_all_aliasing_chains caller.
625 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
627 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
628 Return true for more constants, symbols and label references.
629 (aarch64_valid_floating_const): Remove unused function.
631 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
634 * config/aarch64/aarch64.c (aarch64_layout_frame):
635 Undo forcing of LR at bottom of frame.
637 2017-11-03 Jeff Law <law@redhat.com>
640 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
641 for int_registers_saved.
643 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
644 extracted from tree-ssa-dom.c.
645 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
646 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
647 (record_equivalences_from_incoming_edge): Add additional argument
648 to single_pred_edge_ignoring_loop_edges call.
649 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
650 (uncprop_dom_walker::before_dom_children): Add additional argument
651 to single_pred_edge_ignoring_loop_edges call.
652 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
653 single_pred_edge_ignoring_loop_edges rather than open coding.
654 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
656 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
658 * match.pd (-(-A)): Rewrite.
660 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
662 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
663 (rs6000_emit_int_cmove): New declaration.
664 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
665 (rs6000_emit_sISEL): Delete.
666 (rs6000_emit_int_cmove): Make non-static.
667 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
668 instead of rs6000_emit_sISEL.
670 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
672 * asan.c (create_cond_insert_point): Maintain profile.
673 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
675 * basic-block.h (struct basic_block_def): Remove frequency.
676 (EDGE_FREQUENCY): Use to_frequency
677 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
679 (find_traces): Update to use to_frequency.
680 (find_traces_1_round): Likewise; use only IPA counts.
681 (bb_to_key): Likewise.
682 (connect_traces): Use IPA counts only.
683 (copy_bb_p): Update to use to_frequency.
684 (fix_up_crossing_landing_pad): Likewise.
685 (sanitize_hot_paths): Likewise.
686 * bt-load.c (basic_block_freq): Likewise.
687 * cfg.c (init_flow): Set count_max to uninitialized.
688 (check_bb_profile): Remove frequencies; check counts.
689 (dump_bb_info): Do not dump frequencies.
690 (update_bb_profile_for_threading): Update counts only.
691 (scale_bbs_frequencies_int): Likewise.
692 (MAX_SAFE_MULTIPLIER): Remove.
693 (scale_bbs_frequencies_gcov_type): Update counts only.
694 (scale_bbs_frequencies_profile_count): Update counts only.
695 (scale_bbs_frequencies): Update counts only.
696 * cfg.h (struct control_flow_graph): Add count-max.
697 (update_bb_profile_for_threading): Update prototype.
698 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
699 (find_many_sub_basic_blocks): Likewise.
700 * cfgcleanup.c (try_forward_edges): Likewise.
701 (try_crossjump_to_edge): Likewise.
702 * cfgexpand.c (expand_gimple_cond): Likewise.
703 (expand_gimple_tailcall): Likewise.
704 (construct_init_block): Likewise.
705 (construct_exit_block): Likewise.
706 * cfghooks.c (verify_flow_info): Check consistency of counts.
707 (dump_bb_for_graph): Do not dump frequencies.
708 (split_block_1): Do not update frequencies.
709 (split_edge): Do not update frequencies.
710 (make_forwarder_block): Do not update frequencies.
711 (duplicate_block): Do not update frequencies.
712 (account_profile_record): Do not update frequencies.
713 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
714 for global heuristics.
715 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
716 (expected_loop_iterations_unbounded): Use counts only.
717 * cfgloopmanip.c (scale_loop_profile): Simplify.
718 (create_empty_loop_on_edge): Simplify
720 (duplicate_loop_to_header_edge): Simplify
721 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
722 (update_br_prob_note): Take care of removing note when profile
724 (relink_block_chain): Do not dump frequency.
725 (rtl_account_profile_record): Use to_frequency.
726 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
727 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
728 (cgraph_update_edges_for_call_stmt_node): Likewise.
729 (cgraph_edge::verify_count_and_frequency): Update.
730 (cgraph_node::verify_node): Temporarily disable frequency verification.
731 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
733 (cgraph_edge::rebuild_edges): Convert to ipa counts.
734 * cgraphunit.c (init_lowered_empty_function): Do not initialize
736 (cgraph_node::expand_thunk): Update profile.
737 * except.c (dw2_build_landing_pads): Do not update frequency.
738 * final.c (compute_alignments): Use to_frequency.
739 (dump_basic_block_info): Do not dump frequency.
740 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
741 (dump_gimple_bb_header): Do not dump frequency.
742 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
744 * gimple-streamer-in.c (input_bb): Do not stream frequency.
745 * gimple-streamer-out.c (output_bb): Do not stream frequency.
746 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
747 (init_before_recovery): Do not update frequency.
748 (sched_create_recovery_edges): Do not update frequency.
749 * hsa-gen.c (convert_switch_statements): Do not update frequency.
750 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
751 (ipa_cp_c_finalize): Set max_count to uninitialized.
752 * ipa-fnsummary.c (get_minimal_bb): Use counts.
753 (param_change_prob): Use counts.
754 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
756 * ipa-split.c (consider_split): Use to_frequency.
757 (split_function): Use to_frequency.
758 * ira-build.c (loop_compare_func): Likewise.
759 (mark_loops_for_removal): Likewise.
760 (mark_all_loops_for_removal): Likewise.
761 * loop-doloop.c (doloop_modify): Do not update frequency.
762 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
764 * lto-streamer-in.c (input_function): Update count_max.
765 * omp-expand.c (expand_omp_taskreg): Update count_max.
766 * omp-simd-clone.c (simd_clone_adjust): Update profile.
767 * predict.c (maybe_hot_frequency_p): Use to_frequency.
768 (maybe_hot_count_p): Use ipa counts only.
769 (maybe_hot_bb_p): Simplify.
770 (maybe_hot_edge_p): Simplify.
771 (probably_never_executed): Do not take frequency argument.
772 (probably_never_executed_bb_p): Do not pass frequency.
773 (probably_never_executed_edge_p): Likewise.
774 (combine_predictions_for_bb): Check that profile is nonzero.
775 (propagate_freq): Do not set frequency.
776 (drop_profile): Simplify.
777 (counts_to_freqs): Simplify.
778 (expensive_function_p): Use to_frequency.
779 (propagate_unlikely_bbs_forward): Simplify.
780 (determine_unlikely_bbs): Simplify.
781 (estimate_bb_frequencies): Add hack to silence graphite issues.
782 (compute_function_frequency): Use ipa counts.
783 (pass_profile::execute): Update.
784 (rebuild_frequencies): Use counts only.
785 (force_edge_cold): Use counts only.
786 * profile-count.c (profile_count::dump): Dump new count types.
787 (profile_count::differs_from_p): Check compatiblity.
788 (profile_count::to_frequency): New function.
789 (profile_count::to_cgraph_frequency): New function.
790 * profile-count.h (struct function): Declare.
791 (enum profile_quality): Add profile_guessed_local and
792 profile_guessed_global0.
793 (class profile_proability): Decrease number of bits to 29;
794 update from_reg_br_prob_note and to_reg_br_prob_note.
795 (class profile_count: Update comment; decrease number of bits
796 to 61. Check compatibility.
797 (profile_count::compatible_p): New private member function.
798 (profile_count::ipa_p): New member function.
799 (profile_count::operator<): Handle global zero correctly.
800 (profile_count::operator>): Handle global zero correctly.
801 (profile_count::operator<=): Handle global zero correctly.
802 (profile_count::operator>=): Handle global zero correctly.
803 (profile_count::nonzero_p): New member function.
804 (profile_count::force_nonzero): New member function.
805 (profile_count::max): New member function.
806 (profile_count::apply_scale): Handle IPA scalling.
807 (profile_count::guessed_local): New member function.
808 (profile_count::global0): New member function.
809 (profile_count::ipa): New member function.
810 (profile_count::to_frequency): Declare.
811 (profile_count::to_cgraph_frequency): Declare.
812 * profile.c (OVERLAP_BASE): Delete.
813 (compute_frequency_overlap): Delete.
814 (compute_branch_probabilities): Do not use compute_frequency_overlap.
815 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
816 * sched-ebb.c (rank): Use counts only.
817 * shrink-wrap.c (handle_simple_exit): Use counts only.
818 (try_shrink_wrapping): Use counts only.
819 (place_prologue_for_one_component): Use counts only.
820 * tracer.c (find_best_predecessor): Use to_frequency.
821 (find_trace): Use to_frequency.
822 (tail_duplicate): Use to_frequency.
823 * trans-mem.c (expand_transaction): Do not update frequency.
824 * tree-call-cdce.c: Do not update frequency.
825 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
826 (gimple_merge_blocks): Likewise.
827 (gimple_split_edge): Likewise.
828 (gimple_duplicate_sese_region): Likewise.
829 (gimple_duplicate_sese_tail): Likewise.
830 (move_sese_region_to_fn): Likewise.
831 (gimple_account_profile_record): Likewise.
832 (insert_cond_bb): Likewise.
833 * tree-complex.c (expand_complex_div_wide): Likewise.
834 * tree-eh.c (lower_resx): Update profile.
835 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
837 (initialize_cfun): Do not initialize frequencies
838 (freqs_to_counts): Delete.
839 (copy_cfg_body): Ignore count parameter.
841 (expand_call_inline): Update count_max.
842 (optimize_inline_calls): Update count_max.
843 (tree_function_versioning): Update count_max.
844 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
845 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
847 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
848 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
849 (try_peel_loop): Likewise.
850 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
852 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
853 (tree_transform_and_unroll_loop): Do not use frequencies
854 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
855 Use reliable prediction only.
856 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
857 * tree-ssa-sink.c (select_best_block): Use to_frequency.
858 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
860 * tree-ssa-threadupdate.c (create_block_for_threading): Do
862 (any_remaining_duplicated_blocks): Likewise.
863 (update_profile): Likewise.
864 (estimated_freqs_path): Delete.
865 (freqs_to_counts_path): Delete.
866 (clear_counts_path): Delete.
867 (ssa_fix_duplicate_block_edges): Likewise.
868 (duplicate_thread_path): Likewise.
869 * tree-switch-conversion.c (gen_inbound_check): Use counts.
870 * tree-tailcall.c (decrease_profile): Do not update frequency.
871 (eliminate_tail_call): Likewise.
872 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
873 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
874 (optimize_mask_stores): Likewise.
875 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
876 * ubsan.c (ubsan_expand_null_ifn): Update profile.
877 (ubsan_expand_ptr_ifn): Update profile.
878 * value-prof.c (gimple_ic): Simplify.
879 * value-prof.h (gimple_ic): Update prototype.
880 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
881 * ipa-inline.c (compute_uninlined_call_time): Be sure that
883 (want_inline_self_recursive_call_p): Likewise.
884 (resolve_noninline_speculation): Only cummulate defined counts.
885 (inline_small_functions): Use nonzero_p.
886 (ipa_inline): Do not access freed node.
888 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
890 * config/aarch64/aarch64.c (aarch64_override_options_internal):
891 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
893 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
895 * config/riscv/riscv.c (riscv_legitimize_move): Handle
896 non-legitimate address.
898 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
900 * config/rs6000/rs6000.md (*lt0_disi): Delete.
901 (*lt0_<mode>di, *lt0_<mode>si): New.
903 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
905 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
908 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
909 Jim Wilson <jim.wilson@linaro.org>
911 * config/aarch64/aarch64-cores.def (saphira): New CPU.
912 * config/aarch64/aarch64-tune.md: Regenerated.
913 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
914 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
916 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
918 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
919 (arc_expand_prologue): Restore blink for millicode.
920 * config/arc/linux.h (LINK_EH_SPEC): Defined.
922 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
925 * config/i386/i386.c (ix86_vector_duplicate_value): Use
926 gen_vec_duplicate after forcing the scalar into a register.
928 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
930 * combine (try_combine): Print the insns input to try_combine to the
933 2017-11-02 Steve Ellcey <sellcey@cavium.com>
936 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
937 Remove second argument from aarch64_process_target_attr call.
938 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
940 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
942 (aarch64_handle_attr_arch): Remove second argument.
943 (aarch64_handle_attr_cpu): Ditto.
944 (aarch64_handle_attr_tune): Ditto.
945 (aarch64_handle_attr_isa_flags): Ditto.
946 (aarch64_process_one_target_attr): Ditto.
947 (aarch64_process_target_attr): Ditto.
948 (aarch64_option_valid_attribute_p): Remove second argument.
949 on aarch64_process_target_attr call.
951 2017-11-02 David Malcolm <dmalcolm@redhat.com>
953 * diagnostic.c: Include "selftest-diagnostic.h".
954 (selftest::assert_location_text): New function.
955 (selftest::test_diagnostic_get_location_text): New function.
956 (selftest::diagnostic_c_tests): Call it.
958 2017-11-02 David Malcolm <dmalcolm@redhat.com>
960 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
961 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
962 (class selftest::test_diagnostic_context): Move to...
963 * selftest-diagnostic.c: New file.
964 * selftest-diagnostic.h: New file.
966 2017-11-02 James Bowman <james.bowman@ftdichip.com>
968 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
969 offset range for FT32B.
970 * config/ft32/ft32.h: option "mcompress" enables relaxation.
971 * config/ft32/ft32.md: Add TARGET_NOPM.
972 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
973 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
975 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
977 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
979 2017-11-02 Jeff Law <law@redhat.com>
981 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
982 virtual keyword on FINAL OVERRIDE members.
984 * tree-ssa-propagate.h (ssa_propagation_engine): Group
985 virtuals together. Add virtual destructor.
986 (substitute_and_fold_engine): Similarly.
988 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
990 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
992 2017-11-02 Richard Biener <rguenther@suse.de>
994 PR tree-optimization/82795
995 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
997 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
999 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
1001 (gcc_GAS_CHECK_FEATURE): Remove.
1002 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
1003 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
1004 * configure: Regenerate.
1006 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
1008 * config/arc/arc.c (hwloop_optimize): Account for empty
1011 2017-11-02 Richard Biener <rguenther@suse.de>
1014 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
1015 Truncate ARRAY_REF index and element size.
1017 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1019 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
1021 2017-11-01 Jeff Law <law@redhat.com>
1023 * tree-ssa-ccp.c (ccp_folder): New class derived from
1024 substitute_and_fold_engine.
1025 (ccp_folder::get_value): New member function.
1026 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
1027 (ccp_fold_stmt): Remove prototype.
1028 (ccp_finalize): Call substitute_and_fold from the ccp_class.
1029 * tree-ssa-copy.c (copy_folder): New class derived from
1030 substitute_and_fold_engine.
1031 (copy_folder::get_value): Renamed from get_value.
1032 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
1033 * tree-vrp.c (vrp_folder): New class derived from
1034 substitute_and_fold_engine.
1035 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
1036 (vrp_folder::get_value): New member function.
1037 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
1038 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
1039 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
1040 provide a class interface to folder/substitute routines.
1041 (ssa_prop_fold_stmt_fn): Remove typedef.
1042 (ssa_prop_get_value_fn): Likewise.
1043 (subsitute_and_fold): Remove prototype.
1044 (replace_uses_in): Likewise.
1045 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
1046 Renamed from replace_uses_in. Call the virtual member function
1047 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
1048 (substitute_and_fold_dom_walker): Remove initialization of
1049 data member entries for calbacks. Add substitute_and_fold_engine
1050 member and initialize it.
1051 (substitute_and_fold_dom_walker::before_dom_children0: Use the
1052 member functions for get_value, replace_phi_args_in c
1053 replace_uses_in, and fold_stmt calls.
1054 (substitute_and_fold_engine::substitute_and_fold): Renamed from
1055 substitute_and_fold. Remove assert. Update ctor call.
1057 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
1058 (ssa_prop_visit_phi_fn): Likewise.
1059 (class ssa_propagation_engine): New class to provide an interface
1061 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
1063 (ssa_prop_visit_phi): Likewise.
1064 (ssa_propagation_engine::simulate_stmt): Moved into class.
1065 Call visit_phi/visit_stmt from the class rather than via
1066 file scoped static variables.
1067 (ssa_propagation_engine::simulate_block): Moved into class.
1068 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
1069 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
1070 set file scoped statics for the visit_stmt/visit_phi callbacks.
1071 * tree-complex.c (complex_propagate): New class derived from
1072 ssa_propagation_engine.
1073 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
1074 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
1075 (tree_lower_complex): Call ssa_propagate via the complex_propagate
1077 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
1078 ssa_propagation_engine.
1079 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
1080 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
1081 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
1082 * tree-ssa-copy.c (copy_prop): New class derived from
1083 ssa_propagation_engine.
1084 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
1085 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
1086 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
1087 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
1088 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
1089 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
1090 (execute_vrp): Call ssa_propagate from the vrp_prop class.
1092 2017-11-01 Jakub Jelinek <jakub@redhat.com>
1094 PR rtl-optimization/82778
1095 PR rtl-optimization/82597
1096 * compare-elim.c (struct comparison): Add in_a_setter field.
1097 (find_comparison_dom_walker::before_dom_children): Remove killed
1098 bitmap and df_simulate_find_defs call, instead walk the defs.
1099 Compute last_setter and initialize in_a_setter. Merge definitions
1100 with first initialization for a few variables.
1101 (try_validate_parallel): Use insn_invalid_p instead of
1102 recog_memoized. Return insn rather than just the pattern.
1103 (try_merge_compare): Fix up comment. Don't uselessly test if
1104 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
1106 (execute_compare_elim_after_reload): Remove df_chain_add_problem
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/aarch64.c (aarch64_rtx_costs): Use
1114 aarch64_hard_regno_nregs to get the number of registers
1117 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1118 Alan Hayward <alan.hayward@arm.com>
1119 David Sherwood <david.sherwood@arm.com>
1121 * config/aarch64/constraints.md (Upl): Rename to...
1123 * config/aarch64/aarch64.md
1124 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
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_add_constant_internal)
1132 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
1135 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1136 Alan Hayward <alan.hayward@arm.com>
1137 David Sherwood <david.sherwood@arm.com>
1139 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
1140 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
1141 (aarch64_evpc_dup): Generate rtl direcly, rather than using
1143 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
1144 of a single element.
1145 * config/aarch64/iterators.md: Add a comment above the permute
1146 unspecs to say that they are generated directly by
1147 aarch64_expand_vec_perm_const.
1148 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
1150 2017-11-01 Nathan Sidwell <nathan@acm.org>
1152 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
1154 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1156 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
1157 and medany code models, and describe what they do.
1159 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1161 Revert accidental duplicate:
1163 * combine.c (can_change_dest_mode): Reject changes in
1164 REGMODE_NATURAL_SIZE.
1166 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
1168 PR rtl-optimization/64682
1169 PR rtl-optimization/69567
1170 PR rtl-optimization/69737
1171 PR rtl-optimization/82683
1172 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
1173 register mentioned in the note, drop the note, unless it came from I3,
1174 in which case it should go to I3 again.
1176 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1178 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
1179 and return false if not.
1180 (clear_bytes_written_by, live_bytes_read): Update accordingly.
1182 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1184 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
1185 range is known to be empty.
1187 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1188 Alan Hayward <alan.hayward@arm.com>
1189 David Sherwood <david.sherwood@arm.com>
1191 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
1192 and CONST_VECTOR_NUNITS instead of computing the number of units from
1193 the byte sizes of the vector and element.
1194 (simplify_binary_operation_1): Likewise.
1195 (simplify_const_binary_operation): Likewise.
1196 (simplify_ternary_operation): Likewise.
1198 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1199 Alan Hayward <alan.hayward@arm.com>
1200 David Sherwood <david.sherwood@arm.com>
1202 * var-tracking.c (INT_MEM_OFFSET): Replace with...
1203 (int_mem_offset): ...this new function.
1204 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
1205 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
1206 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
1209 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1210 Alan Hayward <alan.hayward@arm.com>
1211 David Sherwood <david.sherwood@arm.com>
1213 * lower-subreg.c (interesting_mode_p): New function.
1214 (compute_costs, find_decomposable_subregs, decompose_register)
1215 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
1216 (resolve_clobber, dump_choices): Use it.
1218 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1219 Alan Hayward <alan.hayward@arm.com>
1220 David Sherwood <david.sherwood@arm.com>
1222 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
1224 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1225 Alan Hayward <alan.hayward@arm.com>
1226 David Sherwood <david.sherwood@arm.com>
1228 * alias.c (find_base_value, find_base_term): Only process integer
1229 truncations. Check the precision rather than the size.
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 * machmode.h (is_narrower_int_mode): New function
1236 * optabs.c (expand_float, expand_fix): Use it.
1237 * dwarf2out.c (rotate_loc_descriptor): Likewise.
1239 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1240 Alan Hayward <alan.hayward@arm.com>
1241 David Sherwood <david.sherwood@arm.com>
1243 * rtl.h (narrower_subreg_mode): New function.
1244 * ira-color.c (update_costs_from_allocno): Use it.
1246 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1247 Alan Hayward <alan.hayward@arm.com>
1248 David Sherwood <david.sherwood@arm.com>
1250 * optabs-query.h (convert_optab_p): New function, split out from...
1251 (convert_optab_handler): ...here.
1252 (widening_optab_handler): Delete.
1253 (find_widening_optab_handler): Remove permit_non_widening parameter.
1254 (find_widening_optab_handler_and_mode): Likewise. Provide an
1255 override that operates on mode class wrappers.
1256 * optabs-query.c (widening_optab_handler): Delete.
1257 (find_widening_optab_handler_and_mode): Remove permit_non_widening
1258 parameter. Assert that the two modes are the same class and that
1259 the "from" mode is narrower than the "to" mode. Use
1260 convert_optab_handler instead of widening_optab_handler.
1261 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
1262 instead of widening_optab_handler.
1263 * expr.c (expand_expr_real_2): Update calls to
1264 find_widening_optab_handler.
1265 * optabs.c (expand_widen_pattern_expr): Likewise.
1266 (expand_binop_directly): Take the insn_code as a parameter.
1267 (expand_binop): Only call find_widening_optab_handler for
1268 conversion optabs; use optab_handler otherwise. Update calls
1269 to find_widening_optab_handler and expand_binop_directly.
1270 Use convert_optab_handler instead of widening_optab_handler.
1271 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
1272 find_widening_optab_handler and use scalar_mode rather than
1274 (convert_plusminus_to_widen): Likewise.
1276 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1277 Alan Hayward <alan.hayward@arm.com>
1278 David Sherwood <david.sherwood@arm.com>
1280 * machmode.h (fixed_size_mode): New class.
1281 * rtl.h (get_pool_mode): Return fixed_size_mode.
1282 * gengtype.c (main): Add fixed_size_mode.
1283 * target.def (get_raw_result_mode): Return a fixed_size_mode.
1284 (get_raw_arg_mode): Likewise.
1285 * doc/tm.texi: Regenerate.
1286 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
1287 * targhooks.c (default_get_reg_raw_mode): Likewise.
1288 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
1289 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
1290 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
1291 (msp430_get_raw_result_mode): Likewise.
1292 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
1293 * dbxout.c (dbxout_parms): Require fixed-size modes.
1294 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
1295 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
1296 * omp-low.c (lower_oacc_reductions): Likewise.
1297 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
1298 (simplify_subreg): Update accordingly.
1299 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
1300 (force_const_mem): Update accordingly. Return NULL_RTX for modes
1301 that aren't fixed-size.
1302 (get_pool_mode): Return a fixed_size_mode.
1303 (output_constant_pool_2): Take a fixed_size_mode.
1305 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1306 Alan Hayward <alan.hayward@arm.com>
1307 David Sherwood <david.sherwood@arm.com>
1309 * doc/rtl.texi (vec_series): Document.
1310 (const): Say that the operand can be a vec_series.
1311 * rtl.def (VEC_SERIES): New rtx code.
1312 * rtl.h (const_vec_series_p_1): Declare.
1313 (const_vec_series_p): New function.
1314 * emit-rtl.h (gen_const_vec_series): Declare.
1315 (gen_vec_series): Likewise.
1316 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
1317 (gen_vec_series): Likewise.
1318 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
1319 * simplify-rtx.c (simplify_unary_operation): Handle negations
1321 (simplify_binary_operation_series): New function.
1322 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
1323 (test_vector_ops_series): New function.
1324 (test_vector_ops): Call it.
1325 * config/powerpcspe/altivec.md (altivec_lvsl): Use
1326 gen_const_vec_series.
1327 (altivec_lvsr): Likewise.
1328 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
1330 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1331 Alan Hayward <alan.hayward@arm.com>
1332 David Sherwood <david.sherwood@arm.com>
1334 * doc/rtl.texi (const): Update description of address constants.
1335 Say that vector constants are allowed too.
1336 * common.md (E, F): Use CONSTANT_P instead of checking for
1338 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
1339 checking for CONST_VECTOR.
1340 * expmed.c (make_tree): Use build_vector_from_val for a CONST
1342 * expr.c (expand_expr_real_2): Check for vector modes instead
1343 of checking for CONST_VECTOR.
1344 * rtl.h (const_vec_p): New function.
1345 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
1346 (unwrap_const_vec_duplicate): Handle them here too.
1348 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1349 David Malcolm <dmalcolm@redhat.com>
1350 Alan Hayward <alan.hayward@arm.com>
1351 David Sherwood <david.sherwood@arm.com>
1353 * rtl.h (vec_duplicate_p): New function.
1354 * selftest-rtl.c (assert_rtx_eq_at): New function.
1355 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
1356 (assert_rtx_eq_at): Declare.
1357 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
1358 * selftest-run-tests.c (selftest::run_tests): Call it.
1359 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
1360 (simplify_unary_operation_1): Recursively handle vector duplicates.
1361 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
1363 (simplify_subreg): Handle subregs of vector duplicates.
1364 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
1365 (selftest::simplify_rtx_c_tests): New functions.
1367 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1368 Alan Hayward <alan.hayward@arm.com>
1369 David Sherwood <david.sherwood@arm.com>
1371 * emit-rtl.h (gen_const_vec_duplicate): Declare.
1372 (gen_vec_duplicate): Likewise.
1373 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
1375 (gen_const_vector): ...here.
1376 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
1377 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
1378 whose elements are all equal.
1379 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
1380 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
1381 (simplify_relational_operation): Likewise.
1382 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
1384 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
1385 (aarch64_expand_vector_init): Likewise.
1386 * config/arm/arm.c (neon_vdup_constant): Likewise.
1387 (neon_expand_vector_init): Likewise.
1388 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
1389 (arm_block_set_unaligned_vect): Likewise.
1390 (arm_block_set_aligned_vect): Likewise.
1391 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
1392 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
1393 (expand_vec_perm_even_odd_pack): Likewise.
1394 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
1395 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
1396 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
1397 gen_const_vec_duplicate.
1398 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
1399 * config/mips/mips.c (mips_gen_const_int_vector): Use
1400 gen_const_vec_duplicate.
1401 (mips_expand_vector_init): Use CONST0_RTX.
1402 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
1403 (define_split): Use gen_const_vec_duplicate.
1404 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
1405 (define_split): Use gen_const_vec_duplicate.
1406 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
1407 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
1408 * config/spu/spu.c (spu_const): Likewise.
1410 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1411 Alan Hayward <alan.hayward@arm.com>
1412 David Sherwood <david.sherwood@arm.com>
1414 * combine.c (can_change_dest_mode): Reject changes in
1415 REGMODE_NATURAL_SIZE.
1417 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
1419 * configure.ac (--enable-libssp): New.
1420 (gcc_cv_libc_provides_ssp): Check for explicit setting before
1421 trying to determine target-specific default. Adjust indentation.
1422 * configure: Regenerated.
1423 * doc/install.texi (Configuration): Expand --disable-libssp
1426 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
1428 config/i386/i386.c (ix86_expand_epilogue): Correct stack
1431 2017-10-31 Martin Jambor <mjambor@suse.cz>
1434 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
1436 2017-10-31 David Malcolm <dmalcolm@redhat.com>
1438 * auto-profile.c (autofdo_source_profile::read): Use
1439 UNKNOWN_LOCATION rather than 0.
1440 * diagnostic-core.h (warning_at_rich_loc): Rename to...
1441 (warning_at): ...this overload.
1442 (warning_at_rich_loc_n): Rename to...
1443 (warning_n): ...this overload.
1444 (error_at_rich_loc): Rename to...
1445 (error_at): ...this overload.
1446 (pedwarn_at_rich_loc): Rename to...
1447 (pedwarn): ...this overload.
1448 (permerror_at_rich_loc): Rename to...
1449 (permerror): ...this overload.
1450 (inform_at_rich_loc): Rename to...
1451 (inform): ...this overload.
1452 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
1453 (diagnostic_n_impl_richloc): Rename to...
1454 (diagnostic_n_impl): ...this rich_location *-based decl.
1455 (inform_at_rich_loc): Rename to...
1456 (inform): ...this, and add an assertion.
1457 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
1458 (warning_at_rich_loc): Rename to...
1459 (warning_at): ...this, and add an assertion.
1460 (warning_at_rich_loc_n): Rename to...
1461 (warning_n): ...this, and add an assertion.
1462 (warning_n): Update location_t-based implementation for removal of
1463 location_t-based diagnostic_n_impl.
1464 (pedwarn_at_rich_loc): Rename to...
1465 (pedwarn): ...this, and add an assertion.
1466 (permerror_at_rich_loc): Rename to...
1467 (permerror): ...this, and add an assertion.
1468 (error_n): Update for removal of location_t-based diagnostic_n_impl.
1469 (error_at_rich_loc): Rename to...
1470 (error_at): ...this, and add an assertion.
1471 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
1472 (driver::do_spec_on_infiles): Likewise.
1473 * substring-locations.c (format_warning_va): Update for renaming
1474 of inform_at_rich_loc.
1476 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
1478 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
1479 _Float<N>X built-in functions so that the variant without the
1480 "__builtin_" prefix is only enabled for the GNU C and Objective C
1481 languages when they are in non-strict ANSI/ISO mode.
1482 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
1483 * target.def (floatn_builtin_p): Add a target hook to control
1484 whether _Float<N> and _Float<N>X built-in functions without the
1485 "__builtin_" prefix are enabled, and return true for C and
1486 Objective C in the default hook. Include langhooks.h in
1488 * targhooks.h (default_floatn_builtin_p): Likewise.
1489 * targhooks.c (default_floatn_builtin_p): Likewise.
1490 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
1491 floatn_builtin_p target hook.
1492 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
1494 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
1495 Eric Botcazou <ebotcazou@adacore.com>
1497 PR rtl-optimization/81803
1498 * lra-constraints.c (curr_insn_transform): Also reload the whole
1499 register for a strict subreg no wider than a word if this is for
1500 a WORD_REGISTER_OPERATIONS target.
1502 2017-10-31 Jason Merrill <jason@redhat.com>
1504 * gdbinit.in: Skip over inlines from timevar.h.
1506 2017-10-31 Martin Liska <mliska@suse.cz>
1508 * doc/gcov.texi: Document new option.
1509 * gcov.c (print_usage): Likewise print it.
1510 (process_args): Support the argument.
1511 (format_count): New function.
1512 (format_gcov): Use the function.
1514 2017-10-31 Martin Liska <mliska@suse.cz>
1516 * gcov.c (struct name_map): do not use typedef.
1517 Define operator== and operator<.
1518 (name_search): Remove.
1519 (name_sort): Remove.
1520 (main): Do not allocate names.
1521 (process_file): Add vertical space.
1522 (generate_results): Use std::find.
1523 (release_structures): Do not release memory.
1524 (find_source): Use std::find.
1526 2017-10-31 Martin Liska <mliska@suse.cz>
1528 * gcov.c (struct line_info): Remove it's typedef.
1529 (line_info::line_info): Add proper ctor.
1530 (line_info::has_block): Do not use a typedef.
1531 (struct source_info): Do not use typedef.
1532 (circuit): Likewise.
1533 (get_cycles_count): Likewise.
1534 (output_intermediate_file): Iterate via vector iterator.
1535 (add_line_counts): Use std::vector methods.
1536 (accumulate_line_counts): Likewise.
1537 (output_lines): Likewise.
1539 2017-10-31 Martin Liska <mliska@suse.cz>
1541 * gcov.c (struct source_info): Remove typedef.
1542 (source_info::source_info): Add proper ctor.
1543 (accumulate_line_counts): Use struct, not it's typedef.
1544 (output_gcov_file): Likewise.
1545 (output_lines): Likewise.
1546 (main): Do not allocate an array.
1547 (output_intermediate_file): Use size of vector container.
1548 (process_file): Resize the vector.
1549 (generate_results): Do not preallocate, use newly added vector
1551 (release_structures): Do not release sources.
1552 (find_source): Use vector methods.
1553 (add_line_counts): Do not use typedef.
1555 2017-10-31 Martin Liska <mliska@suse.cz>
1557 * doc/gcov.texi: Document that.
1558 * gcov.c (add_line_counts): Mark lines with a non-executed
1560 (output_line_beginning): Handle such lines.
1561 (output_lines): Pass new argument.
1562 (output_intermediate_file): Print it in intermediate format.
1564 2017-10-31 Martin Liska <mliska@suse.cz>
1566 * color-macros.h: New file.
1567 * diagnostic-color.c: Factor out color related to macros to
1569 * doc/gcov.texi: Document -k option.
1570 * gcov.c (INCLUDE_STRING): Include string.h.
1571 (print_usage): Add -k option.
1572 (process_args): Parse it.
1573 (pad_count_string): New function.
1574 (output_line_beginning): Likewise.
1575 (DEFAULT_LINE_START): New macro.
1576 (output_lines): Support color output.
1578 2017-10-31 Martin Liska <mliska@suse.cz>
1580 PR gcov-profile/82633
1581 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
1582 their interaction with GCOV infrastructure.
1583 * configure.ac: Add -fkeep-{inline,static}-functions to
1585 * configure: Regenerate.
1587 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
1590 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
1592 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
1595 * config/rs6000/rs6000.md (allocate_stack): Force update interval
1596 into a register if it does not fit into an immediate offset field.
1598 2017-10-31 Olivier Hainque <hainque@adacore.com>
1600 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
1602 2017-10-31 Julia Koval <julia.koval@intel.com>
1604 * config.gcc: Add gfniintrin.h.
1605 * config/i386/gfniintrin.h: New.
1606 * config/i386/i386-builtin-types.def
1607 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1608 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1609 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1610 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1611 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1612 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
1613 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
1614 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
1615 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
1616 V64QI_FTYPE_V64QI_V64QI_INT): New types.
1617 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1618 * config/i386/immintrin.h: Include gfniintrin.h.
1619 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
1621 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
1623 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
1625 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1627 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
1628 (arm_ashldi3_1bit): Remove pattern.
1629 (ashrdi3): Remove shift by 1 expansion.
1630 (arm_ashrdi3_1bit): Remove pattern.
1631 (lshrdi3): Remove shift by 1 expansion.
1632 (arm_lshrdi3_1bit): Remove pattern.
1633 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
1634 cost of ashldi3 by 1.
1635 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
1636 (<shift>di3_neon): Likewise.
1638 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
1640 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
1641 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
1642 and (*aarch64_simd_mov<VQ:mode>).
1643 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
1644 pattern alternative.
1645 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
1646 attributes to match pattern alternative.
1648 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
1650 * config.gcc (powerpc*-*-*): Add emmintrin.h.
1651 * config/rs6000/emmintrin.h: New file.
1652 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
1654 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1656 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
1657 * (movdi_vfp_cortexa8): Remove pattern.
1659 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1661 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
1663 (Specific, alpha*-dec-osf5.1): Remove.
1664 (Specific, mips-sgi-irix5): Remove.
1665 (Specific, mips-sgi-irix6): Remove.
1667 2017-10-30 Jakub Jelinek <jakub@redhat.com>
1670 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
1671 arguments to clear_bit_region_be.
1673 2017-10-30 Jim Wilson <wilson@tuliptree.org>
1675 * gimplify.c: Include memmodel.h.
1677 2017-10-30 Martin Jambor <mjambor@suse.cz>
1679 * omp-grid.c (grid_attempt_target_gridification): Also insert a
1680 condition whether loop should be executed at all.
1682 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
1684 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1685 gimple folding of vec_madd() intrinsics.
1686 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
1687 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
1688 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
1690 2017-10-30 Richard Biener <rguenther@suse.de>
1692 PR tree-optimization/82762
1694 2017-10-23 Richard Biener <rguenther@suse.de>
1696 PR tree-optimization/82129
1698 2017-08-01 Richard Biener <rguenther@suse.de>
1700 PR tree-optimization/81181
1701 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
1702 (compute_antic): ... end of iteration here.
1704 2017-10-30 Joseph Myers <joseph@codesourcery.com>
1706 * doc/invoke.texi (C Dialect Options): Document -std=c17,
1707 -std=iso9899:2017 and -std=gnu17.
1708 * doc/standards.texi (C Language): Document C17 support.
1709 * doc/cpp.texi (Overview): Mention -std=c17.
1710 (Standard Predefined Macros): Document C11 and C17 values of
1711 __STDC_VERSION__. Do not refer to C99 support as incomplete.
1712 * doc/extend.texi (Inline): Do not list individual options for
1713 standards newer than C99.
1714 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
1716 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
1719 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
1721 * asan.c (asan_finish_file): Align asan globals array by shadow
1724 2017-10-30 Jakub Jelinek <jakub@redhat.com>
1727 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
1728 (struct store_immediate_info): Add bitregion_start and bitregion_end
1730 (store_immediate_info::store_immediate_info): Add brs and bre
1731 arguments and initialize bitregion_{start,end} from those.
1732 (struct merged_store_group): Add bitregion_start, bitregion_end,
1733 align_base and mask fields. Drop unnecessary struct keyword from
1734 struct store_immediate_info. Add do_merge method.
1735 (clear_bit_region_be): Use memset instead of loop storing zeros.
1736 (merged_store_group::do_merge): New method.
1737 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
1738 stores as long as the surrounding bitregions have no gaps.
1739 (merged_store_group::merge_overlapping): Use do_merge.
1740 (merged_store_group::apply_stores): Test that bitregion_{start,end}
1741 is byte aligned, rather than requiring that start and width are
1742 byte aligned. Drop unnecessary struct keyword from
1743 struct store_immediate_info. Allocate and populate also mask array.
1744 Make start of the arrays relative to bitregion_start rather than
1745 start and size them according to bitregion_{end,start} difference.
1746 (struct imm_store_chain_info): Drop unnecessary struct keyword from
1747 struct store_immediate_info.
1748 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
1749 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
1750 struct keyword from struct store_immediate_info.
1751 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
1752 between stores as long as the surrounding bitregions have no gaps.
1754 (struct split_store): Add orig non-static data member.
1755 (split_store::split_store): Initialize orig to false.
1756 (find_constituent_stmts): Return store_immediate_info *, non-NULL
1757 if there is exactly a single original stmt. Change stmts argument
1758 to pointer from reference, if NULL, don't push anything to it. Add
1759 first argument, use it to optimize skipping over orig stmts that
1760 are known to be before bitpos already. Simplify.
1761 (split_group): Return unsigned int count how many stores are or
1762 would be needed rather than a bool. Add allow_unaligned argument.
1763 Change split_stores argument from reference to pointer, if NULL,
1764 only do a dry run computing how many stores would be produced.
1765 Rewritten algorithm to use both alignment and misalign if
1766 !allow_unaligned and handle bitfield stores with gaps.
1767 (imm_store_chain_info::output_merged_store): Set start_byte_pos
1768 from bitregion_start instead of start. Compute allow_unaligned
1769 here, if true, do 2 split_group dry runs to compute which one
1770 produces fewer stores and prefer aligned if equal. Punt if
1771 new count is bigger or equal than original before emitting any
1772 statements, rather than during that. Remove no longer needed
1773 new_ssa_names tracking. Replace num_stmts with
1774 split_stores.length (). Use 32-bit stack allocated entries
1775 in split_stores auto_vec. Try to reuse original store lhs/rhs1
1776 if possible. Handle bitfields with gaps.
1777 (pass_store_merging::execute): Ignore bitsize == 0 stores.
1778 Compute bitregion_{start,end} for the stores and construct
1779 store_immediate_info with that. Formatting fixes.
1781 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
1784 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
1785 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
1787 2017-10-29 Jim Wilson <wilson@tuliptree.org>
1789 * gimplify.c: Include tm_p.h.
1791 * common.opt (gcoff): Re-add as ignored option.
1792 (gcoff1, gcoff2, gcoff3): Likewise.
1794 * Makefile.in (OBJS): Delete sdbout.o.
1795 (GTFILES): Delete $(srcdir)/sdbout.c.
1796 * debug.h: Delete sdb_debug_hooks.
1797 * final.c: Delete sdbout.h include.
1798 (final_scan_insn): Delete SDB_DEBUG check.
1799 (rest_of_clean_state): Likewise.
1800 * output.h: Delete sdb_begin_function_line.
1803 * toplev.c: Delete sdbout.h include.
1804 (process_options): Delete SDB_DEBUG check.
1805 * tree-core.h (tree_type_common): Delete pointer field of
1807 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
1808 TYPE_SYMTAB_POINTER.
1809 * tree.h (TYPE_SYMTAB_POINTER): Delete.
1810 (TYPE_SYMTAB_IS_POINTER): Delete.
1811 (TYPE_SYMTAB_IS_DIE): Renumber.
1812 * xcoffout.c: Refer to former sdbout.c file.
1813 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
1815 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
1816 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
1817 (Debugging Options): Delete -gcoff.
1818 (-gstabs): Delete SDB reference.
1820 (-gcoff@var{level}): Delete.
1821 * doc/passes.texi (Debugging information output): Delete SDB and
1822 sdbout.c references.
1823 * doc/tm.texi: Regenerate.
1824 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
1825 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
1827 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
1828 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
1829 (SDB_DEBUGGING_INFO): Delete.
1830 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
1831 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
1832 * target.def (output_source_filename): Delete COFF reference.
1834 * common.opt (gcoff): Delete.
1835 (gxcoff+): Update Negative chain.
1836 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
1838 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
1839 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
1840 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
1842 (expand_function_start): Change sdb reference to past tense.
1843 (expand_function_end): Change sdb reference to past tense.
1844 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
1845 * opts.c (debug_type_names): Delete coff entry.
1846 (common_handle_option): Delete OPT_gcoff case.
1847 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
1849 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
1850 * config/cris/cris.h: Delete SDB reference in comment.
1851 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
1852 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
1853 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
1854 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
1856 (ix86_expand_prologue): Likewise.
1857 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
1858 * config/ia64/ia64.h: Likewise.
1859 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
1860 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
1862 * config/mmix/mmix.h: Likewise.
1863 * config/nds32/nds32.c: Likewise.
1864 * config/stormy/storym16.h: Likewise.
1865 * config/visium/visium.h: Likewise.
1866 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
1868 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
1870 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
1871 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
1872 FRAME_POINTER_REGNUM point at high end of local var area.
1874 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
1876 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
1877 Move comment around. Do not reset best_edge for a copiable
1878 destination if the copy would cause a partition change.
1879 (better_edge_p): Remove redundant check.
1881 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
1883 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
1885 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
1887 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
1888 for math functions that have _Float<N> and _Float<N>X variants.
1889 (mathfn_built_in_2): Add support for math functions that have
1890 _Float<N> and _Float<N>X variants.
1891 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
1892 (expand_builtin_mathfn_ternary): Add support for fma with
1893 _Float<N> and _Float<N>X variants.
1894 (expand_builtin): Likewise.
1895 (fold_builtin_3): Likewise.
1896 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
1897 create math function _Float<N> and _Float<N>X variants as external
1899 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
1900 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
1901 the __builtin_ prefix and if not strict ansi, without the prefix.
1902 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
1903 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
1904 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
1905 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
1906 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
1907 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
1908 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
1909 function signatures for fma _Float<N> and _Float<N>X variants.
1910 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
1911 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
1912 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
1913 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
1914 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
1915 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
1916 * gencfn-macros.c (print_case_cfn): Add support for math functions
1917 that have _Float<N> and _Float<N>X variants.
1918 (print_define_operator_list): Likewise.
1919 (fltfn_suffixes): Likewise.
1921 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
1922 for math functions that have _Float<N> and _Float<N>X variants.
1923 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
1924 and _Float<N>X variants.
1925 (COPYSIGN): Likewise.
1928 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
1929 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
1931 (integer_valued_read_call_p): Likewise.
1932 * fold-const-call.c (fold_const_call_ss): Likewise.
1933 (fold_const_call_sss): Add support for copysign, fmin, and fmax
1934 _Float<N> and _Float<N>X variants.
1935 (fold_const_call_ssss): Add support for fma _Float<N> and
1936 _Float<N>X variants.
1937 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
1938 support for copysign and fma _Float<N> and _Float<N>X variants.
1939 (backprop::process_builtin_call_use): Likewise.
1940 * tree-call-cdce.c (can_test_argument_range); Add support for
1941 sqrt _Float<N> and _Float<N>X variants.
1942 (edom_only_function): Likewise.
1943 (get_no_error_domain): Likewise.
1944 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
1945 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
1946 copysign _Float<N> and _Float<N>X variants.
1947 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
1948 handled by machine independent code.
1949 (FMAF128): Likewise.
1950 * doc/cpp.texi (Common Predefined Macros): Document defining
1951 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
1952 fma _Float<N> and _Float<N>X variants.
1954 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
1957 * config/i386/i386-modes.def (CCFPU): Remove definition.
1958 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
1959 (ix86_cc_modes_compatible): Ditto.
1960 (ix86_expand_carry_flag_compare): Ditto.
1961 (ix86_expand_int_movcc): Ditto.
1962 (ix86_expand_int_addcc): Ditto.
1963 (ix86_reverse_condition): Ditto.
1964 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
1965 Return true/false for unordered/ordered fp comparisons.
1966 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
1967 (ix86_prepare_fp_compare_args): Update for rename.
1968 (ix86_expand_fp_compare): Update for rename. Generate unordered
1969 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
1970 (ix86_expand_sse_compare_and_jump): Ditto.
1971 * config/i386/predicates.md (fcmov_comparison_operator):
1972 Remove CCFPU mode handling.
1973 (ix86_comparison_operator): Ditto.
1974 (ix86_carry_flag_operator): Ditto.
1975 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
1976 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
1977 (*cmpu<mode>_cc_i387): Ditto.
1978 (FPCMP): Remove mode iterator.
1979 (unord): Remove mode attribute.
1980 (unord_subst): New define_subst transformation
1981 (unord): New define_subst attribute.
1983 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
1984 (*cmpi<unord>xf_i387): Ditto.
1985 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
1986 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
1987 using unord_subst transformation.
1988 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
1989 (round_saeonly): Also handle CCFP mode.
1990 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
1991 Remove UNSPEC_SAHF unspec handling.
1993 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
1995 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
1997 2017-10-27 Jeff Law <law@redhat.com>
1999 * gimple-ssa-sprintf.c: Include domwalk.h.
2000 (class sprintf_dom_walker): New class, derived from dom_walker.
2001 (sprintf_dom_walker::before_dom_children): New function.
2002 (struct call_info): Moved into sprintf_dom_walker class
2003 (compute_formath_length, handle_gimple_call): Likewise.
2004 (sprintf_length::execute): Call the dominator walker rather
2005 than walking the statements.
2007 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
2008 gimple statement locations.
2009 (check_array_bounds): Corresponding changes. Get the statement's
2010 location directly from wi->stmt.
2012 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
2015 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
2017 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
2019 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
2020 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
2022 2017-10-27 Jakub Jelinek <jakub@redhat.com>
2025 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
2026 * config/i386/i386.c (maybe_get_pool_constant): Removed.
2027 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
2028 maybe_get_pool_constant.
2029 * config/i386/predicates.md (zero_extended_scalar_load_operand):
2032 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2034 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
2035 2.26 caveat. Update gas and gld versions.
2036 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
2039 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2041 * cgraph.h (set_malloc_flag): Declare.
2042 * cgraph.c (set_malloc_flag_1): New function.
2043 (set_malloc_flag): Likewise.
2044 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
2045 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
2047 (read_ipa_call_summary): Add support for reading is_return_callee.
2048 (write_ipa_call_summary): Stream is_return_callee.
2049 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
2050 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
2051 ipa-prop.h, ipa-fnsummary.h.
2052 (pure_const_names): Change to static.
2053 (malloc_state_e): Define.
2054 (malloc_state_names): Define.
2055 (funct_state_d): Add field malloc_state.
2056 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
2057 (check_retval_uses): New function.
2058 (malloc_candidate_p): Likewise.
2059 (analyze_function): Add support for malloc attribute.
2060 (pure_const_write_summary): Stream malloc_state.
2061 (pure_const_read_summary): Add support for reading malloc_state.
2062 (dump_malloc_lattice): New function.
2063 (propagate_malloc): New function.
2064 (warn_function_malloc): New function.
2065 (ipa_pure_const::execute): Call propagate_malloc and
2066 ipa_free_fn_summary.
2067 (pass_local_pure_const::execute): Add support for malloc attribute.
2068 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
2069 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
2071 2017-10-27 Martin Liska <mliska@suse.cz>
2073 PR gcov-profile/82457
2074 * doc/invoke.texi: Document that one needs a non-strict ISO mode
2075 for fork-like functions to be properly instrumented.
2077 2017-10-27 Richard Biener <rguenther@suse.de>
2080 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
2081 info when we redirected EH.
2083 2017-10-26 Michael Collison <michael.collison@arm.com>
2085 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
2087 (<optab>_trunchf<GPI:mode>2: New pattern.
2088 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
2089 * config/aarch64/iterators.md (wv): New mode attribute.
2090 (vf, VF): New mode attributes.
2091 (vgp, VGP): New mode attributes.
2092 (s): Update attribute with SImode and DImode prefixes.
2094 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2096 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
2097 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
2098 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
2099 (nios2_option_overide): Initialize it. Don't allow R0-relative
2100 addressing with PIC.
2101 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
2102 (nios2_symbolic_constant_p): Likewise.
2103 (nios2_legitimate_address_p): Likewise.
2104 (nios2_r0rel_section_name_p): New.
2105 (nios2_symbol_ref_in_r0rel_data_p): New.
2106 (nios2_emit_move_sequence): Handle r0rel_constant_p.
2107 (r0rel_constant_p): New.
2108 (nios2_print_operand_address): Handle r0rel_constant_p.
2109 (nios2_cdx_narrow_form_p): Likewise.
2110 * config/nios2/nios2.opt (mr0rel-sec=): New option.
2111 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
2112 (Nios II Options): Document -mr0rel-sec.
2114 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2116 * config/nios2/nios2.c: Include xregex.h.
2117 (nios2_gprel_sec_regex): New.
2118 (nios2_option_overide): Initialize it. Don't allow GP-relative
2119 addressing with PIC.
2120 (nios2_small_section_name_p): Check for regex match.
2121 * config/nios2/nios2.opt (mgprel-sec=): New option.
2122 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
2123 (Nios II Options): Document -mgprel-sec.
2125 2017-10-26 Jim Wilson <wilson@tuliptree.org>
2127 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
2129 2017-10-26 Tom de Vries <tom@codesourcery.com>
2131 PR tree-optimization/82707
2132 * gimple.c (gimple_copy): Fix unsharing of
2133 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
2135 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
2137 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
2138 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
2139 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
2140 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
2141 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
2142 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
2143 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
2144 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
2145 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
2146 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
2147 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
2148 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
2149 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
2150 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
2151 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
2152 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
2153 _mm512_mask_cmpunord_ps_mask): New intrinsics.
2155 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2157 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
2159 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
2160 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
2161 warning to rs6000.c. Remove the Undocumented flag, since it has
2163 (-mabi=ibmlongdouble): Likewise.
2164 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
2165 already set, set the default format for long double.
2166 (rs6000_debug_reg_global): Print whether long double is IBM or
2168 (rs6000_option_override_internal): Rework setting long double
2169 format. Only warn if the user is changing the long double default
2170 and they did not use -Wno-psabi.
2171 * doc/invoke.texi (PowerPC options): Update the documentation for
2172 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
2174 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2175 Alan Hayward <alan.hayward@arm.com>
2176 David Sherwood <david.sherwood@arm.com>
2178 * rtl.h (wider_subreg_mode): New function.
2179 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
2180 rather than an unsigned int *.
2181 * ira-color.c (regno_max_ref_width): Replace with...
2182 (regno_max_ref_mode): ...this new variable.
2183 (coalesced_pseudo_reg_slot_compare): Update accordingly.
2184 Use wider_subreg_mode.
2185 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
2186 rather than an unsigned int *.
2187 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
2188 (process_alt_operands): Likewise.
2189 (invariant_p): Likewise.
2190 * lra-spills.c (assign_mem_slot): Likewise.
2191 (add_pseudo_to_slot): Likewise.
2192 * lra.c (collect_non_operand_hard_regs): Likewise.
2193 (add_regs_to_insn_regno_info): Likewise.
2194 * reload1.c (regno_max_ref_width): Replace with...
2195 (regno_max_ref_mode): ...this new variable.
2196 (reload): Update accordingly. Update call to
2197 ira_sort_regnos_for_alter_reg.
2198 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
2199 (init_eliminable_invariants): Update to use regno_max_ref_mode.
2200 (scan_paradoxical_subregs): Likewise.
2202 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2204 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
2205 (aarch64_frame): Add emit_frame_chain boolean.
2206 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2207 Move eh_return case to aarch64_layout_frame.
2208 (aarch64_layout_frame): Initialize emit_frame_chain.
2209 (aarch64_expand_prologue): Use emit_frame_chain.
2211 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2213 * config/aarch64/aarch64.c (aarch64_layout_frame):
2214 Ensure LR is always stored at the bottom of the callee-saves.
2215 Remove rarely used frame layout which saves callee-saves at top of
2216 frame, so the store of LR can be used as a valid probe in all cases.
2218 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2220 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
2221 Improve unaligned TImode/TFmode base/offset split.
2223 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2224 Alan Hayward <alan.hayward@arm.com>
2225 David Sherwood <david.sherwood@arm.com>
2227 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
2228 * combine.c (find_single_use_1): Likewise.
2229 (expand_field_assignment): Likewise.
2230 (move_deaths): Likewise.
2231 * lra-constraints.c (simplify_operand_subreg): Likewise.
2232 (curr_insn_transform): Likewise.
2233 * lra.c (collect_non_operand_hard_regs): Likewise.
2234 (add_regs_to_insn_regno_info): Likewise.
2235 * rtlanal.c (reg_referenced_p): Likewise.
2236 (covers_regno_no_parallel_p): Likewise.
2238 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2240 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
2241 Don't print any bits outside the precision of the value.
2242 * wide-int.cc (test_printing): Add some new tests.
2244 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2246 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
2247 supports -xbrace_comment option.
2248 * configure: Regenerate.
2249 * config.in: Regenerate.
2250 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
2251 (ASM_CPU_SPEC): Use it.
2253 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2255 * target.def (static_rtx_alignment): New hook.
2256 * targhooks.h (default_static_rtx_alignment): Declare.
2257 * targhooks.c (default_static_rtx_alignment): New function.
2258 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
2259 * doc/tm.texi: Regenerate.
2260 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
2261 instead of targetm.constant_alignment. Remove call to
2263 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2264 (cris_preferred_mininum_alignment): New function, split out from...
2265 (cris_constant_alignment): ...here.
2266 (cris_static_rtx_alignment): New function.
2267 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
2269 (ix86_constant_alignment): ...here.
2270 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2271 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2272 (mmix_static_rtx_alignment): New function.
2273 * config/spu/spu.c (spu_static_rtx_alignment): New function.
2274 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2276 2017-10-26 Tamar Christina <tamar.christina@arm.com>
2279 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
2280 Add flag_trapping_math and flag_fp_int_builtin_inexact.
2282 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
2284 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
2285 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
2286 mark as a sign-extending load.
2287 (local_pic_load_u): Define.
2289 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
2292 * fold-const.c (operand_equal_for_comparison_p): Also return true
2293 if ARG0 is a simple variant of ARG1 with narrower precision.
2294 (fold_ternary_loc): Always pass unstripped operands to the predicate.
2296 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
2298 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
2300 * i386.h (processor_costs): Add gather_static, gather_per_elt,
2301 scatter_static, scatter_per_elt.
2302 * x86-tune-costs.h: Add new cost entries.
2304 2017-10-25 Richard Biener <rguenther@suse.de>
2306 * tree-ssa-sccvn.h (vn_eliminate): Declare.
2307 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
2308 class pass_fre): Move to ...
2309 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
2310 class pass_fre): ... here and adjust for statistics.
2312 2017-10-25 Jakub Jelinek <jakub@redhat.com>
2315 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
2316 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
2317 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
2318 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
2321 2017-10-25 Richard Biener <rguenther@suse.de>
2323 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
2324 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
2325 eliminate_push_avail, eliminate_insert): Move inside...
2326 (class eliminate_dom_walker): ... this class in preparation
2328 (fini_eliminate): Remove by merging with ...
2329 (eliminate): ... this function. Adjust for class changes.
2330 (pass_pre::execute): Remove fini_eliminate call.
2331 (pass_fre::execute): Likewise.
2333 2017-10-24 Jakub Jelinek <jakub@redhat.com>
2336 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
2337 (VPERMI2, VPERMI2I): New mode iterators.
2338 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
2339 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
2341 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
2342 mode iterator. Remove 3 old define_insn patterns.
2343 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
2344 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
2345 VPERMI2 mode iterator, remove the other two expanders.
2346 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
2347 to use VPERMI2 mode iterator, add another alternative for vpermi2*
2348 instructions, remove the other two patterns.
2349 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
2350 mode iterator, remove the other two patterns.
2351 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
2352 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
2353 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
2354 and adjust argument order accordingly.
2355 (ix86_expand_vec_perm): Adjust caller.
2356 (expand_vec_perm_1): Likewise.
2357 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
2358 (expand_vec_perm_vpermt2_vpshub2): ... this.
2359 (ix86_expand_vec_perm_const_1): Adjust caller.
2360 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
2363 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
2364 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
2365 (vec_shl_<mode>): Remove unused expander.
2366 (avx512bw_<shift_insn><mode>3): New define_insn.
2367 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
2368 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
2370 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
2373 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
2376 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
2378 PR rtl-optimization/82396
2379 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
2380 (autopref_multipass_init): Simplify initialization.
2381 (autopref_rank_data): Simplify sort order.
2382 * gcc/sched-int.h (autopref_multipass_data_): Remove
2383 multi_mem_insn_p, min_offset and max_offset.
2385 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
2388 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2389 Check special value of flag_omit_frame_pointer.
2390 (aarch64_can_eliminate): Likewise.
2391 (aarch64_override_options_after_change_1): Simplify handling of
2392 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
2394 2017-10-24 Richard Biener <rguenther@suse.de>
2396 PR tree-optimization/82697
2397 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
2398 zero for conditional load and unconditional store.
2400 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
2402 * doc/install.texi: Document bootstrap-cet.
2404 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
2407 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
2408 ENDBR instruction at function entrance if function is only
2411 2017-10-24 Jakub Jelinek <jakub@redhat.com>
2414 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
2415 patterns to better describe from which operation the CF is computed.
2416 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
2417 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
2418 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
2419 is 0, use _0 suffixed expanders instead of emitting a comparison
2422 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
2424 * config/i386/i386.md(*movsf_internal, *movdf_internal):
2425 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
2427 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
2430 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
2431 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
2432 * loop-iv.c (iv_get_reaching_def): Likewise.
2433 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
2434 variable is promoted and the partition contains undefined values.
2436 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2438 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
2440 (nios2_address_cost): Define.
2441 (nios2_legitimize_address): Recognize (exp + constant) directly.
2442 (TARGET_ADDRESS_COST): Define.
2444 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2446 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
2447 (nios2_symbolic_memory_operand_p): Declare.
2448 (nios2_split_large_constant): Declare.
2449 (nios2_split_symbolic_memory_operand): Declare.
2450 * config/nios2/nios2.c: Adjust includes.
2451 (nios2_symbolic_constant_allowed): New.
2452 (nios2_symbolic_constant_p): New.
2453 (nios2_plus_symbolic_constant_p): New.
2454 (nios2_valid_addr_expr_p): Recognize addresses involving
2456 (nios2_legitimate_address_p): Likewise, also LO_SUM.
2457 (nios2_symbolic_memory_operand_p): New.
2458 (nios2_large_constant_p): New.
2459 (nios2_split_large_constant): New.
2460 (nios2_split_plus_large_constant): New.
2461 (nios2_split_symbolic_memory_operand): New.
2462 (nios2_legitimize_address): Code refactoring. Handle addresses
2463 involving symbolic constants.
2464 (nios2_emit_move_sequence): Likewise.
2465 (nios2_print_operand): Improve error output.
2466 (nios2_print_operand_address): Handle LO_SUM.
2467 (nios2_cdx_narrow_form_p): Likewise.
2468 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
2469 operands involving symbolic constants.
2470 (movhi_internal, movsi_internal): Likewise.
2471 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
2472 (extendhisi2, extendqi<mode>2): Likewise.
2474 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2476 * tree-pass.h (PROP_rtl_split_insns): Define.
2477 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
2479 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2481 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
2483 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2486 * target.def (const_not_ok_for_debug_p): Default to
2487 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
2488 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
2489 * targhooks.c (default_const_not_ok_for_debug_p): New function.
2490 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
2491 which targetm.const_not_ok_for_debug_p returned true.
2492 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
2494 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
2496 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
2497 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
2498 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
2500 (ix86_const_not_ok_for_debug_p): New function.
2501 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
2502 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
2504 2017-10-23 David Malcolm <dmalcolm@redhat.com>
2507 * system.h: Conditionally include "unique-ptr.h" if
2508 INCLUDE_UNIQUE_PTR is defined.
2509 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
2510 of defining INCLUDE_UNIQUE_PTR before including "system.h".
2512 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
2514 * config/rl78/rl78.md: New define_expand "subdi3".
2516 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
2519 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
2520 DF_REF_INSN if DF_REF_INSN_INFO is false.
2522 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
2524 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
2525 xmm_move instead of sse_move.
2526 (sse_store_index): New function.
2527 (ix86_register_move_cost): Be more sensible about mismatch stall;
2528 model AVX moves correctly; make difference between sse->integer and
2530 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
2531 moves; make difference between SSE and AVX.
2532 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
2533 and zmm_move. Increase size of sse load and store tables;
2534 add unaligned load and store tables; add ssemmx_to_integer.
2535 * x86-tune-costs.h: Update all entries according to real
2536 move latencies from Agner Fog's manual and chip documentation.
2538 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2541 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
2542 * config/i386/constraints.md (Wf): New constraint.
2543 * config/i386/i386.md (UNSPEC_SBB): New unspec.
2544 (cmp<dwi>_doubleword): Removed.
2545 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
2546 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
2547 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
2548 expand with cmp<dwi>_doubleword. For LTU and GEU use
2549 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
2551 * common.opt (gcolumn-info): Enable by default.
2552 * doc/invoke.texi (gcolumn-info): Document new default.
2554 2017-10-23 Richard Biener <rguenther@suse.de>
2556 PR tree-optimization/82672
2557 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
2558 Fold the stmt if we propagated into it.
2560 2017-10-23 Richard Biener <rguenther@suse.de>
2562 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
2563 (bitmap_remove_expr_from_set): ... this. All callers call this
2564 for non-constant values.
2565 (bitmap_set_subtract): Rename to...
2566 (bitmap_set_subtract_expressions): ... this. Adjust and
2568 (bitmap_set_contains_value): Remove superfluous check.
2569 (bitmap_set_replace_value): Inline into single caller ...
2570 (bitmap_value_replace_in_set): ... here and simplify.
2571 (dependent_clean): Merge into ...
2572 (clean): ... this using an overload. Adjust.
2573 (prune_clobbered_mems): Adjust.
2574 (compute_antic_aux): Likewise.
2575 (compute_partial_antic_aux): Likewise.
2577 2017-10-23 Richard Biener <rguenther@suse.de>
2579 PR tree-optimization/82129
2581 2017-08-01 Richard Biener <rguenther@suse.de>
2583 PR tree-optimization/81181
2584 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
2585 (compute_antic): ... end of iteration here.
2587 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2589 * target.def (starting_frame_offset): New hook.
2590 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
2591 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
2592 * doc/tm.texi.in: Regenerate.
2593 * hooks.h (hook_hwi_void_0): Declare.
2594 * hooks.c (hook_hwi_void_0): New function.
2595 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
2596 STARTING_FRAME_OFFSET.
2597 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
2598 * reload1.c (reload): Likewise.
2599 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
2600 instead of STARTING_FRAME_OFFSET.
2601 * function.c (try_fit_stack_local): Likewise.
2602 (assign_stack_local_1): Likewise
2603 (instantiate_virtual_regs): Likewise.
2604 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
2605 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
2606 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
2607 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
2608 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
2609 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
2610 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
2611 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
2612 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
2613 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
2614 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
2615 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
2616 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
2617 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
2618 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
2619 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
2620 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
2621 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
2622 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
2623 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
2624 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
2625 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
2626 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
2627 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
2628 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
2629 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
2630 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
2631 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
2632 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
2633 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
2634 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
2635 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
2636 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
2637 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
2638 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
2639 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
2640 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
2641 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
2642 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
2643 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
2644 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
2645 * config/avr/avr.c (avr_starting_frame_offset): Make static and
2646 return a HOST_WIDE_INT.
2647 (avr_builtin_setjmp_frame_value): Use it instead of
2648 STARTING_FRAME_OFFSET.
2649 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2650 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
2651 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
2653 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2654 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
2655 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
2656 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2657 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
2658 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
2659 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2660 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
2661 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
2662 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2663 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
2664 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
2666 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2667 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
2668 * config/mips/mips.c (mips_compute_frame_info): Refer to
2669 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
2670 (mips_starting_frame_offset): New function.
2671 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2672 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
2673 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
2674 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
2675 and return a HOST_WIDE_INT.
2676 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2677 (mmix_initial_elimination_offset): Refer to
2678 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
2679 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
2680 * config/pa/pa.c (pa_starting_frame_offset): New function.
2681 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
2682 (pa_expand_prologue): Likewise.
2683 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2684 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
2685 !FRAME_GROWS_DOWNWARD handling to...
2686 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2687 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
2688 !FRAME_GROWS_DOWNWARD handling to...
2689 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2690 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
2691 !FRAME_GROWS_DOWNWARD handling to...
2692 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2693 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
2695 (rs6000_starting_frame_offset): New function.
2696 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
2697 !FRAME_GROWS_DOWNWARD handling to...
2698 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2699 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
2700 !FRAME_GROWS_DOWNWARD handling to...
2701 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2702 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
2703 !FRAME_GROWS_DOWNWARD handling to...
2704 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2705 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
2706 (rs6000_starting_frame_offset): New function.
2707 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
2708 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
2709 * config/vax/vax.c (vax_starting_frame_offset): New function.
2710 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
2711 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2712 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
2713 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
2714 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2715 * system.h (STARTING_FRAME_OFFSET): Poison.
2717 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2719 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
2720 SCALAR_TYPE_MODE instead of TYPE_MODE.
2722 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2723 Alan Hayward <alan.hayward@arm.com>
2724 David Sherwood <david.sherwood@arm.com>
2726 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
2728 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2729 Alan Hayward <alan.hayward@arm.com>
2730 David Sherwood <david.sherwood@arm.com>
2732 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
2734 2017-10-23 Richard Biener <rguenther@suse.de>
2736 PR tree-optimization/82129
2737 * tree-ssa-pre.c (bitmap_set_and): Remove.
2738 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
2739 canonicalizing expressions in the set to those with lowest
2740 ID rather than taking that from the first edge.
2742 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2744 * combine.c (rtx_equal_for_field_assignment_p): Use
2745 byte_lowpart_offset.
2747 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2748 Alan Hayward <alan.hayward@arm.com>
2749 David Sherwood <david.sherwood@arm.com>
2751 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
2752 to a SUBREG_PROMOTED_VAR.
2754 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2755 Alan Hayward <alan.hayward@arm.com>
2756 David Sherwood <david.sherwood@arm.com>
2758 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
2759 (expand_debug_source_expr): Likewise.
2760 * combine.c (combine_simplify_rtx): Likewise.
2761 * cse.c (fold_rtx): Likewise.
2762 * optabs.c (expand_float): Likewise.
2763 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2764 (simplify_binary_operation_1): Likewise.
2766 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2767 Alan Hayward <alan.hayward@arm.com>
2768 David Sherwood <david.sherwood@arm.com>
2770 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
2771 (record_promoted_value): Likewise.
2772 * expr.c (expand_expr_real_2): Likewise.
2773 * ree.c (update_reg_equal_equiv_notes): Likewise.
2774 (combine_set_extension): Likewise.
2775 * rtlanal.c (low_bitmask_len): Likewise.
2776 * simplify-rtx.c (neg_const_int): Likewise.
2777 (simplify_binary_operation_1): Likewise.
2779 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2780 Alan Hayward <alan.hayward@arm.com>
2781 David Sherwood <david.sherwood@arm.com>
2783 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
2784 * regcprop.c (maybe_mode_change): Likewise.
2785 * reload1.c (alter_reg): Likewise.
2787 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2789 * inchash.h (inchash::hash::add_wide_int): New function.
2790 * lto-streamer-out.c (hash_tree): Use it.
2792 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2794 * inchash.h (inchash::hash::add_wide_int): Rename to...
2795 (inchash::hash::add_hwi): ...this.
2796 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
2797 (polymorphic_call_target_hasher::hash): Likewise.
2798 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
2799 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
2800 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
2801 * lto-streamer-out.c (hash_tree): Likewise.
2802 * optc-save-gen.awk: Likewise.
2803 * tree.c (add_expr): Likewise.
2805 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
2808 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
2809 for ordered inequality comparisons even with TARGET_IEEE_FP.
2811 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
2814 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
2815 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
2816 Expand with cmp<dwi>_doubleword.
2818 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2820 * extend.texi: Add x86 specific to 'nocf_check' attribute.
2821 List CET intrinsics.
2822 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
2823 specific to -fcf-protection option.
2825 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2827 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
2828 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
2829 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
2830 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
2831 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
2832 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
2833 (extra_objs): Add cet.o for Linux/x86 targets.
2834 (tmake_file): Add i386/t-cet for Linux/x86 targets.
2835 * config/i386/cet.c: New file.
2836 * config/i386/cetintrin.h: Likewise.
2837 * config/i386/t-cet: Likewise.
2838 * config/i386/cpuid.h (bit_SHSTK): New.
2839 (bit_IBT): Likewise.
2840 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
2841 pass IBT and SHSTK bits.
2842 * config/i386/i386-builtin-types.def
2843 (VOID_FTYPE_UNSIGNED_PVOID): New.
2844 (VOID_FTYPE_UINT64_PVOID): Likewise.
2845 * config/i386/i386-builtin.def: Add CET intrinsics.
2846 * config/i386/i386-c.c (ix86_target_macros_internal): Add
2847 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
2848 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
2849 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
2851 * config/i386/i386.c (rest_of_insert_endbranch): New.
2852 (pass_data_insert_endbranch): Likewise.
2853 (pass_insert_endbranch): Likewise.
2854 (make_pass_insert_endbranch): Likewise.
2855 (ix86_notrack_prefixed_insn_p): Likewise.
2856 (ix86_target_string): Add -mibt, -mshstk flags.
2857 (ix86_option_override_internal): Add flag_cf_protection
2859 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
2860 (ix86_print_operand): Add 'notrack' prefix output.
2861 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
2862 (ix86_expand_builtin): Expand CET intrinsics.
2863 (x86_output_mi_thunk): Add 'endbranch' instruction.
2864 * config/i386/i386.h (TARGET_IBT): New.
2865 (TARGET_IBT_P): Likewise.
2866 (TARGET_SHSTK): Likewise.
2867 (TARGET_SHSTK_P): Likewise.
2868 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
2869 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
2870 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
2871 (builtin_setjmp_setup): New pattern.
2872 (builtin_longjmp): Likewise.
2873 (rdssp<mode>): Likewise.
2874 (incssp<mode>): Likewise.
2875 (saveprevssp): Likewise.
2876 (rstorssp): Likewise.
2877 (wrss<mode>): Likewise.
2878 (wruss<mode>): Likewise.
2879 (setssbsy): Likewise.
2880 (clrssbsy): Likewise.
2881 (nop_endbr): Likewise.
2882 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
2884 * config/i386/immintrin.h: Include <cetintrin.h>.
2885 * config/i386/linux-common.h
2886 (file_end_indicate_exec_stack_and_cet): New prototype.
2887 (TARGET_ASM_FILE_END): New.
2889 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
2891 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
2892 latencies instead of having separate table; make difference between
2893 integer and float costs.
2894 * i386.h (processor_costs): Remove scalar_stmt_cost,
2895 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
2896 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
2898 * x86-tune-costs.h: Remove entries which has been removed in
2899 procesor_costs from all tables; make cond_taken_branch_cost
2900 and cond_not_taken_branch_cost COST_N_INSNS based.
2902 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
2904 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
2906 2017-10-20 Jakub Jelinek <jakub@redhat.com>
2908 * config/i386/i386.md (isa): Remove fma_avx512f.
2909 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
2910 <avx512>_fmadd_<mode>_mask3<round_name>,
2911 <avx512>_fmsub_<mode>_mask<round_name>,
2912 <avx512>_fmsub_<mode>_mask3<round_name>,
2913 <avx512>_fnmadd_<mode>_mask<round_name>,
2914 <avx512>_fnmadd_<mode>_mask3<round_name>,
2915 <avx512>_fnmsub_<mode>_mask<round_name>,
2916 <avx512>_fnmsub_<mode>_mask3<round_name>,
2917 <avx512>_fmaddsub_<mode>_mask<round_name>,
2918 <avx512>_fmaddsub_<mode>_mask3<round_name>,
2919 <avx512>_fmsubadd_<mode>_mask<round_name>,
2920 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
2921 (*vec_widen_umult_even_v16si<mask_name>,
2922 *vec_widen_smult_even_v16si<mask_name>): Likewise.
2923 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
2925 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2927 * extend.texi: Add 'nocf_check' documentation.
2928 * gimple.texi: Add second parameter to
2929 gimple_build_call_from_tree.
2930 * invoke.texi: Add -fcf-protection documentation.
2931 * rtl.texi: Add REG_CALL_NOTRACK documenation.
2933 2017-10-20 Richard Biener <rguenther@suse.de>
2935 PR tree-optimization/82473
2936 * tree-vect-loop.c (vectorizable_reduction): Properly get at
2937 the largest input type.
2939 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2941 * c-attribs.c (handle_nocf_check_attribute): New function.
2942 (c_common_attribute_table): Add 'nocf_check' handling.
2943 * gimple-parser.c: Add second argument NULL to
2944 gimple_build_call_from_tree.
2945 * attrib.c (comp_type_attributes): Check nocf_check attribute.
2946 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
2948 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
2949 * common.opt: Add fcf-protection flag.
2950 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
2951 * flag-types.h: Add enum cf_protection_level.
2952 * gimple.c (gimple_build_call_from_tree): Add second parameter.
2953 Add 'nocf_check' attribute propagation to gimple call.
2954 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
2955 (gimple_build_call_from_tree): Update prototype.
2956 (gimple_call_nocf_check_p): New function.
2957 (gimple_call_set_nocf_check): Likewise.
2958 * gimplify.c: Add second argument to gimple_build_call_from_tree.
2959 * ipa-icf.c: Add nocf_check attribute in statement hash.
2960 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
2961 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
2962 * toplev.c (process_options): Add flag_cf_protection handling.
2964 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
2966 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
2968 2017-10-20 Richard Biener <rguenther@suse.de>
2970 PR tree-optimization/82603
2971 * tree-if-conv.c (predicate_mem_writes): Make sure to only
2972 remove false predicated stores.
2974 2017-10-20 Richard Biener <rguenther@suse.de>
2976 * graphite-isl-ast-to-gimple.c
2977 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
2978 Remove return value and simplify, dump copied stmt after lhs
2980 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
2981 Reduce dump verbosity.
2982 (gsi_insert_earliest): Likewise.
2983 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
2984 * graphite.c (print_global_statistics): Adjust dumping.
2985 (print_graphite_scop_statistics): Likewise.
2986 (print_graphite_statistics): Do not dump loops here.
2987 (graphite_transform_loops): But here.
2989 2017-10-20 Nicolas Roche <roche@adacore.com>
2991 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
2992 * configure: Regenerate.
2994 2017-10-20 Jakub Jelinek <jakub@redhat.com>
2997 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
2998 functions when optimizing replace GIMPLE_RETURN stmts with
2999 calls to __builtin_unreachable ().
3002 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
3003 for -fsanitize=thread link of executables.
3004 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
3005 link of executables.
3008 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
3010 (<shift_insn><mode>3<mask_name>): Change the last of the 3
3011 define_insns for logical vector shifts to use VI248_AVX512BW
3012 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
3013 condition, useless isa and prefix attributes. Change the first
3014 2 of these define_insns to ...
3015 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
3016 define_insn for avx512vl.
3017 (<shift_insn><mode>3): ... and this, new define_insn without
3018 masking for non-avx512vl.
3021 * config/i386/sse.md (*andnot<mode>3,
3022 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
3023 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
3024 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
3025 not applied use empty suffix even for TARGET_AVX512VL.
3026 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
3027 is applied, supply evex,evex or evex,evex,evex instead of just
3030 2017-10-20 Julia Koval <julia.koval@intel.com>
3032 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
3033 (OPTION_MASK_ISA_GFNI_UNSET): New.
3034 (ix86_handle_option): Handle OPT_mgfni.
3035 * config/i386/cpuid.h (bit_GFNI): New.
3036 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
3037 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
3038 * config/i386/i386.c (ix86_target_string): Add -mgfni.
3039 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
3040 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
3041 * config/i386/i386.opt: Add mgfni.
3043 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
3045 * config/msp430/msp430.c (msp430_option_override): Disable
3046 -fdelete-null-pointer-checks.
3047 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
3049 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3051 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
3052 of x87 and SSE instructions.
3054 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3056 * asan.c (create_cond_insert_point): Do not update edge count.
3057 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
3058 (afdo_propagate_circuit): Likewise.
3059 (afdo_calculate_branch_prob): Likewise.
3060 (afdo_annotate_cfg): Likewise.
3061 * basic-block.h (struct edge_def): Remove count.
3062 (edge_def::count): New accessor.
3063 * bb-reorder.c (rotate_loop): Update.
3064 (find_traces_1_round): Update.
3065 (connect_traces): Update.
3066 (sanitize_hot_paths): Update.
3067 * cfg.c (unchecked_make_edge): Update.
3068 (make_single_succ_edge): Update.
3069 (check_bb_profile): Update.
3070 (dump_edge_info): Update.
3071 (update_bb_profile_for_threading): Update.
3072 (scale_bbs_frequencies_int): Update.
3073 (scale_bbs_frequencies_gcov_type): Update.
3074 (scale_bbs_frequencies_profile_count): Update.
3075 (scale_bbs_frequencies): Update.
3076 * cfganal.c (connect_infinite_loops_to_exit): Update.
3077 * cfgbuild.c (compute_outgoing_frequencies): Update.
3078 (find_many_sub_basic_blocks): Update.
3079 * cfgcleanup.c (try_forward_edges): Update.
3080 (try_crossjump_to_edge): Update
3081 * cfgexpand.c (expand_gimple_cond): Update
3082 (expand_gimple_tailcall): Update
3083 (construct_exit_block): Update
3084 * cfghooks.c (verify_flow_info): Update
3085 (redirect_edge_succ_nodup): Update
3086 (split_edge): Update
3087 (make_forwarder_block): Update
3088 (duplicate_block): Update
3089 (account_profile_record): Update
3090 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
3091 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
3092 * cfgloopmanip.c (scale_loop_profile): Update.
3094 (lv_adjust_loop_entry_edge): Update.
3095 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
3096 (force_nonfallthru_and_redirect): Update.
3097 (purge_dead_edges): Update.
3098 (rtl_flow_call_edges_add): Update.
3099 * cgraphunit.c (init_lowered_empty_function): Update.
3100 (cgraph_node::expand_thunk): Update.
3101 * gimple-pretty-print.c (dump_probability): Update.
3102 (dump_edge_probability): Update.
3103 * gimple-ssa-isolate-paths.c (isolate_path): Update.
3104 * haifa-sched.c (sched_create_recovery_edges): Update.
3105 * hsa-gen.c (convert_switch_statements): Update.
3106 * ifcvt.c (dead_or_predicable): Update.
3107 * ipa-inline-transform.c (inline_transform): Update.
3108 * ipa-split.c (split_function): Update.
3109 * ipa-utils.c (ipa_merge_profiles): Update.
3110 * loop-doloop.c (add_test): Update.
3111 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
3112 * lto-streamer-in.c (input_cfg): Update.
3113 (input_function): Update.
3114 * lto-streamer-out.c (output_cfg): Update.
3115 * modulo-sched.c (sms_schedule): Update.
3116 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
3117 * predict.c (maybe_hot_edge_p): Update.
3118 (unlikely_executed_edge_p): Update.
3119 (probably_never_executed_edge_p): Update.
3120 (dump_prediction): Update.
3121 (drop_profile): Update.
3122 (propagate_unlikely_bbs_forward): Update.
3123 (determine_unlikely_bbs): Update.
3124 (force_edge_cold): Update.
3125 * profile.c (compute_branch_probabilities): Update.
3126 * reg-stack.c (better_edge): Update.
3127 * shrink-wrap.c (handle_simple_exit): Update.
3128 * tracer.c (better_p): Update.
3129 * trans-mem.c (expand_transaction): Update.
3130 (split_bb_make_tm_edge): Update.
3131 * tree-call-cdce.c: Update.
3132 * tree-cfg.c (gimple_find_sub_bbs): Update.
3133 (gimple_split_edge): Update.
3134 (gimple_duplicate_sese_region): Update.
3135 (gimple_duplicate_sese_tail): Update.
3136 (gimple_flow_call_edges_add): Update.
3137 (insert_cond_bb): Update.
3138 (execute_fixup_cfg): Update.
3139 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
3140 * tree-complex.c (expand_complex_div_wide): Update.
3141 * tree-eh.c (lower_resx): Update.
3142 (unsplit_eh): Update.
3143 (cleanup_empty_eh_move_lp): Update.
3144 * tree-inline.c (copy_edges_for_bb): Update.
3145 (freqs_to_counts): Update.
3146 (copy_cfg_body): Update.
3147 * tree-ssa-dce.c (remove_dead_stmt): Update.
3148 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
3149 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
3150 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
3151 (unloop_loops): Update.
3152 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3153 * tree-ssa-loop-split.c (connect_loops): Update.
3154 (split_loop): Update.
3155 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
3156 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
3157 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
3158 * tree-ssa-reassoc.c (branch_fixup): Update.
3159 * tree-ssa-tail-merge.c (replace_block_by): Update.
3160 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
3161 (compute_path_counts): Update.
3162 (update_profile): Update.
3163 (recompute_probabilities): Update.
3164 (update_joiner_offpath_counts): Update.
3165 (estimated_freqs_path): Update.
3166 (freqs_to_counts_path): Update.
3167 (clear_counts_path): Update.
3168 (ssa_fix_duplicate_block_edges): Update.
3169 (duplicate_thread_path): Update.
3170 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
3171 (case_bit_test_cmp): Update.
3172 (collect_switch_conv_info): Update.
3173 (gen_inbound_check): Update.
3174 (do_jump_if_equal): Update.
3175 (emit_cmp_and_jump_insns): Update.
3176 * tree-tailcall.c (decrease_profile): Update.
3177 (eliminate_tail_call): Update.
3178 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
3179 (vect_do_peeling): Update.
3180 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3181 * ubsan.c (ubsan_expand_null_ifn): Update.
3182 (ubsan_expand_ptr_ifn): Update.
3183 * value-prof.c (gimple_divmod_fixed_value): Update.
3184 (gimple_mod_pow2): Update.
3185 (gimple_mod_subtract): Update.
3186 (gimple_ic): Update.
3187 (gimple_stringop_fixed_value): Update.
3189 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3192 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
3194 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
3196 PR rtl-optimization/82395
3197 * ira-color.c (allocno_priority_compare_func): Fix comparison step
3198 based on non_spilled_static_chain_regno_p.
3200 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3202 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
3203 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
3204 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
3206 2017-10-19 Martin Sebor <msebor@redhat.com>
3208 PR tree-optimization/82596
3209 * tree.c (array_at_struct_end_p): Handle STRING_CST.
3211 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3213 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
3214 (get_mem_refs_of_builtin_call): Likewise.
3215 * builtins.c (expand_builtin_apply): Adjust call to
3216 allocate_dynamic_stack_space.
3217 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
3218 the third argument to allocate_dynamic_stack_space, otherwise -1.
3219 (expand_builtin): Deal with all alloca variants.
3220 (is_inexpensive_builtin): Likewise.
3221 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
3222 * calls.c (special_function_p): Deal with all alloca variants.
3223 (initialize_argument_information): Adjust call to
3224 allocate_dynamic_stack_space.
3225 (expand_call): Likewise.
3226 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
3227 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
3228 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
3229 use it for the stack usage computation.
3230 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
3231 * function.c (gimplify_parameters): Call build_alloca_call_expr.
3232 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
3233 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
3234 (in_loop_p): Remove first argument and useless check.
3235 (pass_walloca::execute): Remove useless test and adjust call to above.
3236 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
3237 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
3238 (gimplify_call_expr): Deal with all alloca variants.
3239 * hsa-gen.c (gen_hsa_alloca): Likewise.
3240 (gen_hsa_insns_for_call): Likewise.
3241 * ipa-pure-const.c (special_builtin_state): Likewise.
3242 * tree-chkp.c (chkp_build_returned_bound): Likewise.
3243 * tree-object-size.c (alloc_object_size): Likewise.
3244 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
3245 (call_may_clobber_ref_p_1): Likewise.
3246 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
3247 (ccp_fold_stmt): Likewise.
3248 (optimize_stack_restore): Likewise.
3249 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
3250 (mark_all_reaching_defs_necessary_1): Likewise.
3251 (propagate_necessity): Likewise.
3252 (eliminate_unnecessary_stmts): Likewise.
3253 * tree.c (build_common_builtin_nodes): Build
3254 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
3255 (build_alloca_call_expr): New function.
3256 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
3257 (CASE_BUILT_IN_ALLOCA): Likewise.
3258 (build_alloca_call_expr): Declare.
3259 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
3261 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3264 * dwarf2out.c (new_die_raw): New static inline function.
3265 (new_die): Use it to create the DIE.
3266 (add_AT_external_die_ref): Likewise.
3267 (clone_die): Likewise.
3268 (clone_as_declaration): Likewise.
3269 (dwarf2out_vms_debug_main_pointer): Likewise.
3270 (base_type_die): Likewise. Remove early return for corner cases.
3271 Do not call add_pubtype on the DIE here.
3272 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
3273 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
3274 typedefs for base types with DW_AT_endianity. Make sure a DIE with
3275 native order exists for base types, attach the DIE manually and call
3276 add_pubtype on it. Do not equate a reverse order DIE to the type.
3278 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
3280 * config/arm/arm.c (align_ok_ldrd_strd): New function.
3281 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
3283 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
3285 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3287 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
3288 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
3289 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
3290 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
3291 * opts.c (sanitizer_opts): Add builtin.
3292 * ubsan.c (instrument_builtin): New function.
3293 (pass_ubsan::execute): Call it.
3294 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
3295 * doc/invoke.texi: Document -fsanitize=builtin.
3297 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
3298 builtins, store max (log2 (align), 0) into uchar field instead of
3299 align into uptr field.
3300 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
3301 store uchar 0 field instead of uptr 0 field.
3302 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
3303 instead of passing one address of struct with 2 locations pass
3304 two addresses of structs with 1 location each.
3305 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
3306 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
3307 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
3308 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
3309 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
3310 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
3311 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
3312 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
3314 2017-10-19 Martin Liska <mliska@suse.cz>
3317 * file-find.c (remove_prefix): Remove.
3318 * file-find.h (remove_prefix): Likewise.
3319 * gcc-ar.c: Remove smartness of lookup.
3321 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
3323 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
3324 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
3325 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
3327 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3330 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
3331 (setcc + and to xor + setcc): New peephole2.
3333 2017-10-19 Tom de Vries <tom@codesourcery.com>
3335 * doc/sourcebuild.texi (Test Directives, Variants of
3336 dg-require-support): Add dg-require-stack-size.
3338 2017-10-19 Martin Liska <mliska@suse.cz>
3341 * gimplify.c (gimplify_decl_expr): Do not instrument variables
3342 that have a large alignment.
3343 (gimplify_target_expr): Likewise.
3345 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
3347 PR rtl-optimization/82602
3348 * ira.c (rtx_moveable_p): Return false for volatile asm.
3350 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
3353 * config/i386/i386-modes.def (CCGZ): New CC mode.
3354 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
3355 * config/i386/predicates.md (ix86_comparison_operator):
3357 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
3358 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
3359 with double-word subtraction.
3360 (put_condition_code): Handle CCGZmode.
3362 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
3364 * wide-int.cc (debug (const wide_int &)): New.
3365 (debug (const wide_int *)): New.
3366 (debug (const widest_int &)): New.
3367 (debug (const widest_int *)): New.
3369 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
3372 * lra-constraints.c (curr_insn_transform): Use non-input operand
3373 instead of output one for matched reload.
3375 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3377 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
3378 (tree-ssa-loop-ivopts.h): New header file.
3379 (struct builtin_info): New fields.
3380 (classify_builtin_1): Compute and record base and offset parts for
3381 memset builtin partition by calling strip_offset.
3382 (offset_cmp, fuse_memset_builtins): New functions.
3383 (finalize_partitions): Fuse adjacent memset partitions by calling
3385 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
3386 Expose the interface.
3387 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
3389 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3391 PR tree-optimization/82574
3392 * tree-loop-distribution.c (find_single_drs): New parameter. Check
3393 that data reference must be executed exactly once per iteration
3394 against the outermost loop in nest.
3395 (classify_partition): Update call to above function.
3397 2017-10-18 Richard Biener <rguenther@suse.de>
3399 PR tree-optimization/82591
3400 * graphite.c (graphite_transform_loops): Move code gen message
3402 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3403 Here. Handle scop_to_isl_ast failing.
3404 (scop_to_isl_ast): Limit the number of ISL operations.
3406 2017-10-18 Richard Biener <rguenther@suse.de>
3408 * graphite-isl-ast-to-gimple.c
3409 (translate_isl_ast_to_gimple::set_rename): Simplify.
3410 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
3411 (graphite_copy_stmts_from_block): ... here.
3412 (copy_bb_and_scalar_dependences): Simplify.
3413 (add_parameters_to_ivs_params): Canonicalize.
3414 (generate_entry_out_of_ssa_copies): Simplify.
3415 * graphite-sese-to-poly.c (extract_affine_name): Simplify
3416 by passing in ISL dimension.
3417 (parameter_index_in_region_1): Rename to ...
3418 (parameter_index_in_region): ... this.
3419 (extract_affine): Adjust assert, pass down parameter index.
3420 (add_param_constraints): Use range-info when available.
3421 (build_scop_context): Adjust.
3422 * sese.c (new_sese_info): Adjust.
3423 (free_sese_info): Likewise.
3424 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
3425 Remove unused typedefs.
3426 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
3428 2017-10-18 Martin Liska <mliska@suse.cz>
3430 * combine.c (simplify_compare_const): Add gcc_fallthrough.
3432 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
3434 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
3435 (s390_sched_init): Do not reset s390_sched_state if we entered the
3436 current basic block via a fallthru edge and all others are unlikely.
3438 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
3440 * config/s390/s390.c (NUM_SIDES): New variable.
3441 (LONGRUNNING_THRESHOLD): New variable.
3442 (LATENCY_FACTOR): New variable.
3443 (s390_sched_score): Decrease score for long-running instructions on
3445 (s390_sched_variable_issue): Perform bookkeeping for long-running
3448 2017-10-18 Richard Biener <rguenther@suse.de>
3450 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
3451 Simplify with removal of the parameter rename map.
3452 (set_rename): Likewise.
3453 (should_copy_to_new_region): Likewise.
3454 (graphite_copy_stmts_from_block): Likewise.
3455 (copy_bb_and_scalar_dependences): Remove initialization of
3456 unused copied_bb_map.
3458 (copy_internal_parameters): Likewise.
3459 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
3460 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
3461 Use INTEGRAL_TYPE_P.
3462 (parameter_index_in_region_1): Rename to ...
3463 (assign_parameter_index_in_region): ... this. Assert we have
3464 a parameter we handle.
3465 (scan_tree_for_params): Adjust.
3466 * sese.h (parameter_rename_map_t): Remove.
3467 (struct sese_info_t): Remove unused parameter_rename_map and
3468 copied_bb_map members.
3469 * sese.c (new_sese_info): Adjust.
3470 (free_sese_info): Likewise.
3472 2017-10-18 Martin Liska <mliska@suse.cz>
3475 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
3476 on an abnormal edge.
3478 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
3480 * doc/invoke.texi (ffunction-sections and fdata-sections):
3483 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3485 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
3486 the use statement can throw internally.
3488 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3490 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
3491 any RTX present on the RHS of a SET.
3492 * compare-elim.c (try_eliminate_compare): Restore comment.
3494 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3496 * langhooks.h (struct lang_hooks): Document that tree_size langhook
3497 may be also called on tcc_type nodes.
3498 * langhooks.c (lhd_tree_size): Likewise.
3500 2017-10-17 David Malcolm <dmalcolm@redhat.com>
3502 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
3503 format_warning_at_substring.
3504 (maybe_warn): Convert source_range * param to a location_t. Pass
3505 UNKNOWN_LOCATION rather than NULL to fmtwarn.
3506 (format_directive): Remove code to extract source_ranges and
3507 source_range * in favor of just a location_t.
3508 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
3510 * substring-locations.c (format_warning_va): Convert
3511 source_range * param to a location_t.
3512 (format_warning_at_substring): Likewise.
3513 * substring-locations.h (format_warning_va): Likewise.
3514 (format_warning_at_substring): Likewise.
3516 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
3518 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
3520 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
3521 and scatter/gather ops.
3523 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
3524 vec_gather_load and vec_scatter_store.
3525 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
3526 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
3527 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
3528 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
3529 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
3530 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
3532 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
3534 * reg-stack.c (compare_for_stack_reg): Add bool argument.
3535 Detect FTST instruction and handle its register pops. Only pop
3536 second operand if can_pop_second_op is true.
3537 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
3538 set can_pop_second_op to false in the compare_for_stack_reg call.
3540 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
3541 output_fp_compare for stack register operands.
3542 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
3543 instructions here. Do not emit stack register pops here. Assert
3544 that FCOMPP pops next to top stack register. Rewrite function.
3546 2017-10-17 Nathan Sidwell <nathan@acm.org>
3549 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
3550 use DECL_ASSEMBLER_NAME_RAW.
3553 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
3556 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
3557 Wilco Dijkstra <wilco.dijkstra@arm.com>
3559 * builtins.c (expand_builtin_update_setjmp_buf): Add a
3560 converstion to Pmode from the buf_addr.
3562 2017-10-17 Richard Biener <rguenther@suse.de>
3564 * graphite-dependences.c (scop_get_reads_and_writes): Change
3565 output parameters to references.
3567 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
3569 PR 71026/tree-optimization
3570 * fold-const.c (distribute_real_division): Removed.
3571 (fold_binary_loc): Remove calls to distribute_real_divison.
3573 2017-10-17 Richard Biener <rguenther@suse.de>
3575 * graphite-scop-detection.c
3576 (scop_detection::stmt_has_simple_data_refs_p): Always use
3577 the full nest as region.
3578 (try_generate_gimple_bb): Likewise.
3579 * sese.c (scalar_evolution_in_region): Simplify now that
3580 SCEV can handle instantiation in regions.
3581 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
3582 in the non-loop part of a function if requested.
3584 2017-10-17 Richard Biener <rguenther@suse.de>
3586 PR tree-optimization/82563
3587 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
3589 (graphite_regenerate_ast_isl): Call it.
3590 * graphite-scop-detection.c (build_scops): Remove entry edge split.
3592 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3594 PR tree-optimization/82549
3595 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
3596 Formatting fixes. Instead of calling make_bit_field_ref with negative
3599 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
3601 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
3602 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
3603 _mm_maskz_reduce_ss): New.
3604 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
3605 __builtin_ia32_reducess_mask): Ditto..
3606 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
3607 * config/i386/sse.md (reduces<mode>): Renamed to ...
3608 (reduces<mode><mask_scalar_name>): ... this.
3609 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
3611 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
3612 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
3614 2017-10-16 David Malcolm <dmalcolm@redhat.com>
3616 * Makefile.in (OBJS): Add unique-ptr-tests.o.
3617 * selftest-run-tests.c (selftest::run_tests): Call
3618 selftest::unique_ptr_tests_cc_tests.
3619 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
3620 * unique-ptr-tests.cc: New file.
3622 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
3625 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
3627 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
3628 (make_hard_regno_born, make_hard_regno_dead): Update
3629 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
3631 2017-10-16 Jeff Law <law@redhat.com>
3633 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
3635 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
3637 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
3639 2017-10-16 Olivier Hainque <hainque@adacore.com>
3641 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
3642 with_cpu if we were configured for an e500v2 target cpu name.
3644 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
3646 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
3647 * doc/invoke.texi: Document +nodsp as a valid extension for
3650 2017-10-16 Martin Liska <mliska@suse.cz>
3652 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
3653 (test_set_range): Likewise.
3654 (test_range_functions): Rename to ...
3655 (test_bit_in_range): ... this.
3656 (sbitmap_c_tests): Add new test.
3658 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3660 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
3662 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
3663 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
3665 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3667 * config/aarch64/aarch64-builtins.c
3668 (aarch64_types_quadopu_lane_qualifiers): New.
3669 (TYPES_QUADOPU_LANE): New.
3670 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
3671 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
3672 (aarch64_<sur>dot_laneq<vsi2qi>): New.
3673 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
3674 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
3675 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
3676 (Vdottype, DOTPROD): New.
3677 (sur): Add SDOT and UDOT.
3679 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3681 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
3682 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
3683 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
3685 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
3686 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
3687 Enable TARGET_DOTPROD.
3688 (cortex-a75.cortex-a55): Likewise.
3689 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
3691 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3693 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
3694 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
3696 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
3698 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
3699 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
3700 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
3701 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
3702 * config/arm/types.md (neon_dot, neon_dot_q): New.
3703 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
3705 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3707 * config/arm/arm.h (TARGET_DOTPROD): New.
3708 * config/arm/arm.c (arm_arch_dotprod): New.
3709 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
3710 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
3711 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
3712 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
3713 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
3714 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
3715 * doc/invoke.texi (armv8.2-a): Document dotprod
3717 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
3719 * i386.c (ix86_vec_cost): New function.
3720 (ix86_rtx_costs): Handle vector operations better.
3721 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
3722 * x86-tune-costs.h: Add new costs to all tables.
3724 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
3726 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
3728 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
3729 divsd, sqrtss and sqrtsd
3730 * x86-tune-costs.h: Add new entries to all costs.
3731 (znver1_cost): Fix to match real instruction latencies.
3733 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3734 Michael Collison <michael.collison@arm.com>
3736 * compare-elim.c: Include emit-rtl.h.
3737 (can_merge_compare_into_arith): New function.
3738 (try_validate_parallel): Likewise.
3739 (try_merge_compare): Likewise.
3740 (try_eliminate_compare): Call the above when no previous clobber
3742 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
3745 2017-10-14 Jakub Jelinek <jakub@redhat.com>
3749 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
3750 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
3754 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
3755 to be any operand_equal_p operands. For & (B - 1) require
3756 B to be power of 2. Recognize
3757 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
3759 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3762 * optabs.c (expand_memory_blockage): Fix call of
3763 targetm.have_memory_blockage.
3765 2017-10-14 Jakub Jelinek <jakub@redhat.com>
3768 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
3769 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
3770 objects to extra_objs instead of overwriting it.
3772 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3774 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
3775 Use any_fp_register_operand as operand[3] predicate. Simplify
3776 equality test for operands[2] and operands[4] memory location.
3777 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
3778 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
3779 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
3780 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
3781 any_fp_register_operand as operand[1] predicate. Simplify
3782 equality test for operands[0] and operands[3] memory location.
3783 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
3784 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
3785 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
3787 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3789 * target-insns.def: Add memory_blockage.
3790 * optabs.c (expand_memory_blockage): New function.
3791 (expand_asm_memory_barrier): Rename ...
3792 (expand_asm_memory_blockage): ... to this.
3793 (expand_mem_thread_fence): Call expand_memory_blockage
3794 instead of expand_asm_memory_barrier.
3795 (expand_mem_singnal_fence): Ditto.
3796 (expand_atomic_load): Ditto.
3797 (expand_atomic_store): Ditto.
3798 * doc/md.texi (Standard Pattern Names For Generation):
3799 Document memory_blockage instruction pattern.
3801 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
3803 * config/rl78/rl78.c (rl78_emit_libcall): New function.
3804 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
3805 * config/rl78/rl78.md: New define_expand "adddi3".
3807 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
3809 * cfghooks.c (verify_flow_info): Disable check that all probabilities
3812 2017-10-13 Jeff Law <law@redhat.com>
3814 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
3816 2017-10-13 Jakub Jelinek <jakub@redhat.com>
3819 * internal-fn.c (expand_mul_overflow): If both operands have
3820 the same highpart of -1 or 0 and the topmost bit of lowpart
3821 is different, overflow is if res <= 0 rather than res < 0.
3823 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
3825 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
3826 TARGET_P9_VECTOR code for unaligned_load case.
3828 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
3830 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
3832 2017-10-13 Nathan Sidwell <nathan@acm.org>
3834 * tree-core.h (tree_contains_struct): Make bool.
3835 * tree.c (tree_contains_struct): Likewise.
3836 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
3837 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
3838 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
3839 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
3840 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
3842 2017-10-13 Richard Biener <rguenther@suse.de>
3844 * graphite-isl-ast-to-gimple.c
3845 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
3846 parameters and dominance check.
3847 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
3848 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
3849 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
3850 Do not update SSA form here or do intermediate IL verification.
3851 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
3852 (graphite_initialize): Remove check on the number of loops in
3853 the function and inline into graphite_transform_loops.
3854 (graphite_finalize): Inline into graphite_transform_loops.
3855 (graphite_transform_loops): Perform SSA update and IL verification
3857 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
3859 2017-10-13 Richard Biener <rguenther@suse.de>
3861 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
3862 graphite_expression_type_precision): Avoid global constructor
3864 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
3865 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
3866 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
3867 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
3868 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
3869 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
3871 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
3874 * config/i386/i386.h (ix86_red_zone_size): New.
3875 * config/i386/i386.md (push peephole2s): Replace
3876 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
3878 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3879 Alan Hayward <alan.hayward@arm.com>
3880 David Sherwood <david.sherwood@arm.com>
3882 * combine.c (can_change_dest_mode): Reject changes in
3883 REGMODE_NATURAL_SIZE.
3885 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3886 Alan Hayward <alan.hayward@arm.com>
3887 David Sherwood <david.sherwood@arm.com>
3889 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
3890 (expand_debug_source_expr): Likewise.
3891 * combine.c (combine_simplify_rtx): Likewise.
3892 * cse.c (fold_rtx): Likewise.
3893 * fwprop.c (canonicalize_address): Likewise.
3894 * targhooks.c (default_shift_truncation_mask): Likewise.
3896 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3897 Alan Hayward <alan.hayward@arm.com>
3898 David Sherwood <david.sherwood@arm.com>
3900 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
3901 (widened_mode): Likewise.
3902 (expand_unop): Likewise.
3903 * ree.c (transform_ifelse): Likewise.
3904 (merge_def_and_ext): Likewise.
3905 (combine_reaching_defs): Likewise.
3906 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3908 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3909 Alan Hayward <alan.hayward@arm.com>
3910 David Sherwood <david.sherwood@arm.com>
3912 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
3913 * combine.c (gen_lowpart_for_combine): Likewise.
3914 * dwarf2out.c (rtl_for_decl_location): Likewise.
3915 * final.c (alter_subreg): Likewise.
3916 * rtlhooks.c (gen_lowpart_general): Likewise.
3917 (gen_lowpart_if_possible): Likewise.
3919 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3920 Alan Hayward <alan.hayward@arm.com>
3921 David Sherwood <david.sherwood@arm.com>
3923 * calls.c (expand_call): Use subreg_lowpart_offset.
3924 * cse.c (cse_insn): Likewise.
3925 * regcprop.c (copy_value): Likewise.
3926 (copyprop_hardreg_forward_1): Likewise.
3928 2017-10-13 Jakub Jelinek <jakub@redhat.com>
3931 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
3932 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
3933 =Q constraints to +Q and into insn condition add check
3934 that operands[0] and operands[1] are equal.
3935 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
3936 =Q constraints to +Q and into insn condition add check
3937 that operands[0] is equal to either operands[1] or operands[2].
3940 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
3941 instead of handling MINUS_EXPR twice (once for each argument),
3942 canonicalize operand order and handle just once, use rtype where
3943 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
3946 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
3947 any values of __C while still being pattern recognizable as a simple
3950 2017-10-13 Richard Biener <rguenther@suse.de>
3952 PR tree-optimization/82451
3954 2017-10-02 Richard Biener <rguenther@suse.de>
3956 PR tree-optimization/82355
3957 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
3958 a mapping for the enclosing loop but avoid generating one for
3960 (copy_bb_and_scalar_dependences): Remove premature codegen
3961 error on PHIs in blocks duplicated into multiple places.
3962 * graphite-scop-detection.c
3963 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
3964 in the region use it as loop and nest to analyze the DR in.
3965 (try_generate_gimple_bb): Likewise.
3966 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
3967 (add_loop_constraints): For blocks in a loop not in the region
3968 create a dimension with a single iteration.
3969 * sese.h (gbb_loop_at_index): Remove assert.
3971 * cfgloop.c (loop_preheader_edge): For the loop tree root
3972 return the single successor of the entry block.
3973 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3974 Reset the SCEV hashtable and niters.
3975 * graphite-scop-detection.c
3976 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
3977 assert that we only have POLYNOMIAL_CHREC that vary in loops
3978 contained in the region.
3979 (scop_detection::graphite_can_represent_expr): Adjust.
3980 (scop_detection::stmt_has_simple_data_refs_p): For loops
3981 not in the region set loop to NULL. The nest is now the
3982 entry edge to the region.
3983 (try_generate_gimple_bb): Likewise.
3984 * sese.c (scalar_evolution_in_region): Adjust for
3985 instantiate_scev change.
3986 * tree-data-ref.h (graphite_find_data_references_in_stmt):
3987 Make nest parameter the edge into the region.
3988 (create_data_ref): Likewise.
3989 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
3990 entry edge into a region and adjust instantiate_scev calls.
3991 (create_data_ref): Likewise.
3992 (graphite_find_data_references_in_stmt): Likewise.
3993 (find_data_references_in_stmt): Pass the loop preheader edge
3994 from the nest argument.
3995 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
3996 parameter the edge into the region.
3997 (instantiate_parameters): Use the loop preheader edge as entry.
3998 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
4000 (get_instantiated_value_entry): Make instantiate_below parameter
4001 the edge into the region.
4002 (instantiate_scev_name): Likewise. Adjust dominance checks,
4003 when we cannot use loop-based instantiation instantiate by
4004 walking use-def chains.
4005 (instantiate_scev_poly): Adjust.
4006 (instantiate_scev_binary): Likewise.
4007 (instantiate_scev_convert): Likewise.
4008 (instantiate_scev_not): Likewise.
4009 (instantiate_array_ref): Remove.
4010 (instantiate_scev_3): Likewise.
4011 (instantiate_scev_2): Likewise.
4012 (instantiate_scev_1): Likewise.
4013 (instantiate_scev_r): Do not blindly handle N-operand trees.
4014 Do not instantiate array-refs. Handle all constants and invariants.
4015 (instantiate_scev): Make instantiate_below parameter
4016 the edge into the region.
4017 (resolve_mixers): Use the loop preheader edge for the region
4018 parameter to instantiate_scev_r.
4019 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
4021 2017-10-13 Richard Biener <rguenther@suse.de>
4023 PR tree-optimization/82525
4024 * graphite-isl-ast-to-gimple.c
4025 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
4027 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
4028 Fail code generation when we cannot represent the isl integer.
4029 (binary_op_to_tree): Elide modulo operations that are no-ops
4030 in the type we code generate. Remove now superfluous code
4033 2017-10-13 Richard Biener <rguenther@suse.de>
4035 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
4036 (scop_detection::harmful_loop_in_region): Remove premature
4037 IV type restriction.
4038 (scop_detection::graphite_can_represent_scev): We can handle
4039 pointer IVs just fine.
4041 2017-10-13 Alan Modra <amodra@gmail.com>
4043 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
4044 "Clobbers and Scratch Registers". Add paragraph on
4045 alternative to clobbers for scratch registers and OpenBLAS
4048 2017-10-13 Alan Modra <amodra@gmail.com>
4050 * doc/extend.texi (Clobbers): Correct vax example. Delete old
4051 example of a memory input for a string of known length. Move
4052 commentary out of table. Add a number of new examples
4053 covering array memory inputs.
4055 2017-10-12 Martin Liska <mliska@suse.cz>
4057 PR tree-optimization/82493
4058 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
4059 (test_range_functions): New function.
4060 (sbitmap_c_tests): Likewise.
4061 * selftest-run-tests.c (selftest::run_tests): Run new tests.
4062 * selftest.h (sbitmap_c_tests): New function.
4064 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
4066 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
4068 * config/rs6000/amo.h: Fix spacing issue.
4070 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4073 * config/i386/i386.md (*ashl<mode>3_mask_1,
4074 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
4075 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
4078 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4080 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
4081 (profile_probability): Set max_probability
4082 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
4084 * profile-count.c (profile_probability::differs_from_p): Do not
4085 rely on max_probaiblity == 10000
4087 2017-10-12 Jeff Law <law@redhat.com>
4089 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
4092 2017-10-12 Martin Sebor <msebor@redhat.com>
4096 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
4097 (handle_alias_pairs): Call it.
4098 * common.opt (-Wattribute-alias): New option.
4099 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
4100 * doc/invoke.texi (-Wattribute-alias): Document.
4102 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
4105 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4107 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4109 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4110 (make_hard_regno_born, make_hard_regno_dead): Update
4111 bb_killed_pseudos and bb_gen_pseudos.
4113 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4115 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
4117 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
4119 * config/alpha/alpha.c (alpha_split_conditional_move):
4120 Use std::swap instead of manually swapping.
4121 (alpha_stdarg_optimize_hook): Ditto.
4122 (alpha_canonicalize_comparison): Ditto.
4124 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4126 * tree-loop-distribution.c (struct builtin_info): New struct.
4127 (struct partition): Refactor fields into struct builtin_info.
4128 (partition_free): Free struct builtin_info.
4129 (build_size_arg_loc, build_addr_arg_loc): Delete.
4130 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
4131 information from struct builtin_info.
4132 (find_single_drs): New function refactored from classify_partition.
4133 Also moved builtin validity checks to this function.
4134 (compute_access_range, alloc_builtin): New functions.
4135 (classify_builtin_st, classify_builtin_ldst): New functions.
4136 (classify_partition): Refactor code into functions find_single_drs,
4137 classify_builtin_st and classify_builtin_ldst.
4138 (distribute_loop): Don't do runtime alias check when distributing
4140 (find_seed_stmts_for_distribution): New function.
4141 (pass_loop_distribution::execute): Refactor code finding seed
4142 stmts into above function. Support distribution for the innermost
4143 two-level loop nest. Adjust dump information.
4145 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4147 * tree-loop-distribution.c: Adjust the general comment.
4148 (NUM_PARTITION_THRESHOLD): New macro.
4149 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
4150 (classify_partition): Skip builtin pattern of loop nest's inner loop.
4151 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
4152 in call to build_partition_graph.
4153 (finalize_partitions): New parameter. Make loop distribution more
4154 conservative by fusing more partitions.
4155 (distribute_loop): Don't do runtime alias check in case of loop nest
4157 (find_seed_stmts_for_distribution): New function.
4158 (prepare_perfect_loop_nest): New function.
4159 (pass_loop_distribution::execute): Refactor code finding seed stmts
4160 and loop nest into above functions. Support loop nest distribution.
4161 Adjust dump information accordingly.
4163 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4165 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
4166 and set PTYPE_SEQUENTIAL for merged partition.
4168 2017-10-12 Richard Biener <rguenther@suse.de>
4170 PR tree-optimization/69728
4172 2017-09-19 Richard Biener <rguenther@suse.de>
4174 PR tree-optimization/69728
4175 * graphite-sese-to-poly.c (schedule_error): New global.
4176 (add_loop_schedule): Handle empty domain by failing the
4178 (build_original_schedule): Handle schedule_error.
4180 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
4181 domain by returning an unchanged schedule.
4183 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4185 * genrecog.c (validate_pattern): For VEC_SELECT verify that
4186 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
4188 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
4190 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
4193 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4196 * expr.c (store_field): Don't optimize away bitsize == 0 store
4197 from CALL_EXPR with addressable return type.
4199 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4201 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
4202 * config/rs6000/rs6000.md (sel): Delete mode attribute.
4203 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
4204 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
4205 TARGET_ISEL instead of TARGET_ISEL<sel>.
4207 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
4209 * config/rs6000/rs6000.c
4210 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
4212 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4214 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
4216 (reg_or_cint_operand): Fix comment.
4217 (reg_or_zero_operand): New predicate.
4218 * config/rs6000/rs6000-protos.h (output_isel): Delete.
4219 * config/rs6000/rs6000.c (output_isel): Delete.
4220 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
4221 instead of reg_or_cint_operand. Output instruction directly (not via
4223 (isel_unsigned_<mode>): Ditto.
4224 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
4225 gpc_reg_operand. Add an instruction alternative for this. Output
4226 instruction directly.
4227 (*isel_reversed_unsigned_<mode>): Ditto.
4229 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4231 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
4232 (TARGET_CANONICALIZE_COMPARISON): Define.
4234 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
4237 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
4238 Check whether the dest is REG before adding REG_EQUIV note.
4240 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4243 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4245 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4246 (make_hard_regno_born, make_hard_regno_dead): Update
4247 bb_killed_pseudos and bb_gen_pseudos.
4249 2017-10-11 Nathan Sidwell <nathan@acm.org>
4251 * incpath.h (enum incpath_kind): Name enum, prefix values.
4252 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
4253 * incpath.c (heads, tails): Use INC_MAX.
4254 (add_env_var_paths, add_standard_paths): Use incpath_kind.
4255 (merge_include_chains, split_quote_chain,
4256 register_include_chains): Update incpath_kind names.
4257 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
4258 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
4260 (add_framework_path, darwin_register_objc_includes): Likewise.
4261 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
4263 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4265 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
4266 Do not use float_operator operator predicate.
4267 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
4268 * config/i386/predicates.md (float_operator): Remove predicate.
4270 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4272 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
4273 (*jccxf_i387): Ditto.
4274 (*jcc<mode>_i387): Ditto.
4275 (*jccu<mode>_i387): Ditto.
4276 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
4277 (*jcc_*_i387 splitters): Remove.
4278 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
4279 * config/i386/i386.c (ix86_split_fp_branch): Remove.
4280 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
4283 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4285 * profile-count.h (slow_safe_scale_64bit): New function.
4286 (safe_scale_64bit): New inline.
4287 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
4288 * profile-count.c: Include wide-int.h
4289 (slow_safe_scale_64bit): New.
4291 2017-10-11 Nathan Sidwell <nathan@acm.org>
4293 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
4294 HAS_DECL_ASSEMBLER_NAME_P.
4295 * gimple-expr.c (gimple_decl_printable_name: Check
4296 HAS_DECL_ASSEMBLER_NAME_P too.
4297 * ipa-utils.h (type_in_anonymous_namespace_p): Check
4298 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
4299 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
4300 * passes.c (rest_of_decl_compilation): Check
4301 HAS_DECL_ASSEMBLER_NAME_P too.
4302 * recog.c (verify_changes): Likewise.
4303 * tree-pretty-print.c (dump_decl_name): Likewise.
4304 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
4306 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
4307 (SET_DECL_ASSEMBLER_NAME): Use it.
4308 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
4309 (COPY_DECL_ASSEMBLER_NAME): Likewise.
4310 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
4312 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4314 * config.gcc (i386, x86_64): Add extra objects.
4315 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
4316 (ix86_min_insn_size): Declare.
4317 (ix86_issue_rate): Declare.
4318 (ix86_adjust_cost): Declare.
4319 (ia32_multipass_dfa_lookahead): Declare.
4320 (ix86_macro_fusion_p): Declare.
4321 (ix86_macro_fusion_pair_p): Declare.
4322 (ix86_bd_has_dispatch): Declare.
4323 (ix86_bd_do_dispatch): Declare.
4324 (ix86_core2i7_init_hooks): Declare.
4325 (ix86_atom_sched_reorder): Declare.
4326 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
4327 (COSTS_N_BYTES): Move to x86-tune-costs.h.
4328 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
4329 (rip_relative_addr_p): Rename to ...
4330 (ix86_rip_relative_addr_p): ... this one; export.
4331 (memory_address_length): Update.
4332 (ix86_issue_rate): Move to x86-tune-sched.c.
4333 (ix86_flags_dependent): Move to x86-tune-sched.c.
4334 (ix86_agi_dependent): Move to x86-tune-sched.c.
4335 (exact_dependency_1): Move to x86-tune-sched.c.
4336 (exact_store_load_dependency): Move to x86-tune-sched.c.
4337 (ix86_adjust_cost): Move to x86-tune-sched.c.
4338 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
4339 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
4340 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
4341 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
4342 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
4343 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
4344 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
4345 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
4346 (min_insn_size): Rename to ...
4347 (ix86_min_insn_size): ... this one; export.
4348 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
4349 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
4350 (core2i7_first_cycle_multipass_backtrack): Move to
4351 x86-tune-sched-core.c.
4352 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
4353 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
4354 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
4355 (ix86_avoid_jump_mispredicts): Update.
4356 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
4357 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
4358 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
4359 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
4360 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
4361 (MAX_INSN): Move to ix86-tune-sched-bd.c.
4362 (MAX_IMM): Move to ix86-tune-sched-bd.c.
4363 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
4364 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
4365 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
4366 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
4367 (MAX_STORE): Move to ix86-tune-sched-bd.c.
4368 (BIG): Move to ix86-tune-sched-bd.c.
4369 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
4370 (enum insn_path): Move to ix86-tune-sched-bd.c.
4371 (get_mem_group): Move to ix86-tune-sched-bd.c.
4372 (is_cmp): Move to ix86-tune-sched-bd.c.
4373 (dispatch_violation): Move to ix86-tune-sched-bd.c.
4374 (is_branch): Move to ix86-tune-sched-bd.c.
4375 (is_prefetch): Move to ix86-tune-sched-bd.c.
4376 (init_window): Move to ix86-tune-sched-bd.c.
4377 (allocate_window): Move to ix86-tune-sched-bd.c.
4378 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
4379 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
4380 (process_end_window): Move to ix86-tune-sched-bd.c.
4381 (allocate_next_window): Move to ix86-tune-sched-bd.c.
4382 (find_constant): Move to ix86-tune-sched-bd.c.
4383 (get_num_immediates): Move to ix86-tune-sched-bd.c.
4384 (has_immediate): Move to ix86-tune-sched-bd.c.
4385 (get_insn_path): Move to ix86-tune-sched-bd.c.
4386 (get_insn_group): Move to ix86-tune-sched-bd.c.
4387 (count_num_restricted): Move to ix86-tune-sched-bd.c.
4388 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
4389 (add_insn_window): Move to ix86-tune-sched-bd.c.
4390 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
4391 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
4392 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
4393 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
4394 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
4395 (do_dispatch): Move to ix86-tune-sched-bd.c.
4396 (has_dispatch): Move to ix86-tune-sched-bd.c.
4397 * config/i386/t-i386: Add new object files.
4398 * config/i386/x86-tune-costs.h: New file.
4399 * config/i386/x86-tune-sched-atom.c: New file.
4400 * config/i386/x86-tune-sched-bd.c: New file.
4401 * config/i386/x86-tune-sched-core.c: New file.
4402 * config/i386/x86-tune-sched.c: New file.
4404 2017-10-11 Liu Hao <lh_mouse@126.com>
4406 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
4407 the generic version below instead.
4408 (should_colorize): Recognize Windows consoles as terminals
4410 * pretty-print.c [__MINGW32__] (write_all): New function.
4411 [__MINGW32__] (find_esc_head): Likewise.
4412 [__MINGW32__] (find_esc_terminator): Likewise.
4413 [__MINGW32__] (eat_esc_sequence): Likewise.
4414 [__MINGW32__] (mingw_ansi_fputs): New function that handles
4416 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
4419 2017-10-11 Richard Biener <rguenther@suse.de>
4421 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
4422 Properly call analyze_scalar_evolution with the loop of the stmt.
4424 2017-10-11 Richard Biener <rguenther@suse.de>
4426 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
4427 * tree-core.h (tree_base): Add chrec_var union member.
4428 * tree.h (CHREC_VAR): Remove.
4429 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
4430 * tree-chrec.h (build_polynomial_chrec): Adjust.
4431 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
4432 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
4434 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
4436 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
4437 * match.pd: ... here.
4438 ((T) X == (T) Y): Relax condition.
4440 2017-10-11 Bin Cheng <bin.cheng@arm.com>
4442 PR tree-optimization/82472
4443 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
4445 (break_alias_scc_partitions): Update postorder number.
4447 2017-10-11 Martin Liska <mliska@suse.cz>
4450 * opts.c (parse_no_sanitize_attribute): Do not use error_value
4452 * opts.h (parse_no_sanitize_attribute): Remove last argument.
4454 2017-10-11 Martin Liska <mliska@suse.cz>
4456 * print-rtl.c (print_insn): Move declaration of idbuf
4457 to same scope as name.
4459 2017-10-11 Martin Liska <mliska@suse.cz>
4464 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
4467 2017-10-11 Martin Liska <mliska@suse.cz>
4470 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
4473 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
4475 * config.gcc (default_gnu_indirect_function): Default to yes for
4476 arm*-*-linux* with glibc.
4478 2017-10-11 Richard Biener <rguenther@suse.de>
4480 * tree-scalar-evolution.c (get_scalar_evolution): Handle
4481 default-defs and types we do not want to analyze.
4482 (interpret_loop_phi): Replace unreachable code with an assert.
4483 (compute_scalar_evolution_in_loop): Remove and inline ...
4484 (analyze_scalar_evolution_1): ... here, replacing condition with
4485 what makes the intent clearer. Remove handling of cases
4486 get_scalar_evolution now handles.
4488 2017-10-10 Jim Wilson <wilson@tuliptree.org>
4490 PR rtl-optimization/81434
4491 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
4492 comment for main loop. In sched_group_found if, also add checks for
4493 pass and min_cost_group.
4495 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
4497 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
4498 (rs6000_insn_cost): New function.
4499 * config/rs6000/rs6000.md (cost): New attribute.
4501 2017-10-10 Jakub Jelinek <jakub@redhat.com>
4502 H.J. Lu <hongjiu.lu@intel.com>
4506 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
4507 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
4508 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
4509 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
4510 to OPTION_MASK_ISA_AVX512VL - builtins that have both
4511 OPTION_MASK_ISA_MMX and some other bit set require both
4512 mmx and the ISAs without the mmx bit.
4513 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
4514 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
4515 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
4516 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
4517 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
4518 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
4519 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
4520 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
4521 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
4522 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
4523 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
4524 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
4525 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
4526 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
4527 Add OPTION_MASK_ISA_MMX.
4529 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
4531 * config.gcc (armv7*-*-freebsd*): New target.
4532 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
4534 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
4536 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
4537 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
4540 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
4542 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
4543 a template parameter.
4544 (WIDE_INT_REF_FOR): Update accordingly.
4545 * tree.h (wi::int_traits <const_tree>): Delete.
4546 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
4547 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
4548 (wi::tree_to_wide_ref): New typedef.
4549 (wi::to_wide): New function.
4550 * calls.c (get_size_range): Use wi::to_wide when operating on
4552 * cgraph.c (cgraph_node::create_thunk): Likewise.
4553 * config/i386/i386.c (ix86_data_alignment): Likewise.
4554 (ix86_local_alignment): Likewise.
4555 * dbxout.c (stabstr_O): Likewise.
4556 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
4557 * expr.c (const_vector_from_tree): Likewise.
4558 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
4559 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
4560 (fold_negate_expr_1, int_const_binop_1, const_binop)
4561 (fold_convert_const_int_from_real, optimize_bit_field_compare)
4562 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
4563 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
4564 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
4565 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
4566 (fold_not_const, round_up_loc): Likewise.
4567 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
4568 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
4569 (alloca_call_type): Likewise.
4570 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
4571 * godump.c (go_output_typedef): Likewise.
4572 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
4573 * internal-fn.c (get_min_precision): Likewise.
4574 * ipa-cp.c (ipcp_store_vr_results): Likewise.
4575 * ipa-polymorphic-call.c
4576 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
4577 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
4578 (ipa_modify_call_arguments): Likewise.
4579 * match.pd: Likewise.
4580 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
4581 * print-tree.c (print_node_brief, print_node): Likewise.
4582 * stmt.c (expand_case): Likewise.
4583 * stor-layout.c (layout_type): Likewise.
4584 * tree-affine.c (tree_to_aff_combination): Likewise.
4585 * tree-cfg.c (group_case_labels_stmt): Likewise.
4586 * tree-data-ref.c (dr_analyze_indices): Likewise.
4587 (prune_runtime_alias_test_list): Likewise.
4588 * tree-dump.c (dequeue_and_dump): Likewise.
4589 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
4590 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
4591 * tree-pretty-print.c (dump_generic_node): Likewise.
4592 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
4593 (simple_iv_with_niters): Likewise.
4594 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
4595 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
4596 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
4597 * tree-ssa-loop-niter.c (split_to_var_and_offset)
4598 (refine_value_range_using_guard, number_of_iterations_ne_max)
4599 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
4600 (get_cst_init_from_scev, record_nonwrapping_iv)
4601 (scev_var_range_cant_overflow): Likewise.
4602 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
4603 * tree-ssa-pre.c (compute_avail): Likewise.
4604 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
4605 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
4606 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
4607 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
4608 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
4609 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
4610 (dump_case_nodes, try_switch_expansion): Likewise.
4611 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
4612 (vect_do_peeling): Likewise.
4613 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
4614 * tree-vect-stmts.c (vectorizable_load): Likewise.
4615 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
4616 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
4617 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
4618 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
4619 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
4620 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
4621 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
4622 (evrp_dom_walker::before_dom_children): Likewise.
4623 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
4624 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
4625 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
4626 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
4627 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
4628 (get_range_pos_neg): Likewise.
4629 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
4630 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
4631 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
4632 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
4633 * config/avr/avr.c (avr_fold_builtin): Likewise.
4634 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
4635 * config/msp430/msp430.c (msp430_attr): Likewise.
4636 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
4637 * config/powerpcspe/powerpcspe-c.c
4638 (altivec_resolve_overloaded_builtin): Likewise.
4639 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
4640 (rs6000_expand_ternop_builtin): Likewise.
4641 * config/rs6000/rs6000-c.c
4642 (altivec_resolve_overloaded_builtin): Likewise.
4643 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
4644 (rs6000_expand_ternop_builtin): Likewise.
4645 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
4647 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4649 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
4650 when copying loop nest with only one inner loop.
4652 2017-10-10 Richard Biener <rguenther@suse.de>
4654 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
4655 blocks if SCEV is active.
4656 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
4658 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
4659 (scev_initialize): Assert we are not yet initialized.
4661 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4663 * tree-loop-distribution.c (generate_loops_for_partition): Remove
4664 inner loop's exit stmt by making it always exit the loop, otherwise
4665 we would generate an infinite empty loop.
4667 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4669 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
4670 renaming variables in new preheader if it's deleted.
4672 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4674 * tree-loop-distribution.c (struct partition): Remove unused field
4675 loops of the structure.
4676 (partition_alloc, partition_free): Ditto.
4677 (build_rdg_partition_for_vertex): Ditto.
4679 2017-10-09 Jeff Law <law@redhat.com>
4681 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
4682 return type to match prototype and documentation.
4684 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4686 * config/rs6000/rs6000.c (processor_costs): Move to ...
4687 * config/rs6000/rs6000.h: ... here.
4688 (rs6000_cost): Declare.
4690 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
4692 * except.c (setjmp_fn): New global variable.
4693 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
4694 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
4695 if DONT_USE_BUILTIN_SETJMP is defined.
4697 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4699 * target.def (insn_cost): New hook.
4700 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
4701 * doc/tm.texi: Regenerate.
4702 * rtlanal.c (insn_cost): Use the new hook.
4704 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4706 * combine.c (combine_validate_cost): Compute the new insn_cost,
4707 not just pattern_cost.
4708 (try_combine): Adjust comment.
4710 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4712 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
4714 * combine.c (uid_insn_cost): Adjust comment.
4715 (combine_validate_cost): Adjust comment. Use pattern_cost instead
4717 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
4718 * dse.c (find_shift_sequence): Ditto.
4719 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
4720 (bb_valid_for_noce_process_p): Use pattern_cost.
4721 * rtl.h (insn_rtx_cost): Delete.
4722 (pattern_cost): New prototype.
4723 (insn_cost): New prototype.
4724 * rtlanal.c (insn_rtx_cost): Rename to...
4725 (pattern_cost): ... this.
4728 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
4730 * config/i386/i386.md (*jcc_2): Remove insn pattern.
4731 (*jcc<mode>_0_r_i387): Ditto.
4732 (*jccxf_r_i387): Ditto.
4733 (*jcc<mode>_r_i387): Ditto.
4734 (*jccu<mode>_r_i387): Ditto.
4735 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
4736 (*jcc): Rename from *jcc_1.
4738 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4740 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
4741 deferred rescans after the lvx/stvx recombination pre-pass.
4743 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
4745 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
4746 memory operation instruction support.
4747 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
4748 (rs6000-ibm-aix[789]*): Likewise.
4749 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
4750 Document new functions.
4752 2017-10-09 Richard Biener <rguenther@suse.de>
4754 PR tree-optimization/82397
4755 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
4756 equality only for semantically equal trees.
4758 2017-10-09 Richard Biener <rguenther@suse.de>
4760 PR tree-optimization/82449
4761 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
4762 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
4763 allow constant addresses.
4764 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
4767 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4769 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
4772 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4775 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
4778 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4781 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
4783 2017-10-09 Jakub Jelinek <jakub@redhat.com>
4786 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
4787 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
4789 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
4791 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
4792 (WI_BINARY_PREDICATE_RESULT): Likewise.
4793 (wi::binary_traits::operator_result): New type.
4794 (wi::binary_traits::predicate_result): Likewise.
4795 (generic_wide_int::operator~, unary generic_wide_int::operator-)
4796 (generic_wide_int::operator==, generic_wide_int::operator!=)
4797 (generic_wide_int::operator&, generic_wide_int::and_not)
4798 (generic_wide_int::operator|, generic_wide_int::or_not)
4799 (generic_wide_int::operator^, generic_wide_int::operator+
4800 (binary generic_wide_int::operator-, generic_wide_int::operator*):
4802 (operator~, unary operator-, operator==, operator!=, operator&)
4803 (operator|, operator^, operator+, binary operator-, operator*): New
4805 * expr.c (get_inner_reference): Use wi::bit_and_not.
4806 * fold-const.c (fold_binary_loc): Likewise.
4807 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
4808 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
4809 (bit_value_binop): Likewise.
4810 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
4811 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
4812 (extract_range_from_binary_expr_1): Likewise.
4813 (masked_increment): Likewise.
4814 (simplify_bit_ops_using_ranges): Likewise.
4816 2017-10-09 Martin Jambor <mjambor@suse.cz>
4819 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
4820 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
4821 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
4823 (hsa_fixup_mov_insn_type): New function.
4824 (hsa_op_with_type::get_in_type): Use it.
4825 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
4826 immediates in an assert.
4827 (hsa_op_with_type::extend_int_to_32bit): New method.
4828 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
4829 types. Convert to dest type if necessary.
4830 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
4831 (reg_for_gimple_ssa): Pass false as min32int to
4832 hsa_type_for_scalar_tree_type.
4833 (gen_hsa_addr): Fixup type when creating addresable temporary.
4834 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
4835 (gen_hsa_unary_operation): Extend operands and convert to dest type if
4836 necessary. Call hsa_fixup_mov_insn_type.
4837 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
4838 extend operands and convert to dest type if necessary.
4839 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
4840 to dest type if necessary.
4841 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
4842 if conversion nt necessary and size matches.
4843 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
4844 to dest type if necessary.
4845 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
4846 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
4848 (gen_hsa_clrsb): Likewise.
4849 (gen_hsa_ffs): Likewise.
4850 (gen_hsa_divmod): Extend operands and convert to dest type if
4852 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
4854 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
4856 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
4857 Remove empty default arguments. Use a brace block as output
4859 (conditional return): Ditto.
4861 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
4862 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
4864 (group_ending_nop): Ditto.
4865 (doloop_end): Ditto.
4866 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
4867 (splitters for those): Ditto.
4869 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
4871 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
4872 a conditional jump (and the compare for it) so that pc_rtx is the
4874 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
4875 for the deleted and renamed ctr<mode>_internal[234] patterns.
4876 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
4877 Delete second conditional return pattern.
4878 (ctr<mode>_internal2): Delete this second bdnz pattern.
4879 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
4880 (ctr<mode>_internal4): Delete this second bdz pattern.
4882 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
4884 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
4885 (always_initialized_rtx_for_ssa_name_p): New predicate.
4886 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
4887 (finish_out_of_ssa): Free new field of SA.
4888 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
4889 * tree-ssa-coalesce.c: Include tree-ssa.h.
4890 (get_parm_default_def_partitions): Remove extern keyword.
4891 (get_undefined_value_partitions): New function.
4892 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
4893 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
4895 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
4897 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
4899 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
4901 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
4903 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
4904 for targets that preffer 128bit.
4906 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
4908 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
4910 2017-10-08 Olivier Hainque <hainque@adacore.com>
4912 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
4913 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
4915 (thumb_set_return_address): Likewise.
4917 2017-10-08 Olivier Hainque <hainque@adacore.com>
4919 * common/config/arm/arm-common.c (arm_except_unwind_info):
4920 Handle DWARF2_UNWIND_INFO.
4922 2017-10-07 Michael Collison <michael.collison@arm.com>
4924 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
4927 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
4929 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
4930 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
4931 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
4932 defined, force the creation of a new block for a dispatch label.
4934 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
4936 * invoke.texi (Wsuggest-attribute=cold): Document.
4937 * common.opt (Wsuggest-attribute=cold): New
4938 * ipa-pure-const.c (warn_function_cold): New function.
4939 * predict.c (compute_function_frequency): Use it.
4940 * predict.h (warn_function_cold): Declare.
4942 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
4944 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
4947 2017-10-06 Martin Liska <mliska@suse.cz>
4949 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
4950 keyword for member functions.
4951 (struct sanopt_tree_couple): New struct.
4952 (struct sanopt_tree_couple_hash): New function.
4953 (struct sanopt_ctx): Add new hash_map.
4954 (has_dominating_ubsan_ptr_check): New function.
4955 (record_ubsan_ptr_check_stmt): Likewise.
4956 (maybe_optimize_ubsan_ptr_ifn): Likewise.
4957 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
4958 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
4960 2017-10-06 Sudakshina Das <sudi.das@arm.com>
4963 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
4964 aarch64_simd_valid_immediate on CONST_VECTORs.
4965 (aarch64_reg_or_bic_imm): Likewise.
4967 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
4969 PR rtl-optimization/82396
4970 * haifa-sched.c (ready_sort_real): Disable qsort checking.
4972 2017-10-06 Sebastian Pop <sebpop@gmail.com>
4974 * graphite-dependences.c (scop_get_reads): Move code to...
4975 (scop_get_must_writes): Move code to...
4976 (scop_get_may_writes): Move code to...
4977 (scop_get_reads_and_writes): ... here.
4978 (scop_get_dependences): Call scop_get_reads_and_writes.
4980 2017-10-06 Jakub Jelinek <jakub@redhat.com>
4982 PR tree-optimization/82434
4983 * fold-const.h (can_native_encode_type_p,
4984 can_native_encode_string_p): Remove.
4985 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
4986 don't encode anything, just return what would be otherwise returned.
4987 (native_encode_fixed, native_encode_complex, native_encode_vector):
4989 (native_encode_string): Likewise. Inline by hand
4990 can_native_encode_string_p.
4991 (can_native_encode_type_p): Remove.
4992 (can_native_encode_string_p): Remove.
4993 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
4994 STRING_CSTs using can_native_encode_string_p, test all
4995 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
4996 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
4997 argument from native_encode_expr.
4998 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
4999 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
5002 2017-10-06 Richard Biener <rguenther@suse.de>
5004 PR tree-optimization/82397
5005 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
5006 operand_equal_p but rely on data_ref_compare_tree for detecting
5008 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
5009 to match up with dr_group_sort_cmp.
5011 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5014 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
5016 * config/s390/s390-builtin-types.def: Regenerate.
5018 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5021 * config/s390/s390-builtin-types.def: Regenerate.
5022 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
5023 Change flag from B_VXE to B_VX.
5024 (s390_vec_min_dbl): Remove B_VXE flag.
5026 2017-10-06 Richard Biener <rguenther@suse.de>
5028 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
5029 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
5030 translate_isl_ast_to_gimple::is_valid_rename,
5031 translate_isl_ast_to_gimple::get_rename,
5032 translate_isl_ast_to_gimple::get_def_bb_for_const,
5033 translate_isl_ast_to_gimple::get_new_name,
5034 translate_isl_ast_to_gimple::collect_all_ssa_names,
5035 translate_isl_ast_to_gimple::copy_loop_phi_args,
5036 translate_isl_ast_to_gimple::collect_all_ssa_names,
5037 translate_isl_ast_to_gimple::copy_loop_phi_args,
5038 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
5039 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
5040 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
5041 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
5042 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
5043 translate_isl_ast_to_gimple::copy_cond_phi_args,
5044 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
5045 translate_isl_ast_to_gimple::edge_for_new_close_phis,
5046 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
5047 translate_isl_ast_to_gimple::rename_uses,
5048 translate_isl_ast_to_gimple::rename_all_uses): Remove.
5049 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
5050 (set_rename_for_each_def): Likewise.
5051 (graphite_copy_stmts_from_block): Handle debug stmt resetting
5052 here. Handle rewriting SCEV analyzable uses here.
5053 (copy_bb_and_scalar_dependences): Generate code for PHI
5055 (graphite_regenerate_ast_isl): Adjust.
5056 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
5057 (add_write, add_read): New functions.
5058 (build_cross_bb_scalars_def): Use it and simplify.
5059 (build_cross_bb_scalars_use): Likewise.
5060 (graphite_find_cross_bb_scalar_vars): Inline into...
5061 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
5062 simulating out-of-SSA. Compute liveout and add dependencies.
5063 (build_scops): Force an empty entry block.
5064 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
5066 (sese_build_liveouts): Declare.
5067 (sese_trivially_empty_bb_p): Likewise.
5068 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
5069 compute liveout and debug_liveout.
5070 (sese_bad_liveouts_use): Remove.
5071 (sese_reset_debug_liveouts_bb): Likewise.
5072 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
5073 (sese_build_liveouts): Build liveout and debug_liveout and store
5075 (new_sese_info): Adjust.
5076 (free_sese_info): Likewise.
5077 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
5078 do not build liveout here.
5079 (move_sese_in_condition): Adjust region entry.
5080 (scev_analyzable_p): Match up with chrec_apply requirements.
5081 (sese_trivially_empty_bb_p): New.
5082 * tree-into-ssa.c (get_reaching_def): Properly support generating
5083 default-defs for incremental rewrite of anonymous names.
5085 2017-10-06 Richard Biener <rguenther@suse.de>
5087 * graphite-sese-to-poly.c (extract_affine): For casts increasing
5088 precision do not perform modulo reduction.
5090 2017-10-06 Richard Biener <rguenther@suse.de>
5092 PR tree-optimization/82436
5093 * tree-vect-slp.c (vect_supported_load_permutation_p): More
5094 conservatively choose the vectorization factor when checking
5095 whether we can perform the required load permutation.
5096 (vect_transform_slp_perm_load): Assert when we may not fail.
5098 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
5100 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
5101 message for incompatible -msdata=* and -mcall-* options.
5103 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5105 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
5106 rate for post-reload scheduling.
5108 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5110 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
5112 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5114 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
5115 to improve monte carlo in scimark.
5117 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5119 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
5120 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
5121 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
5122 pentium4_cost, nocona_cost): Set reassociation width to 1.
5123 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
5124 width to 2 for fp operations and 1 otherwise.
5125 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
5127 (atom_cost): Set reassociation width to 2.
5128 (slm_cost, generic_cost): Set fp reassociation width
5129 to 2 and 1 otherwise.
5130 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
5131 (core_cost): Set fp reassociation width to 4 and vector to 2.
5132 (ix86_reassociation_width): Rewrite using cost table; special case
5133 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
5134 and TARGET_AVX128_OPTIMAL.
5135 * config/i386/i386.h (processor_costs): Add
5136 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
5137 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
5138 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
5139 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
5140 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
5141 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
5143 2017-10-05 Nathan Sidwell <nathan@acm.org>
5145 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
5147 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5149 * config/arm/arm.c (arm_test_fpu_data): New.
5150 (arm_run_selftests): Call arm_test_fpu_data.
5152 2017-10-04 Nathan Sidwell <nathan@acm.org>
5154 * toplev.c (toplev::main): Remove excess parens on pretty_printer
5156 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
5158 2017-10-04 Sudakshina Das <sudi.das@arm.com>
5160 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
5161 check type for aarch64_simd_valid_immediate.
5162 (aarch64_output_simd_mov_immediate): Update prototype.
5163 (aarch64_simd_valid_immediate): Update prototype.
5164 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
5165 support for ORR-immediate.
5166 (and<mode>3): modified pattern to add support for BIC-immediate.
5167 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
5168 now checks for valid immediate for BIC and ORR based on new enum
5170 (aarch64_output_simd_mov_immediate): Function now used to output
5171 BIC/ORR imm as well based on new enum argument.
5172 * config/aarch64/constraints.md (Do): New vector immediate constraint.
5174 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
5175 (aarch64_reg_or_bic_imm): Likewise.
5177 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5179 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
5180 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
5182 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5186 PR rtl-optimization/82396
5187 * haifa-sched.c (autopref_multipass_init): Simplify
5189 (autopref_rank_data): Simplify sort order.
5190 * sched-int.h (autopref_multipass_data_): Remove
5191 multi_mem_insn_p, min_offset and max_offset.
5193 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5195 * doc/sourcebuild.texi: Document vect_peeling_profitable.
5197 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5199 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
5202 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5204 * doc/sourcebuild.texi: Document vect_long_mult.
5206 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
5208 PR tree-optimization/82413
5209 * fold-const.c (build_range_check): Use widest_int when comparing
5210 the maximum ETYPE value with HIGH.
5212 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5214 PR rtl-optimization/82396
5215 * haifa-sched.c (autopref_multipass_init): Simplify
5217 (autopref_rank_data): Simplify sort order.
5218 * sched-int.h (autopref_multipass_data_): Remove
5219 multi_mem_insn_p, min_offset and max_offset.
5221 2017-10-04 Jakub Jelinek <jakub@redhat.com>
5223 PR tree-optimization/82381
5224 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
5225 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
5228 PR tree-optimization/82374
5229 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
5230 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
5231 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
5232 current_function_decl to the new decl.
5234 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
5236 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
5237 helper macro for IEEE float128 hardware built-in functions.
5238 (SQRTF128_ODD): Add built-in functions with the round-to-odd
5240 (TRUNCF128_ODD): Likewise.
5241 (ADDF128_ODD): Likewise.
5242 (SUBF128_ODD): Likewise.
5243 (MULF128_ODD): Likewise.
5244 (DIVF128_ODD): Likewise.
5245 (FMAF128_ODD): Likewise.
5246 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
5247 UNSPEC_TRUNC_ROUND_TO_ODD.
5248 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
5249 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
5250 floating point round to odd instructions.
5251 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
5252 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
5253 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
5254 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
5255 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
5256 (trunc<mode>sf2_hw): Change the truncate with round to odd
5257 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
5258 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
5259 to odd hardware instructions.
5260 (sub<mode>3_odd): Likewise.
5261 (mul<mode>3_odd): Likewise.
5262 (div<mode>3_odd): Likewise.
5263 (sqrt<mode>2_odd): Likewise.
5264 (fma<mode>4_odd): Likewise.
5265 (fms<mode>4_odd): Likewise.
5266 (nfma<mode>4_odd): Likewise.
5267 (nfms<mode>4_odd): Likewise.
5268 (trunc<mode>df2_odd): Change the truncate with round to odd
5269 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
5271 * doc/extend.texi (PowerPC built-in functions): Update documentation
5272 for existing IEEE float128-bit built-in functions. Add built-in
5273 functions that generate the IEEE 128-bit floating point round to
5276 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
5278 PR rtl-optimization/77729
5279 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
5280 to (X&(C1&~C2))|C2 transformations.
5282 2017-10-03 Martin Jambor <mjambor@suse.cz>
5284 PR tree-optimization/82363
5285 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
5286 mismatch, mark lacc written regardless of racc.
5288 2017-10-03 Jakub Jelinek <jakub@redhat.com>
5290 PR tree-optimization/82381
5291 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
5292 stmt_to_insert nor wheather SSA_NAMEs are default defs.
5293 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
5294 fallthrough into reassoc_stmt_dominates_stmt_p.
5297 * combine.c (combine_instructions): Don't combine in unreachable
5300 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
5303 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
5304 function to not use the have_cpu variable. Do not set cpu_index,
5305 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
5307 (rs6000_valid_attribute_p): Remove duplicate initializations of
5308 old_optimize and func_optimize.
5309 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
5310 (rs6000_activate_target_options): Make global.
5311 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
5314 2017-10-02 Jakub Jelinek <jakub@redhat.com>
5316 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
5317 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
5318 Set *poffset to 0 and *psize and *pmax_size to -1 if
5319 *poffset + *psize overflows in HOST_WIDE_INT.
5321 PR tree-optimization/82387
5322 PR tree-optimization/82388
5323 PR tree-optimization/82389
5324 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
5325 instead of live_bytes non-NULL.
5327 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
5330 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
5333 2017-10-02 Richard Biener <rguenther@suse.de>
5335 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
5336 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
5337 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
5339 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
5341 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
5342 requested precision matches the type's.
5343 * calls.c (alloc_max_size): Calculate the new candidate size as
5344 a widest_int and use wi::to_widest when comparing it with the
5345 current candidate size.
5346 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
5347 zero rather than integer_zero_node.
5348 * match.pd: Check for a no-op conversion before using wi::add
5349 rather than after. Use tree_to_uhwi when summing small shift
5350 counts into an unsigned int.
5352 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
5353 Alan Hayward <alan.hayward@arm.com>
5354 David Sherwood <david.sherwood@arm.com>
5357 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
5358 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
5359 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
5360 POINTER_AND_FP_REGS.
5362 2017-10-02 Richard Biener <rguenther@suse.de>
5364 PR tree-optimization/82355
5365 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5366 a mapping for the enclosing loop but avoid generating one for
5368 (copy_bb_and_scalar_dependences): Remove premature codegen
5369 error on PHIs in blocks duplicated into multiple places.
5370 * graphite-scop-detection.c
5371 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5372 in the region use it as loop and nest to analyze the DR in.
5373 (try_generate_gimple_bb): Likewise.
5374 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5375 (add_loop_constraints): For blocks in a loop not in the region
5376 create a dimension with a single iteration.
5377 * sese.h (gbb_loop_at_index): Remove assert.
5379 2017-10-01 Kevin Buettner <kevinb@redhat.com>
5381 * omp-expand.c (adjust_context_scope): New function.
5382 (expand_parallel_call): Call adjust_context_scope.
5384 2017-10-01 Jeff Law <law@redhat.com>
5386 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
5387 dom_opt_dom_walker class with direct access to private members.
5388 Add comments. Call test_for_singularity.
5389 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
5390 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
5391 m_dummy_cond anymore.
5392 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
5394 (pass_dominator:execute): Build the dummy_cond here and pass it
5395 to the dom_opt_dom_walker ctor.
5396 (test_for_singularity): New function.
5398 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
5399 Maya Rashish <coypu@sdf.org>
5401 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
5402 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
5403 (alpha*-*-netbsd*) Use nbsd_tm_file.
5404 (arm*-*-netbsdelf*) Likewise.
5405 (i[34567]86-*-netbsdelf*) Likewise.
5406 (x86_64-*-netbsd*) Likewise.
5407 (mips*-*-netbsd*) Likewise.
5408 (powerpc-*-netbsd*) Likewise.
5409 (sh*-*-netbsd*) Likewise.
5410 (sparc-*-netbsdelf*) Likewise.
5411 (sparc64-*-netbsd*) Likewise.
5412 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
5414 (vax-*-netbsdelf*) Likewise.
5415 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
5416 (UINT_FAST8_TYPE) Likewise.
5417 (INT_FAST16_TYPE) Check CHAR_FAST16.
5418 (UINT_FAST16_TYPE) Likewise.
5420 2017-09-30 Jakub Jelinek <jakub@redhat.com>
5423 * config/i386/i386.md
5424 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
5425 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
5426 *divmodsi4_zext_2): New define_insn_and_split.
5427 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
5428 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
5429 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
5430 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
5431 New define_insn_and_split.
5432 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
5433 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
5434 operands[1] having DImode when mode is SImode.
5436 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
5437 always SImode for DIV and MOD in REG_EQUAL notes.
5439 2017-09-29 Yury Gribov <tetra2005@gmail.com>
5442 * match.pd: Fix handling of NaNs in pattern.
5444 2017-09-29 Jeff Law <law@redhat.com>
5446 * sbitmap.c (bitmap_bit_in_range_p): New function.
5447 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
5448 * tree-ssa-dse.c (live_bytes_read): New function.
5449 (dse_classify_store): Ignore reads of dead bytes.
5451 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
5452 typos and whitespace errors.
5453 * config/i386/predicates.md (address_no_seg_operand): Likewise.
5454 * config/s390/s390.c (s390_emit_prologue): Likewise.
5456 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
5459 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
5460 with a symbol for LRA.
5462 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
5464 PR rtl-optimization/82338
5465 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
5467 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
5469 * genmodes.c (calc_wider_mode): Suppress qsort macro.
5470 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
5471 (qsort_chk): Declare.
5472 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
5473 (qsort_chk): New function.
5475 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5477 PR tree-optimization/82337
5478 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
5479 phi definition if the PHI result appears in an abnormal PHI.
5480 (find_basis_for_base_expr): Don't record a basis if the LHS of the
5481 basis appears in an abnormal PHI.
5483 2017-09-29 Richard Biener <rguenther@suse.de>
5485 * graphite-isl-ast-to-gimple.c
5486 (translate_isl_ast_to_gimple::set_codegen_error): New function.
5487 (binary_op_to_tree): Use it.
5488 (get_rename_from_scev): Likewise.
5489 (copy_loop_phi_nodes): Likewise.
5490 (copy_bb_and_scalar_dependences): Likewise.
5491 (translate_pending_phi_nodes): Likewise.
5493 2017-09-29 Jakub Jelinek <jakub@redhat.com>
5496 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
5497 for movabsq $(i32 << shift), r64.
5499 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
5501 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
5502 index when encoding %esp as %rsp to avoid 0x67 prefix.
5504 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
5506 * config/i386/i386.md (*movsf_internal, *movdf_internal):
5507 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
5509 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
5511 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
5512 Extensions with more than 16 double VFP registers.
5513 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
5514 to_clear_mask and all code related to it. Replace the remaining
5515 entry by a sbitmap and adapt code accordingly.
5517 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
5519 * brig-builtins.def: Change pure attributes to const.
5521 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5523 * config.gcc (default_gnu_indirect_function): Default to yes for
5524 sparc*-*-linux* with glibc.
5526 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5528 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
5529 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
5530 when creating .init_array and .fini_array sections with priority
5533 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
5536 * config/aarch64/aarch64.c
5537 (aarch64_builtin_support_vector_misalignment): Always return false
5538 when misalignment is unknown.
5540 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
5542 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
5543 this function to return false if the definition used by the swap
5544 instruction is artificial, or if the memory address from which the
5545 constant value is loaded is not represented by a base address held
5546 in a register or if the base address register is a frame or stack
5547 pointer. Additionally, return false if the base address of the
5548 loaded constant is a SYMBOL_REF but is not considered to be a
5550 (replace_swapped_load_constant): New function.
5551 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
5552 loaded constant vector with a load of a swapped constant vector.
5554 2017-09-27 Carl Love <cel@us.ibm.com>
5556 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
5557 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
5558 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
5561 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
5563 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
5564 first, always call autopref_rank_data otherwise.
5566 2017-09-27 Richard Biener <rguenther@suse.de>
5568 * graphite-scop-detection.c (find_scop_parameters): Move
5569 loop bound handling ...
5570 (gather_bbs::before_dom_children): ... here, avoiding the need
5571 to build scop_info->loop_nest.
5572 (record_loop_in_sese): Remove.
5573 * sese.h (sese_info_t::loop_nest): Remove.
5574 * sese.c (new_sese_info): Do not allocate loop_nest.
5575 (free_sese_info): Do not free loop_nest.
5577 2017-09-27 Jakub Jelinek <jakub@redhat.com>
5580 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
5581 lhs from calls if the lhs has addressable type.
5583 2017-09-27 Richard Biener <rguenther@suse.de>
5585 * graphite.h (scop::max_alias_set): New member.
5586 * graphite-scop-detection.c: Remove references to non-existing
5587 --param in comments.
5588 (build_alias_sets): Record the maximum alias set used for drs.
5589 (build_scops): Support zero as unlimited for
5590 --param graphite-max-arrays-per-scop.
5591 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
5593 (build_poly_sr_1): ... here. Compute alias set based on the
5594 maximum alias set used for drs rather than
5595 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
5597 2017-09-27 Richard Biener <rguenther@suse.de>
5599 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
5600 --param loop-block-tile-size=0 to disable tiling.
5602 2017-09-27 Richard Biener <rguenther@suse.de>
5604 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
5605 (graphite-max-nb-scop-params): Document special value zero.
5606 * domwalk.h (dom_walker::STOP): New symbolical constant.
5607 (dom_walker::dom_walker): Add optional parameter for bb to
5609 (dom_walker::~dom_walker): Declare.
5610 (dom_walker::before_dom_children): Document STOP return value.
5611 (dom_walker::m_user_bb_to_rpo): New member.
5612 (dom_walker::m_bb_to_rpo): Likewise.
5613 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
5614 mapping here if not provided by the user.
5615 (dom_walker::~dom_walker): Free bb to RPO mapping if not
5616 provided by the user.
5617 (dom_walker::STOP): Define.
5618 (dom_walker::walk): Do not compute bb to RPO mapping here.
5619 Support STOP return value from before_dom_children to stop
5621 * graphite-optimize-isl.c (optimize_isl): If the schedule
5622 is the same still generate code if -fgraphite-identity
5623 or -floop-parallelize-all are given.
5624 * graphite-scop-detection.c: Include cfganal.h.
5625 (gather_bbs::gather_bbs): Get and pass through bb to RPO
5627 (gather_bbs::before_dom_children): Return STOP for BBs
5629 (build_scops): Compute bb to RPO mapping and pass it to
5630 the domwalk. Treat --param graphite-max-nb-scop-params=0
5631 as not limiting the number of params.
5632 * graphite.c (graphite_initialize): Remove limit on the
5633 number of basic-blocks in a function.
5634 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
5635 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
5636 default value of 10.
5638 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5640 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
5641 Adjust code to eliminate needing to do the shift right 32-bits
5642 operation after XSCVDPSPN.
5644 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5646 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
5647 ((X / Y) != 0 -> X >= Y): Likewise.
5649 2017-09-26 Carl Love <cel@us.ibm.com>
5651 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
5652 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
5653 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
5654 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
5655 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
5656 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
5657 definitions and overloading.
5658 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
5659 statement for P9V_BUILTIN_XST_LEN_R.
5660 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
5661 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
5662 define_expand and define_insn for the instructions and builtins.
5663 * doc/extend.texi: Update the built-in documentation file for the new
5665 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
5666 define_insn for the instructions
5668 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
5671 * gcc/config/netbsd-protos.h: New file.
5672 * gcc/config/netbsd.c: New file.
5673 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
5674 * gcc/config/t-netbsd: New file.
5675 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
5676 (tmake_file) Add t-netbsd.
5677 (extra_objs) Add netbsd.o.
5679 2017-09-26 Janus Weil <janus@gcc.gnu.org>
5683 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
5685 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5687 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
5688 sign extension from a vector register to a GPR by doing a 32-bit
5689 direct move and then an EXTSW.
5690 (extendsi<mode>2 splitter): Likewise.
5691 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
5692 right or vector extract after doing XSCVDPSPN. Use
5693 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
5695 (movdi_from_sf_zero_ext): Likewise.
5696 (reload_gpr_from_vsxsf): Likewise.
5697 (p8_mfvsrd_4_disf): Delete, no longer used.
5698 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
5699 then needing to move the SFmode to a GPR to use the XSCVDPSP
5700 instruction instead of FRSP and XSCVDPSPN.
5701 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
5702 it is adjacent to the other XSCVSPDP insns.
5703 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
5704 SFmode to be in traditional Altivec registers.
5705 (vsx_xscvdpspn): Eliminate useless alternative constraint.
5706 (vsx_xscvspdpn): Likewise.
5707 (vsx_xscvspdpn_scalar): Likewise.
5709 2017-09-26 Martin Jambor <mjambor@suse.cz>
5711 * tree-sra.c (compare_access_positions): Put integral types first,
5712 stabilize sorting of integral types, remove conditions putting
5713 non-full-precision integers last.
5714 (sort_and_splice_var_accesses): Disable scalarization if a
5715 non-integert would be represented by a non-full-precision integer.
5717 2017-09-26 Joseph Myers <joseph@codesourcery.com>
5719 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
5720 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
5721 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
5722 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
5723 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
5724 conditionals inside the function instead of around it. Call
5725 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
5726 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
5728 2017-09-26 Richard Biener <rguenther@suse.de>
5730 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
5732 (scop_detection::build_scop_breadth): ... this. Removed.
5733 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
5734 (scop_detection::harmful_stmt_in_bb): Likewise.
5735 (scop_detection::graphite_can_represent_stmt): Likewise.
5736 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
5737 (scop_detection::can_represent_loop): Remove recursion, fold in ...
5738 (scop_detection::can_represent_loop_1): ... this. Removed.
5739 (scop_detection::harmful_loop_in_region): Simplify after inlining
5740 the above and remove more quadraticness.
5741 (build_scops): Adjust.
5742 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
5745 2017-09-26 Jakub Jelinek <jakub@redhat.com>
5748 * config/i386/i386.c (ix86_print_operand_address_as): Only test
5749 REGNO (base) == SP_REG if base is a REG.
5752 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
5753 if it is different SSA_NAME.
5754 (optimize_range_tests_cmp_bitwise): New function.
5755 (optimize_range_tests): Call it.
5757 2017-09-26 Richard Biener <rguenther@suse.de>
5759 PR tree-optimization/82321
5760 * graphite.c (canonicalize_loop_closed_ssa): Properly check
5761 for the def being inside the loop.
5763 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5765 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
5767 * config/s390/s390-builtins.def: Fix constraint on op4.
5769 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5771 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
5772 independent expanders.
5773 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
5774 ("vec_ordered", "vec_unordered"): New expanders.
5776 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5778 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
5781 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5783 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
5784 vec_unpacks_lo_v16qi.
5785 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
5787 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5789 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
5790 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
5791 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
5793 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5795 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
5797 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
5799 ("*vec_slb<mode>"): New insn pattern.
5800 ("vec_shr_<mode>"): New expander.
5801 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
5802 and force the shift count operand to V16QImode.
5803 ("vec_srb<mode>"): Set shift count mode to V16QI.
5805 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5807 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
5808 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
5809 ("vec_widen_smult_hi_<mode>"): New expander definitions.
5811 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
5814 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
5816 2017-09-26 Richard Biener <rguenther@suse.de>
5818 PR tree-optimization/82320
5819 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
5822 2017-09-25 Jeff Law <law@redhat.com>
5824 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
5825 prototype for new argument.
5826 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
5827 mostly extracted from rs6000_emit_allocate_stack.
5828 (rs6000_emit_probe_stack_range_stack_clash): New function.
5829 (rs6000_emit_allocate_stack): Call
5830 rs6000_emit_probe_stack_range_stack_clash as needed.
5831 (rs6000_emit_probe_stack_range): Add additional argument
5832 to call to gen_probe_stack_range{si,di}.
5833 (output_probe_stack_range): New.
5834 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
5835 (output_probe_stack_range_stack_clash): New.
5836 (rs6000_emit_prologue): Emit notes into dump file as requested.
5837 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
5838 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
5839 Add additional operand and pass it to output_probe_stack_range.
5841 2017-09-25 Bin Cheng <bin.cheng@arm.com>
5843 PR tree-optimization/82163
5844 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
5845 (checking_verify_loop_closed_ssa): New parameter.
5846 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
5847 (check_loop_closed_ssa_stmt): Delete.
5848 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
5849 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
5850 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
5853 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
5855 * brig-builtins.def: Treat HSAIL barrier builtins as
5856 setjmp/longjump style functions.
5858 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
5860 * target.def (constant_alignment): New hook.
5861 * defaults.h (CONSTANT_ALIGNMENT): Delete.
5862 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
5863 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
5864 * doc/tm.texi: Regenerate.
5865 * targhooks.h (default_constant_alignment): Declare.
5866 (constant_alignment_word_strings): Likewise.
5867 * targhooks.c (default_constant_alignment): New function.
5868 (constant_alignment_word_strings): Likewise.
5869 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
5870 instead of CONSTANT_ALIGNMENT.
5871 * varasm.c (align_variable, get_variable_align, build_constant_desc)
5872 (force_const_mem): Likewise.
5873 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
5874 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
5875 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
5876 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5877 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
5879 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
5880 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5881 constant_alignment_word_strings.
5882 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
5883 (CONSTANT_ALIGNMENT): Likewise.
5884 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5885 (arm_constant_alignment): New function.
5886 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
5887 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5888 constant_alignment_word_strings.
5889 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
5890 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5891 constant_alignment_word_strings.
5892 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
5893 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5894 (cris_constant_alignment): New function.
5895 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
5896 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5897 (epiphany_constant_alignment): New function.
5898 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
5899 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5900 constant_alignment_word_strings.
5901 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
5902 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5903 constant_alignment_word_strings.
5904 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
5905 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5906 constant_alignment_word_strings.
5907 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
5908 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
5909 * config/i386/i386.c (ix86_constant_alignment): Make static.
5910 Use the same interface as the target hook.
5911 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5912 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
5913 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5914 constant_alignment_word_strings.
5915 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
5916 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
5917 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5918 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
5919 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5920 constant_alignment_word_strings.
5921 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
5922 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5923 constant_alignment_word_strings.
5924 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
5925 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5926 constant_alignment_word_strings.
5927 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
5928 * config/microblaze/microblaze.c (microblaze_constant_alignment):
5930 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5931 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
5932 * config/mips/mips.c (mips_constant_alignment): New function.
5933 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5934 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
5935 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
5936 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5937 (mmix_constant_alignment): Make static. Use the same interface
5939 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
5940 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5941 constant_alignment_word_strings.
5942 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
5943 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5944 constant_alignment_word_strings.
5945 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
5946 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5947 constant_alignment_word_strings.
5948 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
5949 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5950 (rs6000_constant_alignment): New function.
5951 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
5952 * config/riscv/riscv.c (riscv_constant_alignment): New function.
5953 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5954 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
5955 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5956 (rs6000_constant_alignment): New function.
5957 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
5958 * config/s390/s390.c (s390_constant_alignment): New function.
5959 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5960 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
5961 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5962 constant_alignment_word_strings.
5963 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
5964 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5965 (sparc_constant_alignment): New function.
5966 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
5967 * config/spu/spu.c (spu_constant_alignment): New function.
5968 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5969 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
5970 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5971 constant_alignment_word_strings.
5972 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
5973 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5974 constant_alignment_word_strings.
5975 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
5976 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5977 constant_alignment_word_strings.
5978 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
5979 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5980 (visium_constant_alignment): New function.
5981 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
5982 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5983 (xtensa_constant_alignment): New function.
5984 * system.h (CONSTANT_ALIGNMENT): Poison.
5986 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
5988 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
5989 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
5990 (rs6000_builtin_valid_without_lhs): New helper function.
5991 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5992 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
5994 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
5996 * target.h (vec_perm_indices): Use unsigned short rather than
5998 (auto_vec_perm_indices): Likewise.
5999 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
6000 Use unsigned int rather than unsigned char.
6001 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
6003 2017-09-25 Richard Biener <rguenther@suse.de>
6005 * cfgloop.h (sort_sibling_loops): Declare.
6006 * cfgloop.c (sort_sibling_loops_cmp): New helper.
6007 (sort_sibling_loops): New function sorting the sibling loop list
6009 * graphite.c (graphite_transform_loops): Sort sibling loops.
6011 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
6013 * target.def (vec_perm_const_ok): Change sel parameter to
6015 * optabs-query.c (can_vec_perm_p): Update accordingly.
6016 * doc/tm.texi: Regenerate.
6017 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
6018 auto_vec_perm_indices and remove separate nelt field.
6019 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
6020 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
6021 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
6022 (aarch64_expand_vec_perm_const): Update accordingly.
6023 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
6024 to vec_perm_indices.
6025 * config/arm/arm.c (expand_vec_perm_d): Change perm to
6026 auto_vec_perm_indices and remove separate nelt field.
6027 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
6028 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
6029 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
6031 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
6032 to vec_perm_indices.
6033 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
6034 sel to vec_perm_indices.
6035 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
6036 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
6037 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
6039 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
6042 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
6045 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
6046 on the FUNCTION_DECL function context if it has a DIE that is a
6049 2017-09-25 Richard Biener <rguenther@suse.de>
6051 PR tree-optimization/82285
6052 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
6055 2017-09-25 Tom de Vries <tom@codesourcery.com>
6059 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
6062 2017-09-25 Richard Biener <rguenther@suse.de>
6064 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
6065 ISL errors other than isl_error_quota happen. Dump if the
6066 schedule is the same.
6067 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
6068 errors instead of aborting inside ISL.
6070 2017-09-25 Iain Sandoe <iain@codesourcery.com>
6073 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
6074 of libgcc_eh for m64.
6075 * config/i386/darwin64.h: Likewise.
6077 2017-09-25 Richard Biener <rguenther@suse.de>
6080 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
6081 attribute for incomplete types nor twice for complete ones.
6083 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
6086 * config/i386/i386.c (ix86_print_operand_address_as): Encode
6087 %esp as %rsp to avoid 0x67 prefix if there is no index or base
6090 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
6093 * config/i386/i386.opt (mprefer-avx256): Use
6094 ix86_target_flags variable.
6095 * config/i386/i386.c (ix86_target_string): Move
6096 -mprefer-avx256 to flag2_opts.
6098 2017-09-22 Jakub Jelinek <jakub@redhat.com>
6101 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
6102 and x != -1 | y != -1 into (x & y) != -1.
6104 2017-09-22 Steve Ellcey <sellcey@cavium.com>
6106 * config.gcc: Add new case statement to set
6107 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
6108 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
6109 s390x-*-linux* case statements. Added aarch64 to the list of
6110 supported architectures.
6112 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6114 PR tree-optimization/82289
6115 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
6116 STMT_VINFO_RELEVANT_P.
6118 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6119 Alan Hayward <alan.hayward@arm.com>
6120 David Sherwood <david.sherwood@arm.com>
6122 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
6123 for VR_RANGE only; don't allow VR_ANTI_RANGE.
6124 (extract_range_from_binary_expr_1): Don't call
6125 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
6127 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6128 Alan Hayward <alan.hayward@arm.com>
6129 David Sherwood <david.sherwood@arm.com>
6131 * target.def (preferred_vector_alignment): New hook.
6132 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
6134 * doc/tm.texi: Regenerate.
6135 * targhooks.h (default_preferred_vector_alignment): Declare.
6136 * targhooks.c (default_preferred_vector_alignment): New function.
6137 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
6139 (DR_TARGET_ALIGNMENT): New macro.
6140 (aligned_access_p): Update commentary.
6141 (vect_known_alignment_in_bytes): New function.
6142 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
6144 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
6145 Calculate the misalignment based on the target alignment rather than
6147 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
6148 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
6149 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
6150 the target alignment, rather than masking the element misalignment
6151 with the number of elements in a vector. Also use the target
6152 alignment when calculating the maximum number of peels.
6153 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
6154 instead of TYPE_ALIGN_UNIT.
6155 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
6156 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
6157 (vect_create_addr_base_for_vector_ref): Update call accordingly.
6158 (vect_create_data_ref_ptr): Likewise.
6159 (vect_setup_realignment): Realign by ANDing with
6160 -DR_TARGET_MISALIGNMENT.
6161 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
6162 the number of peels based on DR_TARGET_ALIGNMENT.
6163 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
6164 with the guaranteed alignment boundary when deciding whether
6166 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
6167 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
6168 (ensure_base_align): Remove stmt_info parameter. Get the
6169 target base alignment from DR_TARGET_ALIGNMENT.
6170 (vectorizable_store): Update call accordingly. Interpret
6171 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
6173 (vectorizable_load): Likewise.
6175 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6176 Alan Hayward <alan.hayward@arm.com>
6177 David Sherwood <david.sherwood@arm.com>
6179 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
6180 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
6181 (vect_enhance_data_refs_alignment): Likewise.
6183 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6185 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
6186 error. Only quit immediately if parsing is complete.
6187 (BEGIN): Initialize fatal_err and parse_done.
6188 (begin fpu, end fpu): Check number of arguments.
6189 (begin arch, end arch): Likewise.
6190 (begin cpu, end cpu): Likewise.
6191 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
6192 (optalias): Likewise.
6194 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6196 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
6197 * config/arm/arm-isa.h: Delete. Move definitions to ...
6198 * arm-cpus.in: ... here. Use new feature and fgroup values.
6199 * config/arm/arm.c (arm_option_override): Use lower case for feature
6201 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
6202 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
6203 * config/arm/parsecpu.awk (END): Add new command 'isa'.
6205 (print_isa_bits_for): New function.
6206 (gen_isa): New function.
6207 (gen_comm_data): Use print_isa_bits_for.
6208 (define feature): New keyword.
6209 (define fgroup): New keyword.
6210 * config/arm/t-arm (TM_H): Remove.
6211 (GTM_H): Add arm-isa.h.
6212 (arm-isa.h): Add rule to generate file.
6213 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
6214 case for feature bit names.
6216 2017-09-22 Richard Biener <rguenther@suse.de>
6218 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
6220 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
6221 print of no dependency loops ...
6222 * graphite.c (graphite_transform_loops): ... here.
6223 (canonicalize_loop_closed_ssa_form): Work from inner to outer
6225 (same_close_phi_node, remove_duplicate_close_phi,
6226 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
6227 (canonicalize_loop_closed_ssa): ... here and simplify.
6228 * graphite-optimize-isl.c: Include tree-vectorizer.h.
6229 (optimize_isl): Use dump_printf_loc to tell when we stopped
6230 optimizing because of an ISL timeout.
6232 2017-09-22 Richard Biener <rguenther@suse.de>
6234 PR tree-optimization/82291
6235 * tree-if-conv.c (predicate_mem_writes): Make sure to
6236 remove writes in blocks predicated with false.
6238 2017-09-22 Richard Biener <rguenther@suse.de>
6240 * sese.c: Include cfganal.h.
6241 (if_region_set_false_region): Remove.
6242 (create_if_region_on_edge): Likewise.
6243 (move_sese_in_condition): Re-implement without destroying
6246 2017-09-22 Richard Biener <rguenther@suse.de>
6248 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
6249 Verify both BBs contain loop PHI nodes before dispatching to
6251 (graphite_regenerate_ast_isl): Do not recompute dominators,
6252 do not verify three times. Restructure for clarity.
6253 * graphite-scop-detection.c (same_close_phi_node,
6254 remove_duplicate_close_phi, make_close_phi_nodes_unique,
6255 defined_in_loop_p, canonicalize_loop_closed_ssa,
6256 canonicalize_loop_closed_ssa_form): Simplify, remove excess
6257 checking and SSA rewrite, move to ...
6258 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
6259 (graphite_initialize): Do not pass in ctx, do not reset the
6260 SCEV cache, compute only dominators.
6261 (graphite_transform_loops): Allocate ISL ctx after
6262 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
6263 Maintain post-dominators only around build_scops.
6264 * sese.c (if_region_set_false_region): Make static. Free
6265 and recompute dominators.
6266 (move_sese_in_condition): Assert we don't get called with
6267 post-dominators computed.
6268 * sese.h (if_region_set_false_region): Remove.
6270 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
6272 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
6273 mode attribute for TARGET_AVX512VL.
6275 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
6277 * config/i386/i386.opt (mprefer-avx256): New option.
6278 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
6280 (ix86_preferred_simd_mode): Return 256-bit AVX modes
6281 for TARGET_PREFER_AVX256.
6282 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
6284 2017-09-21 Jeff Law <law@redhat.com>
6286 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
6287 Fix dump output if the only stack space is for pushed registers.
6289 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6291 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
6294 2017-09-21 Martin Sebor <msebor@redhat.com>
6297 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
6298 code (in C++) or code that triggers warnings.
6300 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
6302 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
6304 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
6306 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
6307 * sched-rgn.c: Ditto.
6308 * sel-sched-ir.c: Ditto.
6310 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
6312 * toplev.h (set_random_seed): Adjust return type.
6313 * toplev.c (init_local_tick): Move eager initialization of random_seed
6314 to get_random_seed. Adjust comment.
6315 (init_random_seed): Inline to get_random_seed, delete.
6316 (get_random_seed): Initialize random_seed lazily.
6317 (set_random_seed): Do not return previous value.
6318 (print_switch_value): Do not call get_random_seed.
6320 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
6322 * cgraph.c (delete_function_version): New, broken out from...
6323 (cgraph_node::delete_function_version): ...here. Rename to
6324 cgraph_node::delete_function_version_by_decl. Update all uses.
6325 (cgraph_node::remove): Call delete_function_version.
6327 2017-09-21 Jakub Jelinek <jakub@redhat.com>
6330 * tree-inline.c (expand_call_inline): Emit clobber stmts for
6331 VAR_DECLs to which addressable non-volatile parameters are mapped
6332 and for id->retvar after the return value assignment. Clear
6333 id->retval and id->retbnd after inlining.
6335 2017-09-21 Richard Biener <rguenther@suse.de>
6337 PR tree-optimization/82276
6338 PR tree-optimization/82244
6339 * tree-vrp.c (build_assert_expr_for): Set
6340 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
6342 (remove_range_assertions): Revert earlier change.
6344 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
6347 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
6349 2017-09-21 Richard Biener <rguenther@suse.de>
6351 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
6352 Restore valid IL after code generation errors.
6353 * graphite.c (graphite_transform_loops): Diagnose code
6354 generation issues as MSG_MISSED_OPTIMIZATION and continue
6355 with processing SCOPs.
6357 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6358 Alan Hayward <alan.hayward@arm.com>
6359 David Sherwood <david.sherwood@arm.com>
6361 * calls.c (compute_argument_addresses): Use simplify_gen_binary
6362 rather than choosing between plus_constant and gen_rtx_<CODE>.
6363 * expr.c (emit_push_insn): Likewise.
6364 (expand_expr_real_2): Likewise.
6366 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6367 Alan Hayward <alan.hayward@arm.com>
6368 David Sherwood <david.sherwood@arm.com>
6370 * loop-unroll.c (split_iv): Call copy_rtx on the step.
6372 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6373 Alan Hayward <alan.hayward@arm.com>
6374 David Sherwood <david.sherwood@arm.com>
6376 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
6377 calling tree_to_uhwi.
6379 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6380 Alan Hayward <alan.hayward@arm.com>
6381 David Sherwood <david.sherwood@arm.com>
6383 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
6384 INTEGER_CST rather than a negative test for ADDR_EXPR.
6386 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6387 Alan Hayward <alan.hayward@arm.com>
6388 David Sherwood <david.sherwood@arm.com>
6390 * tree-vrp.c (extract_range_from_binary_expr_1): Check
6391 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
6393 2017-09-21 Richard Biener <rguenther@suse.de>
6395 PR tree-optimization/71351
6396 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
6397 graphite_create_new_loop_guard): Remove, fold remaining parts
6399 (translate_isl_ast_node_for): ... here and simplify.
6401 2017-09-21 Jakub Jelinek <jakub@redhat.com>
6404 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
6405 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
6406 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
6407 alternative always use QI mode, for -Os imov (=R,R) alternative
6408 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
6411 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6412 Jeff Law <law@redhat.com>
6414 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
6415 (allocate_stack_space): New function, partially extracted from
6417 (s390_emit_prologue): Track offset to most recent stack probe.
6418 Code to allocate space moved into allocate_stack_space.
6419 Dump actions when no stack is allocated.
6420 (s390_prologue_plus_offset): New function.
6421 (s390_emit_stack_probe): Likewise.
6423 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
6425 * common.opt (Wa, Wl, Wp, g, gz=): Add
6427 (gno-column-info): Remove.
6428 (gcolumn-info): Drop RejectNegative.
6430 (gno-record-gcc-switches): Remove.
6431 (grecord-gcc-switches): Drop RejectNegative.
6432 (gno-split-dwarf): Remove.
6433 (gsplit-dwarf): Drop RejectNegative.
6434 (gno-strict-dwarf): Remove.
6435 (gstrict-dwarf): Drop RejectNegative.
6436 * config/darwin.opt (gfull, gused): Add RejectNegative.
6437 * dwarf2out.c (gen_producer_string): Drop
6438 gno-record-gcc-switches handler.
6439 * optc-gen.awk: Add g to prefixes with negative forms.
6440 * opts-common.c (remapping_prefix_p): New.
6441 (find_opt): Check it.
6442 (generate_canonical_option): Test g prefix.
6443 (option_map): Add -gno- mapping.
6444 (add_misspelling_candidates): Check remapping_prefix_p.
6446 2017-09-20 Jeff Law <law@redhat.com>
6448 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
6449 thinko in stack clash protection support.
6451 * explow.c (compute_stack_clash_protection_loop_data): Use
6452 CONST_INT_P instead of explicit test. Verify object is a
6453 CONST_INT_P before looking at INTVAL.
6454 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
6455 instead of explicit test.
6457 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
6460 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
6461 address instead of to r1 and r11.
6463 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
6465 * config.gcc: Support "knm".
6466 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
6467 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
6469 * config/i386/i386.c (m_KNM): Define.
6470 (processor_target_table): Add "knm".
6472 (ix86_option_override_internal): Add "knm".
6473 (ix86_issue_rate): Add PROCESSOR_KNM.
6474 (ix86_adjust_cost): Ditto.
6475 (ia32_multipass_dfa_lookahead): Ditto.
6476 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
6477 (fold_builtin_cpu): Add M_INTEL_KNM.
6478 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
6479 (processor_type): Add PROCESSOR_KNM.
6480 * config/i386/x86-tune.def: Add m_KNM.
6481 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
6483 2017-09-20 Richard Biener <rguenther@suse.de>
6485 PR tree-optimization/80213
6486 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
6487 are allowed in empty BBs as well.
6488 (canonicalize_loop_closed_ssa): Also look for other complex
6490 (scop_detection::get_sese): Include the loop-closed PHI block
6492 (scop_detection::merge_sese): Remove code adding extra blocks.
6493 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
6494 (build_scops): Assert the final returned scop is invalid.
6496 2017-09-20 Richard Biener <rguenther@suse.de>
6498 PR tree-optimization/82264
6499 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
6501 (vn_phi_lookup): Likewise.
6502 (vn_phi_insert): Likewise.
6504 2017-09-20 Jakub Jelinek <jakub@redhat.com>
6506 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
6507 that fits into uhwi or shwi, add DW_AT_const_value regardless
6508 of early_dwarf without going through RTL, using add_AT_unsigned
6511 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
6512 (DEBUG_LTO_ABBREV_SECTION): Likewise.
6513 (DEBUG_LTO_MACINFO_SECTION): Likewise.
6514 (DEBUG_MACRO_SECTION): Likewise.
6515 (DEBUG_LTO_MACRO_SECTION): Likewise.
6516 (DEBUG_STR_DWO_SECTION): Likewise.
6517 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
6518 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
6519 (DEBUG_LTO_DWO_LINE_SECTION): Define.
6520 (DEBUG_LTO_LINE_STR_SECTION): Define.
6521 (init_sections_and_labels): Initialize debug_line_str_section
6522 variable. Initialize debug_loc_section for -gdwarf-5 to
6523 DEBUG_LOCLISTS_SECTION. Formatting fixes.
6525 2017-09-20 Richard Biener <rguenther@suse.de>
6527 * graphite-sese-to-poly.c (extract_affine): Properly handle
6528 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
6530 2017-09-20 Richard Biener <rguenther@suse.de>
6532 PR tree-optimization/81373
6533 * graphite-scop-detection.c (build_cross_bb_scalars_def):
6534 Force SESE live-out defs to be handled even if they are
6537 2017-09-19 Jeff Law <law@redhat.com>
6539 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
6540 nothing for stack adjustments with REG_STACK_CHECK.
6541 * sched-deps.c (parse_add_or_inc): Reject insns with
6542 REG_STACK_CHECK from dependency breaking.
6543 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
6544 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
6545 * reg-notes.def (STACK_CHECK): New note.
6547 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
6548 (ix86_expand_prologue): Dump stack clash info as needed.
6549 Call ix86_adjust_stack_and_probe_stack_clash as needed.
6551 * function.c (dump_stack_clash_frame_info): New function.
6552 * function.h (dump_stack_clash_frame_info): Prototype.
6553 (enum stack_clash_probes): New enum.
6555 * config/alpha/alpha.c (alpha_expand_prologue): Also check
6556 flag_stack_clash_protection.
6557 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
6558 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
6559 (arm_frame_pointer_required): Likewise.
6560 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
6561 (ia64_expand_prologue): Likewise.
6562 * config/mips/mips.c (mips_expand_prologue): Likewise.
6563 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
6564 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6565 (sparc_flat_expand_prologue): Likewise.
6566 * config/spu/spu.c (spu_expand_prologue): Likewise.
6568 * explow.c: Include "params.h".
6569 (anti_adjust_stack_and_probe_stack_clash): New function.
6570 (get_stack_check_protect): Likewise.
6571 (compute_stack_clash_protection_loop_data): Likewise.
6572 (emit_stack_clash_protection_loop_start): Likewise.
6573 (emit_stack_clash_protection_loop_end): Likewise.
6574 (allocate_dynamic_stack_space): Use get_stack_check_protect.
6575 Use anti_adjust_stack_and_probe_stack_clash.
6576 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
6577 (emit_stack_clash_protection_loop_start): Likewise.
6578 (emit_stack_clash_protection_loop_end): Likewise.
6579 * rtl.h (get_stack_check_protect): Prototype.
6580 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
6581 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
6582 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
6584 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
6586 * doc/tm.texi: Rebuilt.
6587 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
6588 get_stack_check_protect.
6589 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
6590 * config/arm/arm.c (arm_expand_prologue): Likewise.
6591 (arm_frame_pointer_required): Likewise.
6592 * config/i386/i386.c (ix86_expand_prologue): Likewise.
6593 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
6594 * config/mips/mips.c (mips_expand_prologue): Likewise.
6595 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
6596 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
6597 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6598 (sparc_flat_expand_prologue): Likewise.
6600 * common.opt (-fstack-clash-protection): New option.
6601 * flag-types.h (enum stack_check_type): Note difference between
6602 -fstack-check= and -fstack-clash-protection.
6603 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
6604 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
6605 * toplev.c (process_options): Issue warnings/errors for cases
6606 not handled with -fstack-clash-protection.
6607 * doc/invoke.texi (-fstack-clash-protection): Document new option.
6608 (-fstack-check): Note additional problem with -fstack-check=generic.
6609 Note that -fstack-check is primarily for Ada and refer users
6610 to -fstack-clash-protection for stack-clash-protection.
6611 Document new params for stack clash protection.
6613 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
6615 * config/i386/i386.c (ix86_split_long_move): Do not handle
6616 address used for LEA in a special way.
6618 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
6620 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
6622 2017-09-19 Martin Sebor <msebor@redhat.com>
6625 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
6626 of incompatible types.
6628 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
6630 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6631 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
6632 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6633 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
6635 2017-09-19 Richard Biener <rguenther@suse.de>
6637 PR tree-optimization/82244
6638 * tree-vrp.c (remove_range_assertions): Do not propagate
6639 a constant to abnormals but replace the assert with a copy.
6641 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
6643 PR rtl-optimization/57878
6644 PR rtl-optimization/68988
6645 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
6646 avoidance test involving non_reload_pseudos. Move frequency test
6647 below the general fragmentation avoidance test.
6649 2017-09-19 Richard Biener <rguenther@suse.de>
6651 PR tree-optimization/69728
6652 * graphite-sese-to-poly.c (schedule_error): New global.
6653 (add_loop_schedule): Handle empty domain by failing the
6655 (build_original_schedule): Handle schedule_error.
6657 2017-09-19 Richard Biener <rguenther@suse.de>
6659 * graphite-scop-detection.c (scop_detection::can_represent_loop):
6660 Do not iterate to sibling loops but only to siblings of inner
6663 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
6666 * config/m68k/m68k.md (moveq feeding equality comparison): Check
6667 that the registers are different.
6669 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
6671 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
6672 to processor_model and "amdfam17h" to arch_names_table.
6673 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
6675 2017-09-18 Jakub Jelinek <jakub@redhat.com>
6678 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
6680 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6681 Alan Hayward <alan.hayward@arm.com>
6682 David Sherwood <david.sherwood@arm.com>
6684 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
6686 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
6688 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
6689 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
6690 vect_schedule_slp_instance.
6691 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
6692 Update call to vect_slp_analyze_node_operations. Simplify return
6694 (vect_slp_analyze_bb_1): Update call accordingly.
6695 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
6696 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
6697 (vect_schedule_slp): Update call accordingly.
6699 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6700 Alan Hayward <alan.hayward@arm.com>
6701 David Sherwood <david.sherwood@arm.com>
6703 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
6704 with types that aren't in fact scalar.
6706 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6708 * tree-vect-slp.c (vect_record_max_nunits): New function,
6710 (vect_build_slp_tree_1): ...here.
6711 (vect_build_slp_tree_2): Call it for phis too.
6713 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6715 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
6716 to vect_get_vec_def_for_operand when getting the mask operand.
6718 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6719 Alan Hayward <alan.hayward@arm.com>
6720 David Sherwood <david.sherwood@arm.com>
6722 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
6725 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6726 Alan Hayward <alan.hayward@arm.com>
6727 David Sherwood <david.sherwood@arm.com>
6729 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
6730 calculation for vector booleans.
6732 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6733 Alan Hayward <alan.hayward@arm.com>
6734 David Sherwood <david.sherwood@arm.com>
6736 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
6738 (vect_transform_stmt): ...here.
6739 (vect_analyze_stmt): Use it instead of calling
6740 vectorizable_live_operation directly.
6742 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
6744 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
6745 non-SIMT targets in acc vector loops.
6747 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
6749 * configure.ac: Add arc and check if assembler supports gdwarf2.
6750 * configure: Regenerate.
6752 2017-09-18 Richard Biener <rguenther@suse.de>
6754 PR tree-optimization/82220
6755 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
6756 epilogue niters from the min_profitable_iters compute.
6758 2017-09-18 Jakub Jelinek <jakub@redhat.com>
6761 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
6762 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
6763 (ix86_init_pic_reg): Revert 2017-09-01 changes.
6765 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
6768 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
6769 switching to a new text section.
6771 2017-09-18 Richard Biener <rguenther@suse.de>
6773 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
6775 (build_alias_set): Reject aliases with no access function.
6777 2017-09-18 Richard Biener <rguenther@suse.de>
6779 PR tree-optimization/79622
6780 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
6782 (build_cross_bb_scalars_use): Likewise.
6784 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
6786 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
6788 2017-09-18 Alan Modra <amodra@gmail.com>
6791 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
6792 stack_pointer_rtx for count 0. Update comments. Break up
6793 large rtl expression.
6795 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
6797 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
6798 Increase to 20 bytes.
6799 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
6800 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
6801 or avx version of the stub.
6803 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
6806 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
6807 compute the minimum stack alignment. Also update preferred stack
6808 boundary for leaf functions.
6810 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
6812 PR tree-optimization/82228
6813 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
6816 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
6818 * common/config/nds32/nds32-common.c
6819 (nds32_option_optimization_table): Refine formatting.
6820 (nds32_option_optimization_table): Use -fsched-pressure and
6821 -fomit-frame-pointer for specific optimization level.
6823 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
6825 * config/nds32/nds32.c: Refine formatting and comments.
6826 * config/nds32/nds32.h: Likewise.
6827 * config/nds32/nds32.md: Likewise.
6828 * config/nds32/nds32-cost.c: Likewise.
6829 * config/nds32/nds32-isr.c: Likewise.
6830 * config/nds32/nds32-md-auxiliary.c: Likewise.
6831 * config/nds32/nds32-multiple.md: Likewise.
6832 * config/nds32/nds32-predicates.c: Likewise.
6834 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
6835 Jakub Jelinek <jakub@redhat.com>
6837 Add support for -std=c++2a.
6838 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
6840 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
6842 2017-09-15 Steve Ellcey <sellcey@cavium.com>
6845 * doc/extend.texi (Common Function Attributes): Add
6846 references to ARM, AArch64, and S/390 specific attributes.
6847 (Function Specific Option Pragmas): Add AArch64 and S/390
6848 to list of back ends that support the target pragma.
6850 2017-09-15 Nathan Sidwell <nathan@acm.org>
6852 * doc/standards.texi: Fix C++17 description. Update URLs for
6855 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
6857 * common.opt (Wcast-align=strict): New warning option.
6858 * doc/invoke.texi: Document -Wcast-align=strict.
6860 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
6862 * cgraph.h (cgraph_thunk_info): Add comments.
6863 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
6864 assert for VIRTUAL_* arguments stricter.
6866 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
6868 PR tree-optimization/71026
6869 * match.pd: Move RDIV patterns from fold-const.c
6870 * fold-const.c (distribute_real_division): Removed.
6871 (fold_binary_loc): Remove calls to distribute_real_divison.
6873 2017-09-15 Jakub Jelinek <jakub@redhat.com>
6875 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
6876 c++1z and gnu++1z as deprecated. Change other references to
6877 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
6878 Change -Wc++1z-compat to -Wc++17-compat.
6879 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
6880 * dwarf2out.c (highest_c_language): Handle C++17.
6881 (gen_compile_unit_die): Likewise.
6883 2017-09-15 Jakub Jelinek <jakub@redhat.com>
6885 PR rtl-optimization/82192
6886 * combine.c (make_extraction): Don't look through non-paradoxical
6887 SUBREGs or TRUNCATE if pos + len is or might be bigger than
6890 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
6891 Alan Hayward <alan.hayward@arm.com>
6892 David Sherwood <david.sherwood@arm.com>
6894 * target.def (function_arg_offset): New hook.
6895 * targhooks.h (default_function_arg_offset): Declare.
6896 * targhooks.c (default_function_arg_offset): New function.
6897 * function.c (locate_and_pad_parm): Use
6898 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
6899 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
6900 (TARGET_FUNCTION_ARG_OFFSET): ...this.
6901 * doc/tm.texi: Regenerate.
6902 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
6903 * config/spu/spu.c (spu_function_arg_offset): New function.
6904 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
6905 * system.h (FUNCTION_ARG_OFFSET): Poison.
6907 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
6908 Alan Hayard <alan.hayward@arm.com>
6909 David Sherwood <david.sherwood@arm.com>
6911 * target.def (truly_noop_truncation): New hook.
6912 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
6913 than TRULY_NOOP_TRUNCATION.
6914 * hooks.h (hook_bool_uint_uint_true): Declare.
6915 * hooks.c (hook_bool_uint_uint_true): New function.
6916 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
6917 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
6918 * doc/tm.texi: Regenerate.
6919 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
6920 rather than TRULY_NOOP_TRUNCATION in comments.
6921 (simplify_comparison): Likewise.
6922 (record_truncated_value): Likewise.
6923 * expmed.c (extract_bit_field_1): Likewise.
6924 (extract_split_bit_field): Likewise.
6925 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
6926 instead of TRULY_NOOP_TRUNCATION.
6927 * function.c (assign_parm_setup_block): Likewise.
6928 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
6929 * rtlhooks.c: Include target.h.
6930 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
6931 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
6932 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
6933 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
6934 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
6935 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
6936 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
6937 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
6938 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
6939 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
6940 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
6941 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
6942 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
6943 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
6944 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
6945 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
6946 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
6947 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
6948 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
6949 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
6950 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
6951 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
6952 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
6953 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
6954 * config/mips/mips.c (mips_truly_noop_truncation): New function.
6955 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
6956 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
6957 rather than TRULY_NOOP_TRUNCATION in comments.
6958 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
6959 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
6960 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
6961 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
6962 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
6963 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
6964 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
6965 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
6966 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
6967 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
6968 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
6969 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
6970 rather than TRULY_NOOP_TRUNCATION in comments.
6971 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
6972 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
6973 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
6974 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
6975 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
6976 TRULY_NOOP_TRUNCATION condition.
6977 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
6978 (TRULY_NOOP_TRUNCATION): Delete.
6979 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
6980 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
6981 * config/spu/spu.c (spu_truly_noop_truncation): New function.
6982 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
6983 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
6984 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
6985 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
6986 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
6987 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
6988 rather than TRULY_NOOP_TRUNCATION in comments.
6989 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
6990 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
6991 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
6992 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
6993 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
6994 * system.h (TRULY_NOOP_TRUNCATION): Poison.
6996 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
6999 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
7000 (*cmp_ior): Likewise.
7001 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
7002 (*ior_scc_scc_cmp): Likewise.
7003 (*and_scc_scc): Likewise.
7004 (*and_scc_scc_cmp): Likewise.
7006 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7007 Alan Hayard <alan.hayward@arm.com>
7008 David Sherwood <david.sherwood@arm.com>
7010 * target.def (can_change_mode_class): New hook.
7011 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
7012 (hard_regno_nregs): Likewise.
7013 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
7014 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
7015 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
7016 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
7017 (LOAD_EXTEND_OP): Update accordingly.
7018 * doc/tm.texi: Regenerate.
7019 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
7020 CANNOT_CHANGE_MODE_CLASS.
7021 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
7022 (REG_CAN_CHANGE_MODE_P): ...this new macro.
7023 * combine.c (simplify_set): Update accordingly.
7024 * emit-rtl.c (validate_subreg): Likewise.
7025 * recog.c (general_operand): Likewise.
7026 * regcprop.c (mode_change_ok): Likewise.
7027 * reload1.c (choose_reload_regs): Likewise.
7028 (inherit_piecemeal_p): Likewise.
7029 * rtlanal.c (simplify_subreg_regno): Likewise.
7030 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
7031 instead of CANNOT_CHANGE_MODE_CLASS.
7032 (reload_cse_simplify_operands): Likewise.
7033 * reload.c (push_reload): Use targetm.can_change_mode_class
7034 instead of CANNOT_CHANGE_MODE_CLASS.
7035 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
7036 REG_CANNOT_CHANGE_MODE_P.
7037 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7038 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
7039 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7040 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7041 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7042 (arm_can_change_mode_class): New function.
7043 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7044 than CANNOT_CHANGE_MODE_CLASS in comments.
7045 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7046 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
7047 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
7048 (ix86_can_change_mode_class): ...this new function, inverting the
7049 sense of the return value.
7050 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7051 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7052 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7053 (ia64_can_change_mode_class): New function.
7054 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7055 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
7056 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
7057 (m32c_can_change_mode_class): ...this new function, inverting the
7058 sense of the return value.
7059 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7060 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7061 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
7062 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
7063 (mips_can_change_mode_class): ...this new function, inverting the
7064 sense of the return value.
7065 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7066 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7067 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7068 (msp430_can_change_mode_class): New function.
7069 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7070 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
7071 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7072 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7073 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7074 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
7075 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7076 (pa_cannot_change_mode_class): Replace with...
7077 (pa_can_change_mode_class): ...this new function, inverting the
7078 sense of the return value.
7079 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7080 than CANNOT_CHANGE_MODE_CLASS in comments.
7081 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7082 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
7083 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7084 (pdp11_cannot_change_mode_class): Replace with...
7085 (pdp11_can_change_mode_class): ...this new function, inverting the
7086 sense of the return value.
7087 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7088 * config/powerpcspe/powerpcspe-protos.h
7089 (rs6000_cannot_change_mode_class_ptr): Delete.
7090 * config/powerpcspe/powerpcspe.c
7091 (rs6000_cannot_change_mode_class_ptr): Delete.
7092 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7093 (rs6000_option_override_internal): Assign to
7094 targetm.can_change_mode_class instead of
7095 rs6000_cannot_change_mode_class_ptr.
7096 (rs6000_cannot_change_mode_class): Replace with...
7097 (rs6000_can_change_mode_class): ...this new function, inverting the
7098 sense of the return value.
7099 (rs6000_debug_cannot_change_mode_class): Replace with...
7100 (rs6000_debug_can_change_mode_class): ...this new function.
7101 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7102 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
7103 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7104 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7105 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
7107 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
7108 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7109 (rs6000_option_override_internal): Assign to
7110 targetm.can_change_mode_class instead of
7111 rs6000_cannot_change_mode_class_ptr.
7112 (rs6000_cannot_change_mode_class): Replace with...
7113 (rs6000_can_change_mode_class): ...this new function, inverting the
7114 sense of the return value.
7115 (rs6000_debug_cannot_change_mode_class): Replace with...
7116 (rs6000_debug_can_change_mode_class): ...this new function.
7117 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7118 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
7119 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
7120 (s390_can_change_mode_class): ...this new function, inverting the
7121 sense of the return value.
7122 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7123 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7124 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
7125 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7126 (sh_cannot_change_mode_class): Replace with...
7127 (sh_can_change_mode_class): ...this new function, inverting the
7128 sense of the return value.
7129 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7130 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7131 (sparc_can_change_mode_class): New function.
7132 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7133 * config/spu/spu.c (spu_can_change_mode_class): New function.
7134 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7135 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7136 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7137 (visium_can_change_mode_class): New function.
7138 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
7140 2017-09-15 Richard Biener <rguenther@suse.de>
7142 PR tree-optimization/82217
7143 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
7144 but not undefined case.
7146 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7149 * postreload.c (reload_cse_simplify_operands): Skip
7150 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
7152 2017-09-15 Richard Biener <rguenther@suse.de>
7154 PR tree-optimization/68823
7155 * graphite-scop-detection.c (build_alias_set): If we have a
7156 possible dependence check whether we can handle them by just
7157 looking at the DRs DR_ACCESS_FNs.
7158 (build_scops): If build_alias_set fails, fail the SCOP.
7160 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
7162 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
7163 to support float128 built-in functions that require the ISA 3.0
7165 (BU_FLOAT128_3_HW): Likewise.
7166 (SQRTF128): Add support for the IEEE 128-bit square root and fma
7168 (FMAF128): Likewise.
7170 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
7171 support for built-in functions that need the ISA 3.0 IEEE 128-bit
7172 floating point instructions.
7173 (rs6000_invalid_builtin): Likewise.
7174 (rs6000_builtin_mask_names): Likewise.
7175 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
7176 (RS6000_BTM_FLOAT128_HW): Likewise.
7177 (RS6000_BTM_COMMON): Likewise.
7178 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
7180 * doc/extend.texi (RS/6000 built-in functions): Document the
7181 IEEE 128-bit floating point square root and fused multiply-add
7184 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
7186 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
7187 reg (r2) isn't in the set of registers defined in the prologue.
7189 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7190 Alan Hayward <alan.hayward@arm.com>
7191 David Sherwood <david.sherwood@arm.com>
7193 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
7194 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
7195 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
7196 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
7197 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
7199 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
7200 max_vectorization_factor.
7201 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
7203 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7204 Alan Hayward <alan.hayward@arm.com>
7205 David Sherwood <david.sherwood@arm.com>
7207 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
7208 (vect_worthwhile_without_simd_p): Declare.
7209 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
7210 (vectorizable_reduction): Use it.
7211 * tree-vect-stmts.c (vectorizable_shift): Likewise.
7212 (vectorizable_operation): Likewise.
7214 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7215 Alan Hayward <alan.hayward@arm.com>
7216 David Sherwood <david.sherwood@arm.com>
7218 * tree-vectorizer.h (vect_get_num_copies): New function.
7219 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
7220 * tree-vect-loop.c (vectorizable_reduction): Likewise.
7221 (vectorizable_induction): Likewise.
7222 (vectorizable_live_operation): Likewise.
7223 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7224 (vectorizable_bswap): Likewise.
7225 (vectorizable_call): Likewise.
7226 (vectorizable_conversion): Likewise.
7227 (vectorizable_assignment): Likewise.
7228 (vectorizable_shift): Likewise.
7229 (vectorizable_operation): Likewise.
7230 (vectorizable_store): Likewise.
7231 (vectorizable_load): Likewise.
7232 (vectorizable_condition): Likewise.
7233 (vectorizable_comparison): Likewise.
7234 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
7236 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7237 Alan Hayward <alan.hayward@arm.com>
7238 David Sherwood <david.sherwood@arm.com>
7240 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
7241 of vect_init_vector.
7243 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7244 Alan Hayward <alan.hayward@arm.com>
7245 David Sherwood <david.sherwood@arm.com>
7247 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
7248 an inline wrapper that provides a location.
7249 (gimple_build_vector): Likewise.
7250 * gimple-fold.c (gimple_build_vector_from_val): New function.
7251 (gimple_build_vector): Likewise.
7252 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
7253 functions to build the initial value. Always return a gimple value.
7254 (get_initial_defs_for_reduction): Likewise. Only compute
7256 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
7257 vect_init_vector on the results from get_initial_def(s)_for_reduction.
7258 (vectorizable_induction): Use gimple_build_vector rather than
7261 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7262 Alan Hayward <alan.hayward@arm.com>
7263 David Sherwood <david.sherwood@arm.com>
7265 * target.h (vec_perm_indices): New typedef.
7266 (auto_vec_perm_indices): Likewise.
7267 * optabs-query.h: Include target.h
7268 (can_vec_perm_p): Take a vec_perm_indices *.
7269 * optabs-query.c (can_vec_perm_p): Likewise.
7270 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
7271 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7272 * tree-vect-generic.c (lower_vec_perm): Likewise.
7273 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
7274 (vect_grouped_load_supported): Likewise.
7275 (vect_shift_permute_load_chain): Likewise.
7276 (vect_permute_store_chain): Use auto_vec_perm_indices.
7277 (vect_permute_load_chain): Likewise.
7278 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
7279 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
7280 Update uses of can_vec_perm_p.
7281 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
7282 mode with a number of elements. Take a vec_perm_indices *.
7283 (vect_create_epilog_for_reduction): Update accordingly.
7284 Use auto_vec_perm_indices.
7285 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
7286 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
7287 (vect_transform_slp_perm_load): Likewise.
7288 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
7289 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
7290 (vect_gen_perm_mask_checked): Likewise.
7291 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
7292 (vect_gen_perm_mask_checked): Likewise.
7293 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
7294 (vectorizable_store): Likewise.
7295 (vectorizable_load): Likewise.
7296 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
7297 (vectorizable_bswap): Likewise.
7299 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7300 Alan Hayward <alan.hayward@arm.com>
7301 David Sherwood <david.sherwood@arm.com>
7303 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
7304 * tree.c (build_vector): Likewise.
7305 (build_vector_from_ctor): Update accordingly.
7306 (build_vector_from_val): Likewise.
7307 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
7308 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7309 * tree-vect-generic.c (add_rshift): Likewise.
7310 (expand_vector_divmod): Likewise.
7311 (optimize_vector_constructor): Likewise.
7312 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
7313 (vect_transform_slp_perm_load): Likewise.
7314 (vect_schedule_slp_instance): Likewise.
7315 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
7316 (vectorizable_call): Likewise.
7317 (vect_gen_perm_mask_any): Likewise. Add elements in order.
7318 * expmed.c (make_tree): Likewise.
7319 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
7320 a vector passed to build_vector.
7321 (fold_convert_const): Likewise.
7322 (exact_inverse): Likewise.
7323 (fold_ternary_loc): Likewise.
7324 (fold_relational_const): Likewise.
7325 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
7326 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
7327 (const_unop): Likewise. Store the reduction accumulator in a
7328 variable rather than an array.
7329 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
7330 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
7331 the new vector, rather than constructing it after the input arrays.
7332 (native_interpret_vector): Use auto_vec<tree> when building
7333 a vector passed to build_vector. Add elements in order.
7334 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
7335 auto_vec<tree> when building a vector passed to build_vector.
7336 (vect_create_epilog_for_reduction): Likewise.
7337 (vectorizable_induction): Likewise.
7338 (get_initial_def_for_reduction): Likewise. Fix indentation of
7340 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
7342 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
7343 passed to build_vector.
7345 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7346 Alan Hayward <alan.hayward@arm.com>
7347 David Sherwood <david.sherwood@arm.com>
7349 * tree-core.h (tree_base::u): Add an "nelts" field.
7350 (tree_vector): Use VECTOR_CST_NELTS as the length.
7351 * tree.c (tree_size): Likewise.
7352 (make_vector): Initialize VECTOR_CST_NELTS.
7353 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
7354 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
7355 TYPE_VECTOR_SUBPARTS.
7356 * expr.c (const_vector_mask_from_tree): Consistently use "units"
7357 as the number of units, setting it from VECTOR_CST_NELTS.
7358 (const_vector_from_tree): Likewise.
7359 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
7360 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
7361 (fold_negate_expr_1): Likewise.
7362 (fold_convert_const): Likewise.
7363 (const_binop): Likewise. Differentiate the number of output and
7365 (const_unop): Likewise.
7366 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
7367 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
7368 in cases that did the opposite.
7370 2017-09-14 Richard Biener <rguenther@suse.de>
7372 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
7375 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
7377 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
7379 2017-09-14 Jakub Jelinek <jakub@redhat.com>
7382 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
7383 if and where to split a bb, except for splitting before debug insn
7384 sequences followed by non-label real insn. Delete debug insns
7385 in between basic blocks.
7387 * combine.c (make_compound_operation_int): Formatting fixes.
7389 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
7390 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7391 * config/netbsd.h (LINK_EH_SPEC): Likewise.
7392 * config/sol2.h (LINK_EH_SPEC): Likewise.
7393 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7394 * config/s390/linux.h (LINK_SPEC): Likewise.
7395 * config/freebsd.h (LINK_EH_SPEC): Likewise.
7396 * config/openbsd.h (LINK_EH_SPEC): Likewise.
7397 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7398 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
7399 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
7400 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7401 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
7402 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
7404 2017-09-13 Jakub Jelinek <jakub@redhat.com>
7406 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
7408 (ENDFILE_LINUX_SPEC): Likewise.
7409 (LINK_EH_SPEC): Likewise.
7410 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
7411 (LINK_OS_LINUX_SPEC32): Likewise.
7412 (LINK_OS_LINUX_SPEC64): Likewise.
7413 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
7414 (LINK_OS_LINUX_SPEC): Likewise.
7416 2017-09-13 Martin Liska <mliska@suse.cz>
7419 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
7420 CASE_HIGH is NULL_TREE.
7422 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
7423 Alan Hayward <alan.hayward@arm.com>
7424 David Sherwood <david.sherwood@arm.com>
7426 * target.def (secondary_memory_needed): New hook.
7427 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
7428 instead of SECONDARY_MEMORY_NEEDED.
7429 (secondary_memory_needed_mode): Likewise.
7430 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
7431 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
7432 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
7433 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
7434 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
7435 * doc/tm.texi: Regenerate.
7436 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
7437 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
7438 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7439 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
7440 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
7441 * config/i386/i386.c (inline_secondary_memory_needed): Put the
7442 mode argument first and change the reg_class arguments to reg_class_t.
7443 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
7444 Make static. Update the call to inline_secondary_memory_needed.
7445 (ix86_register_move_cost): Update the call to
7446 inline_secondary_memory_needed.
7447 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7448 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
7450 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
7451 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7453 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
7454 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
7455 * config/mips/mips.c (mips_secondary_memory_needed): Make static
7456 and match hook interface. Add comment from mips.h.
7457 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7458 * config/mmix/mmix.md (truncdfsf2): Refer to
7459 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7461 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
7462 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
7463 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7464 (pa_secondary_memory_needed): New function.
7465 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
7466 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
7467 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7468 (pdp11_secondary_memory_needed): Make static and match hook interface.
7469 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
7470 * config/powerpcspe/powerpcspe-protos.h
7471 (rs6000_secondary_memory_needed_ptr): Delete.
7472 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
7474 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7475 (rs6000_option_override_internal): Assign to
7476 targetm.secondary_memory_needed rather than
7477 rs6000_secondary_memory_needed_ptr.
7478 (rs6000_secondary_memory_needed): Match hook interface.
7479 (rs6000_debug_secondary_memory_needed): Likewise.
7480 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
7481 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
7482 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
7483 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7484 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
7485 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
7487 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
7488 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7489 (rs6000_option_override_internal): Assign to
7490 targetm.secondary_memory_needed rather than
7491 rs6000_secondary_memory_needed_ptr.
7492 (rs6000_secondary_memory_needed): Match hook interface.
7493 (rs6000_debug_secondary_memory_needed): Likewise.
7494 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
7495 * config/s390/s390.c (s390_secondary_memory_needed): New function.
7496 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7497 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
7498 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7499 (sparc_secondary_memory_needed): New function.
7500 * lra-constraints.c (check_and_process_move): Refer to
7501 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7503 (curr_insn_transform): Likewise.
7504 (process_alt_operands): Use targetm.secondary_memory_needed
7505 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7506 (check_secondary_memory_needed_p): Likewise.
7507 (choose_split_class): Likewise.
7508 * reload.c: Unconditionally include code that was previously
7509 conditional on SECONDARY_MEMORY_NEEDED.
7510 (push_secondary_reload): Use targetm.secondary_memory_needed
7511 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7512 (push_reload): Likewise.
7513 * reload1.c: Unconditionally include code that was previously
7514 conditional on SECONDARY_MEMORY_NEEDED.
7515 (choose_reload_regs): Use targetm.secondary_memory_needed
7516 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7517 (gen_reload): Likewise.
7518 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
7520 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
7521 Alan Hayward <alan.hayward@arm.com>
7522 David Sherwood <david.sherwood@arm.com>
7524 * target.def (secondary_memory_needed_mode): New hook:
7525 * targhooks.c (default_secondary_memory_needed_mode): Declare.
7526 * targhooks.h (default_secondary_memory_needed_mode): New function.
7527 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
7528 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
7529 * doc/tm.texi: Regenerate.
7530 * lra-constraints.c (check_and_process_move): Use
7531 targetm.secondary_memory_needed_mode instead of
7532 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
7533 (curr_insn_transform): Likewise.
7534 * reload.c (get_secondary_mem): Likewise.
7535 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7536 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
7538 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7539 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7540 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
7541 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7542 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
7544 * config/powerpcspe/powerpcspe-protos.h
7545 (rs6000_secondary_memory_needed_mode): Delete.
7546 * config/powerpcspe/powerpcspe.c
7547 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7548 (rs6000_secondary_memory_needed_mode): Make static.
7549 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7550 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
7552 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7554 (rs6000_secondary_memory_needed_mode): Make static.
7555 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7556 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
7557 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7558 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7559 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7561 (sparc_secondary_memory_needed_mode): New function.
7562 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
7564 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
7566 * config/aarch64/constraints.md (Umq): New constraint.
7567 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
7569 (mov<mode>): Update condition.
7571 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7573 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
7574 when bitposition is the same.
7576 2017-09-13 Richard Biener <rguenther@suse.de>
7578 * dwarf2out.c (output_die_symbol): Remove.
7579 (output_die): Do not output a DIEs symbol.
7581 2017-09-13 Richard Biener <rguenther@suse.de>
7584 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
7585 default-def to avoid breaking iterator update with the weird
7586 interaction with cgraph_update_edges_for_call_stmt_node.
7588 2017-09-13 Richard Biener <rguenther@suse.de>
7590 * tree-cfg.c (verify_gimple_assign_binary): Add verification
7591 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
7592 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
7593 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
7595 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
7597 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
7598 Disable pc relative literal load irrespective of
7599 TARGET_FIX_ERR_A53_84341 for default.
7601 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
7603 * config/sparc/sparc.c (output_return): Output the source location of
7604 the insn in the delay slot, if any.
7605 (output_sibcall): Likewise.
7607 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
7610 * common.opt (-static-pie): New alias.
7611 (shared): Negate static-pie.
7612 (-no-pie): Update help text.
7614 (static-pie): New option.
7615 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
7616 -static-pie support.
7617 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
7618 (LINK_EH_SPEC): Likewise.
7619 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7620 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7621 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7622 * gcc.c (LINK_COMMAND_SPEC): Likewise.
7623 (init_gcc_specs): Likewise.
7624 (init_spec): Likewise.
7625 (display_help): Update help message for -pie.
7626 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
7629 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
7631 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
7632 (movdi_aarch64): Likewise.
7633 (movti_aarch64): Likewise.
7635 2017-09-12 Simon Wright <simon@pushface.org>
7638 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
7639 calculation of the minor version, always output as 0.
7641 2017-09-12 Jakub Jelinek <jakub@redhat.com>
7644 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
7645 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
7646 on it early, rather than manual conversion late. For
7647 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
7648 instead of performing manual conversion.
7650 2017-09-12 Carl Love <cel@us.ibm.com>
7652 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
7653 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
7655 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7656 VMULOSW): Add definitions.
7657 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7658 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7659 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7660 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7661 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7663 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
7665 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
7667 (movti_aarch64): Likewise.
7668 (movdf_aarch64): Likewise.
7669 (movtf_aarch64): Likewise.
7670 (load_pairdi): Likewise.
7671 (store_pairdi): Likewise.
7672 (load_pairdf): Likewise.
7673 (store_pairdf): Likewise.
7674 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
7675 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
7676 (ldr_got_small_<mode>): Likewise.
7677 (ldr_got_small_28k_<mode>): Likewise.
7678 (ldr_got_tiny): Likewise.
7679 * config/aarch64/iterators.md (ldst_sz): New.
7680 (ldpstp_sz): Likewise.
7681 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
7683 (thunderx_load): Split load_8 to load_16.
7684 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
7686 (thunderx2t99_storepair_basic): Split store_8 to store_16.
7687 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
7688 (xgene1_store_pair): Split store_8 to store_16.
7689 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
7690 (falkor_st_0_st_sd): Split store_8 to store_16.
7692 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
7694 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
7695 and store1/2/3/4 to store_4/8/12/16.
7696 * config/aarch64/aarch64.md: Update for rename.
7697 * config/arm/arm.md: Likewise.: Likewise.
7698 * config/arm/arm.c: Likewise.
7699 * config/arm/thumb1.md: Likewise.
7700 * config/arm/thumb2.md: Likewise.
7701 * config/arm/vfp.md: Likewise.
7702 * config/arm/arm-generic.md: Likewise.
7703 * config/arm/arm1020e.md: Likewise.
7704 * config/arm/arm1026ejs.md: Likewise.
7705 * config/arm/arm1136jfs.md: Likewise.
7706 * config/arm/arm926ejs.md: Likewise.
7707 * config/arm/cortex-a15.md: Likewise.
7708 * config/arm/cortex-a17.md: Likewise.
7709 * config/arm/cortex-a5.md: Likewise.
7710 * config/arm/cortex-a53.md: Likewise.
7711 * config/arm/cortex-a57.md: Likewise.
7712 * config/arm/cortex-a7.md: Likewise.
7713 * config/arm/cortex-a8.md: Likewise.
7714 * config/arm/cortex-a9.md: Likewise.
7715 * config/arm/cortex-m4.md: Likewise.
7716 * config/arm/cortex-m7.md: Likewise.
7717 * config/arm/cortex-r4.md: Likewise.
7718 * config/arm/exynos-m1.md: Likewise.
7719 * config/arm/fa526.md: Likewise.
7720 * config/arm/fa606te.md: Likewise.
7721 * config/arm/fa626te.md: Likewise.
7722 * config/arm/fa726te.md: Likewise.
7723 * config/arm/fmp626.md: Likewise.
7724 * config/arm/iwmmxt.md: Likewise.
7725 * config/arm/ldmstm.md: Likewise.
7726 * config/arm/marvell-pj4.md: Likewise.
7727 * config/arm/xgene1.md: Likewise.
7728 * config/aarch64/thunderx.md: Likewise.
7729 * config/aarch64/thunderx2t99.md: Likewise.
7730 * config/aarch64/falkor.md: Likewise.
7732 2017-09-12 Martin Liska <mliska@suse.cz>
7734 * attribs.c (private_lookup_attribute): New function.
7735 * attribs.h (private_lookup_attribute): Declared here.
7736 (lookup_attribute): Called from this place.
7738 2017-09-12 Richard Biener <rguenther@suse.de>
7740 PR tree-optimization/82157
7741 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
7742 stmts with side-effects.
7744 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7745 Alan Hayward <alan.hayward@arm.com>
7746 David Sherwood <david.sherwood@arm.com>
7748 * target.def (hard_regno_nregs): New hook.
7749 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
7750 * targhooks.h (default_hard_regno_nregs): Declare.
7751 * targhooks.c (default_hard_regno_nregs): New function.
7752 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
7753 (TARGET_HARD_REGNO_NREGS): ...this hook.
7754 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
7755 (CLASS_MAX_NREGS): Likewise.
7756 * doc/tm.texi: Regenerate.
7757 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
7758 instead of HARD_REGNO_NREGS.
7759 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
7760 HARD_REGNO_NREGS in the comment.
7761 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
7762 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
7763 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
7764 Return an unsigned int.
7765 (TARGET_HARD_REGNO_NREGS): Redefine.
7766 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
7767 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
7768 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
7769 (arc_hard_regno_nregs): New function.
7770 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
7771 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
7772 (arm_hard_regno_nregs): New function.
7773 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
7774 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
7775 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
7776 (TARGET_HARD_REGNO_NREGS): Redefine.
7777 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
7778 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
7779 (HARD_REGNO_NREGS): Delete.
7780 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
7781 (cr16_hard_regno_nregs): New function.
7782 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
7783 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
7784 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
7785 (cris_hard_regno_nregs): New function.
7786 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
7787 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
7788 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
7789 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
7790 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
7791 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
7792 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
7793 (frv_hard_regno_nregs): Make static. Take and return an
7795 (frv_class_max_nregs): Remove outdated copy of documentation.
7796 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
7797 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
7798 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
7799 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
7800 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
7801 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
7802 (TARGET_HARD_REGNO_NREGS): Redefine.
7803 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
7804 (CLASS_MAX_NREGS): Update comment.
7805 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
7806 (ia64_hard_regno_nregs): New function.
7807 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
7808 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
7809 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
7810 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
7811 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
7813 (m32c_hard_regno_nregs): Likewise. Make static.
7814 (TARGET_HARD_REGNO_NREGS): Redefine.
7815 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
7816 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
7817 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
7818 (m68k_hard_regno_nregs): New function.
7819 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
7820 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
7821 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
7822 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
7823 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
7824 Take and return an unsigned int.
7825 (TARGET_HARD_REGNO_NREGS): Redefine.
7826 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
7827 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
7828 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
7829 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
7830 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
7831 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
7832 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
7833 (msp430_hard_regno_nregs): Make static. Take and return an
7835 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
7836 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
7837 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
7838 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
7839 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
7840 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
7841 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
7842 (TARGET_HARD_REGNO_NREGS): Redefine.
7843 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
7844 (PA_HARD_REGNO_NREGS): ...this.
7845 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
7846 (PA_HARD_REGNO_NREGS): ...this.
7847 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
7848 (pa_hard_regno_nregs): New function.
7849 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
7850 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
7851 (pdp11_hard_regno_nregs): New function.
7852 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
7853 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
7854 (rs6000_hard_regno_nregs_hook): New function.
7855 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
7856 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
7857 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
7858 Take and return an unsigned int. Move earlier in file.
7859 (TARGET_HARD_REGNO_NREGS): Redefine.
7860 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
7861 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
7862 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
7863 (rl78_hard_regno_nregs): Make static. Take and return an
7865 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
7866 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
7867 (rs6000_hard_regno_nregs_hook): New function.
7868 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
7869 * config/rx/rx.c (rx_hard_regno_nregs): New function.
7870 (TARGET_HARD_REGNO_NREGS): Redefine.
7871 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
7872 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
7873 instead of HARD_REGNO_NREGS.
7874 (s390_hard_regno_nregs): New function.
7875 (s390_hard_regno_mode_ok): Add comment from s390.h.
7876 (TARGET_HARD_REGNO_NREGS): Redefine.
7877 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
7878 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
7879 (sh_hard_regno_nregs): New function.
7880 (sh_pass_in_reg_p): Use it.
7881 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
7882 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
7883 (sparc_hard_regno_nregs): New function.
7884 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
7885 * config/spu/spu.c (spu_hard_regno_nregs): New function.
7886 (spu_function_arg_advance): Use it, supplying a valid register number.
7887 (TARGET_HARD_REGNO_NREGS): Redefine.
7888 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
7889 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
7890 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
7891 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
7892 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
7893 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
7894 (CLASS_MAX_NREGS): Remove copy of old documentation.
7895 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
7896 (visium_hard_regno_nregs): New function.
7897 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
7898 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
7899 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
7900 xtensa_hard_regno_nregs): New function.
7901 * system.h (HARD_REGNO_NREGS): Poison.
7903 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7905 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
7906 hard_regno_nregs instead of HARD_REGNO_NREGS.
7907 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
7908 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
7909 (c6x_expand_epilogue): Likewise.
7910 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
7911 (frv_read_iacc_argument): Likewise.
7912 * config/sh/sh.c: Include regs.h.
7913 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
7914 (regs_used): Likewise.
7915 (output_stack_adjust): Likewise.
7916 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
7917 * expmed.c: Include regs.h.
7918 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
7919 * ree.c: Include regs.h.
7920 (combine_reaching_defs): Use hard_regno_nregs instead of
7922 (add_removable_extension): Likewise.
7924 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7926 * regs.h (hard_regno_nregs): Turn into a function.
7927 (end_hard_regno): Update accordingly.
7928 * caller-save.c (setup_save_areas): Likewise.
7929 (save_call_clobbered_regs): Likewise.
7930 (replace_reg_with_saved_mem): Likewise.
7931 (insert_restore): Likewise.
7932 (insert_save): Likewise.
7933 * combine.c (can_change_dest_mode): Likewise.
7934 (move_deaths): Likewise.
7935 (distribute_notes): Likewise.
7936 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
7937 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
7938 (rs6000_split_multireg_move): Likewise.
7939 (rs6000_register_move_cost): Likewise.
7940 (rs6000_memory_move_cost): Likewise.
7941 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
7942 (rs6000_split_multireg_move): Likewise.
7943 (rs6000_register_move_cost): Likewise.
7944 (rs6000_memory_move_cost): Likewise.
7945 * cselib.c (cselib_reset_table): Likewise.
7946 (cselib_lookup_1): Likewise.
7947 * emit-rtl.c (set_mode_and_regno): Likewise.
7948 * function.c (aggregate_value_p): Likewise.
7949 * ira-color.c (setup_profitable_hard_regs): Likewise.
7950 (check_hard_reg_p): Likewise.
7951 (calculate_saved_nregs): Likewise.
7952 (assign_hard_reg): Likewise.
7953 (improve_allocation): Likewise.
7954 (calculate_spill_cost): Likewise.
7955 * ira-emit.c (modify_move_list): Likewise.
7956 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
7957 (ira_hard_reg_in_set_p): Likewise.
7958 * ira.c (setup_reg_mode_hard_regset): Likewise.
7959 (clarify_prohibited_class_mode_regs): Likewise.
7960 (check_allocation): Likewise.
7961 * lra-assigns.c (find_hard_regno_for_1): Likewise.
7962 (lra_setup_reg_renumber): Likewise.
7963 (setup_try_hard_regno_pseudos): Likewise.
7964 (spill_for): Likewise.
7965 (assign_hard_regno): Likewise.
7966 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
7967 * lra-constraints.c (in_class_p): Likewise.
7968 (lra_constraint_offset): Likewise.
7969 (simplify_operand_subreg): Likewise.
7970 (lra_constraints): Likewise.
7971 (split_reg): Likewise.
7972 (split_if_necessary): Likewise.
7973 (invariant_p): Likewise.
7974 (inherit_in_ebb): Likewise.
7975 * lra-lives.c (process_bb_lives): Likewise.
7976 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
7977 (get_hard_regs): Likewise.
7978 (do_remat): Likewise.
7979 * lra-spills.c (assign_spill_hard_regs): Likewise.
7980 * mode-switching.c (create_pre_exit): Likewise.
7981 * postreload.c (reload_combine_recognize_pattern): Likewise.
7982 * recog.c (peep2_find_free_register): Likewise.
7983 * regcprop.c (kill_value_regno): Likewise.
7984 (set_value_regno): Likewise.
7985 (copy_value): Likewise.
7986 (maybe_mode_change): Likewise.
7987 (find_oldest_value_reg): Likewise.
7988 (copyprop_hardreg_forward_1): Likewise.
7989 * regrename.c (check_new_reg_p): Likewise.
7990 (regrename_do_replace): Likewise.
7991 * reload.c (push_reload): Likewise.
7992 (combine_reloads): Likewise.
7993 (find_dummy_reload): Likewise.
7994 (operands_match_p): Likewise.
7995 (find_reloads): Likewise.
7996 (find_equiv_reg): Likewise.
7997 (reload_adjust_reg_for_mode): Likewise.
7998 * reload1.c (count_pseudo): Likewise.
7999 (count_spilled_pseudo): Likewise.
8000 (find_reg): Likewise.
8001 (clear_reload_reg_in_use): Likewise.
8002 (free_for_value_p): Likewise.
8003 (allocate_reload_reg): Likewise.
8004 (choose_reload_regs): Likewise.
8005 (reload_adjust_reg_for_temp): Likewise.
8006 (emit_reload_insns): Likewise.
8007 (delete_output_reload): Likewise.
8008 * rtlanal.c (subreg_get_info): Likewise.
8009 * sched-deps.c (sched_analyze_reg): Likewise.
8010 * sel-sched.c (init_regs_for_mode): Likewise.
8011 (mark_unavailable_hard_regs): Likewise.
8012 (choose_best_reg_1): Likewise.
8013 (verify_target_availability): Likewise.
8014 * valtrack.c (dead_debug_insert_temp): Likewise.
8015 * var-tracking.c (track_loc_p): Likewise.
8016 (emit_note_insn_var_location): Likewise.
8017 * varasm.c (make_decl_rtl): Likewise.
8018 * reginfo.c (choose_hard_reg_mode): Likewise.
8019 (init_reg_modes_target): Refer directly to
8020 this_target_regs->x_hard_regno_nregs.
8022 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8024 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
8025 instead of hard_regno_nregs.
8027 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8029 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
8030 end_hard_regno instead of hard_regno_nregs.
8031 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
8032 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
8033 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
8034 * ira-color.c (improve_allocation): Likewise.
8035 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8036 * lra-lives.c (mark_regno_live): Likewise.
8037 (mark_regno_dead): Likewise.
8038 * lra-remat.c (operand_to_remat): Likewise.
8039 * lra.c (collect_non_operand_hard_regs): Likewise.
8040 * postreload.c (reload_combine_note_store): Likewise.
8041 (move2add_valid_value_p): Likewise.
8042 * reload.c (regno_clobbered_p): Likewise.
8044 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8046 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
8048 * config/v850/v850.c (v850_reorg): Likewise.
8049 * reload.c (refers_to_regno_for_reload_p): Likewise.
8050 (find_equiv_reg): Likewise.
8051 * reload1.c (reload_reg_reaches_end_p): Likewise.
8053 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8055 * caller-save.c (add_used_regs): Use REG_NREGS instead of
8057 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
8058 * config/arm/arm.c (output_move_neon): Likewise.
8059 (arm_attr_length_move_neon): Likewise.
8060 (neon_split_vcombine): Likewise.
8061 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
8062 (c6x_mark_reg_written): Likewise.
8063 (c6x_dwarf_register_span): Likewise.
8064 * config/i386/i386.c (ix86_save_reg): Likewise.
8065 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
8066 (rws_access_reg): Likewise.
8067 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
8068 * mode-switching.c (create_pre_exit): Likewise.
8069 * ree.c (combine_reaching_defs): Likewise.
8070 (add_removable_extension): Likewise.
8071 * regcprop.c (find_oldest_value_reg): Likewise.
8072 (copyprop_hardreg_forward_1): Likewise.
8073 * reload.c (reload_inner_reg_of_subreg): Likewise.
8074 (push_reload): Likewise.
8075 (combine_reloads): Likewise.
8076 (find_dummy_reload): Likewise.
8077 (reload_adjust_reg_for_mode): Likewise.
8078 * reload1.c (find_reload_regs): Likewise.
8079 (forget_old_reloads_1): Likewise.
8080 (reload_reg_free_for_value_p): Likewise.
8081 (reload_adjust_reg_for_temp): Likewise.
8082 (emit_reload_insns): Likewise.
8083 (delete_output_reload): Likewise.
8084 * sel-sched.c (choose_best_reg_1): Likewise.
8085 (choose_best_pseudo_reg): Likewise.
8087 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8088 Alan Hayward <alan.hayward@arm.com>
8089 David Sherwood <david.sherwood@arm.com>
8091 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
8092 * target.def (slow_unaligned_access): New hook.
8093 * targhooks.h (default_slow_unaligned_access): Declare.
8094 * targhooks.c (default_slow_unaligned_access): New function.
8095 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
8096 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
8097 * doc/tm.texi: Regenerate.
8098 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
8099 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
8100 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
8102 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
8103 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
8105 (rs6000_slow_unaligned_access): New function.
8106 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8107 (expand_block_compare): Likewise.
8108 (expand_strn_compare): Likewise.
8109 (rs6000_rtx_costs): Likewise.
8110 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
8111 (riscv_slow_unaligned_access): Likewise.
8112 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
8113 (riscv_slow_unaligned_access_p): ...this and make static.
8114 (riscv_option_override): Update accordingly.
8115 (riscv_slow_unaligned_access): New function.
8116 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8117 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
8118 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8119 (rs6000_slow_unaligned_access): New function.
8120 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8121 (rs6000_rtx_costs): Likewise.
8122 * config/rs6000/rs6000-string.c (expand_block_compare)
8123 (expand_strn_compare): Use targetm.slow_unaligned_access instead
8124 of SLOW_UNALIGNED_ACCESS.
8125 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
8126 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
8127 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
8128 of SLOW_UNALIGNED_ACCESS.
8129 * expmed.c (simple_mem_bitfield_p): Likewise.
8130 * expr.c (alignment_for_piecewise_move): Likewise.
8131 (emit_group_load_1): Likewise.
8132 (emit_group_store): Likewise.
8133 (copy_blkmode_from_reg): Likewise.
8134 (emit_push_insn): Likewise.
8135 (expand_assignment): Likewise.
8136 (store_field): Likewise.
8137 (expand_expr_real_1): Likewise.
8138 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
8139 * lra-constraints.c (simplify_operand_subreg): Likewise.
8140 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
8141 * gimple-ssa-store-merging.c: Likewise in block comment at start
8143 * tree-ssa-strlen.c: Include target.h.
8144 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
8145 of SLOW_UNALIGNED_ACCESS.
8146 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
8148 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8150 PR rtl-optimization/82185
8151 * expmed.c (emit_store_flag_int): Only test tem if it has been
8154 2017-09-12 Richard Biener <rguenther@suse.de>
8157 * match.pd ((FTYPE) N CMP CST): Fix typo.
8159 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
8161 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
8163 (mips_near_type_p): Add 'short_call' attribute as a synonym
8165 * doc/extend.texi (short_call): Document new function attribute.
8167 2017-09-12 Jakub Jelinek <jakub@redhat.com>
8170 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
8171 assertion check that in the condition.
8172 (get_atomic_generic_size): Likewise. Before testing if parameter
8173 has pointer type, if it has array type, call for C++
8174 default_conversion to perform array-to-pointer conversion.
8176 2017-09-12 Richard Biener <rguenther@suse.de>
8178 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
8179 for operations we cannot scalarize.
8181 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
8183 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
8184 vectors heap vectors. Clean up comments.
8185 Make visited_bbs a reference.
8186 (profitable_jump_thread_path): Make GC
8187 vectors heap vectors. Clean up comments.
8189 (convert_and_register_jump_thread_path): Make GC vectors heap
8191 (check_subpath_and_update_thread_path): Same. Clean up comments.
8192 Make visited_bbs a reference.
8193 (handle_phi): Abstract common code to to
8194 register_jump_thread_path_if_profitable.
8195 Rename VAR_BB to DEF_BB.
8197 Make GC vectors heap vectors.
8198 Make visited_bbs a reference.
8199 (handle_assignment): Same.
8200 (register_jump_thread_path_if_profitable): New.
8201 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
8203 Make GC vectors heap vectors. Clean up comments.
8204 Make visited_bbs a reference.
8205 (find_jump_threads_backwards): Make visited_bbs live in the stack.
8206 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
8209 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
8212 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
8213 words of E_DImode object are reachable by xtensa_uimm8x4 access.
8215 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
8217 Revert r251800 and r251799.
8219 2017-09-11 Martin Jambor <mjambor@suse.cz>
8222 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
8223 arguments in advance.
8224 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
8225 use it to find predecessor edges.
8226 (naive_outof_ssa): Collect vector of predecessors.
8228 2017-09-08 Jason Merrill <jason@redhat.com>
8230 PR c++/70029 - ICE with ref-qualifier and -flto
8231 * langhooks.h (struct lang_hooks_for_types): Add
8232 copy_lang_qualifiers.
8233 * attribs.c (build_type_attribute_qual_variant): Use it.
8234 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
8236 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
8237 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
8239 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
8242 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
8243 (*mulsi3_sp64): New instruction.
8244 (mulsi3): New expander.
8246 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
8248 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
8250 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8252 * sancov.c: Include memmodel.h.
8254 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
8257 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
8260 2017-09-07 Carl Love <cel@us.ibm.com>
8262 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
8263 the sldi instruction.
8265 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
8267 * sancov.c: Include tm_p.h.
8269 2017-09-07 Jakub Jelinek <jakub@redhat.com>
8272 * output.h (switch_to_other_text_partition): New declaration.
8273 * varasm.c (switch_to_other_text_partition): New function.
8274 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
8275 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
8276 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
8277 to the other text partition before emitting LCL label and switch back
8278 after emitting the word after it.
8280 2017-09-07 Richard Biener <rguenther@suse.de>
8282 * passes.def (pass_split_crit_edges): Remove instance before PRE.
8283 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
8284 critical edges here, after loop init.
8285 (pass_data_pre): Remove PROP_no_crit_edges flags.
8286 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
8287 for valueization of call args to avoid leaking VN_TOP.
8288 (visit_use): Assert we do not visit default defs.
8289 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
8290 Use error_mark_node to more easily detect leaking VN_TOP.
8291 All default-defs are varying, not VN_TOP. Mark them visited.
8292 (run_scc_vn): Make code match comment.
8294 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8296 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
8297 OPTION_MASK_FLOAT128_KEYWORD.
8298 (POWERPC_MASKS): Likewise.
8299 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
8300 support for the -mfloat128-type option, and make -mfloat128
8301 default on PowerPC Linux systems. Define or undefine
8302 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
8303 Define __float128 to be __ieee128 if IEEE 128-bit support is
8304 enabled, or undefine it.
8305 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
8306 Delete defining __FLOAT128_TYPE__.
8307 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
8308 -mfloat128-type option and make -mfloat128 default on PowerPC
8310 (TARGET_FLOAT128_TYPE): Likewise.
8311 (-mfloat128-type): Likewise.
8312 * config/rs6000/rs6000.c (rs6000_option_override_internal):
8313 Delete the -mfloat128-type option and make -mfloat128 default on
8314 PowerPC Linux systems. Always use __ieee128 to be the keyword for
8315 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
8316 128-bit floating point is enabled. Change tests from using
8317 -mfloat128-type to -mfloat128.
8318 (rs6000_mangle_type): Use the correct mangling for the __float128
8319 type even if normal long double is restricted to 64-bits.
8320 (floatn_mode): Enable the _Float128 type by default on VSX Linux
8322 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
8323 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
8324 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
8326 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
8327 documentation for -mfloat128.
8329 2017-09-06 Olivier Hainque <hainque@adacore.com>
8331 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
8333 2017-09-06 Wish Wu <wishwu007@gmail.com>
8334 Jakub Jelinek <jakub@redhat.com>
8336 * asan.c (initialize_sanitizer_builtins): Add
8337 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
8338 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
8339 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
8340 BT_FN_VOID_UINT64_PTR variables.
8341 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
8342 (BT_FN_VOID_UINT16_UINT16): Likewise.
8343 (BT_FN_VOID_UINT32_UINT32): Likewise.
8344 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
8345 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
8346 (BT_FN_VOID_UINT64_PTR): Likewise.
8347 * common.opt (flag_sanitize_coverage): New variable.
8348 (fsanitize-coverage=trace-pc): Remove.
8349 (fsanitize-coverage=): Add.
8350 * flag-types.h (enum sanitize_coverage_code): New enum.
8351 * fold-const.c (fold_range_test): Disable non-short-circuit
8352 optimization if flag_sanitize_coverage.
8353 (fold_truth_andor): Likewise.
8354 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
8355 * opts.c (COVERAGE_SANITIZER_OPT): Define.
8356 (coverage_sanitizer_opts): New array.
8357 (get_closest_sanitizer_option): Add OPTS argument, handle also
8358 OPT_fsanitize_coverage_.
8359 (parse_sanitizer_options): Adjusted to also handle
8360 OPT_fsanitize_coverage_.
8361 (common_handle_option): Add OPT_fsanitize_coverage_.
8362 * sancov.c (instrument_comparison, instrument_switch): New function.
8363 (sancov_pass): Add trace-cmp support.
8364 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
8365 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
8366 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
8367 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
8368 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
8369 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
8370 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
8371 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
8372 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
8373 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
8375 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
8377 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
8378 error. Only quit immediately if parsing is complete.
8379 (BEGIN): Initialize fatal_err and parse_done.
8380 (begin fpu, end fpu): Check number of arguments.
8381 (begin arch, end arch): Likewise.
8382 (begin cpu, end cpu): Likewise.
8383 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
8384 (optalias): Likewise.
8386 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
8388 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
8389 * config/arm/arm-isa.h: Delete. Move definitions to ...
8390 * arm-cpus.in: ... here. Use new feature and fgroup values.
8391 * config/arm/arm.c (arm_option_override): Use lower case for feature
8393 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
8394 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
8395 * config/arm/parsecpu.awk (END): Add new command 'isa'.
8397 (print_isa_bits_for): New function.
8398 (gen_isa): New function.
8399 (gen_comm_data): Use print_isa_bits_for.
8400 (define feature): New keyword.
8401 (define fgroup): New keyword.
8402 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
8403 (arm-isa.h): Add rule to generate file.
8404 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
8405 case for feature bit names.
8407 2017-09-06 Richard Biener <rguenther@suse.de>
8409 * tree-ssa-pre.c (NECESSARY): Remove.
8410 (create_expression_by_pieces): Do not touch pass-local flags.
8411 (insert_into_preds_of_block): Likewise.
8412 (do_pre_regular_insertion): Likewise.
8413 (eliminate_insert): Likewise.
8414 (eliminate_dom_walker::before_dom_children): Likewise.
8415 (fini_eliminate): Do not look at inserted_exprs.
8416 (mark_operand_necessary): Remove.
8417 (remove_dead_inserted_code): Replace with simple work-list
8418 algorithm based on inserted_exprs and SSA uses.
8419 (pass_pre::execute): Re-order fini_eliminate and
8420 remove_dead_inserted_code.
8422 2017-09-06 Olivier Hainque <hainque@adacore.com>
8424 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
8425 for VxWorks 7. Adjust surrounding comments.
8427 2017-09-06 Richard Biener <rguenther@suse.de>
8429 * gimple-ssa-strength-reduction.c
8430 (find_candidates_dom_walker::before_dom_children): Also allow
8433 2017-09-06 Richard Biener <rguenther@suse.de>
8435 PR tree-optimization/82108
8436 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
8437 for gap in the non-permutation SLP case.
8439 2017-09-06 Martin Jambor <mjambor@suse.cz>
8441 PR tree-optimization/82078
8442 * tree-sra.c (sort_and_splice_var_accesses): Move call to
8443 add_access_to_work_queue...
8444 (build_accesses_from_assign): ...here.
8445 (propagate_all_subaccesses): Make sure racc is the group
8446 representative, if there is one.
8448 2017-09-06 Jakub Jelinek <jakub@redhat.com>
8451 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
8454 2017-09-06 Richard Biener <rguenther@suse.de>
8456 * gimple-ssa-strength-reduction.c
8457 (find_candidates_dom_walker::before_doom_children): Use a
8458 type and not a mode check.
8460 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
8463 * config/arm/predicates.md (arm_general_adddi_operand): Create new
8465 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
8467 2017-09-05 Jeff Law <law@redhat.com>
8469 PR tree-optimization/64910
8470 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
8471 cases where we have 3 or more operands.
8473 2017-09-05 Jakub Jelinek <jakub@redhat.com>
8476 * omp-low.c (lower_omp_for): Recompute tree invariant if
8477 gimple_omp_for_initial/final is ADDR_EXPR.
8480 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
8481 into gimple val before gimplification fo the COND_EXPR.
8483 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
8485 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
8486 REGION_COPY argument.
8487 (thread_through_all_blocks): Remove unused argument to
8488 duplicate_thread_path.
8490 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8491 Alan Hayward <alan.hayward@arm.com>
8492 David Sherwood <david.sherwood@arm.com>
8494 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
8495 Take a scalar_mode rather than a machine_mode.
8496 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8497 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
8498 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8499 (aarch64_gen_adjusted_ldpstp): Likewise.
8500 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
8502 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8503 Alan Hayward <alan.hayward@arm.com>
8504 David Sherwood <david.sherwood@arm.com>
8506 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
8507 Take a scalar_int_mode instead of a machine_mode.
8508 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8509 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8510 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8511 (aarch64_simd_attr_length_rglist): Delete.
8512 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
8513 a scalar_int_mode instead of a machine_mode.
8514 (aarch64_add_offset): Likewise.
8515 (aarch64_internal_mov_immediate): Likewise
8516 (aarch64_add_constant_internal): Likewise.
8517 (aarch64_add_constant): Likewise.
8518 (aarch64_movw_imm): Likewise.
8519 (aarch64_rtx_arith_op_extract_p): Likewise.
8520 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8521 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8522 Remove assert that the mode isn't a vector.
8523 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8524 (aarch64_expand_mov_immediate): Update calls after above changes.
8525 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
8526 (aarch64_and_bitmask_imm): Check for scalar integer modes.
8527 (aarch64_move_imm): Likewise.
8528 (aarch64_can_const_movi_rtx_p): Likewise.
8529 (aarch64_strip_extend): Likewise.
8530 (aarch64_extr_rtx_p): Likewise.
8531 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
8532 a CONST_INT when the mode parameter is VOIDmode.
8533 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
8535 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8537 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
8538 * stor-layout.c (bitwise_mode_for_mode): Likewise.
8539 (bitwise_type_for_mode): Update accordingly.
8541 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8543 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
8544 * stor-layout.c (mode_for_size_tree): Likewise.
8545 (mode_for_array): Update accordingly.
8546 (layout_decl): Likewise.
8547 (compute_record_mode): Likewise. Only set the mode once.
8549 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8551 * target.def (get_mask_mode): Change return type to opt_mode.
8553 * doc/tm.texi: Regenerate.
8554 * targhooks.h (default_get_mask_mode): Return an opt_mode.
8555 * targhooks.c (default_get_mask_mode): Likewise.
8556 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
8557 * optabs-query.c (can_vec_mask_load_store_p): Update use of
8558 targetm.get_mask_mode.
8559 * tree.c (build_truth_vector_type): Likewise.
8561 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8563 * machmode.h (mode_for_vector): Return an opt_mode.
8564 * stor-layout.c (mode_for_vector): Likewise.
8565 (mode_for_int_vector): Update accordingly.
8566 (layout_type): Likewise.
8567 * config/i386/i386.c (emit_memmov): Likewise.
8568 (ix86_expand_set_or_movmem): Likewise.
8569 (ix86_expand_vector_init): Likewise.
8570 (ix86_get_mask_mode): Likewise.
8571 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
8573 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
8574 * expmed.c (extract_bit_field_1): Likewise.
8575 * expr.c (expand_expr_real_2): Likewise.
8576 * optabs-query.c (can_vec_perm_p): Likewise.
8577 (can_vec_mask_load_store_p): Likewise.
8578 * optabs.c (expand_vec_perm): Likewise.
8579 * targhooks.c (default_get_mask_mode): Likewise.
8580 * tree-vect-stmts.c (vectorizable_store): Likewise.
8581 (vectorizable_load): Likewise.
8582 (get_vectype_for_scalar_type_and_size): Likewise.
8584 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8586 * machmode.h (mode_for_int_vector): New function.
8587 * stor-layout.c (mode_for_int_vector): Likewise.
8588 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
8589 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
8590 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
8591 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
8592 (s390_expand_vcond): Likewise.
8594 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8596 * machmode.h (opt_machine_mode): New type.
8597 (opt_mode<T>): Allow construction from anything that can be
8599 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
8600 (mode_for_size): Return an opt_machine_mode.
8601 * stor-layout.c (mode_for_size): Likewise.
8602 (mode_for_size_tree): Update call accordingly.
8603 (bitwise_mode_for_mode): Likewise.
8604 (make_fract_type): Likewise.
8605 (make_accum_type): Likewise.
8606 * caller-save.c (replace_reg_with_saved_mem): Update call
8608 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8609 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8610 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8611 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8612 * expmed.c (extract_bit_field_1): Likewise.
8613 * reload.c (get_secondary_mem): Likewise.
8614 * varasm.c (assemble_integer): Likewise.
8615 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
8618 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8620 * machmode.h (decimal_float_mode_for_size): New function.
8621 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
8622 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
8623 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
8624 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
8625 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
8626 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
8628 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8630 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
8631 (get_builtin_sync_mode): Likewise.
8632 (expand_ifn_atomic_compare_exchange): Likewise.
8633 (expand_builtin_atomic_clear): Likewise.
8634 (expand_builtin_atomic_test_and_set): Likewise.
8635 (fold_builtin_atomic_always_lock_free): Likewise.
8636 * calls.c (compute_argument_addresses): Likewise.
8637 (emit_library_call_value_1): Likewise.
8638 (store_one_arg): Likewise.
8639 * combine.c (combine_instructions): Likewise.
8640 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
8641 * config/arm/arm.c (arm_function_value): Likewise.
8642 (aapcs_allocate_return_reg): Likewise.
8643 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
8644 * config/i386/i386.c (construct_container): Likewise.
8645 (ix86_gimplify_va_arg): Likewise.
8646 (ix86_expand_sse_cmp): Likewise.
8647 (emit_memmov): Likewise.
8648 (emit_memset): Likewise.
8649 (expand_small_movmem_or_setmem): Likewise.
8650 (ix86_expand_pextr): Likewise.
8651 (ix86_expand_pinsr): Likewise.
8652 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
8653 * config/microblaze/microblaze.c (microblaze_block_move_straight):
8655 * config/mips/mips.c (mips_function_value_1) Likewise.
8656 (mips_block_move_straight): Likewise.
8657 (mips_expand_ins_as_unaligned_store): Likewise.
8658 * config/powerpcspe/powerpcspe.c
8659 (rs6000_darwin64_record_arg_advance_flush): Likewise.
8660 (rs6000_darwin64_record_arg_flush): Likewise.
8661 * config/rs6000/rs6000.c
8662 (rs6000_darwin64_record_arg_advance_flush): Likewise.
8663 (rs6000_darwin64_record_arg_flush): Likewise.
8664 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
8665 (sparc_function_value_1): Likewise.
8666 * config/spu/spu.c (adjust_operand): Likewise.
8667 (spu_emit_branch_or_set): Likewise.
8668 (arith_immediate_p): Likewise.
8669 * emit-rtl.c (gen_lowpart_common): Likewise.
8670 * expr.c (expand_expr_real_1): Likewise.
8671 * function.c (assign_parm_setup_block): Likewise.
8672 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
8673 * reload1.c (alter_reg): Likewise.
8674 * stor-layout.c (mode_for_vector): Likewise.
8675 (layout_type): Likewise.
8677 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8679 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
8680 (spu_convert_move): Likewise.
8681 * lower-subreg.c (resolve_simple_move): Likewise.
8683 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8686 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
8687 define_insn to a define_expand.
8688 (altivec_vsum2sws_direct): New define_insn.
8689 (altivec_vsumsws): Convert from a define_insn to a define_expand.
8691 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
8693 * config/arm/arm.c (arm_option_params_internal): Improve setting of
8696 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
8701 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
8702 realign stack if stack alignment needed is less than incoming
8705 2017-09-05 Marek Polacek <polacek@redhat.com>
8708 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
8711 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
8713 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
8715 2017-09-05 Richard Biener <rguenther@suse.de>
8717 PR tree-optimization/82084
8718 * fold-const.c (can_native_encode_string_p): Handle wide characters.
8720 2017-09-05 Richard Biener <rguenther@suse.de>
8722 PR tree-optimization/82102
8723 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
8725 2017-09-05 Martin Liska <mliska@suse.cz>
8727 PR tree-optimization/82032
8728 * tree-cfg.c (generate_range_test): New function.
8729 * tree-cfg.h (generate_range_test): Declared here.
8730 * tree-cfgcleanup.c (convert_single_case_switch): New function.
8731 (cleanup_control_expr_graph): Use it.
8732 * tree-switch-conversion.c (try_switch_expansion): Remove
8734 (emit_case_nodes): Use generate_range_test.
8736 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
8739 * config/i386/i386.md (*<btsc><mode>_mask): Add
8740 TARGET_USE_BT to insn constraint.
8741 (*btr<mode>_mask): Ditto.
8743 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
8745 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
8746 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
8748 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
8751 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
8752 TARGET_NEON and TARGET_IWMMXT.
8753 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
8754 TARGET_NEON and TARGET_IWMMXT.
8755 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
8757 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
8759 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
8760 (ix86_rewrite_tls_address): Ditto.
8761 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
8762 (ix86_rewrite_tls_address_1): Ditto.
8763 (ix86_rewrite_tls_address): Ditto.
8764 * config/i386/predicates.md (tls_address_pattern): New predicate.
8765 * config/i386/i386.md (TLS address splitter): New splitter.
8767 2017-09-04 Richard Biener <rguenther@suse.de>
8769 PR tree-optimization/82084
8770 * fold-const.h (can_native_encode_string_p): Declare.
8771 * fold-const.c (can_native_encode_string_p): Factor out from ...
8772 (native_encode_string): ... here.
8773 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
8774 vectorizing stores from constants we later cannot handle.
8776 2017-09-04 Marek Polacek <polacek@redhat.com>
8779 * doc/invoke.texi: Update -Wtautological-compare documentation.
8781 2017-09-04 Jeff Law <law@redhat.com>
8783 PR tree-optimization/64910
8784 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
8785 swap the first and last operand if the last is a constant.
8787 2017-09-04 Marek Polacek <polacek@redhat.com>
8790 * convert.c (do_narrow): When sanitizing signed integer overflows,
8791 bail out for signed types.
8792 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
8794 2017-09-04 Richard Biener <rguenther@suse.de>
8796 PR tree-optimization/82060
8797 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
8798 Move devirtualization after stmt folding and before EH/AB/noreturn
8799 cleanup to get the stmt refs canonicalized. Use a bool instead
8800 of gimple_modified_p since that doesn't work for NOPs. Schedule
8801 NOPs generated by folding for removal.
8803 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8804 Alan Hayward <alan.hayward@arm.com>
8805 David Sherwood <david.sherwood@arm.com>
8807 * coretypes.h (pad_direction): New enum.
8808 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
8809 (FUNCTION_ARG_PADDING): Likewise.
8810 * target.def (function_arg_padding): New hook.
8811 * targhooks.h (default_function_arg_padding): Declare.
8812 * targhooks.c (default_function_arg_padding): New function.
8813 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
8814 (TARGET_FUNCTION_ARG_PADDING): ...this.
8815 * doc/tm.texi: Regenerate.
8816 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
8817 instead of direction.
8818 (compute_argument_addresses): Likewise.
8819 (load_register_parameters): Likewise.
8820 (emit_library_call_value_1): Likewise.
8821 (store_one_arg): Use targetm.calls.function_arg_padding instead
8822 of FUNCTION_ARG_PADDING.
8823 (must_pass_in_stack_var_size_or_pad): Likewise.
8824 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
8825 (emit_group_store): Likewise.
8826 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
8827 instead of FUNCTION_ARG_PADDING.
8828 (emit_push_insn): Likewise, and propagate enum change throughout
8830 * function.h (direction): Delete.
8831 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
8833 * function.c (assign_parm_find_stack_rtl): Likewise.
8834 (assign_parm_setup_block_p): Likewise.
8835 (assign_parm_setup_block): Likewise.
8836 (gimplify_parameters): Likewise.
8837 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
8838 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
8840 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
8841 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8842 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
8843 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
8844 (aarch64_function_arg_padding): ...this new function.
8845 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
8846 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8847 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
8848 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8849 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
8850 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8851 (arm_pad_arg_upward): Replace with...
8852 (arm_function_arg_padding): ...this new function.
8853 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
8855 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
8856 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
8857 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8858 (ia64_hpux_function_arg_padding): Replace with...
8859 (ia64_function_arg_padding): ...this new function. Use pad_direction
8860 instead of direction. Check for TARGET_HPUX.
8861 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
8862 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8863 (iq2000_function_arg_padding): New function.
8864 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
8865 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
8866 (mips_function_arg_padding): ...this new function.
8867 (mips_pad_reg_upward): Update accordingly.
8868 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8869 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
8870 targetm.calls.function_arg_padding.
8871 (FUNCTION_ARG_PADDING): Delete.
8872 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8873 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
8874 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8875 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
8876 (nios2_block_reg_padding): Return pad_direction instead of direction.
8877 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
8878 instead of direction.
8879 (nios2_function_arg_padding): Likewise. Make static.
8880 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8881 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
8882 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
8883 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
8884 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8885 (pa_function_arg_padding): Make static. Return pad_direction instead
8887 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
8888 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8889 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
8890 instead of direction. Use targetm.calls.function_arg_padding.
8891 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
8892 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
8893 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
8894 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
8895 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
8897 (function_arg_padding): Rename to...
8898 (rs6000_function_arg_padding): ...this. Make static. Return
8899 pad_direction instead of direction.
8900 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
8901 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
8902 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8903 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
8904 instead of direction. Use targetm.calls.function_arg_padding.
8905 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
8906 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
8907 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
8908 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
8909 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8910 (function_arg_padding): Rename to...
8911 (rs6000_function_arg_padding): ...this. Make static. Return
8912 pad_direction instead of direction.
8913 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
8914 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
8915 * config/s390/s390.c (s390_function_arg_padding): New function.
8916 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8917 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
8918 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
8919 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8920 (function_arg_padding): Rename to...
8921 (sparc_function_arg_padding): ...this. Make static. Return
8922 pad_direction instead of direction.
8923 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
8924 * config/spu/spu.c (spu_function_arg_padding): New function.
8925 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8926 * system.h (FUNCTION_ARG_PADDING): Poison.
8928 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8929 Alan Hayward <alan.hayward@arm.com>
8930 David Sherwood <david.sherwood@arm.com>
8932 * target.def (modes_tieable_p): New hook.
8933 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
8934 (TARGET_MODES_TIEABLE_P): ...this.
8935 * doc/tm.texi.in: Regenerate.
8936 * hooks.h (hook_bool_mode_mode_true): Declare.
8937 * hooks.c (hook_bool_mode_mode_true): New function.
8938 * combine.c (subst): Use targetm.modes_tieable_p instead of
8940 * dse.c (find_shift_sequence): Likewise.
8941 * expmed.c (extract_low_bits): Likewise.
8942 * lower-subreg.c: Include target.h.
8943 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
8945 * rtlanal.c (rtx_cost): Likewise.
8946 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
8947 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
8948 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
8949 (TARGET_MODES_TIEABLE_P): Redefine.
8950 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
8951 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
8952 (TARGET_MODES_TIEABLE_P): Redefine.
8953 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
8954 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
8955 (arc_modes_tieable_p): New function.
8956 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
8957 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
8958 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
8959 (arm_modes_tieable_p): Make static.
8960 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
8961 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
8962 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
8963 (TARGET_MODES_TIEABLE_P): Redefine.
8964 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
8965 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
8966 (TARGET_MODES_TIEABLE_P): Redefine.
8967 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
8968 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
8969 (cr16_modes_tieable_p): New function.
8970 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
8971 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
8972 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
8973 (TRULY_NOOP_TRUNCATION): Update comment.
8974 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
8975 (TRULY_NOOP_TRUNCATION): Update comment.
8976 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
8977 (frv_modes_tieable_p): New function.
8978 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
8979 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
8980 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
8981 (TARGET_MODES_TIEABLE_P): Redefine.
8982 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
8983 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
8984 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
8985 (TARGET_MODES_TIEABLE_P): Redefine.
8986 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
8987 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
8988 (ia64_modes_tieable_p): New function.
8989 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
8990 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
8991 (iq2000_modes_tieable_p): New function.
8992 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
8993 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
8994 (lm32_modes_tieable_p): New function.
8995 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
8996 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
8997 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
8998 (TARGET_MODES_TIEABLE_P): Redefine.
8999 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
9000 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
9001 (m32r_modes_tieable_p): New function.
9002 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
9003 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
9004 (m68k_modes_tieable_p): New function.
9005 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
9006 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
9007 (mcore_modes_tieable_p): New function.
9008 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
9009 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
9011 (TARGET_MODES_TIEABLE_P): Redefine.
9012 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
9013 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
9014 * config/mips/mips.c (mips_modes_tieable_p): Make static.
9015 (TARGET_MODES_TIEABLE_P): Redefine.
9016 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
9017 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
9018 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
9019 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
9020 (mn10300_modes_tieable_p): ...this and make static.
9021 (TARGET_MODES_TIEABLE_P): Redefine.
9022 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
9023 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
9024 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
9025 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
9026 (msp430_modes_tieable_p): Make static.
9027 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
9028 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
9029 (TARGET_MODES_TIEABLE_P): Redefine.
9030 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
9031 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
9032 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
9033 (TARGET_MODES_TIEABLE_P): Redefine.
9034 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
9035 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
9036 * config/pa/pa.c (pa_modes_tieable_p): Make static.
9037 (TARGET_MODES_TIEABLE_P): Redefine.
9038 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
9039 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
9040 (pdp11_modes_tieable_p): New function.
9041 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
9042 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
9043 (rs6000_modes_tieable_p): New function.
9044 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9045 * config/powerpcspe/powerpcspe.md: Update comment.
9046 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
9047 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
9048 (TARGET_MODES_TIEABLE_P): Redefine.
9049 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
9050 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
9051 (rl78_modes_tieable_p): New function.
9052 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
9053 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
9054 (rs6000_modes_tieable_p): New function.
9055 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9056 * config/rs6000/rs6000.md: Update comment.
9057 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
9058 * config/rx/rx.c (rx_modes_tieable_p): New function.
9059 (TARGET_MODES_TIEABLE_P): Redefine.
9060 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
9061 * config/s390/s390.c (s390_modes_tieable_p): New function.
9062 (TARGET_MODES_TIEABLE_P): Redefine.
9063 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
9064 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
9065 (sh_modes_tieable_p): New function.
9066 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
9067 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
9068 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
9069 (sparc_modes_tieable_p): Make static.
9070 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
9071 * config/spu/spu.c (spu_modes_tieable_p): New function.
9072 (TARGET_MODES_TIEABLE_P): Redefine.
9073 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
9074 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
9075 (TARGET_MODES_TIEABLE_P): Redefine.
9076 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
9077 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
9078 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
9079 * config/v850/v850.c (v850_modes_tieable_p): New function.
9080 (TARGET_MODES_TIEABLE_P): Redefine.
9081 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
9082 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
9083 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
9084 (visium_modes_tieable_p): New function.
9085 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
9086 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
9087 (xtensa_modes_tieable_p): New function.
9088 * system.h (MODES_TIEABLE_P): Poison.
9090 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9091 Alan Hayward <alan.hayward@arm.com>
9092 David Sherwood <david.sherwood@arm.com>
9094 * target.def (hard_regno_mode_ok): New hook.
9095 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
9096 (TARGET_HARD_REGNO_MODE_OK): ...this.
9097 * doc/tm.texi.in: Regenerate.
9098 * hooks.h (hook_bool_uint_mode_true): Declare.
9099 * hooks.c (hook_bool_uint_mode_true): New function.
9100 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
9102 * genpreds.c (write_insn_preds_c): Add an include of target.h.
9103 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
9104 instead of HARD_REGNO_MODE_OK.
9105 * caller-save.c: Include target.h.
9106 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
9108 * combine.c (can_combine_p): Likewise.
9109 (combinable_i3pat): Likewise.
9110 (can_change_dest_mode): Likewise.
9111 * expr.c (init_expr_target): Likewise.
9112 (convert_move): Likewise.
9113 (convert_modes): Likewise.
9114 * ira.c (setup_prohibited_class_mode_regs): Likewise.
9115 (setup_prohibited_mode_move_regs): Likewise.
9116 * ira.h (target_ira): Likewise.
9117 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9118 * lra-constraints.c (process_alt_operands): Likewise.
9119 (split_reg): Likewise.
9120 * recog.c (peep2_find_free_register): Likewise.
9121 * ree.c (combine_reaching_defs): Likewise.
9122 * regcprop.c (maybe_mode_change): Likewise.
9123 * reginfo.c (init_reg_sets_1): Likewise.
9124 (choose_hard_reg_mode): Likewise.
9125 (simplifiable_subregs): Likewise.
9126 * regrename.c (check_new_reg_p): Likewise.
9127 * reload.c (find_valid_class): Likewise.
9128 (find_valid_class_1): Likewise.
9129 (reload_inner_reg_of_subreg): Likewise.
9130 (push_reload): Likewise.
9131 (combine_reloads): Likewise.
9132 (find_dummy_reload): Likewise.
9133 (find_reloads): Likewise.
9134 * reload1.c (find_reg): Likewise.
9135 (set_reload_reg): Likewise.
9136 (allocate_reload_reg): Likewise.
9137 (choose_reload_regs): Likewise.
9138 (reload_adjust_reg_for_temp): Likewise.
9139 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
9140 (simplify_subreg_regno): Likewise.
9141 * sel-sched.c (init_regs_for_mode): Likewise.
9142 * varasm.c (make_decl_rtl): Likewise.
9143 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
9144 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
9146 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
9147 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
9148 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9149 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
9150 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
9151 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9152 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
9153 (arc_mode_class): Delete.
9154 (HARD_REGNO_MODE_OK): Delete.
9155 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9156 (arc_hard_regno_mode_ok): Rename old array to...
9157 (arc_hard_regno_mode_ok_modes): ...this.
9158 (arc_conditional_register_usage): Update accordingly.
9159 (arc_mode_class): Make static.
9160 (arc_hard_regno_mode_ok): New function.
9161 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
9162 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
9163 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9164 (arm_hard_regno_mode_ok): Make static.
9165 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
9167 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
9168 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
9169 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
9171 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9172 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
9173 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
9174 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
9175 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
9176 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9177 * config/bfin/predicates.md (valid_reg_operand): Use
9178 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
9179 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
9180 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
9181 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9182 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
9183 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
9184 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9185 (cr16_hard_regno_mode_ok): Make static and return a bool.
9186 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
9187 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9188 (cris_hard_regno_mode_ok): New function.
9189 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
9190 (epiphany_mode_class): Delete.
9191 (HARD_REGNO_MODE_OK): Delete.
9192 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
9193 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9194 (hard_regno_mode_ok): Rename to...
9195 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
9196 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
9197 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
9199 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
9200 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
9201 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9202 (frv_hard_regno_mode_ok): Make static and return a bool.
9203 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
9205 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
9206 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
9207 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
9208 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
9210 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9211 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
9212 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
9213 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
9215 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9216 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
9217 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9218 (ia64_hard_regno_mode_ok): New function.
9219 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
9220 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9221 (iq2000_hard_regno_mode_ok): New function.
9222 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
9223 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9224 (lm32_hard_regno_mode_ok): New function.
9225 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
9226 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
9227 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
9228 instead of HARD_REGNO_MODE_OK.
9229 (m32c_hard_regno_ok): Rename to...
9230 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
9231 (m32c_cannot_change_mode_class): Update accordingly.
9232 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9233 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
9234 (m32r_mode_class): Delete.
9235 (HARD_REGNO_MODE_OK): Delete.
9236 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9237 (m32r_hard_regno_mode_ok): Rename to...
9238 (m32r_hard_regno_modes): ...this.
9239 (m32r_mode_class): Make static.
9240 (m32r_hard_regno_mode_ok): New function.
9241 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
9242 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
9243 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9244 (m68k_hard_regno_mode_ok): Make static.
9245 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
9246 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9247 (mcore_hard_regno_mode_ok): New function.
9248 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
9249 (HARD_REGNO_MODE_OK): Delete.
9250 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
9252 (microblaze_hard_regno_mode_ok_p): ...this and make static.
9253 (microblaze_hard_regno_mode_ok): New function.
9254 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9255 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
9256 (mips_hard_regno_mode_ok): Delete.
9257 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
9258 (mips_hard_regno_mode_ok_p): ...this and make static.
9259 (mips_hard_regno_mode_ok_p): Rename to...
9260 (mips_hard_regno_mode_ok_uncached): ...this.
9261 (mips_hard_regno_mode_ok): New function.
9262 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
9263 of HARD_REGNO_MODE_OK.
9264 (mips_option_override): Update after above name changes.
9265 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9266 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
9267 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
9268 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
9269 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
9270 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9271 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
9272 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
9273 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
9274 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9275 (msp430_hard_regno_mode_ok): Make static and return a bool.
9276 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
9277 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
9278 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
9280 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9281 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
9282 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
9283 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
9284 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
9285 (PA_HARD_REGNO_MODE_OK): ...this
9286 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
9287 (PA_HARD_REGNO_MODE_OK): ...this.
9288 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9289 (pa_hard_regno_mode_ok): New function.
9290 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
9291 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9292 (pdp11_hard_regno_mode_ok): New function.
9293 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
9294 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
9296 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
9298 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9299 (rs6000_hard_regno_mode_ok): Rename to...
9300 (rs6000_hard_regno_mode_ok_uncached): ...this.
9301 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9302 (rs6000_hard_regno_mode_ok): New function.
9303 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
9304 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
9305 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
9306 (riscv_hard_regno_mode_ok): ...this and make static.
9307 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9308 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
9309 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
9310 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9311 (rl78_hard_regno_mode_ok): Make static and return bool.
9312 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
9313 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
9315 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
9316 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9317 (rs6000_hard_regno_mode_ok): Rename to...
9318 (rs6000_hard_regno_mode_ok_uncached): ...this.
9319 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9320 (rs6000_hard_regno_mode_ok): New function.
9321 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
9322 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
9323 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9324 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
9325 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
9326 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
9327 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9328 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
9329 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
9330 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9331 (sh_hard_regno_mode_ok): Make static.
9332 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
9333 instead of HARD_REGNO_MODE_OK.
9334 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
9335 (sparc_mode_class): Delete.
9336 (HARD_REGNO_MODE_OK): Delete.
9337 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9338 (hard_regno_mode_classes): Make static.
9339 (sparc_mode_class): Likewise.
9340 (sparc_hard_regno_mode_ok): New function.
9341 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
9342 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
9343 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
9345 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9346 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
9347 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
9348 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
9349 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
9350 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9351 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
9352 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
9353 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9354 (visium_hard_regno_mode_ok): New function.
9355 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
9356 instead of HARD_REGNO_MODE_OK.
9357 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
9358 (HARD_REGNO_MODE_OK): Delete.
9359 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
9360 (xtensa_hard_regno_mode_ok_p): ...this and make static.
9361 (xtensa_option_override): Update accordingly.
9362 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9363 (xtensa_hard_regno_mode_ok): New function.
9364 * system.h (HARD_REGNO_MODE_OK): Poison.
9366 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9367 Alan Hayward <alan.hayward@arm.com>
9368 David Sherwood <david.sherwood@arm.com>
9370 * target.def (hard_regno_call_part_clobbered): New hook.
9371 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
9372 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
9373 * doc/tm.texi: Regenerate.
9374 * hooks.h (hook_bool_uint_mode_false): Declare.
9375 * hooks.c (hook_bool_uint_mode_false): New function.
9376 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9377 * cselib.c (cselib_process_insn): Use
9378 targetm.hard_regno_call_part_clobbered instead of
9379 HARD_REGNO_CALL_PART_CLOBBERED.
9380 * ira-conflicts.c (ira_build_conflicts): Likewise.
9381 * ira-costs.c (ira_tune_allocno_costs): Likewise.
9382 * lra-constraints.c (need_for_call_save_p): Likewise.
9383 * lra-lives.c: Include target.h.
9384 (check_pseudos_live_through_calls): Use
9385 targetm.hard_regno_call_part_clobbered instead of
9386 HARD_REGNO_CALL_PART_CLOBBERED.
9387 * regcprop.c: Include target.h.
9388 (copyprop_hardreg_forward_1): Use
9389 targetm.hard_regno_call_part_clobbered instead of
9390 HARD_REGNO_CALL_PART_CLOBBERED.
9391 * reginfo.c (choose_hard_reg_mode): Likewise.
9392 * regrename.c (check_new_reg_p): Likewise.
9393 * reload.c (find_equiv_reg): Likewise.
9394 * reload1.c (emit_reload_insns): Likewise.
9395 * sched-deps.c (deps_analyze_insn): Likewise.
9396 * sel-sched.c (init_regs_for_mode): Likewise.
9397 (mark_unavailable_hard_regs): Likewise.
9398 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
9399 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9400 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
9402 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9403 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9404 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
9406 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
9408 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9409 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9410 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
9412 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9413 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9414 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
9416 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9417 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
9419 * config/powerpcspe/powerpcspe.c
9420 (rs6000_hard_regno_call_part_clobbered): New function.
9421 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9422 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9423 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
9425 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9426 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9427 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
9429 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9430 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9431 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
9433 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9434 Alan Hayward <alan.hayward@arm.com>
9435 David Sherwood <david.sherwood@arm.com>
9437 * rtl.h (subreg_memory_offset): Declare.
9438 * emit-rtl.c (subreg_memory_offset): New function.
9439 * expmed.c (store_bit_field_1): Use it.
9440 * expr.c (undefined_operand_subword_p): Likewise.
9441 * simplify-rtx.c (simplify_subreg): Likewise.
9443 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
9445 PR rtl-optimization/57448
9448 * optabs.c (expand_atomic_load): Place compiler memory barriers if
9449 using atomic_load pattern.
9450 (expand_atomic_store): Likewise.
9452 2017-09-04 Jakub Jelinek <jakub@redhat.com>
9455 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
9456 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
9457 handling. Use replace_call_with_value with NULL instead of
9458 gsi_replace, unlink_stmt_vdef and release_defs.
9460 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
9463 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
9465 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9468 * rtl.h (emit_library_call_value_1): Declare.
9469 (emit_library_call): Replace declaration with a series of overloads.
9470 Remove the parameter count argument.
9471 (emit_library_call_value): Likewise.
9472 * calls.c (emit_library_call_value_1): Make global. Replace varargs
9473 with an "rtx_mode_t *".
9474 (emit_library_call_value): Delete.
9475 (emit_library_call): Likewise.
9476 * asan.c (asan_emit_stack_protection): Update calls accordingly.
9477 (asan_emit_allocas_unpoison): Likewise.
9478 * builtins.c (expand_builtin_powi): Likewise.
9479 (expand_asan_emit_allocas_unpoison): Likewise.
9480 * cfgexpand.c (expand_main_function): Likewise.
9481 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
9482 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
9483 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
9484 * config/arm/arm.c (arm_trampoline_init): Likewise.
9485 (arm_call_tls_get_addr): Likewise.
9486 (arm_expand_divmod_libfunc): Likewise.
9487 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
9488 (smulsi3_highpart): Likewise.
9489 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
9490 (c6x_expand_compare): Likewise.
9491 (c6x_expand_movmem): Likewise.
9492 * config/frv/frv.c (frv_trampoline_init): Likewise.
9493 * config/i386/i386.c (ix86_trampoline_init): Likewise.
9494 (ix86_expand_divmod_libfunc): Likewise.
9495 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
9496 (ia64_expand_compare): Likewise.
9497 (ia64_profile_hook): Likewise.
9498 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
9499 (nonlocal_goto): Likewise.
9500 (restore_stack_nonlocal): Likewise.
9501 * config/m32r/m32r.c (block_move_call): Likewise.
9502 (m32r_trampoline_init): Likewise.
9503 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
9504 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
9505 (m68k_call_m68k_read_tp): Likewise.
9506 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
9507 (microblaze_expand_divide): Likewise.
9508 * config/mips/mips.h (mips_args): Likewise.
9509 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
9510 (MIPS_ICACHE_SYNC): Likewise.
9511 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
9512 (nios2_trampoline_init): Likewise.
9513 * config/pa/pa.c (hppa_tls_call): Likewise.
9514 (pa_trampoline_init): Likewise.
9515 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
9516 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
9517 (expand_strn_compare): Likewise.
9518 (rs6000_generate_compare): Likewise.
9519 (rs6000_expand_float128_convert): Likewise.
9520 (output_profile_hook): Likewise.
9521 (rs6000_trampoline_init): Likewise.
9522 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
9523 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
9524 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
9525 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
9526 (rs6000_generate_compare): Likewise.
9527 (rs6000_expand_float128_convert): Likewise.
9528 (output_profile_hook): Likewise.
9529 (rs6000_trampoline_init): Likewise.
9530 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
9531 * config/sh/sh.c (sh_trampoline_init): Likewise.
9532 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
9533 (sparc_emit_float_lib_cmp): Likewise.
9534 (sparc32_initialize_trampoline): Likewise.
9535 (sparc64_initialize_trampoline): Likewise.
9536 (sparc_profile_hook): Likewise.
9537 * config/spu/spu.c (ea_load_store): Likewise.
9538 * config/spu/spu.md (floatunssidf2): Likewise.
9539 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
9540 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
9541 * config/visium/visium.c (expand_block_move_4): Likewise.
9542 (expand_block_move_2): Likewise.
9543 (expand_block_move_1): Likewise.
9544 (expand_block_set_4): Likewise.
9545 (expand_block_set_2): Likewise.
9546 (expand_block_set_1): Likewise.
9547 (visium_trampoline_init): Likewise.
9548 (visium_profile_hook): Likewise.
9549 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
9550 (xtensa_setup_frame_addresses): Likewise.
9551 (xtensa_trampoline_init): Likewise.
9552 * except.c (sjlj_emit_function_enter): Likewise.
9553 (sjlj_emit_function_exit): Likewise.
9554 * explow.c (allocate_dynamic_stack_space): Likewise.
9555 (probe_stack_range): Likewise.
9556 * expr.c (convert_mode_scalar): Likewise.
9557 * optabs.c (expand_binop): Likewise.
9558 (expand_twoval_binop_libfunc): Likewise.
9559 (expand_unop): Likewise.
9560 (prepare_cmp_insn): Likewise.
9561 (prepare_float_lib_cmp): Likewise.
9562 (expand_float): Likewise.
9563 (expand_fix): Likewise.
9564 (expand_fixed_convert): Likewise.
9565 (maybe_emit_sync_lock_test_and_set): Likewise.
9566 (expand_atomic_compare_and_swap): Likewise.
9567 (expand_mem_thread_fence): Likewise.
9568 (expand_atomic_fetch_op): Likewise.
9570 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
9572 * doc/generic.texi (OpenACC): Adjust URL.
9573 * doc/invoke.texi (C Dialect Options): Ditto.
9575 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
9577 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
9578 predicate for operand 1. Add (m,<S>) constraint.
9579 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
9580 Prevent memory operand 1 with register operand 2.
9582 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
9584 PR rtl-optimization/82024
9585 * combine.c (try_combine): If the combination result is a PARALLEL,
9586 and we only need to retain the SET in there that would be placed
9587 at I2, check that we can place that at I3 instead, before doing so.
9589 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9592 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
9594 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
9595 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
9598 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
9599 Jeff Law <law@redhat.com>
9601 * varasm.c (bss_initializer_p): Do not put constants into .bss
9602 (categorize_decl_for_section): Handle bss_initializer_p returning
9603 false when DECL_INITIAL is NULL.
9605 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9608 * config/s390/s390.c (s390_can_inline_p): New function.
9610 2017-09-01 Jeff Law <law@redhat.com>
9612 PR tree-optimization/82052
9613 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
9614 Always initialize the returned slot after a hash table miss
9615 when INSERT is true.
9617 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
9619 * config/s390/s390.md (mem_signal_fence): Remove.
9620 * doc/md.texi (mem_signal_fence): Remove.
9621 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
9623 * target-insns.def (mem_signal_fence): Remove.
9625 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9628 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
9631 * asan.c (create_odr_indicator): Strip name encoding from assembler
9632 name before appending it after __odr_asan_.
9634 2017-09-01 Martin Liska <mliska@suse.cz>
9636 PR tree-optimization/82059
9637 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
9638 frequency only when an edge is redirected.
9640 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9642 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
9643 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
9644 (arc_conditional_register_usage): Remove ARC600 lp_count
9646 (arc_file_start): Emit Tag_ARC_CPU_variation.
9647 (arc_can_use_doloop_p): New conditions to use ZOLs.
9648 (hwloop_fail): New function.
9649 (hwloop_optimize): Likewise.
9650 (hwloop_pattern_reg): Likewise.
9651 (arc_doloop_hooks): New struct, to be used with reorg_loops.
9652 (arc_reorg_loops): New function, calls reorg_loops.
9653 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
9654 (arc600_corereg_hazard): Remove ZOL checking, case handled by
9656 (arc_loop_hazard): Remove function, functionality moved into
9658 (arc_hazard): Remove arc_loop_hazard call.
9659 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
9660 into hwloop_optimize.
9661 (arc_label_align): Remove ZOL handling.
9662 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
9663 * config/arc/arc.md (doloop_begin): Remove pattern.
9664 (doloop_begin_i): Likewise.
9665 (doloop_end_i): Likewise.
9666 (doloop_fallback): Likewise.
9667 (doloop_fallback_m): Likewise.
9668 (doloop_end): Reimplement expand.
9669 (arc_lp): New pattern for LP instruction.
9670 (loop_end): New pattern.
9671 (loop_fail): Likewise.
9672 (decrement_and_branch_until_zero): Likewise.
9673 * config/arc/arc.opt (mlpc-width): New option.
9674 * doc/invoke.texi (mlpc-width): Document option.
9676 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9678 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
9679 (arc_ccfsm_advance): Fix checking for delay slots.
9680 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
9682 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9684 * config/arc/arc.md (movqi_insn): Add stores to save constant long
9686 (movhi_insn): Update store instruction constraint which are saving
9687 6-bit short immediates.
9688 (movsi_insn): Consider also short scaled load operations.
9689 (zero_extendhisi2_i): Use Usd constraint instead of T.
9690 (extendhisi2_i): Add q constraint.
9691 (arc_clzsi2): Add type and length attributes.
9692 (arc_ctzsi2): Likewise.
9693 * config/arc/constraints.md (Usc): Update constraint, the
9694 assembler can parse two relocations for a single instruction.
9696 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9698 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
9699 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
9701 2017-08-31 Olivier Hainque <hainque@adacore.com>
9703 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
9704 match as powerpc-wrs-vxworks*.
9706 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
9708 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
9709 register constraint for by-element operand.
9710 (aarch64_mls_elt_merge<mode>): Likewise.
9712 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9714 * config/arc/arc.c (arc_can_follow_jump): Check for short
9717 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9719 * config.gcc: Use g.opt for arc.
9720 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
9721 functionality moved to ...
9722 (legitimate_scaled_address_p): New function, ...here.
9723 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
9724 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
9725 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
9727 (arc_override_options): Handle G option.
9728 (arc_output_pic_addr_const): Correct function definition.
9729 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
9730 (arc_decl_anon_ns_mem_p): Delete.
9731 (arc_in_small_data_p): Overhaul this function to take into
9732 consideration the value given via G option.
9733 (arc_rewrite_small_data_1): Renamed and corrected old
9734 arc_rewrite_small_data function.
9735 (arc_rewrite_small_data): New function.
9736 (small_data_pattern): Don't use pic_offset_table_rtx.
9737 * config/arc/arc.h (CC1_SPEC): Recognize G option.
9738 * config/arc/simdext.md (movmisalignv2hi): Use
9739 prepare_move_operands function.
9741 (movmisalign*): Likewise.
9742 * doc/invoke.texi (ARC options): Document -G option.
9744 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9746 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
9748 * config/arc/arc.c (arc_print_operand): Output scalled address for
9749 sdata whenever is possible.
9750 (arc_in_small_data_p): Allow sdata for 64bit datum when double
9751 load/stores are available.
9752 (compact_sda_memory_operand): Check for the alignment required by
9753 code density instructions.
9754 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
9756 * config/arc/constraints.md (Usd): Update constraint.
9757 (Us0): New constraint.
9758 (Usc): Update constraint.
9760 2017-08-31 Richard Biener <rguenther@suse.de>
9763 * dwarf2out.c (dwarf2out_early_global_decl): Process each
9766 2017-08-31 Tamar Christina <tamar.christina@arm.com>
9768 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
9769 Resize type_signature.
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/aarch64.c (aarch64_base_register_rtx_p): Only allow
9776 subregs whose inner modes can be stored in GPRs.
9777 (aarch64_classify_index): Likewise.
9779 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9780 Alan Hayward <alan.hayward@arm.com>
9781 David Sherwood <david.sherwood@arm.com>
9783 * config/aarch64/iterators.md (V_cmp_result): Rename to...
9784 (V_INT_EQUIV): ...this.
9785 (v_cmp_result): Rename to...
9786 (v_int_equiv): ...this.
9787 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
9788 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
9789 (copysign<mode>3): Likewise.
9790 (aarch64_simd_bsl<mode>_internal): Likewise.
9791 (aarch64_simd_bsl<mode>): Likewise.
9792 (vec_cmp<mode><mode>): Likewise.
9793 (vcond<mode><mode>): Likewise.
9794 (vcond<v_cmp_mixed><mode>): Likewise.
9795 (vcondu<mode><v_cmp_mixed>): Likewise.
9796 (aarch64_cm<optab><mode>): Likewise.
9797 (aarch64_cmtst<mode>): Likewise.
9798 (aarch64_fac<optab><mode>): Likewise.
9799 (vec_perm_const<mode>): Likewise.
9800 (vcond_mask_<mode><v_cmp_result>): Rename to...
9801 (vcond_mask_<mode><v_int_equiv>): ...this.
9802 (vec_cmp<mode><v_cmp_result>): Rename to...
9803 (vec_cmp<mode><v_int_equiv>): ...this.
9805 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9806 Alan Hayward <alan.hayward@arm.com>
9807 David Sherwood <david.sherwood@arm.com>
9809 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
9811 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
9812 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
9813 (UNSPEC_LD4_DREG): New unspecs.
9814 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
9815 (aarch64_ld2<mode>_dreg_be): Replace with...
9816 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
9818 (aarch64_ld3<mode>_dreg_le)
9819 (aarch64_ld3<mode>_dreg_be): Replace with...
9820 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
9822 (aarch64_ld4<mode>_dreg_le)
9823 (aarch64_ld4<mode>_dreg_be): Replace with...
9824 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
9827 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9829 PR tree-optimization/81987
9830 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
9831 insert an initializer in a location not dominated by the stride
9834 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
9836 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
9837 on the entire header of the finally block in the fallthru case.
9839 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
9841 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
9843 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
9845 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
9846 rs6000_emit_move_from_cr and call renamed function.
9847 (rs6000_emit_prologue): Call renamed functions.
9848 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
9849 movesi_from_cr, remove volatile CRs.
9851 2017-08-30 Jon Beniston <jon@beniston.com>
9852 Richard Biener <rguenther@suse.de>
9854 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
9855 of VECTOR_MODE_P check.
9856 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
9857 element vector types.
9859 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9861 * df.h (df_read_modify_subreg_p): Remove in favor of...
9862 * rtl.h (read_modify_subreg_p): ...this new function. Take a
9863 const_rtx instead of an rtx.
9864 * cprop.c (local_cprop_find_used_regs): Update accordingly.
9865 * df-problems.c (df_word_lr_mark_ref): Likewise.
9866 * ira-lives.c (mark_pseudo_reg_live): Likewise.
9867 (mark_pseudo_reg_dead): Likewise.
9868 (mark_ref_dead): Likewise.
9869 * reginfo.c (init_subregs_of_mode): Likewise.
9870 * sched-deps.c (sched_analyze_1): Likewise.
9871 * df-scan.c (df_def_record_1): Likewise.
9872 (df_uses_record): Likewise.
9873 (df_read_modify_subreg_p): Remove in favor of...
9874 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
9875 const_rtx instead of an rtx.
9877 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9878 Alan Hayward <alan.hayward@arm.com>
9879 David Sherwood <david.sherwood@arm.com>
9881 * rtl.h (partial_subreg_p): New function.
9882 * caller-save.c (save_call_clobbered_regs): Use it.
9883 * calls.c (expand_call): Likewise.
9884 * combine.c (combinable_i3pat): Likewise.
9885 (simplify_set): Likewise.
9886 (make_extraction): Likewise.
9887 (make_compound_operation_int): Likewise.
9888 (gen_lowpart_or_truncate): Likewise.
9889 (force_to_mode): Likewise.
9890 (make_field_assignment): Likewise.
9891 (reg_truncated_to_mode): Likewise.
9892 (record_truncated_value): Likewise.
9893 (move_deaths): Likewise.
9894 * cse.c (record_jump_cond): Likewise.
9895 (cse_insn): Likewise.
9896 * cselib.c (cselib_lookup_1): Likewise.
9897 * expmed.c (extract_bit_field_using_extv): Likewise.
9898 * function.c (assign_parm_setup_reg): Likewise.
9899 * ifcvt.c (noce_convert_multiple_sets): Likewise.
9900 * ira-build.c (create_insn_allocnos): Likewise.
9901 * lra-coalesce.c (merge_pseudos): Likewise.
9902 * lra-constraints.c (match_reload): Likewise.
9903 (simplify_operand_subreg): Likewise.
9904 (curr_insn_transform): Likewise.
9905 * lra-lives.c (process_bb_lives): Likewise.
9906 * lra.c (new_insn_reg): Likewise.
9907 (lra_substitute_pseudo): Likewise.
9908 * regcprop.c (mode_change_ok): Likewise.
9909 (maybe_mode_change): Likewise.
9910 (copyprop_hardreg_forward_1): Likewise.
9911 * reload.c (push_reload): Likewise.
9912 (find_reloads): Likewise.
9913 (find_reloads_subreg_address): Likewise.
9914 * reload1.c (alter_reg): Likewise.
9915 (eliminate_regs_1): Likewise.
9916 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9918 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
9920 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
9921 back to if statements, including unpack.
9923 2017-08-30 Martin Liska <mliska@suse.cz>
9926 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
9928 (func_checker::compare_asm_inputs_outputs): ... this function.
9929 (func_checker::compare_gimple_asm): Use the function to compare
9930 also ASM constrains.
9931 * ipa-icf-gimple.h: Rename the function.
9933 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9934 Alan Hayward <alan.hayward@arm.com>
9935 David Sherwood <david.sherwood@arm.com>
9937 * coretypes.h (complex_mode): New type.
9938 * gdbhooks.py (build_pretty_printer): Handle it.
9939 * machmode.h (complex_mode): New class.
9940 (complex_mode::includes_p): New function.
9941 (is_complex_int_mode): Likewise.
9942 (is_complex_float_mode): Likewise.
9943 * genmodes.c (get_mode_class): Handle complex mode classes.
9944 * function.c (expand_function_end): Use is_complex_int_mode.
9946 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9947 Alan Hayward <alan.hayward@arm.com>
9948 David Sherwood <david.sherwood@arm.com>
9950 * coretypes.h (scalar_mode_pod): New typedef.
9951 * gdbhooks.py (build_pretty_printer): Handle it.
9952 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
9953 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
9954 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
9955 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
9958 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9959 Alan Hayward <alan.hayward@arm.com>
9960 David Sherwood <david.sherwood@arm.com>
9962 * machmode.h (mode_for_vector): Take a scalar_mode instead
9964 * stor-layout.c (mode_for_vector): Likewise.
9965 * explow.c (promote_mode): Use as_a <scalar_mode>.
9966 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
9968 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9969 Alan Hayward <alan.hayward@arm.com>
9970 David Sherwood <david.sherwood@arm.com>
9972 * target.def (preferred_simd_mode): Take a scalar_mode
9973 instead of a machine_mode.
9974 * targhooks.h (default_preferred_simd_mode): Likewise.
9975 * targhooks.c (default_preferred_simd_mode): Likewise.
9976 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
9977 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
9978 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
9979 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
9980 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
9981 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
9982 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
9983 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
9985 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
9986 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
9987 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
9988 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
9989 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
9990 * doc/tm.texi: Regenerate.
9991 * optabs-query.c (can_vec_mask_load_store_p): Return false for
9994 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9995 Alan Hayward <alan.hayward@arm.com>
9996 David Sherwood <david.sherwood@arm.com>
9998 * target.def (scalar_mode_supported_p): Take a scalar_mode
9999 instead of a machine_mode.
10000 * targhooks.h (default_scalar_mode_supported_p): Likewise.
10001 * targhooks.c (default_scalar_mode_supported_p): Likewise.
10002 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
10003 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
10004 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
10005 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
10006 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
10007 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
10008 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
10009 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
10010 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
10011 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
10012 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
10013 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
10015 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
10016 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
10017 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
10018 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
10019 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
10021 * doc/tm.texi: Regenerate.
10023 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10024 Alan Hayward <alan.hayward@arm.com>
10025 David Sherwood <david.sherwood@arm.com>
10027 * coretypes.h (opt_scalar_mode): New typedef.
10028 * gdbhooks.py (build_pretty_printers): Handle it.
10029 * machmode.h (mode_iterator::get_2xwider): Add overload for
10031 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
10033 * expr.c (convert_mode_scalar): Likewise.
10034 * omp-low.c (omp_clause_aligned_alignment): Likewise.
10035 * optabs.c (expand_float): Likewise.
10036 (expand_fix): Likewise.
10037 * tree-vect-stmts.c (vectorizable_conversion): 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 * optabs.c (expand_float): Explicitly check for scalars before
10044 using a branching expansion.
10045 (expand_fix): Likewise.
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 * expr.c (convert_mode): Split scalar handling out into...
10052 (convert_mode_scalar): ...this new function. Treat the modes
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 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
10061 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
10063 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10064 Alan Hayward <alan.hayward@arm.com>
10065 David Sherwood <david.sherwood@arm.com>
10067 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
10068 rather than a machine_mode.
10069 (fixed_from_string): Likewise.
10070 (fixed_convert): Likewise.
10071 (fixed_convert_from_int): Likewise.
10072 (fixed_convert_from_real): Likewise.
10073 (real_convert_from_fixed): Likewise.
10074 * fixed-value.c (fixed_from_double_int): Likewise.
10075 (fixed_from_string): Likewise.
10076 (fixed_convert): Likewise.
10077 (fixed_convert_from_int): Likewise.
10078 (fixed_convert_from_real): Likewise.
10079 (real_convert_from_fixed): Likewise.
10080 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
10082 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10083 Alan Hayward <alan.hayward@arm.com>
10084 David Sherwood <david.sherwood@arm.com>
10086 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
10087 of separate mode class checks. Do not allow vector modes here.
10088 (immed_wide_int_const): Use as_a <scalar_mode>.
10089 * explow.c (trunc_int_for_mode): Likewise.
10090 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
10091 (wi::shwi): Likewise.
10092 (wi::min_value): Likewise.
10093 (wi::max_value): Likewise.
10094 * dwarf2out.c (loc_descriptor): Likewise.
10095 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
10096 for CONST_WIDE_INT.
10098 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10099 Alan Hayward <alan.hayward@arm.com>
10100 David Sherwood <david.sherwood@arm.com>
10102 * tree.h (SCALAR_TYPE_MODE): New macro.
10103 * expr.c (expand_expr_addr_expr_1): Use it.
10104 (expand_expr_real_2): Likewise.
10105 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
10106 (fold_convert_const_fixed_from_int): Likewise.
10107 (fold_convert_const_fixed_from_real): Likewise.
10108 (native_encode_fixed): Likewise
10109 (native_encode_complex): Likewise
10110 (native_encode_vector): Likewise.
10111 (native_interpret_fixed): Likewise.
10112 (native_interpret_real): Likewise.
10113 (native_interpret_complex): Likewise.
10114 (native_interpret_vector): Likewise.
10115 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
10116 (simd_clone_adjust_argument_types): Likewise.
10117 (simd_clone_init_simd_arrays): Likewise.
10118 (simd_clone_adjust): Likewise.
10119 * stor-layout.c (layout_type): Likewise.
10120 * tree.c (build_minus_one_cst): Likewise.
10121 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
10122 * tree-inline.c (estimate_move_cost): Likewise.
10123 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
10124 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
10125 (vectorizable_reduction): Likewise.
10126 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
10127 (vect_recog_mixed_size_cond_pattern): Likewise.
10128 (check_bool_pattern): Likewise.
10129 (adjust_bool_pattern): Likewise.
10130 (search_type_for_mask_1): Likewise.
10131 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
10132 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10133 (vectorizable_load): Likewise.
10134 (vectorizable_store): Likewise.
10135 * ubsan.c (ubsan_encode_value): Likewise.
10136 * varasm.c (output_constant): Likewise.
10138 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10139 Alan Hayward <alan.hayward@arm.com>
10140 David Sherwood <david.sherwood@arm.com>
10142 * coretypes.h (scalar_mode): New class.
10143 * machmode.h (scalar_mode): Likewise.
10144 (scalar_mode::includes_p): New function.
10145 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
10146 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
10147 * genmodes.c (get_mode_class): Handle remaining scalar modes.
10148 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
10149 * expmed.c (store_bit_field_1): Likewise.
10150 (extract_bit_field_1): Likewise.
10151 * expr.c (write_complex_part): Likewise.
10152 (read_complex_part): Likewise.
10153 (emit_move_complex_push): Likewise.
10154 (expand_expr_real_2): Likewise.
10155 * function.c (assign_parm_setup_reg): Likewise.
10156 (assign_parms_unsplit_complex): Likewise.
10157 * optabs.c (expand_binop): Likewise.
10158 * rtlanal.c (subreg_get_info): Likewise.
10159 * simplify-rtx.c (simplify_immed_subreg): Likewise.
10160 * varasm.c (output_constant_pool_2): Likewise.
10162 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10163 Alan Hayward <alan.hayward@arm.com>
10164 David Sherwood <david.sherwood@arm.com>
10166 * expmed.c (extract_high_half): Use scalar_int_mode and remove
10168 (expmed_mult_highpart_optab): Likewise.
10169 (expmed_mult_highpart): Likewise.
10171 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10172 Alan Hayward <alan.hayward@arm.com>
10173 David Sherwood <david.sherwood@arm.com>
10175 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
10176 instead of a machine_mode.
10177 (builtin_memset_read_str): Likewise.
10178 * builtins.c (c_readstr): Likewise.
10179 (builtin_memcpy_read_str): Likewise.
10180 (builtin_strncpy_read_str): Likewise.
10181 (builtin_memset_read_str): Likewise.
10182 (builtin_memset_gen_str): Likewise.
10183 (expand_builtin_signbit): Use scalar_int_mode for local variables.
10184 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
10185 instead of a machine_mode.
10186 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
10188 (make_extraction): Likewise.
10189 (try_widen_shift_mode): Take and return scalar_int_modes instead
10191 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
10192 a scalar_int_mode instead of a machine_mode.
10193 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
10194 (avr_addr_space_pointer_mode): Likewise.
10195 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
10196 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
10197 (msp430_unwind_word_mode): Likewise.
10198 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
10199 (spu_addr_space_pointer_mode): Likewise.
10200 (spu_addr_space_address_mode): Likewise.
10201 (spu_libgcc_cmp_return_mode): Likewise.
10202 (spu_libgcc_shift_count_mode): Likewise.
10203 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
10204 (rl78_addr_space_pointer_mode): Likewise.
10205 (fl78_unwind_word_mode): Likewise.
10206 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
10208 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
10209 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
10210 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
10211 (mips_valid_pointer_mode): Likewise.
10212 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
10213 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
10214 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
10216 (ft32_addr_space_address_mode): Likewise.
10217 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
10218 scalar_int_mode instead of a machine_mode.
10219 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
10221 (m32c_addr_space_address_mode): Likewise.
10222 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
10223 (rs6000_eh_return_filter_mode): Likewise.
10224 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
10225 (rs6000_eh_return_filter_mode): Likewise.
10226 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
10227 (s390_libgcc_shift_count_mode): Likewise.
10228 (s390_unwind_word_mode): Likewise.
10229 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
10231 * target.def (mode_rep_extended): Likewise.
10232 (valid_pointer_mode): Likewise.
10233 (addr_space.valid_pointer_mode): Likewise.
10234 (eh_return_filter_mode): Return a scalar_int_mode rather than
10236 (libgcc_cmp_return_mode): Likewise.
10237 (libgcc_shift_count_mode): Likewise.
10238 (unwind_word_mode): Likewise.
10239 (addr_space.pointer_mode): Likewise.
10240 (addr_space.address_mode): Likewise.
10241 * doc/tm.texi: Regenerate.
10242 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
10244 (do_jump): Use scalar_int_mode for local variables.
10245 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
10246 rather than a machine_mode.
10247 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
10248 (scompare_loc_descriptor_wide): Likewise.
10249 (scompare_loc_descriptor_narrow): Likewise.
10250 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
10252 * except.c (sjlj_emit_dispatch_table): Likewise.
10253 (expand_builtin_eh_copy_values): Likewise.
10254 * explow.c (convert_memory_address_addr_space_1): Likewise.
10255 Take a scalar_int_mode rather than a machine_mode.
10256 (convert_memory_address_addr_space): Take a scalar_int_mode rather
10257 than a machine_mode.
10258 (memory_address_addr_space): Use scalar_int_mode for local variables.
10259 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
10260 rather than a machine_mode.
10261 * expmed.c (mask_rtx): Likewise.
10262 (init_expmed_one_conv): Likewise.
10263 (expand_mult_highpart_adjust): Likewise.
10264 (extract_high_half): Likewise.
10265 (expmed_mult_highpart_optab): Likewise.
10266 (expmed_mult_highpart): Likewise.
10267 (expand_smod_pow2): Likewise.
10268 (expand_sdiv_pow2): Likewise.
10269 (emit_store_flag_int): Likewise.
10270 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
10272 (extract_low_bits): Likewise.
10273 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
10275 * expr.c (pieces_addr::adjust): Likewise.
10276 (can_store_by_pieces): Likewise.
10277 (store_by_pieces): Likewise.
10278 (clear_by_pieces_1): Likewise.
10279 (expand_expr_addr_expr_1): Likewise.
10280 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
10281 (expand_expr_real_1): Likewise.
10282 (try_casesi): Likewise.
10283 * final.c (shorten_branches): Likewise.
10284 * fold-const.c (fold_convert_const_int_from_fixed): Change the
10285 type of "mode" to machine_mode.
10286 * internal-fn.c (expand_arith_overflow_result_store): Take a
10287 scalar_int_mode rather than a machine_mode.
10288 (expand_mul_overflow): Use scalar_int_mode for local variables.
10289 * loop-doloop.c (doloop_modify): Likewise.
10290 (doloop_optimize): Likewise.
10291 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
10292 than a machine_mode.
10293 (expand_doubleword_shift_condmove): Likewise.
10294 (expand_doubleword_shift): Likewise.
10295 (expand_doubleword_clz): Likewise.
10296 (expand_doubleword_popcount): Likewise.
10297 (expand_doubleword_parity): Likewise.
10298 (expand_absneg_bit): Use scalar_int_mode for local variables.
10299 (prepare_float_lib_cmp): Likewise.
10300 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
10301 rather than a machine_mode.
10302 (convert_memory_address_addr_space): Likewise.
10303 (get_mode_bounds): Likewise.
10304 (get_address_mode): Return a scalar_int_mode rather than a
10306 * rtlanal.c (get_address_mode): Likewise.
10307 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
10308 than a machine_mode.
10309 * targhooks.c (default_mode_rep_extended): Likewise.
10310 (default_valid_pointer_mode): Likewise.
10311 (default_addr_space_valid_pointer_mode): Likewise.
10312 (default_eh_return_filter_mode): Return a scalar_int_mode rather
10313 than a machine_mode.
10314 (default_libgcc_cmp_return_mode): Likewise.
10315 (default_libgcc_shift_count_mode): Likewise.
10316 (default_unwind_word_mode): Likewise.
10317 (default_addr_space_pointer_mode): Likewise.
10318 (default_addr_space_address_mode): Likewise.
10319 * targhooks.h (default_eh_return_filter_mode): Likewise.
10320 (default_libgcc_cmp_return_mode): Likewise.
10321 (default_libgcc_shift_count_mode): Likewise.
10322 (default_unwind_word_mode): Likewise.
10323 (default_addr_space_pointer_mode): Likewise.
10324 (default_addr_space_address_mode): Likewise.
10325 (default_mode_rep_extended): Take a scalar_int_mode rather than
10327 (default_valid_pointer_mode): Likewise.
10328 (default_addr_space_valid_pointer_mode): Likewise.
10329 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
10331 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
10332 rather than a machine_mode.
10333 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
10334 for local variables.
10335 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10336 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
10337 than a machine_mode.
10339 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10340 Alan Hayward <alan.hayward@arm.com>
10341 David Sherwood <david.sherwood@arm.com>
10343 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
10344 the mode argument to scalar_int_mode.
10345 (do_jump_by_parts_zero_rtx): Likewise.
10346 (do_jump_by_parts_equality_rtx): Likewise.
10347 (do_jump_by_parts_greater): Take a mode argument.
10348 (do_jump_by_parts_equality): Likewise.
10349 (do_jump_1): Update calls accordingly.
10351 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10352 Alan Hayward <alan.hayward@arm.com>
10353 David Sherwood <david.sherwood@arm.com>
10355 * is-a.h (safe_dyn_cast): New function.
10356 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
10357 (jump_table_for_label): Likewise.
10358 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
10359 instead of an rtx_insn *.
10360 (shorten_branches): Use dyn_cast instead of LABEL_P and
10361 JUMP_TABLE_DATA_P. Use jump_table_for_label and
10362 rtx_jump_table_data::get_data_mode.
10363 (final_scan_insn): Likewise.
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 * combine.c (try_combine): Use is_a <scalar_int_mode> when
10370 trying to combine a full-register integer set with a subreg
10373 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10374 Alan Hayward <alan.hayward@arm.com>
10375 David Sherwood <david.sherwood@arm.com>
10377 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
10378 that is always either address_mode or pointer_mode.
10380 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10381 Alan Hayward <alan.hayward@arm.com>
10382 David Sherwood <david.sherwood@arm.com>
10384 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
10385 when the two are known to be equal.
10387 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10388 Alan Hayward <alan.hayward@arm.com>
10389 David Sherwood <david.sherwood@arm.com>
10391 * simplify-rtx.c (simplify_const_unary_operation): Use
10392 is_a <scalar_int_mode> instead of checking for a nonzero
10393 precision. Forcibly convert op_mode to a scalar_int_mode
10394 in that case. More clearly differentiate the operand and
10395 result modes and use the former when deciding what the value
10396 of a count-bits operation should be. Use is_int_mode instead
10397 of checking for a MODE_INT. Remove redundant check for whether
10398 this mode has a zero precision.
10400 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10401 Alan Hayward <alan.hayward@arm.com>
10402 David Sherwood <david.sherwood@arm.com>
10404 * optabs.c (widen_leading): Change the type of the mode argument
10405 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
10406 (widen_bswap): Likewise.
10407 (expand_parity): Likewise.
10408 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
10409 (expand_ffs): Likewise.
10410 (epand_unop): Check for scalar integer modes before calling the
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 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
10419 (expand_expr_real_1): Update call accordingly.
10421 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10422 Alan Hayward <alan.hayward@arm.com>
10423 David Sherwood <david.sherwood@arm.com>
10425 * expmed.c (store_bit_field_using_insv): Add op0_mode and
10426 value_mode arguments. Use scalar_int_mode internally.
10427 (store_bit_field_1): Rename the new integer mode from imode
10428 to op0_mode and use it instead of GET_MODE (op0). Update calls
10429 to store_split_bit_field, store_bit_field_using_insv and
10430 store_fixed_bit_field.
10431 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
10432 Use scalar_int_mode internally. Use a bit count rather than a mode
10433 when calculating the largest bit size for get_best_mode.
10434 Update calls to store_split_bit_field and store_fixed_bit_field_1.
10435 (store_fixed_bit_field_1): Add mode and value_mode arguments.
10436 Remove assertion that OP0 has a scalar integer mode.
10437 (store_split_bit_field): Add op0_mode and value_mode arguments.
10438 Update calls to extract_fixed_bit_field.
10439 (extract_bit_field_using_extv): Add an op0_mode argument.
10440 Use scalar_int_mode internally.
10441 (extract_bit_field_1): Rename the new integer mode from imode to
10442 op0_mode and use it instead of GET_MODE (op0). Update calls to
10443 extract_split_bit_field, extract_bit_field_using_extv and
10444 extract_fixed_bit_field.
10445 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
10446 to extract_split_bit_field and extract_fixed_bit_field_1.
10447 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
10448 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
10449 on the target mode.
10450 (extract_split_bit_field): Add an op0_mode argument. Update call
10451 to extract_fixed_bit_field.
10453 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10454 Alan Hayward <alan.hayward@arm.com>
10455 David Sherwood <david.sherwood@arm.com>
10457 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
10458 * explow.c (hard_function_value): Likewise.
10459 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
10460 convert_to_mode call outside the loop.
10461 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
10462 for the mode iterator. Require the mode specified by max_pieces
10464 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
10466 (copy_blkmode_to_reg): Likewise.
10467 (set_storage_via_setmem): Likewise.
10468 * optabs.c (prepare_cmp_insn): Likewise.
10469 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
10470 * stor-layout.c (finish_bitfield_representative): Likewise.
10472 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10473 Alan Hayward <alan.hayward@arm.com>
10474 David Sherwood <david.sherwood@arm.com>
10476 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
10477 * expr.c (convert_move): Use them.
10478 (convert_modes): Likewise.
10479 (store_expr_with_bounds): Likewise.
10481 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10482 Alan Hayward <alan.hayward@arm.com>
10483 David Sherwood <david.sherwood@arm.com>
10485 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
10486 parameter for the mode of "x". Remove the "known_x", "known_mode"
10487 and "known_ret" arguments. Change the type of the mode argument
10488 to scalar_int_mode.
10489 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
10490 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
10491 (reg_num_sign_bit_copies_for_combine): Likewise.
10492 * rtlanal.c (nonzero_bits1): Likewise.
10493 (num_sign_bit_copies1): Likewise.
10494 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
10495 (reg_num_sign_bit_copies_general): Likewise.
10496 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
10497 (reg_nonzero_bits_general): Likewise.
10499 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10500 Alan Hayward <alan.hayward@arm.com>
10501 David Sherwood <david.sherwood@arm.com>
10503 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
10504 than in subroutines. Return 1 for non-integer modes.
10505 (cached_num_sign_bit_copies): Change the type of the mode parameter
10506 to scalar_int_mode.
10507 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
10508 classes. Handle CONST_INT_P first and then check whether X also
10509 has a scalar integer mode. Check the same thing for inner registers
10510 of a SUBREG and for values that are being extended or truncated.
10512 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10513 Alan Hayward <alan.hayward@arm.com>
10514 David Sherwood <david.sherwood@arm.com>
10516 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
10517 in subroutines. Return the mode mask for non-integer modes.
10518 (cached_nonzero_bits): Change the type of the mode parameter
10519 to scalar_int_mode.
10520 (nonzero_bits1): Likewise. Remove early exit for other mode
10521 classes. Handle CONST_INT_P first and then check whether X
10522 also has a scalar integer mode.
10524 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10525 Alan Hayward <alan.hayward@arm.com>
10526 David Sherwood <david.sherwood@arm.com>
10528 * expr.c (widest_int_mode_for_size): Make the comment match the code.
10529 Return a scalar_int_mode and assert that the size is greater than
10531 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
10532 (op_by_pieces_d::op_by_pieces_d): Likewise.
10533 (op_by_pieces_d::run): Likewise.
10534 (can_store_by_pieces): Likewise.
10536 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10537 Alan Hayward <alan.hayward@arm.com>
10538 David Sherwood <david.sherwood@arm.com>
10540 * combine.c (extract_left_shift): Add a mode argument and update
10542 (make_compound_operation_int): Change the type of the mode parameter
10543 to scalar_int_mode and update the call to extract_left_shift.
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_and_const_int): Change the type of the mode
10550 parameter to scalar_int_mode.
10551 (simplify_and_const_int_1): Likewise. Update recursive call.
10553 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10554 Alan Hayward <alan.hayward@arm.com>
10555 David Sherwood <david.sherwood@arm.com>
10557 * combine.c (simplify_compare_const): Check that the mode is a
10558 scalar_int_mode (rather than VOIDmode) before testing its
10560 (simplify_comparison): Move COMPARISON_P handling out of the
10561 loop and restrict the latter part of the loop to scalar_int_modes.
10562 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
10563 and when considering SUBREG_REGs. Use is_int_mode instead of
10564 checking GET_MODE_CLASS against MODE_INT.
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 (try_widen_shift_mode): Move check for equal modes to...
10571 (simplify_shift_const_1): ...here. Use scalar_int_mode for
10572 shift_unit_mode and for modes involved in scalar shifts.
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 * combine.c (force_int_to_mode): New function, split out from...
10579 (force_to_mode): ...here. Keep xmode up-to-date and use it
10580 instead of GET_MODE (x).
10582 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10583 Alan Hayward <alan.hayward@arm.com>
10584 David Sherwood <david.sherwood@arm.com>
10586 * optabs-query.h (extraction_insn::struct_mode): Change type to
10587 opt_scalar_int_mode and update comment.
10588 (extraction_insn::field_mode): Change type to scalar_int_mode.
10589 (extraction_insn::pos_mode): Likewise.
10590 * combine.c (make_extraction): Update accordingly.
10591 * optabs-query.c (get_traditional_extraction_insn): Likewise.
10592 (get_optab_extraction_insn): Likewise.
10593 * recog.c (simplify_while_replacing): Likewise.
10594 * expmed.c (narrow_bit_field_mem): Change the type of the mode
10595 parameter to opt_scalar_int_mode.
10597 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10598 Alan Hayward <alan.hayward@arm.com>
10599 David Sherwood <david.sherwood@arm.com>
10601 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
10602 to a scalar_int_mode instead of a machine_mode.
10603 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
10604 (get_best_mode): Return a boolean and use a pointer argument to store
10605 the selected mode. Replace the limit mode parameter with a bit limit.
10606 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
10607 for the values returned by bit_field_mode_iterator::next_mode.
10608 (store_bit_field): Update call to get_best_mode.
10609 (store_fixed_bit_field): Likewise.
10610 (extract_fixed_bit_field): Likewise.
10611 * expr.c (optimize_bitfield_assignment_op): Likewise.
10612 * fold-const.c (optimize_bit_field_compare): Likewise.
10613 (fold_truth_andor_1): Likewise.
10614 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
10615 Update for new type of m_mode.
10616 (get_best_mode): As above.
10618 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10619 Alan Hayward <alan.hayward@arm.com>
10620 David Sherwood <david.sherwood@arm.com>
10622 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
10623 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
10624 (store_bit_field): Check is_a <scalar_int_mode> before calling
10625 strict_volatile_bitfield_p.
10626 (extract_bit_field): Likewise.
10628 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10629 Alan Hayward <alan.hayward@arm.com>
10630 David Sherwood <david.sherwood@arm.com>
10632 * target.def (cstore_mode): Return a scalar_int_mode.
10633 * doc/tm.texi: Regenerate.
10634 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
10635 * targhooks.h (default_cstore_mode): Likewise.
10636 * targhooks.c (default_cstore_mode): Likewise, using a forced
10638 * expmed.c (emit_cstore): Expect the target of the cstore to be
10641 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10642 Alan Hayward <alan.hayward@arm.com>
10643 David Sherwood <david.sherwood@arm.com>
10645 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
10647 (niter_desc): Likewise mode.
10648 (iv_analyze): Add a mode parameter.
10650 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
10651 and change its type to scalar_int_mode.
10652 * loop-iv.c: Update commentary at head of file.
10653 (iv_constant): Pass the mode paraeter before the rtx it describes
10654 and change its type to scalar_int_mode. Remove VOIDmode handling.
10655 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
10656 (iv_extend): Likewise.
10657 (shorten_into_mode): Likewise.
10658 (iv_add): Use scalar_int_mode.
10659 (iv_mult): Likewise.
10660 (iv_shift): Likewise.
10661 (canonicalize_iv_subregs): Likewise.
10662 (get_biv_step_1): Pass the outer_mode parameter before the rtx
10663 it describes and change its mode to scalar_int_mode. Also change
10664 the type of the returned inner_mode to scalar_int_mode.
10665 (get_biv_step): Likewise, turning outer_mode from a pointer
10666 into a direct parameter. Update call to get_biv_step_1.
10667 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
10668 iv_constant and get_biv_step.
10669 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
10670 and change its type to scalar_int_mode. Don't initialise iv->mode
10671 to VOIDmode and remove later checks for its still being VOIDmode.
10672 Update calls to iv_analyze_op and iv_analyze_expr. Check
10673 is_a <scalar_int_mode> when changing the mode under consideration.
10674 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
10675 Update call to iv_analyze_expr.
10676 (iv_analyze_op): Add a mode parameter. Reject subregs whose
10677 inner register is not also a scalar_int_mode. Update call to
10679 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
10680 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
10681 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
10682 separate mode class checks. Update calls to iv_analyze. Remove
10683 fix-up of VOIDmodes after iv_analyze_biv.
10684 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
10685 don't have a scalar_int_mode. Update call to biv_p.
10687 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10688 Alan Hayward <alan.hayward@arm.com>
10689 David Sherwood <david.sherwood@arm.com>
10691 * cfgexpand.c (convert_debug_memory_address): Use
10692 as_a <scalar_int_mode>.
10693 * combine.c (expand_compound_operation): Likewise.
10694 (make_extraction): Likewise.
10695 (change_zero_ext): Likewise.
10696 (simplify_comparison): Likewise.
10697 * cse.c (cse_insn): Likewise.
10698 * dwarf2out.c (minmax_loc_descriptor): Likewise.
10699 (mem_loc_descriptor): Likewise.
10700 (loc_descriptor): Likewise.
10701 * expmed.c (init_expmed_one_mode): Likewise.
10702 (synth_mult): Likewise.
10703 (emit_store_flag_1): Likewise.
10704 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
10705 of a comparison with size.
10706 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
10707 (reduce_to_bit_field_precision): Likewise.
10708 * function.c (expand_function_end): Likewise.
10709 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
10710 * loop-doloop.c (doloop_modify): Likewise.
10711 * optabs.c (expand_binop): Likewise.
10712 (expand_unop): Likewise.
10713 (expand_copysign_absneg): Likewise.
10714 (prepare_cmp_insn): Likewise.
10715 (maybe_legitimize_operand): Likewise.
10716 * recog.c (const_scalar_int_operand): Likewise.
10717 * rtlanal.c (get_address_mode): Likewise.
10718 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10719 (simplify_cond_clz_ctz): Likewise.
10720 * tree-nested.c (get_nl_goto_field): Likewise.
10721 * tree.c (build_vector_type_for_mode): Likewise.
10722 * var-tracking.c (use_narrower_mode): Likewise.
10724 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10725 Alan Hayward <alan.hayward@arm.com>
10726 David Sherwood <david.sherwood@arm.com>
10728 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
10729 * builtins.c (expand_builtin_signbit): Use it.
10730 * cfgexpand.c (expand_debug_expr): Likewise.
10731 * dojump.c (do_jump): Likewise.
10732 (do_compare_and_jump): Likewise.
10733 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
10734 * expmed.c (make_tree): Likewise.
10735 * expr.c (expand_expr_real_2): Likewise.
10736 (expand_expr_real_1): Likewise.
10737 (try_casesi): Likewise.
10738 * fold-const-call.c (fold_const_call_ss): Likewise.
10739 * fold-const.c (unextend): Likewise.
10740 (extract_muldiv_1): Likewise.
10741 (fold_single_bit_test): Likewise.
10742 (native_encode_int): Likewise.
10743 (native_encode_string): Likewise.
10744 (native_interpret_int): Likewise.
10745 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
10746 * internal-fn.c (expand_addsub_overflow): Likewise.
10747 (expand_neg_overflow): Likewise.
10748 (expand_mul_overflow): Likewise.
10749 (expand_arith_overflow): Likewise.
10750 * match.pd: Likewise.
10751 * stor-layout.c (layout_type): Likewise.
10752 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
10753 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
10754 * tree-ssanames.c (get_range_info): Likewise.
10755 * tree-switch-conversion.c (array_value_type) Likewise.
10756 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
10757 (vect_recog_divmod_pattern): Likewise.
10758 (vect_recog_mixed_size_cond_pattern): Likewise.
10759 * tree-vrp.c (extract_range_basic): Likewise.
10760 (simplify_float_conversion_using_ranges): Likewise.
10761 * tree.c (int_fits_type_p): Likewise.
10762 * ubsan.c (instrument_bool_enum_load): Likewise.
10763 * varasm.c (mergeable_string_section): Likewise.
10764 (narrowing_initializer_constant_valid_p): Likewise.
10765 (output_constant): Likewise.
10767 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10768 Alan Hayward <alan.hayward@arm.com>
10769 David Sherwood <david.sherwood@arm.com>
10771 * machmode.h (NARROWEST_INT_MODE): New macro.
10772 * expr.c (alignment_for_piecewise_move): Use it instead of
10773 GET_CLASS_NARROWEST_MODE (MODE_INT).
10774 (push_block): Likewise.
10775 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
10777 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10779 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10780 Alan Hayward <alan.hayward@arm.com>
10781 David Sherwood <david.sherwood@arm.com>
10783 * postreload.c (move2add_valid_value_p): Change the type of the
10784 mode parameter to scalar_int_mode.
10785 (move2add_use_add2_insn): Add a mode parameter and use it instead
10787 (move2add_use_add3_insn): Likewise.
10788 (reload_cse_move2add): Update accordingly.
10790 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10791 Alan Hayward <alan.hayward@arm.com>
10792 David Sherwood <david.sherwood@arm.com>
10794 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
10796 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
10797 * optabs.c (expand_unop): Likewise.
10799 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10800 Alan Hayward <alan.hayward@arm.com>
10801 David Sherwood <david.sherwood@arm.com>
10803 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
10804 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
10805 (popcount_loc_descriptor): Likewise.
10806 (bswap_loc_descriptor): Likewise.
10807 (rotate_loc_descriptor): Likewise.
10808 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
10809 calling the functions above.
10811 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10812 Alan Hayward <alan.hayward@arm.com>
10813 David Sherwood <david.sherwood@arm.com>
10815 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
10817 (try_combine): Likewise.
10818 (simplify_if_then_else): Likewise.
10819 * cse.c (cse_insn): Likewise.
10820 * dwarf2out.c (mem_loc_descriptor): Likewise.
10821 * emit-rtl.c (gen_lowpart_common): Likewise.
10822 * simplify-rtx.c (simplify_truncation): Likewise.
10823 (simplify_binary_operation_1): Likewise.
10824 (simplify_const_relational_operation): Likewise.
10825 (simplify_ternary_operation): Likewise.
10826 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): 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 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
10833 * reload.c (push_reload): Likewise.
10834 (find_reloads): Likewise.
10836 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10837 Alan Hayward <alan.hayward@arm.com>
10838 David Sherwood <david.sherwood@arm.com>
10840 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
10841 (make_compound_operation_int): Likewise.
10842 (change_zero_ext): Likewise.
10843 * expr.c (convert_move): Likewise.
10844 (convert_modes): Likewise.
10845 * fwprop.c (forward_propagate_subreg): Likewise.
10846 * loop-iv.c (get_biv_step_1): Likewise.
10847 * optabs.c (widen_operand): Likewise.
10848 * postreload.c (move2add_valid_value_p): Likewise.
10849 * recog.c (simplify_while_replacing): Likewise.
10850 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10851 (simplify_binary_operation_1): Likewise. Remove redundant
10852 mode equality check.
10854 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10855 Alan Hayward <alan.hayward@arm.com>
10856 David Sherwood <david.sherwood@arm.com>
10858 * combine.c (combine_simplify_rtx): Add checks for
10859 is_a <scalar_int_mode>.
10860 (simplify_if_then_else): Likewise.
10861 (make_field_assignment): Likewise.
10862 (simplify_comparison): Likewise.
10863 * ifcvt.c (noce_try_bitop): Likewise.
10864 * loop-invariant.c (canonicalize_address_mult): Likewise.
10865 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10867 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10868 Alan Hayward <alan.hayward@arm.com>
10869 David Sherwood <david.sherwood@arm.com>
10871 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
10872 is_a <scalar_int_mode> instead of != BLKmode.
10874 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10875 Alan Hayward <alan.hayward@arm.com>
10876 David Sherwood <david.sherwood@arm.com>
10878 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
10879 instead of != VOIDmode.
10880 * combine.c (if_then_else_cond): Likewise.
10881 (change_zero_ext): Likewise.
10882 * dwarf2out.c (mem_loc_descriptor): Likewise.
10883 (loc_descriptor): Likewise.
10884 * rtlanal.c (canonicalize_condition): Likewise.
10885 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
10887 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10888 Alan Hayward <alan.hayward@arm.com>
10889 David Sherwood <david.sherwood@arm.com>
10891 * simplify-rtx.c (simplify_binary_operation_1): Use
10892 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
10894 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10895 Alan Hayward <alan.hayward@arm.com>
10896 David Sherwood <david.sherwood@arm.com>
10898 * wide-int.h (int_traits<unsigned char>) New class.
10899 (int_traits<unsigned short>) Likewise.
10900 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
10901 Use GET_MODE_UNIT_PRECISION and remove redundant test for
10903 * combine.c (set_nonzero_bits_and_sign_copies): Use
10904 is_a <scalar_int_mode>.
10905 (find_split_point): Likewise.
10906 (combine_simplify_rtx): Likewise.
10907 (simplify_logical): Likewise.
10908 (expand_compound_operation): Likewise.
10909 (expand_field_assignment): Likewise.
10910 (make_compound_operation): Likewise.
10911 (extended_count): Likewise.
10912 (change_zero_ext): Likewise.
10913 (simplify_comparison): Likewise.
10914 * dwarf2out.c (scompare_loc_descriptor): Likewise.
10915 (ucompare_loc_descriptor): Likewise.
10916 (minmax_loc_descriptor): Likewise.
10917 (mem_loc_descriptor): Likewise.
10918 (loc_descriptor): Likewise.
10919 * expmed.c (init_expmed_one_mode): Likewise.
10920 * lra-constraints.c (lra_constraint_offset): Likewise.
10921 * optabs.c (prepare_libcall_arg): Likewise.
10922 * postreload.c (move2add_note_store): Likewise.
10923 * reload.c (operands_match_p): Likewise.
10924 * rtl.h (load_extend_op): Likewise.
10925 * rtlhooks.c (gen_lowpart_general): Likewise.
10926 * simplify-rtx.c (simplify_truncation): Likewise.
10927 (simplify_unary_operation_1): Likewise.
10928 (simplify_binary_operation_1): Likewise.
10929 (simplify_const_binary_operation): Likewise.
10930 (simplify_const_relational_operation): Likewise.
10931 (simplify_subreg): Likewise.
10932 * stor-layout.c (bitwise_mode_for_mode): Likewise.
10933 * var-tracking.c (adjust_mems): Likewise.
10934 (prepare_call_arguments): Likewise.
10936 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10937 Alan Hayward <alan.hayward@arm.com>
10938 David Sherwood <david.sherwood@arm.com>
10940 * machmode.h (is_int_mode): New fuction.
10941 * combine.c (find_split_point): Use it.
10942 (combine_simplify_rtx): Likewise.
10943 (simplify_if_then_else): Likewise.
10944 (simplify_set): Likewise.
10945 (simplify_shift_const_1): Likewise.
10946 (simplify_comparison): Likewise.
10947 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
10948 * cse.c (notreg_cost): Likewise.
10949 (cse_insn): Likewise.
10950 * cselib.c (cselib_lookup_1): Likewise.
10951 * dojump.c (do_jump_1): Likewise.
10952 (do_compare_rtx_and_jump): Likewise.
10953 * dse.c (get_call_args): Likewise.
10954 * dwarf2out.c (rtl_for_decl_init): Likewise.
10955 (native_encode_initializer): Likewise.
10956 * expmed.c (emit_store_flag_1): Likewise.
10957 (emit_store_flag): Likewise.
10958 * expr.c (convert_modes): Likewise.
10959 (store_field): Likewise.
10960 (expand_expr_real_1): Likewise.
10961 * fold-const.c (fold_read_from_constant_string): Likewise.
10962 * gimple-ssa-sprintf.c (get_format_string): Likewise.
10963 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
10964 * optabs.c (expand_binop): Likewise.
10965 (expand_unop): Likewise.
10966 (expand_abs_nojump): Likewise.
10967 (expand_one_cmpl_abs_nojump): Likewise.
10968 * simplify-rtx.c (mode_signbit_p): Likewise.
10969 (val_signbit_p): Likewise.
10970 (val_signbit_known_set_p): Likewise.
10971 (val_signbit_known_clear_p): Likewise.
10972 (simplify_relational_operation_1): Likewise.
10973 * tree.c (vector_type_mode): Likewise.
10975 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10976 Alan Hayward <alan.hayward@arm.com>
10977 David Sherwood <david.sherwood@arm.com>
10979 * machmode.h (smallest_mode_for_size): Fix formatting.
10980 (smallest_int_mode_for_size): New function.
10981 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
10982 instead of smallest_mode_for_size.
10983 * combine.c (make_extraction): Likewise.
10984 * config/arc/arc.c (arc_expand_movmem): Likewise.
10985 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
10986 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
10987 * config/s390/s390.c (s390_expand_insv): Likewise.
10988 * config/sparc/sparc.c (assign_int_registers): Likewise.
10989 * config/spu/spu.c (spu_function_value): Likewise.
10990 (spu_function_arg): Likewise.
10991 * coverage.c (get_gcov_type): Likewise.
10992 (get_gcov_unsigned_t): Likewise.
10993 * dse.c (find_shift_sequence): Likewise.
10994 * expmed.c (store_bit_field_1): Likewise.
10995 * expr.c (convert_move): Likewise.
10996 (store_field): Likewise.
10997 * internal-fn.c (expand_arith_overflow): Likewise.
10998 * optabs-query.c (get_best_extraction_insn): Likewise.
10999 * optabs.c (expand_twoval_binop_libfunc): Likewise.
11000 * stor-layout.c (layout_type): Likewise.
11001 (initialize_sizetypes): Likewise.
11002 * targhooks.c (default_get_mask_mode): Likewise.
11003 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
11005 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11006 Alan Hayward <alan.hayward@arm.com>
11007 David Sherwood <david.sherwood@arm.com>
11009 * machmode.h (opt_mode::else_blk): New function.
11010 (int_mode_for_mode): Declare.
11011 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
11012 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
11014 * cfgexpand.c (expand_debug_expr): Likewise.
11015 * combine.c (gen_lowpart_or_truncate): Likewise.
11016 (gen_lowpart_for_combine): Likewise.
11017 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
11018 * config/avr/avr.c (avr_to_int_mode): Likewise.
11019 (avr_out_plus_1): Likewise.
11020 (avr_out_plus): Likewise.
11021 (avr_out_round): Likewise.
11022 * config/i386/i386.c (ix86_split_to_parts): Likewise.
11023 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
11024 (s390_expand_vcond): Likewise.
11025 * config/spu/spu.c (spu_split_immediate): Likewise.
11026 (spu_expand_mov): Likewise.
11027 * dse.c (get_stored_val): Likewise.
11028 * expmed.c (store_bit_field_1): Likewise.
11029 (convert_extracted_bit_field): Use int_mode_for_mode instead of
11031 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
11032 (extract_low_bits): Likewise.
11033 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
11034 handling rather than repeating the check.
11035 (emit_group_store): Likewise.
11036 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
11037 * optabs.c (expand_absneg_bit): Likewise.
11038 (expand_copysign_absneg): Likewise.
11039 (expand_copysign_bit): Likewise.
11040 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
11041 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
11042 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
11043 * var-tracking.c (prepare_call_arguments): Likewise.
11044 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
11045 int_mode_for_mode instead of mode_for_size.
11046 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
11048 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11049 Alan Hayward <alan.hayward@arm.com>
11050 David Sherwood <david.sherwood@arm.com>
11052 * machmode.h (int_mode_for_size): New function.
11053 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
11054 instead of mode_for_size.
11055 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
11057 * combine.c (expand_field_assignment): Use int_mode_for_size
11058 instead of mode_for_size.
11059 (make_extraction): Likewise.
11060 (simplify_shift_const_1): Likewise.
11061 (simplify_comparison): Likewise.
11062 * dojump.c (do_jump): Likewise.
11063 * dwarf2out.c (mem_loc_descriptor): Likewise.
11064 * emit-rtl.c (init_derived_machine_modes): Likewise.
11065 * expmed.c (flip_storage_order): Likewise.
11066 (convert_extracted_bit_field): Likewise.
11067 * expr.c (copy_blkmode_from_reg): Likewise.
11068 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
11069 * internal-fn.c (expand_mul_overflow): Likewise.
11070 * lower-subreg.c (simple_move): Likewise.
11071 * optabs-libfuncs.c (init_optabs): Likewise.
11072 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11073 * tree.c (vector_type_mode): Likewise.
11074 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
11075 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
11076 * tree-vect-generic.c (expand_vector_parallel): Likewise.
11077 * tree-vect-stmts.c (vectorizable_load): Likewise.
11078 (vectorizable_store): Likewise.
11080 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11081 Alan Hayward <alan.hayward@arm.com>
11082 David Sherwood <david.sherwood@arm.com>
11084 * coretypes.h (pod_mode): New type.
11085 (scalar_int_mode_pod): New typedef.
11086 * machmode.h (pod_mode): New class.
11087 (int_n_data_t::m): Change type to scalar_int_mode_pod.
11088 * genmodes.c (emit_mode_int_n): Update accordingly.
11089 * lower-subreg.h (target_lower_subreg): Change type to
11090 scalar_int_mode_pod.
11091 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
11092 scalar_int_mode_pod.
11094 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11095 Alan Hayward <alan.hayward@arm.com>
11096 David Sherwood <david.sherwood@arm.com>
11098 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
11099 machine_mode to scalar_int_mode.
11100 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
11101 (rs6000_option_override_internal): Remove cast to int.
11102 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
11103 machine_mode to scalar_int_mode.
11104 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
11105 (rs6000_option_override_internal): Remove cast to int.
11106 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
11107 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
11109 * config/s390/s390.c (s390_expand_builtin): Likewise.
11110 * coretypes.h (scalar_int_mode): New type.
11111 (opt_scalar_int_mode): New typedef.
11112 * machmode.h (scalar_int_mode): New class.
11113 (scalar_int_mode::includes_p): New function.
11114 (byte_mode): Change type to scalar_int_mode.
11115 (word_mode): Likewise.
11116 (ptr_mode): Likewise.
11117 * emit-rtl.c (byte_mode): Likewise.
11118 (word_mode): Likewise.
11119 (ptr_mode): Likewise.
11120 (init_derived_machine_modes): Update accordingly.
11121 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
11122 and MODE_PARTIAL_INT.
11123 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
11124 opt_scalar_int_mode.
11126 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11127 Alan Hayward <alan.hayward@arm.com>
11128 David Sherwood <david.sherwood@arm.com>
11130 * target.def (libgcc_floating_mode_supported_p): Take a
11132 * doc/tm.texi: Regenerate.
11133 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
11135 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
11136 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
11139 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11140 Alan Hayward <alan.hayward@arm.com>
11141 David Sherwood <david.sherwood@arm.com>
11143 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
11144 * doc/tm.texi: Regenerate.
11145 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
11146 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
11147 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
11148 * targhooks.h (default_floatn_mode): Likewise.
11149 * targhooks.c (default_floatn_mode): Likewise.
11150 * tree.c (build_common_tree_nodes): Update accordingly.
11152 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11153 Alan Hayward <alan.hayward@arm.com>
11154 David Sherwood <david.sherwood@arm.com>
11156 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
11157 (mode_iterator::iterate_p): Likewise.
11158 (mode_iterator::get_wider): Likewise.
11159 * expr.c (init_expr_target): Use opt_scalar_float_mode.
11161 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11162 Alan Hayward <alan.hayward@arm.com>
11163 David Sherwood <david.sherwood@arm.com>
11165 * coretypes.h (opt_scalar_float_mode): New typedef.
11166 * machmode.h (float_mode_for_size): New function.
11167 * emit-rtl.c (double_mode): Delete.
11168 (init_emit_once): Use float_mode_for_size.
11169 * stor-layout.c (layout_type): Likewise.
11170 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
11172 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11173 Alan Hayward <alan.hayward@arm.com>
11174 David Sherwood <david.sherwood@arm.com>
11176 * output.h (assemble_real): Take a scalar_float_mode.
11177 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
11178 * config/arm/arm.md (consttable_4): Likewise.
11179 (consttable_8): Likewise.
11180 (consttable_16): Likewise.
11181 * config/mips/mips.md (consttable_float): Likewise.
11182 * config/s390/s390.c (s390_output_pool_entry): Likewise.
11183 * varasm.c (assemble_real): Take a scalar_float_mode.
11184 (output_constant_pool_2): Update accordingly.
11185 (output_constant): Likewise.
11187 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11188 Alan Hayward <alan.hayward@arm.com>
11189 David Sherwood <david.sherwood@arm.com>
11191 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
11192 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
11193 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
11194 (native_encode_real): Likewise.
11195 (native_interpret_real): Likewise.
11196 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
11197 * tree-vrp.c (simplify_float_conversion_using_ranges): 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 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
11204 <scalar_float_mode>. Simplify.
11205 (gen_extend_conv_libfunc): Likewise.
11207 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11208 Alan Hayward <alan.hayward@arm.com>
11209 David Sherwood <david.sherwood@arm.com>
11211 * coretypes.h (scalar_float_mode): New type.
11212 * machmode.h (mode_traits::from_int): Use machine_mode if
11213 USE_ENUM_MODES is defined.
11214 (is_a): New function.
11216 (dyn_cast): Likewise.
11217 (scalar_float_mode): New class.
11218 (scalar_float_mode::includes_p): New function.
11219 (is_float_mode): Likewise.
11220 * gdbhooks.py (MachineModePrinter): New class.
11221 (build_pretty_printer): Use it for scalar_float_mode.
11222 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
11223 (format_helper::format_helper): Turn into a template.
11224 * genmodes.c (get_mode_class): New function.
11225 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
11226 or machine_mode if none.
11227 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
11228 as_a <scalar_float_mode>.
11229 * dwarf2out.c (mem_loc_descriptor): Likewise.
11230 (insert_float): Likewise.
11231 (add_const_value_attribute): Likewise.
11232 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11233 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
11234 (expand_unop): Update accordingly.
11235 (expand_abs_nojump): Likewise.
11236 (expand_copysign_absneg): Take a scalar_float_mode.
11237 (expand_copysign_bit): Likewise.
11238 (expand_copysign): Update accordingly.
11240 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11241 Alan Hayward <alan.hayward@arm.com>
11242 David Sherwood <david.sherwood@arm.com>
11244 * coretypes.h (opt_mode): New class.
11245 * machmode.h (opt_mode): Likewise.
11246 (opt_mode::else_void): New function.
11247 (opt_mode::require): Likewise.
11248 (opt_mode::exists): Likewise.
11249 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
11250 (GET_MODE_2XWIDER_MODE): Likewise.
11251 (mode_iterator::get_wider): Update accordingly.
11252 (mode_iterator::get_2xwider): Likewise.
11253 (mode_iterator::get_known_wider): Likewise, turning into a template.
11254 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
11255 forcing a wider mode to exist.
11256 * config/cr16/cr16.h (LONG_REG_P): Likewise.
11257 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11258 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
11259 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
11260 * lower-subreg.c (init_lower_subreg): Likewise.
11261 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
11262 on the final iteration.
11263 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
11264 a wider mode exists before asking for a move pattern.
11265 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
11266 forcing a wider mode to exist.
11267 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
11268 returning false if no such mode exists.
11269 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
11270 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
11271 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
11272 Avoid checking for a MODE_INT if we already know the mode is not a
11274 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
11275 forcing a wider mode to exist.
11276 (expmed_mult_highpart_optab): Likewise.
11277 (expmed_mult_highpart): Likewise.
11278 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
11280 * lto-streamer-in.c (lto_input_mode_table): Likewise.
11281 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
11282 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
11283 * internal-fn.c (expand_mul_overflow): Update use of
11284 GET_MODE_2XWIDER_MODE.
11285 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11286 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
11287 GET_MODE_WIDER_MODE.
11288 (convert_plusminus_to_widen): Likewise.
11289 * tree-switch-conversion.c (array_value_type): Likewise.
11290 * var-tracking.c (emit_note_insn_var_location): Likewise.
11291 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11292 Return false inside rather than outside the loop if no wider mode
11294 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
11295 and GET_MODE_2XWIDER_MODE
11296 (can_compare_p): Use else_void.
11297 * gdbhooks.py (OptMachineModePrinter): New class.
11298 (build_pretty_printer): Use it for opt_mode.
11300 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11301 Alan Hayward <alan.hayward@arm.com>
11302 David Sherwood <david.sherwood@arm.com>
11304 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
11305 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
11307 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11308 Alan Hayward <alan.hayward@arm.com>
11309 David Sherwood <david.sherwood@arm.com>
11311 * machmode.h (mode_traits): New structure.
11312 (get_narrowest_mode): New function.
11313 (mode_iterator::start): Likewise.
11314 (mode_iterator::iterate_p): Likewise.
11315 (mode_iterator::get_wider): Likewise.
11316 (mode_iterator::get_known_wider): Likewise.
11317 (mode_iterator::get_2xwider): Likewise.
11318 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
11319 (FOR_EACH_MODE): Likewise.
11320 (FOR_EACH_MODE_FROM): Likewise.
11321 (FOR_EACH_MODE_UNTIL): Likewise.
11322 (FOR_EACH_WIDER_MODE): Likewise.
11323 (FOR_EACH_2XWIDER_MODE): Likewise.
11324 * builtins.c (expand_builtin_strlen): Use new mode iterators.
11325 * combine.c (simplify_comparison): Likewise
11326 * config/i386/i386.c (type_natural_mode): Likewise.
11327 * cse.c (cse_insn): Likewise.
11328 * dse.c (find_shift_sequence): Likewise.
11329 * emit-rtl.c (init_derived_machine_modes): Likewise.
11330 (init_emit_once): Likewise.
11331 * explow.c (hard_function_value): Likewise.
11332 * expmed.c (extract_fixed_bit_field_1): Likewise.
11333 (extract_bit_field_1): Likewise.
11334 (expand_divmod): Likewise.
11335 (emit_store_flag_1): Likewise.
11336 * expr.c (init_expr_target): Likewise.
11337 (convert_move): Likewise.
11338 (alignment_for_piecewise_move): Likewise.
11339 (widest_int_mode_for_size): Likewise.
11340 (emit_block_move_via_movmem): Likewise.
11341 (copy_blkmode_to_reg): Likewise.
11342 (set_storage_via_setmem): Likewise.
11343 (compress_float_constant): Likewise.
11344 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11345 * optabs-query.c (get_best_extraction_insn): Likewise.
11346 * optabs.c (expand_binop): Likewise.
11347 (expand_twoval_unop): Likewise.
11348 (expand_twoval_binop): Likewise.
11349 (widen_leading): Likewise.
11350 (widen_bswap): Likewise.
11351 (expand_parity): Likewise.
11352 (expand_unop): Likewise.
11353 (prepare_cmp_insn): Likewise.
11354 (prepare_float_lib_cmp): Likewise.
11355 (expand_float): Likewise.
11356 (expand_fix): Likewise.
11357 (expand_sfix_optab): Likewise.
11358 * postreload.c (move2add_use_add2_insn): Likewise.
11359 * reg-stack.c (reg_to_stack): Likewise.
11360 * reginfo.c (choose_hard_reg_mode): Likewise.
11361 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11362 * stor-layout.c (mode_for_size): Likewise.
11363 (smallest_mode_for_size): Likewise.
11364 (mode_for_vector): Likewise.
11365 (finish_bitfield_representative): Likewise.
11366 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
11367 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
11368 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11369 * var-tracking.c (prepare_call_arguments): Likewise.
11371 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11372 Alan Hayward <alan.hayward@arm.com>
11373 David Sherwood <david.sherwood@arm.com>
11375 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
11376 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
11377 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
11378 * machmode.h (mode_size): Move earlier in file.
11379 (mode_precision): Likewise.
11380 (mode_inner): Likewise.
11381 (mode_nunits): Likewise.
11382 (mode_unit_size): Likewise.
11383 (unit_unit_precision): Likewise.
11384 (mode_wider): Likewise.
11385 (mode_2xwider): Likewise.
11386 (machine_mode): New class.
11387 (mode_to_bytes): New function.
11388 (mode_to_bits): Likewise.
11389 (mode_to_precision): Likewise.
11390 (mode_to_inner): Likewise.
11391 (mode_to_unit_size): Likewise.
11392 (mode_to_unit_precision): Likewise.
11393 (mode_to_nunits): Likewise.
11394 (GET_MODE_SIZE): Use mode_to_bytes.
11395 (GET_MODE_BITSIZE): Use mode_to_bits.
11396 (GET_MODE_PRECISION): Use mode_to_precision.
11397 (GET_MODE_INNER): Use mode_to_inner.
11398 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
11399 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
11400 (GET_MODE_NUNITS): Use mode_to_nunits.
11401 * system.h (ALWAYS_INLINE): New macro.
11402 * config/powerpcspe/powerpcspe-c.c
11403 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
11404 int for arg1_mode and arg2_mode.
11406 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11407 Alan Hayward <alan.hayward@arm.com>
11408 David Sherwood <david.sherwood@arm.com>
11410 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
11411 Prefix mode names with E_ in case statements.
11412 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11413 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
11414 (aarch64_split_simd_move): Likewise.
11415 (aarch64_gen_storewb_pair): Likewise.
11416 (aarch64_gen_loadwb_pair): Likewise.
11417 (aarch64_gen_store_pair): Likewise.
11418 (aarch64_gen_load_pair): Likewise.
11419 (aarch64_get_condition_code_1): Likewise.
11420 (aarch64_constant_pool_reload_icode): Likewise.
11421 (get_rsqrte_type): Likewise.
11422 (get_rsqrts_type): Likewise.
11423 (get_recpe_type): Likewise.
11424 (get_recps_type): Likewise.
11425 (aarch64_gimplify_va_arg_expr): Likewise.
11426 (aarch64_simd_container_mode): Likewise.
11427 (aarch64_emit_load_exclusive): Likewise.
11428 (aarch64_emit_store_exclusive): Likewise.
11429 (aarch64_expand_compare_and_swap): Likewise.
11430 (aarch64_gen_atomic_cas): Likewise.
11431 (aarch64_emit_bic): Likewise.
11432 (aarch64_emit_atomic_swap): Likewise.
11433 (aarch64_emit_atomic_load_op): Likewise.
11434 (aarch64_evpc_trn): Likewise.
11435 (aarch64_evpc_uzp): Likewise.
11436 (aarch64_evpc_zip): Likewise.
11437 (aarch64_evpc_ext): Likewise.
11438 (aarch64_evpc_rev): Likewise.
11439 (aarch64_evpc_dup): Likewise.
11440 (aarch64_gen_ccmp_first): Likewise.
11441 (aarch64_gen_ccmp_next): Likewise.
11442 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11443 (alpha_emit_xfloating_libcall): Likewise.
11444 (emit_insxl): Likewise.
11445 (alpha_arg_type): Likewise.
11446 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
11447 (arc_preferred_simd_mode): Likewise.
11448 (arc_secondary_reload): Likewise.
11449 (get_arc_condition_code): Likewise.
11450 (arc_print_operand): Likewise.
11451 (arc_legitimate_constant_p): Likewise.
11452 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11453 * config/arc/arc.md (casesi_load): Likewise.
11454 (casesi_compact_jump): Likewise.
11455 * config/arc/predicates.md (proper_comparison_operator): Likewise.
11456 (cc_use_register): Likewise.
11457 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11458 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
11459 (arm_init_iwmmxt_builtins): Likewise.
11460 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
11461 (neon_expand_vector_init): Likewise.
11462 (arm_attr_length_move_neon): Likewise.
11463 (maybe_get_arm_condition_code): Likewise.
11464 (arm_emit_vector_const): Likewise.
11465 (arm_preferred_simd_mode): Likewise.
11466 (arm_output_iwmmxt_tinsr): Likewise.
11467 (thumb1_output_casesi): Likewise.
11468 (thumb2_output_casesi): Likewise.
11469 (arm_emit_load_exclusive): Likewise.
11470 (arm_emit_store_exclusive): Likewise.
11471 (arm_expand_compare_and_swap): Likewise.
11472 (arm_evpc_neon_vuzp): Likewise.
11473 (arm_evpc_neon_vzip): Likewise.
11474 (arm_evpc_neon_vrev): Likewise.
11475 (arm_evpc_neon_vtrn): Likewise.
11476 (arm_evpc_neon_vext): Likewise.
11477 (arm_validize_comparison): Likewise.
11478 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
11479 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
11480 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
11481 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
11482 (c6x_preferred_simd_mode): Likewise.
11483 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
11484 (epiphany_rtx_costs): Likewise.
11485 * config/epiphany/predicates.md (proper_comparison_operator):
11487 * config/frv/frv.c (condexec_memory_operand): Likewise.
11488 (frv_emit_move): Likewise.
11489 (output_move_single): Likewise.
11490 (output_condmove_single): Likewise.
11491 (frv_hard_regno_mode_ok): Likewise.
11492 (frv_matching_accg_mode): Likewise.
11493 * config/h8300/h8300.c (split_adds_subs): Likewise.
11494 (h8300_rtx_costs): Likewise.
11495 (h8300_print_operand): Likewise.
11496 (compute_mov_length): Likewise.
11497 (output_logical_op): Likewise.
11498 (compute_logical_op_length): Likewise.
11499 (compute_logical_op_cc): Likewise.
11500 (h8300_shift_needs_scratch_p): Likewise.
11501 (output_a_shift): Likewise.
11502 (compute_a_shift_length): Likewise.
11503 (compute_a_shift_cc): Likewise.
11504 (expand_a_rotate): Likewise.
11505 (output_a_rotate): Likewise.
11506 * config/i386/i386.c (classify_argument): Likewise.
11507 (function_arg_advance_32): Likewise.
11508 (function_arg_32): Likewise.
11509 (function_arg_64): Likewise.
11510 (function_value_64): Likewise.
11511 (ix86_gimplify_va_arg): Likewise.
11512 (ix86_legitimate_constant_p): Likewise.
11513 (put_condition_code): Likewise.
11514 (split_double_mode): Likewise.
11515 (ix86_avx256_split_vector_move_misalign): Likewise.
11516 (ix86_expand_vector_logical_operator): Likewise.
11517 (ix86_split_idivmod): Likewise.
11518 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
11519 (ix86_build_const_vector): Likewise.
11520 (ix86_build_signbit_mask): Likewise.
11521 (ix86_match_ccmode): Likewise.
11522 (ix86_cc_modes_compatible): Likewise.
11523 (ix86_expand_branch): Likewise.
11524 (ix86_expand_sse_cmp): Likewise.
11525 (ix86_expand_sse_movcc): Likewise.
11526 (ix86_expand_int_sse_cmp): Likewise.
11527 (ix86_expand_vec_perm_vpermi2): Likewise.
11528 (ix86_expand_vec_perm): Likewise.
11529 (ix86_expand_sse_unpack): Likewise.
11530 (ix86_expand_int_addcc): Likewise.
11531 (ix86_split_to_parts): Likewise.
11532 (ix86_vectorize_builtin_gather): Likewise.
11533 (ix86_vectorize_builtin_scatter): Likewise.
11534 (avx_vpermilp_parallel): Likewise.
11535 (inline_memory_move_cost): Likewise.
11536 (ix86_tieable_integer_mode_p): Likewise.
11537 (x86_maybe_negate_const_int): Likewise.
11538 (ix86_expand_vector_init_duplicate): Likewise.
11539 (ix86_expand_vector_init_one_nonzero): Likewise.
11540 (ix86_expand_vector_init_one_var): Likewise.
11541 (ix86_expand_vector_init_concat): Likewise.
11542 (ix86_expand_vector_init_interleave): Likewise.
11543 (ix86_expand_vector_init_general): Likewise.
11544 (ix86_expand_vector_set): Likewise.
11545 (ix86_expand_vector_extract): Likewise.
11546 (emit_reduc_half): Likewise.
11547 (ix86_emit_i387_round): Likewise.
11548 (ix86_mangle_type): Likewise.
11549 (ix86_expand_round_sse4): Likewise.
11550 (expand_vec_perm_blend): Likewise.
11551 (canonicalize_vector_int_perm): Likewise.
11552 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
11553 (expand_vec_perm_1): Likewise.
11554 (expand_vec_perm_interleave3): Likewise.
11555 (expand_vec_perm_even_odd_pack): Likewise.
11556 (expand_vec_perm_even_odd_1): Likewise.
11557 (expand_vec_perm_broadcast_1): Likewise.
11558 (ix86_vectorize_vec_perm_const_ok): Likewise.
11559 (ix86_expand_vecop_qihi): Likewise.
11560 (ix86_expand_mul_widen_hilo): Likewise.
11561 (ix86_expand_sse2_abs): Likewise.
11562 (ix86_expand_pextr): Likewise.
11563 (ix86_expand_pinsr): Likewise.
11564 (ix86_preferred_simd_mode): Likewise.
11565 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
11566 * config/i386/sse.md (*andnot<mode>3): Likewise.
11567 (<mask_codefor><code><mode>3<mask_name>): Likewise.
11568 (*<code><mode>3): Likewise.
11569 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
11570 (ia64_expand_atomic_op): Likewise.
11571 (ia64_arg_type): Likewise.
11572 (ia64_mode_to_int): Likewise.
11573 (ia64_scalar_mode_supported_p): Likewise.
11574 (ia64_vector_mode_supported_p): Likewise.
11575 (expand_vec_perm_broadcast): Likewise.
11576 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
11577 (iq2000_function_arg_advance): Likewise.
11578 (iq2000_function_arg): Likewise.
11579 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
11580 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
11581 (m68k_libcall_value): Likewise.
11582 (m68k_function_value): Likewise.
11583 (sched_attr_op_type): Likewise.
11584 * config/mcore/mcore.c (mcore_output_move): Likewise.
11585 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
11587 (microblaze_function_arg): Likewise.
11588 * config/mips/mips.c (mips16_build_call_stub): Likewise.
11589 (mips_print_operand): Likewise.
11590 (mips_mode_ok_for_mov_fmt_p): Likewise.
11591 (mips_vector_mode_supported_p): Likewise.
11592 (mips_preferred_simd_mode): Likewise.
11593 (mips_expand_vpc_loongson_even_odd): Likewise.
11594 (mips_expand_vec_unpack): Likewise.
11595 (mips_expand_vi_broadcast): Likewise.
11596 (mips_expand_vector_init): Likewise.
11597 (mips_expand_vec_reduc): Likewise.
11598 (mips_expand_msa_cmp): Likewise.
11599 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
11600 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
11601 (cc_flags_for_mode): Likewise.
11602 * config/msp430/msp430.c (msp430_print_operand): Likewise.
11603 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
11604 (nds32_output_casesi_pc_relative): Likewise.
11605 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11606 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
11607 (nvptx_gen_unpack): Likewise.
11608 (nvptx_gen_pack): Likewise.
11609 (nvptx_gen_shuffle): Likewise.
11610 (nvptx_gen_wcast): Likewise.
11611 (nvptx_preferred_simd_mode): Likewise.
11612 * config/pa/pa.c (pa_secondary_reload): Likewise.
11613 * config/pa/predicates.md (base14_operand): Likewise.
11614 * config/powerpcspe/powerpcspe-c.c
11615 (altivec_resolve_overloaded_builtin): Likewise.
11616 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
11618 (rs6000_preferred_simd_mode): Likewise.
11619 (output_vec_const_move): Likewise.
11620 (rs6000_expand_vector_extract): Likewise.
11621 (rs6000_split_vec_extract_var): Likewise.
11622 (reg_offset_addressing_ok_p): Likewise.
11623 (rs6000_legitimate_offset_address_p): Likewise.
11624 (rs6000_legitimize_address): Likewise.
11625 (rs6000_emit_set_const): Likewise.
11626 (rs6000_const_vec): Likewise.
11627 (rs6000_emit_move): Likewise.
11628 (spe_build_register_parallel): Likewise.
11629 (rs6000_darwin64_record_arg_recurse): Likewise.
11630 (swap_selector_for_mode): Likewise.
11631 (spe_init_builtins): Likewise.
11632 (paired_init_builtins): Likewise.
11633 (altivec_init_builtins): Likewise.
11634 (do_load_for_compare): Likewise.
11635 (rs6000_generate_compare): Likewise.
11636 (rs6000_expand_float128_convert): Likewise.
11637 (emit_load_locked): Likewise.
11638 (emit_store_conditional): Likewise.
11639 (rs6000_output_function_epilogue): Likewise.
11640 (rs6000_handle_altivec_attribute): Likewise.
11641 (rs6000_function_value): Likewise.
11642 (emit_fusion_gpr_load): Likewise.
11643 (emit_fusion_p9_load): Likewise.
11644 (emit_fusion_p9_store): Likewise.
11645 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
11646 (fusion_gpr_mem_load): Likewise.
11647 (fusion_addis_mem_combo_load): Likewise.
11648 (fusion_addis_mem_combo_store): Likewise.
11649 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
11650 (fusion_gpr_mem_load): Likewise.
11651 (fusion_addis_mem_combo_load): Likewise.
11652 (fusion_addis_mem_combo_store): Likewise.
11653 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11655 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
11656 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
11657 (rs6000_preferred_simd_mode): Likewise.
11658 (output_vec_const_move): Likewise.
11659 (rs6000_expand_vector_extract): Likewise.
11660 (rs6000_split_vec_extract_var): Likewise.
11661 (reg_offset_addressing_ok_p): Likewise.
11662 (rs6000_legitimate_offset_address_p): Likewise.
11663 (rs6000_legitimize_address): Likewise.
11664 (rs6000_emit_set_const): Likewise.
11665 (rs6000_const_vec): Likewise.
11666 (rs6000_emit_move): Likewise.
11667 (rs6000_darwin64_record_arg_recurse): Likewise.
11668 (swap_selector_for_mode): Likewise.
11669 (paired_init_builtins): Likewise.
11670 (altivec_init_builtins): Likewise.
11671 (rs6000_expand_float128_convert): Likewise.
11672 (emit_load_locked): Likewise.
11673 (emit_store_conditional): Likewise.
11674 (rs6000_output_function_epilogue): Likewise.
11675 (rs6000_handle_altivec_attribute): Likewise.
11676 (rs6000_function_value): Likewise.
11677 (emit_fusion_gpr_load): Likewise.
11678 (emit_fusion_p9_load): Likewise.
11679 (emit_fusion_p9_store): Likewise.
11680 * config/rx/rx.c (rx_gen_move_template): Likewise.
11681 (flags_from_mode): Likewise.
11682 * config/s390/predicates.md (s390_alc_comparison): Likewise.
11683 (s390_slb_comparison): Likewise.
11684 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
11685 (s390_vector_mode_supported_p): Likewise.
11686 (s390_cc_modes_compatible): Likewise.
11687 (s390_match_ccmode_set): Likewise.
11688 (s390_canonicalize_comparison): Likewise.
11689 (s390_emit_compare_and_swap): Likewise.
11690 (s390_branch_condition_mask): Likewise.
11691 (s390_rtx_costs): Likewise.
11692 (s390_secondary_reload): Likewise.
11693 (__SECONDARY_RELOAD_CASE): Likewise.
11694 (s390_expand_cs): Likewise.
11695 (s390_preferred_simd_mode): Likewise.
11696 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
11697 * config/sh/sh.c (sh_print_operand): Likewise.
11698 (dump_table): Likewise.
11699 (sh_secondary_reload): Likewise.
11700 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11701 * config/sh/sh.md (casesi_worker_1): Likewise.
11702 (casesi_worker_2): Likewise.
11703 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
11704 (fcc_comparison_operator): Likewise.
11705 * config/sparc/sparc.c (sparc_expand_move): Likewise.
11706 (emit_soft_tfmode_cvt): Likewise.
11707 (sparc_preferred_simd_mode): Likewise.
11708 (output_cbranch): Likewise.
11709 (sparc_print_operand): Likewise.
11710 (sparc_expand_vec_perm_bmask): Likewise.
11711 (vector_init_bshuffle): Likewise.
11712 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11713 (spu_vector_mode_supported_p): Likewise.
11714 (spu_expand_insv): Likewise.
11715 (spu_emit_branch_or_set): Likewise.
11716 (spu_handle_vector_attribute): Likewise.
11717 (spu_builtin_splats): Likewise.
11718 (spu_builtin_extract): Likewise.
11719 (spu_builtin_promote): Likewise.
11720 (spu_expand_sign_extend): Likewise.
11721 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11722 (tilegx_simd_int): Likewise.
11723 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
11724 (tilepro_simd_int): Likewise.
11725 * config/v850/v850.c (const_double_split): Likewise.
11726 (v850_print_operand): Likewise.
11727 (ep_memory_offset): Likewise.
11728 * config/vax/vax.c (vax_rtx_costs): Likewise.
11729 (vax_output_int_move): Likewise.
11730 (vax_output_int_add): Likewise.
11731 (vax_output_int_subtract): Likewise.
11732 * config/visium/predicates.md (visium_branch_operator): Likewise.
11733 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
11734 (visium_print_operand_address): Likewise.
11735 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11736 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
11737 (xtensa_expand_conditional_branch): Likewise.
11738 (xtensa_copy_incoming_a7): Likewise.
11739 (xtensa_output_literal): Likewise.
11740 * dfp.c (decimal_real_maxval): Likewise.
11741 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
11743 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11744 Alan Hayward <alan.hayward@arm.com>
11745 David Sherwood <david.sherwood@arm.com>
11747 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
11748 (mode_nunits_inline): Likewise.
11749 (mode_inner_inline): Likewise.
11750 (mode_unit_size_inline): Likewise.
11751 (mode_unit_precision_inline): Likewise.
11752 (emit_insn_modes_h): Likewise. Also emit a #define of the
11754 (emit_mode_wider): Add an E_ prefix to mode names.
11755 (emit_mode_complex): Likewise.
11756 (emit_mode_inner): Likewise.
11757 (emit_mode_adjustments): Likewise.
11758 (emit_mode_int_n): Likewise.
11759 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
11760 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
11761 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
11762 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
11763 (CRC32_BUILTIN, ENTRY): Likewise.
11764 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
11765 (aarch64_pop_regs): Likewise.
11766 (aarch64_process_components): Likewise.
11767 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
11768 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
11769 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
11770 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
11771 * config/arm/arm.c (arm_init_libfuncs): Likewise.
11772 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
11774 * config/i386/i386-builtin.def (pcmpestr): Likewise.
11775 (pcmpistr): Likewise.
11776 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
11777 * config/mmix/mmix.c (mmix_output_condition): Likewise.
11778 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
11780 * config/rl78/rl78.c (mduc_regs): Likewise.
11781 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
11782 (htm_expand_builtin): Likewise.
11783 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
11784 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
11785 * config/xtensa/xtensa.c (print_operand): Likewise.
11786 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
11787 (NUM_MODE_VECTOR_INT): Likewise.
11788 * genoutput.c (null_operand): Likewise.
11789 (output_operand_data): Likewise.
11790 * genrecog.c (print_parameter_value): Likewise.
11791 * lra.c (debug_operand_data): Likewise.
11793 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11795 * dwarf2out.c (scompare_loc_descriptor_wide)
11796 (scompare_loc_descriptor_narrow): New functions, split out from...
11797 (scompare_loc_descriptor): ...here.
11798 * expmed.c (emit_store_flag_int): New function, split out from...
11799 (emit_store_flag): ...here.
11801 2017-08-30 Richard Biener <rguenther@suse.de>
11803 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
11804 (dwarf2out_early_finish): Move setting of AT_pubnames from
11805 early debug output to early finish.
11807 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11809 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
11810 and -mdata-region to the assembler.
11812 2017-08-30 Richard Biener <rguenther@suse.de>
11814 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
11816 (gen_subprogram_die): Add DW_AT_object_pointer only early.
11817 (dwarf2out_early_global_decl): Only generate a DIE for the
11818 abstract origin if it doesn't already exist or is a declaration DIE.
11819 (resolve_addr): Do not add the linkage name twice when
11820 generating a stub DIE for the DW_TAG_GNU_call_site target.
11822 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11824 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11825 Use machine_mode rather than int for arg1_mode.
11827 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
11830 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
11831 that the second argument of the built-in functions to unpack
11832 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
11833 switch statement instead a lot of if statements.
11834 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
11835 Allow 64-bit values to be in Altivec registers as well as
11836 traditional floating point registers.
11837 (pack<mode>, FMOVE128_VSX iterator): Likewise.
11839 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
11841 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
11842 MAX_REGS_PER_ADDRESS == 1.
11844 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
11846 * config/i386/i386.opt (flag_fentry): Do not init to -1.
11847 * config/i386/i386.c (ix86_option_override_internal): Simplify
11848 setting of opts->x_flag_entry.
11850 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11851 Jakub Jelinek <jakub@redhat.com>
11852 Richard Biener <rguenther@suse.de>
11854 PR tree-optimization/81503
11855 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
11856 folded constant fits in the target type; reorder tests for clarity.
11858 2017-08-29 Martin Liska <mliska@suse.cz>
11860 * passes.def: Include pass_lower_switch.
11861 * stmt.c (dump_case_nodes): Remove and move to
11862 tree-switch-conversion.
11863 (case_values_threshold): Likewise.
11864 (expand_switch_as_decision_tree_p): Likewise.
11865 (emit_case_decision_tree): Likewise.
11866 (expand_case): Likewise.
11867 (balance_case_nodes): Likewise.
11868 (node_has_low_bound): Likewise.
11869 (node_has_high_bound): Likewise.
11870 (node_is_bounded): Likewise.
11871 (emit_case_nodes): Likewise.
11872 (struct simple_case_node): New struct.
11873 (add_case_node): Remove.
11874 (emit_case_dispatch_table): Use vector instead of case_list.
11875 (reset_out_edges_aux): Remove.
11876 (compute_cases_per_edge): Likewise.
11877 (expand_case): Build list of simple_case_node.
11878 (expand_sjlj_dispatch_table): Use it.
11879 * tree-switch-conversion.c (struct case_node): Moved from
11880 stmt.c and adjusted.
11881 (emit_case_nodes): Likewise.
11882 (node_has_low_bound): Likewise.
11883 (node_has_high_bound): Likewise.
11884 (node_is_bounded): Likewise.
11885 (case_values_threshold): Likewise.
11886 (reset_out_edges_aux): Likewise.
11887 (compute_cases_per_edge): Likewise.
11888 (add_case_node): Likewise.
11889 (dump_case_nodes): Likewise.
11890 (balance_case_nodes): Likewise.
11891 (expand_switch_as_decision_tree_p): Likewise.
11892 (emit_jump): Likewise.
11893 (emit_case_decision_tree): Likewise.
11894 (try_switch_expansion): Likewise.
11895 (do_jump_if_equal): Likewise.
11896 (emit_cmp_and_jump_insns): Likewise.
11897 (fix_phi_operands_for_edge): New function.
11898 (record_phi_operand_mapping): Likewise.
11899 (class pass_lower_switch): New pass.
11900 (pass_lower_switch::execute): New function.
11901 (make_pass_lower_switch): Likewise.
11902 (conditional_probability):
11903 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
11904 * tree-pass.h: Add make_pass_lower_switch.
11906 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11909 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
11912 2017-08-29 Richard Biener <rguenther@suse.de>
11914 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
11915 we do not add a DW_AT_inline attribute twice.
11916 (gen_subprogram_die): Remove code setting DW_AT_inline on
11917 DECL_ABSTRACT_P nodes.
11919 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
11921 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
11922 calls to internal functions.
11923 (gimplify_modify_expr): Likewise.
11924 * tree-call-cdce.c (use_internal_fn): Likewise.
11925 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
11926 (convert_to_divmod): Set the nothrow flag.
11927 * tree-if-conv.c (predicate_mem_writes): Likewise.
11928 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
11929 (vectorizable_call): Likewise.
11930 (vectorizable_store): Likewise.
11931 (vectorizable_load): Likewise.
11932 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
11933 (vect_recog_mask_conversion_pattern): Likewise.
11935 2017-08-29 Martin Liska <mliska@suse.cz>
11938 * gcc.c (driver_handle_option): Add new argument.
11939 * opts-common.c (handle_option): Pass
11940 target_option_override_hook.
11941 * opts-global.c (lang_handle_option): Add new option.
11942 (set_default_handlers): Add new argument.
11943 (decode_options): Likewise.
11944 * opts.c (target_handle_option): Likewise.
11945 (common_handle_option): Call target_option_override_hook.
11946 * opts.h (struct cl_option_handler_func): Add hook for
11947 target option override.
11948 (struct cl_option_handlers): Likewise.
11949 (set_default_handlers): Add new argument.
11950 (decode_options): Likewise.
11951 (common_handle_option): Likewise.
11952 (target_handle_option): Likewise.
11953 * toplev.c (toplev::main): Pass targetm.target_option.override
11956 2017-08-29 Richard Biener <rguenther@suse.de>
11957 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
11959 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
11960 life to the active subtree.
11962 2017-08-28 Jeff Law <law@redhat.com>
11964 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
11965 derive_equivalences.
11966 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
11967 Code moved into....
11968 (edge_info::derive_equivalences): New private member function
11970 * tree-ssa-dom.c (class edge_info): Changed from a struct
11971 to a class. Add ctor/dtor, methods and data members.
11972 (edge_info::edge_info): Renamed from allocate_edge_info.
11973 Initialize additional members.
11974 (edge_info::~edge_info): New.
11975 (free_dom_edge_info): Delete the edge info.
11976 (record_edge_info): Use new class & associated member functions.
11977 Tighten forms for testing for edge equivalences.
11978 (record_temporary_equivalences): Iterate over the simple
11979 equivalences rather than assuming there's only one per edge.
11980 (cprop_into_successor_phis): Iterate over the simple
11981 equivalences rather than assuming there's only one per edge.
11982 (optimize_stmt): Use operand_equal_p rather than pointer
11983 equality for mini-DSE code.
11985 2017-08-28 Nathan Sidwell <nathan@acm.org>
11987 * gcc.c (execute): Fold SIGPIPE handling into switch
11988 statement. Adjust internal error message.
11990 2017-08-28 Richard Biener <rguenther@suse.de>
11993 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
11994 Do nothing for removed DIEs.
11996 2017-08-28 Richard Biener <rguenther@suse.de>
11998 PR tree-optimization/81977
11999 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
12002 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
12005 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
12006 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
12007 using targetm.gen_mem_thread_fence.
12009 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
12012 * config/i386/i386.md (*<btsc><mode>): Change operand 2
12013 predicate to register_operand. Reorder operands.
12014 (*btr<mode>): Ditto.
12015 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
12016 (*btr<mode>_mask): Ditto.
12018 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
12020 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
12021 * config/rs6000/xmmintrin.h: New file.
12022 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
12024 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12027 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
12028 parameter and_insn and return it.
12029 (recombine_lvx_pattern): Insert a copy to ensure availability of
12030 the base register of the copied masking operation at the point of
12031 the instruction replacement.
12032 (recombine_stvx_pattern): Likewise.
12034 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
12036 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
12037 undocumented switches.
12038 (-mpower9-dform-vector): Likewise.
12039 (-mpower9-dform): Likewise.
12040 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
12041 comments to delete references to -mpower9-dform* switches.
12042 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
12043 Delete reference to -mpower9-dform* switches, test for
12044 -mpower9-vector instead.
12045 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
12046 (OTHER_P9_VECTOR_MASKS): Likewise.
12047 (POWERPC_MASKS): Likewise.
12048 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
12049 tests against -mpower9-dform* to -mpower9-vector. Delete code
12050 that checked for -mpower9-dform* consistancy with other options.
12051 Add test for -mpower9-misc to enable other power9 switches.
12052 (rs6000_init_hard_regno_mode_ok): Likewise.
12053 (rs6000_option_override_internal): Likewise.
12054 (rs6000_emit_prologue): Likewise.
12055 (rs6000_emit_epilogue): Likewise.
12056 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
12057 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
12058 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
12060 (emit_fusion_p9_store): Likewise.
12061 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
12062 resetting these macros if the assembler does not support ISA 3.0
12064 (TARGET_P9_DFORM_VECTOR): Likewise.
12065 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
12066 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
12068 2017-08-25 Alan Modra <amodra@gmail.com>
12070 PR rtl-optimization/81747
12071 * cse.c (cse_extended_basic_block): Don't attempt to record
12072 equivalences for degenerate conditional branches that branch
12073 to their fall-through.
12075 2017-08-24 Martin Sebor <msebor@redhat.com>
12077 PR middle-end/81908
12078 * gimple-fold.c (size_must_be_zero_p): New function.
12079 (gimple_fold_builtin_memory_op): Call it.
12081 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
12083 * config/rs6000/mm_malloc.h: New file.
12085 2017-08-24 Bin Cheng <bin.cheng@arm.com>
12087 PR tree-optimization/81913
12088 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
12089 analysis when either IVs in condition can wrap.
12091 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
12093 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
12094 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
12096 2017-08-24 Richard Biener <rguenther@suse.de>
12099 * targhooks.c (default_target_can_inline_p): Properly
12100 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
12101 is present and always compare.
12102 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
12103 infer -mfpmath=sse from TARGET_SSE_P.
12104 (ix86_can_inline_p): Properly use target_option_default_node when
12105 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
12107 2017-08-24 Richard Biener <rguenther@suse.de>
12110 * dwarf2out.c (output_die): Handle flag_generate_offload like
12112 (output_comp_unit): Likewise.
12113 (gen_array_type_die): Likewise.
12114 (dwarf2out_early_finish): Likewise.
12115 (note_variable_value_in_expr): Likewise.
12116 (dwarf2out_finish): Likewise. Adjust assert.
12117 * cgraphunit.c (symbol_table::compile): Move setting of
12118 flag_generate_offload earlier ...
12119 (symbol_table::finalize_compilation_unit): ... here, before
12120 early debug finalization.
12122 2017-08-24 Richard Biener <rguenther@suse.de>
12124 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
12125 and ipa-fnsummary.h.
12126 (ix86_can_inline_p): When ix86_fpmath flags do not match
12127 check whether the callee uses FP math at all.
12129 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
12131 PR middle-end/81931
12132 * tree-ssanames.c (get_nonzero_bits): Use element_precision
12133 instead of TYPE_PRECISION.
12135 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
12136 Alan Hayward <alan.hayward@arm.com>
12137 David Sherwood <david.sherwood@arm.com>
12139 * combine.c (make_extraction): Use subreg_offset_from_lsb.
12141 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
12143 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
12145 (ix86_frame::stack_realign_allocate): New field.
12146 (struct machine_frame_state): Modify comments.
12147 (machine_frame_state::sp_realigned_fp_end): New field.
12148 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
12149 layout calculation.
12150 (sp_valid_at): Add assertion to assure no attempt to access invalid
12151 offset of a realigned stack.
12152 (fp_valid_at): Likewise.
12153 (choose_baseaddr): Modify comments.
12154 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
12155 ix86_expand_prologue.
12156 (ix86_expand_prologue): Modify stack realignment and allocation.
12157 (ix86_expand_epilogue): Modify comments.
12158 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
12159 avx2_runtime, avx512f, and avx512f_runtime.
12161 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
12163 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
12164 (mstackrealign): Do not init to -1.
12165 * config/i386/i386.c (ix86_option_override_internal):
12166 Check opts_set, not opts when setting default value of
12167 opts->x_ix86_force_align_arg_pointer.
12169 2017-08-23 Richard Biener <rguenther@suse.de>
12171 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
12172 lang_hooks.decl_printable_name.
12173 * print-rtl-function.c (print_rtx_function): Likewise.
12174 * tree-pretty-print.c (dump_function_header): Likewise.
12176 2017-08-23 Richard Biener <rguenther@suse.de>
12179 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
12180 -g0 at compile-time.
12182 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12184 PR middle-end/19706
12185 * doc/sourcebuild.texi (Other hardware attributes):
12188 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12190 PR middle-end/19706
12191 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
12192 Fix single-use check.
12194 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
12196 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
12198 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12200 * doc/install.texi: Modify to add more details on running selected
12203 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12205 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
12206 is combined with -mabi=ms.
12207 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
12210 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12212 PR tree-optimization/81488
12213 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
12214 and cached_basis fields.
12215 (MAX_SPREAD): New constant.
12216 (alloc_cand_and_find_basis): Initialize new fields.
12217 (clear_visited): New function.
12218 (create_phi_basis_1): Rename from create_phi_basis, set visited
12219 and cached_basis fields.
12220 (create_phi_basis): New wrapper function.
12221 (phi_add_costs_1): Rename from phi_add_costs, add spread
12222 parameter, set visited field, short-circuit when limits reached.
12223 (phi_add_costs): New wrapper function.
12224 (record_phi_increments_1): Rename from record_phi_increments, set
12226 (record_phi_increments): New wrapper function.
12227 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
12228 (phi_incr_cost): New wrapper function.
12229 (all_phi_incrs_profitable_1): Rename from
12230 all_phi_incrs_profitable, set visited field.
12231 (all_phi_incrs_profitable): New wrapper function.
12233 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12234 Alan Hayward <alan.hayward@arm.com>
12235 David Sherwood <david.sherwood@arm.com>
12237 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
12238 that takes the outer and inner modes.
12239 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
12240 comparison as the canonical test for a paradoxical subreg.
12241 * combine.c (simplify_set): Use paradoxical_subreg_p.
12242 (make_extraction): Likewise.
12243 (force_to_mode): Likewise.
12244 (rtx_equal_for_field_assignment_p): Likewise.
12245 (gen_lowpart_for_combine): Likewise.
12246 (simplify_comparison): Likewise.
12247 * cse.c (equiv_constant): Likewise.
12248 * expmed.c (store_bit_field_1): Likewise.
12249 * final.c (alter_subreg): Likewise.
12250 * fwprop.c (propagate_rtx): Likewise.
12251 (forward_propagate_subreg): Likewise.
12252 * ira-conflicts.c (ira_build_conflicts): Likewise.
12253 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
12254 * lra-constraints.c (curr_insn_transform): Likewise.
12255 (split_reg): Likewise.
12256 * lra-eliminations.c (move_plus_up): Likewise.
12257 (lra_eliminate_regs_1): Likewise.
12258 * recog.c (general_operand): Likewise.
12259 * ree.c (combine_reaching_defs): Likewise.
12260 * reload.c (push_reload): Likewise.
12261 (find_reloads): Likewise.
12262 * reload1.c (elimination_effects): Likewise.
12263 (compute_reload_subreg_offset): Likewise.
12264 (choose_reload_regs): Likewise.
12265 * rtlanal.c (subreg_lsb_1): Likewise.
12266 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12267 (simplify_subreg): Likewise.
12268 * var-tracking.c (track_loc_p): Likewise.
12269 * emit-rtl.c (byte_lowpart_offset): Likewise.
12270 (paradoxical_subreg_p): Delete out-of-line definition.
12272 2017-08-22 Jeff Law <law@redhat.com>
12274 PR tree-optimization/81741
12275 PR tree-optimization/71947
12276 * tree-ssa-dom.c: Include tree-inline.h.
12277 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
12278 equivalences if one is more expensive to compute than the other.
12279 * tree-ssa-scopedtables.h (class const_or_copies): Make
12280 record_const_or_copy_raw method private.
12281 (class avail_exprs_stack): New method simplify_binary_operation.
12282 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
12283 avail_exprs_stack::simplify_binary_operation as needed.
12284 (avail_exprs_stack::simplify_binary_operation): New function.
12286 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
12288 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
12289 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
12290 (DOT_SYMBOLS): Likewise.
12291 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
12292 (RELOCATABLE_NEEDS_FIXUP): Likewise.
12293 (RS6000_ABI_NAME): Likewise.
12294 (TARGET_CMODEL): Likewise.
12295 (TOC_SECTION_ASM_OP): Likewise.
12296 (SET_CMODEL): New macro.
12297 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
12299 2017-08-22 Richard Biener <rguenther@suse.de>
12301 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
12302 to allow for free-lang-data replacements similar to verify_type_variant.
12304 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12305 Alan Hayward <alan.hayward@arm.com>
12306 David Sherwood <david.sherwood@arm.com>
12308 * config/aarch64/aarch64.md (casesi): Use DImode rather than
12309 VOIDmode for the LABEL_REF.
12311 2017-08-22 Richard Biener <rguenther@suse.de>
12313 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
12315 2017-08-22 Richard Biener <rguenther@suse.de>
12317 * common.opt (feliminate-dwarf2-dups): Ignore.
12318 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
12319 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
12320 same_die_p_wrap, compute_section_prefix,
12321 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
12322 (comdat_symbol_id, comdat_symbol_number): Likewise.
12323 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
12325 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
12326 (output_die): Mark unreachable path unreachable.
12327 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
12328 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
12329 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
12330 (dwarf2out_early_finish): Likewise.
12332 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
12334 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
12336 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
12339 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
12340 not VAR_P. Filter attribute warnings with OPT_Wattributes.
12341 (avr_attribute_table) <io, io_low, address>: Initialize
12342 .decl_required with true.
12344 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
12346 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
12347 undocumented debugging options.
12348 (-mvsx-scalar-double): Likewise.
12349 (-mallow-df-permute): Likewise.
12350 (-mvectorize-builtins): Likewise.
12351 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12352 (rs6000_builtin_vectorized_function): Likewise.
12353 (rs6000_builtin_md_vectorized_function): Likewise.
12354 (rs6000_opt_vars): Likewise.
12356 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
12359 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
12360 (*btrq_imm): Rename from *btrq.
12361 (*btcq_imm): Rename from *btcq.
12362 (btsc): New code attribute.
12363 (*<btsc><mode>): New insn pattern.
12364 (*btr<mode>): Ditto.
12365 (*<btsc><mode>_mask): New insn_and_split pattern.
12366 (*btr<mode>_mask): Ditto.
12368 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12369 Alan Hayward <alan.hayward@arm.com>
12370 David Sherwood <david.sherwood@arm.com>
12372 * function.c (pad_below): Simplify padding calculation.
12374 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12375 Alan Hayward <alan.hayward@arm.com>
12376 David Sherwood <david.sherwood@arm.com>
12378 * target.def (function_prologue): Remove frame size argument.
12379 (function_epilogue): Likewise.
12380 * doc/tm.texi: Regenerate.
12381 * final.c (final_start_function): Update call to function_prologue.
12382 (final_end_function): Update call to function_epilogue.
12383 (default_function_pro_epilogue): Remove frame size argument.
12384 * output.h (default_function_pro_epilogue): Likewise.
12385 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
12386 (arm_output_function_prologue): Likewise.
12387 * config/frv/frv.c (frv_function_prologue): Likewise.
12388 (frv_function_epilogue): Likewise.
12389 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
12390 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
12391 (ia64_output_function_epilogue): Likewise.
12392 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
12393 (m32r_output_function_epilogue): Likewise.
12394 * config/microblaze/microblaze.c (microblaze_function_prologue)
12395 (microblaze_function_epilogue): Likewise.
12396 * config/mips/mips.c (mips_output_function_prologue): Likewise.
12397 (mips_output_function_epilogue): Likewise.
12398 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
12399 (mmix_target_asm_function_epilogue): Likewise.
12400 * config/msp430/msp430.c (msp430_start_function): Likewise.
12401 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
12402 (nds32_asm_function_epilogue): Likewise.
12403 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
12404 * config/pa/pa.c (pa_output_function_prologue): Likewise.
12405 (pa_output_function_epilogue): Likewise.
12406 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
12407 (rs6000_output_function_epilogue): Likewise.
12408 * config/rl78/rl78.c (rl78_start_function): Likewise.
12409 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
12410 (rs6000_output_function_epilogue): Likewise.
12411 * config/rx/rx.c (rx_output_function_prologue): Likewise.
12412 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
12413 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
12414 (sparc_asm_function_epilogue): Likewise.
12416 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12418 * tree.h (type_has_mode_precision_p): New function.
12419 * convert.c (convert_to_integer_1): Use it.
12420 * expr.c (expand_expr_real_2): Likewise.
12421 (expand_expr_real_1): Likewise.
12422 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
12423 * match.pd: Likewise.
12424 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
12425 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
12426 * tree-tailcall.c (process_assignment): Likewise.
12427 * tree-vect-loop.c (vectorizable_reduction): Likewise.
12428 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
12429 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
12430 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12431 (vectorizable_assignment): Likewise.
12432 (vectorizable_shift): Likewise.
12433 (vectorizable_operation): Likewise.
12434 * tree-vrp.c (register_edge_assert_for_2): Likewise.
12436 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
12438 * match.pd: Add pow (C, x) simplification.
12440 2017-08-21 Richard Biener <rguenther@suse.de>
12442 PR tree-optimization/81900
12443 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
12444 for blocks with abnormal predecessors.
12445 (compute_antic): Do not set visited flag prematurely.
12447 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
12450 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
12452 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12454 * stor-layout.h (vector_type_mode): Move to...
12455 * tree.h (vector_type_mode): ...here.
12456 * stor-layout.c (vector_type_mode): Move to...
12457 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
12459 2017-08-21 Richard Biener <rguenther@suse.de>
12461 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
12462 register_external_die hooks.
12463 (debug_false_tree_charstarstar_uhwistar): Declare.
12464 (debug_nothing_tree_charstar_uhwi): Likewise.
12465 * debug.c (do_nothing_debug_hooks): Adjust.
12466 (debug_false_tree_charstarstar_uhwistar): New do nothing.
12467 (debug_nothing_tree_charstar_uhwi): Likewise.
12468 * dbxout.c (dbx_debug_hooks): Adjust.
12469 (xcoff_debug_hooks): Likewise.
12470 * sdbout.c (sdb_debug_hooks): Likewise.
12471 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
12472 * dwarf2out.c (macinfo_label_base): New global.
12473 (dwarf2out_register_external_die): New function for the
12474 register_external_die hook.
12475 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
12476 (dwarf2_debug_hooks): Use them.
12477 (dwarf2_lineno_debug_hooks): Adjust.
12478 (struct die_struct): Add with_offset flag.
12479 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
12480 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
12481 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
12482 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
12483 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
12484 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
12485 defining section names for the early LTO debug variants.
12486 (reset_indirect_string): New helper.
12487 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
12488 (print_dw_val): Add support for offsetted symbol references.
12489 (get_ultimate_context): Split out from is_cxx.
12490 (is_cxx): Use get_ultimate_context.
12491 (is_fortran): Add decl overload.
12492 (compute_comp_unit_symbol): Split out worker from
12493 compute_section_prefix.
12494 (compute_section_prefix): Call compute_comp_unit_symbol and
12495 set comdat_type_p here.
12496 (output_die): Skip DIE symbol output for the LTO added one.
12497 Handle DIE symbol references with offset.
12498 (output_comp_unit): Guard section name mangling properly.
12499 For LTO debug sections emit a symbol at the section beginning
12500 which we use to refer to its DIEs.
12501 (add_abstract_origin_attribute): For DIEs registered via
12502 dwarf2out_register_external_die directly refer to the early
12503 DIE rather than indirectly through the shadow one we created.
12504 Remove obsolete call to dwarf2out_abstract_function for
12505 non-function/block origins.
12506 (gen_array_type_die): When generating early LTO debug do
12507 not emit DW_AT_string_length.
12508 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
12509 late when in LTO. As suggested place a gcc_unreachable for
12510 the DECL_ABSTRACT_P case.
12511 (gen_subprogram_die): Avoid another specification DIE
12512 for early built declarations/definitions for the late LTO case.
12513 (gen_variable_die): Add type references for late duplicated VLA dies
12515 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
12516 we have the abstract instance already.
12517 (process_scope_var): Adjust decl DIE contexts in LTO which
12518 first puts them in limbo.
12519 (gen_decl_die): Do not generate type DIEs late apart from
12520 types for VLAs or for decls we do not yet have a DIE. Do not
12521 call dwarf2out_abstract_function late.
12522 (dwarf2out_early_global_decl): Make sure to create DIEs
12523 for abstract instances of a decl first.
12524 (dwarf2out_late_global_decl): Adjust comment.
12525 (output_macinfo_op): With multiple macro sections use
12526 macinfo_label_base to distinguish labels.
12527 (output_macinfo): Likewise. Update macinfo_label_base.
12528 Pass in the line info label.
12529 (note_variable_value_in_expr): When generating LTO resolve
12530 all variable values here by generating DIEs as needed.
12531 (init_sections_and_labels): Add early LTO debug flag parameter
12532 and generate different sections and names if set. Add generation
12533 counter for the labels so we can have multiple of them.
12534 (reset_dies): Helper to allow DIEs to be output multiple times.
12535 (dwarf2out_finish): When outputting DIEs to the fat part of an
12536 LTO object first reset DIEs.
12537 (dwarf2out_early_finish): Output early DIEs when generating LTO.
12538 (modified_type_die): Check for decl_ultimate_origin being self
12540 (gen_type_die_with_usage): Likewise.
12541 (gen_typedef_die): Allow decl_ultimate_origin being self.
12542 (set_decl_abstract_flags): Remove.
12543 (set_block_abstract_flags): Likewise.
12544 (dwarf2out_abstract_function): Treat the early generated DIEs
12545 as the abstract copy and only add DW_AT_inline and
12546 DW_AT_artificial here and call set_decl_origin_self.
12547 If the DIE has an abstract origin don't do anything.
12548 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
12549 if we have none yet (Go fails to build one, PR78628).
12550 (variably_modified_type_p): Prevent endless recursion for Ada
12551 cyclic pointer types.
12552 * lto-streamer-in.c: Include debug.h.
12553 (dref_queue): New global.
12554 (lto_read_tree_1): Stream in DIE references.
12555 (lto_input_tree): Register DIE references.
12556 (input_function): Stream DECL_DEBUG_ARGS.
12557 * lto-streamer-out.c: Include debug.h.
12558 (lto_write_tree_1): Output DIE references.
12559 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
12560 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
12561 (output_function): Stream DECL_DEBUG_ARGS.
12562 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
12563 Stream DECL_ABSTRACT_ORIGIN.
12564 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
12565 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
12566 DECL_CONTEXT for file-scope decls.
12567 * lto-streamer.h (struct dref_entry): Declare.
12568 (dref_queue): Likewise.
12569 * cfgexpand.c (pass_expand::execute): Do not call the
12570 outlining_inline_function hook here.
12571 * lto-wrapper.c (debug_obj): New global.
12572 (tool_cleanup): Unlink it if required.
12573 (debug_objcopy): New function.
12574 (run_gcc): Handle early debug sections in the IL files by
12575 extracting them to separate files, partially linkin them and
12576 feeding the result back as result to the linker.
12577 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
12578 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
12579 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
12580 sections into a separate segment.
12581 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
12583 (darwin_asm_dwarf_section): Likewise.
12584 (darwin_asm_output_dwarf_offset): Likewise.
12585 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
12587 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12588 Alan Hayward <alan.hayward@arm.com>
12589 David Sherwood <david.sherwood@arm.com>
12591 * read-md.h (md_reader::record_potential_iterator_use): Replace
12592 pointer argument with an rtx and an index.
12593 * read-rtl.c (iterator_group::apply_iterator): Likewise.
12594 (apply_mode_iterator): Likewise.
12595 (apply_code_iterator): Likewise.
12596 (apply_int_iterator): Likewise.
12597 (apply_subst_iterator): Likewise.
12598 (record_iterator_use): Likewise.
12599 (record_attribute_use): Likewise.
12600 (md_reader::record_potential_iterator_use): Likewise. Update calls
12601 to record_iterator_use and apply_iterator.
12602 (iterator_use): Replace ptr with x and index.
12603 (attribute_use): Likewise.
12604 (apply_attribute_uses): Update calls to apply_iterator.
12605 (apply_iterators): Likewise. Update initialization of iterator_use.
12606 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
12607 and record_potential_iterator_use.
12608 (rtx_reader::read_rtx_operand): Likewise.
12610 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12611 Alan Hayward <alan.hayward@arm.com>
12612 David Sherwood <david.sherwood@arm.com>
12614 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
12617 2017-08-21 Richard Biener <rguenther@suse.de>
12619 PR middle-end/81884
12620 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
12621 at struct end conservatively when comparing common bases.
12623 2017-08-21 Richard Biener <rguenther@suse.de>
12625 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
12626 (mem_ref_in_stmt): Remove.
12627 (determine_max_movement): Use ref index to get at the reference.
12628 (invariantness_dom_walker::before_dom_children): Deal with
12629 lim data already initialized.
12630 (gather_mem_refs_stmt): Initialize lim data and record ref index.
12632 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12634 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
12635 (TARGET_ISA_ROUND): Ditto.
12636 (TARGET_ROUND): Ditto.
12637 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
12638 * config/i386/i386.md: Ditto.
12639 * config/i386/sse.md: Ditto.
12640 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
12641 with OPTION_MASK_ISA_SSE4_1.
12643 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12646 * doc/extend.texi (x86 Built-in Functions): Correct the name of
12647 __builtin_ia32_lzcnt_u16.
12649 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
12652 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
12653 (rs6000_set_current_function): Rewrite function to use it.
12655 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
12658 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
12659 and TYPE_WARN_IF_NOT_ALIGN.
12660 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
12661 (handle_warn_if_not_align): New.
12662 (place_union_field): Call handle_warn_if_not_align.
12663 (place_field): Call handle_warn_if_not_align.
12664 Copy TYPE_WARN_IF_NOT_ALIGN.
12665 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
12666 (layout_type): Likewise.
12667 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
12669 (tree_decl_common): Add warn_if_not_align.
12670 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
12671 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
12672 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
12673 (DECL_WARN_IF_NOT_ALIGN): Likewise.
12674 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
12675 * doc/extend.texi: Document warn_if_not_aligned attribute.
12676 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
12678 2017-08-17 Martin Liska <mliska@suse.cz>
12681 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
12682 (get_data_dependence): Use it as pointer type.
12683 (distribute_loop): Likewise.
12685 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12687 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
12688 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
12689 (p8_vmrgow_<mode>_direct): New define_insn.
12690 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
12691 handle endianness for vmrgew and vmrgow permute patterns.
12693 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
12695 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
12696 * config/rs6000/rs6000-cpus.def: Remove comment.
12697 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
12698 (POWERPC_MASKS): Likewise.
12699 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
12700 use of TARGET_VSX_TIMODE.
12701 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
12702 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
12703 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
12704 (rs6000_option_override_internal): Remove dead code.
12705 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
12706 (rs6000_legitimize_reload_address): Likewise.
12707 (rs6000_legitimate_address_p): Likewise.
12708 (rs6000_opt_masks): Delete "vsx-timode".
12709 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
12710 from function comment.
12711 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
12712 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
12713 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
12715 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
12716 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
12717 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
12718 (VSX_TI): Likewise.
12720 (define_peephole2): Likewise.
12722 2017-08-17 Martin Sebor <msebor@redhat.com>
12725 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
12726 past the end of an array.
12727 (test_pp_format): Add test cases.
12729 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
12731 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
12732 missing ECF_NOTHROW flags.
12734 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
12737 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
12738 operands residing in integer registers.
12739 (*vsx_le_perm_load_<mode>): Likewise.
12740 (*vsx_le_perm_store_<mode>): Likewise.
12741 (define_peephole2): Add peepholes to optimize the above.
12743 2017-08-17 Marek Polacek <polacek@redhat.com>
12745 PR middle-end/81814
12746 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
12747 to mimic what shorten_compare did. Change the return type to bool.
12748 (fold_cond_expr_with_comparison): Update call to
12749 operand_equal_for_comparison_p.
12750 (fold_ternary_loc): Likewise.
12752 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
12754 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
12756 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
12758 2017-08-17 Richard Biener <rguenther@suse.de>
12760 * tree-ssa-structalias.c (solve_graph): When propagating
12761 to successors update the graphs succ edges and avoid duplicate work.
12763 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
12766 * config/i386/i386.c (ix86_option_override_internal): Save target
12767 specific options after ix86_stack_protector_guard_reg was changed.
12769 2017-08-17 Richard Biener <rguenther@suse.de>
12771 PR tree-optimization/81827
12772 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
12773 (new_var_info): Initialize it conservatively.
12774 (get_call_vi): Mark register vars.
12775 (new_scalar_tmp_constraint_exp): Likewise.
12776 (handle_rhs_call): Likewise.
12777 (handle_const_call): Likewise.
12778 (create_function_info_for): Likewise.
12779 (solve_constraints): Sort varinfos to separate register from
12780 non-register vars to pack points-to solution bitmaps during
12783 2017-08-17 Marek Polacek <polacek@redhat.com>
12785 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
12787 2017-08-17 Richard Biener <rguenther@suse.de>
12789 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
12790 to true when overflow is undefined and we saturated the result.
12792 2017-08-17 Alan Modra <amodra@gmail.com>
12795 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
12796 Don't use store multiple if only one reg needs saving.
12797 (interesting_frame_related_regno): New function.
12798 (rs6000_frame_related): Don't emit frame info for regs that
12800 (rs6000_emit_epilogue): Likewise.
12802 2017-08-16 Nathan Sidwell <nathan@acm.org>
12804 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
12805 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
12806 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
12807 * tree.c (free_lang_data_in_type): Use else-if chain. Always
12808 clear TYPE_LANG_1. Remove obsolete member-function stripping.
12809 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
12810 (verify_type): Adjust for TYPE_BINFO move.
12811 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
12812 process TYPE_BINFO directly.
12813 (hash_tree): Likewise.
12814 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
12816 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
12819 2017-08-16 David Malcolm <dmalcolm@redhat.com>
12821 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
12823 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
12826 * config/i386/i386.md (*anddi_1_btr): Change predicates of
12827 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
12828 Add ix86_binary_operator_ok to insn constraint.
12829 (*iordi_1_bts): Ditto.
12830 (*xordi_1_btc): Ditto.
12831 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
12832 Update corresponding peephole2 pattern.
12836 2017-08-16 Bin Cheng <bin.cheng@arm.com>
12838 PR tree-optimization/81832
12839 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
12840 copy loop header which has IFN_LOOP_DIST_ALIAS call.
12842 2017-08-16 Marek Polacek <polacek@redhat.com>
12845 * fold-const.c (fold_indirect_ref_1): Restore original behavior
12846 regarding size_zero_node.
12848 2017-08-16 Martin Liska <mliska@suse.cz>
12851 * config.gcc: Respect previously set extra_objs in case
12854 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
12856 PR tree-optimization/81835
12857 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
12858 the phi SSA_NAME. Check that the condition in a COND_EXPR does
12859 not depend on the phi.
12861 2017-08-16 Alan Modra <amodra@gmail.com>
12863 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
12866 2017-08-16 Alan Modra <amodra@gmail.com>
12868 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
12869 (save_reg_p): ..into this. Update all callers.
12870 (first_reg_to_save): Simplify.
12872 2017-08-16 Alan Modra <amodra@gmail.com>
12874 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
12877 2017-08-15 Joseph Myers <joseph@codesourcery.com>
12881 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
12882 constant count if that count is less than 32.
12884 2017-08-15 Nathan Sidwell <nathan@acm.org>
12886 * gcc.c (execute): Emit friendlier message if inferior is killed
12887 by an external cause.
12889 2017-08-15 Richard Biener <rguenther@suse.de>
12891 PR tree-optimization/81790
12892 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
12893 CONSTRUCTORs from simplifying and VN.
12895 2017-08-14 Martin Sebor <msebor@redhat.com>
12897 * builtin-attrs.def: Add comments.
12899 2017-08-14 Martin Sebor <msebor@redhat.com>
12902 * doc/extend.texi (attribute nonstring): Document new attribute.
12904 2017-08-14 Martin Sebor <msebor@redhat.com>
12907 * tree-diagnostic.c (default_tree_printer): Handle %G.
12908 * gimple-pretty-print.h (percent_G_format): Declare new function.
12909 * gimple-pretty-print.c (percent_G_format): Define.
12910 * tree-pretty-print.c (percent_K_format): Add argument.
12912 2017-08-14 Martin Sebor <msebor@redhat.com>
12914 PR translation/79998
12915 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
12916 Remove a stray space.
12918 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
12921 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
12922 (*iordi_1_bts): Ditto.
12923 (*xordi_1_btc): Ditto.
12925 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12928 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
12929 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12931 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
12932 quoted strings, and make more translator-friendly.
12933 (darwin_rs6000_override_options): Likewise.
12934 (rs6000_option_override_internal): Likewise.
12935 (rs6000_return_in_memory): Fix overlong line.
12936 (init_cmulative_args): Use quoted strings, and make more
12937 translator-friendly.
12938 (rs6000_pass_by_reference): Fix overlong line.
12939 (def_builtin): Use quoted strings.
12940 (altivec_expand_predicate_builtin): Use quoted strings, and make
12941 more translator-friendly.
12942 (htm_expand_builtin): Use quoted strings.
12943 (cpu_expand_builtin): Use quoted strings, and make more
12944 translator-friendly.
12945 (altivec_expand_builtin): Likewise.
12946 (paired_expand_predicate_builtin): Likewise.
12947 (rs6000_invalid_builtin): Likewise.
12948 (builtin_function_type): Use quoted strings.
12949 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
12950 more translator-friendly.
12951 (rs6000_trampoline_init): Likewise.
12952 (rs6000_handle_altivec_attribute): Likewise.
12953 (rs6000_inner_target_options): Use quoted strings.
12954 (rs6000_disable_incompatible_switches): Likewise.
12955 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
12956 strings, and make more translator-friendly.
12957 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
12959 2017-08-14 Bin Cheng <bin.cheng@arm.com>
12961 PR tree-optimization/81799
12962 * tree-loop-distribution.c (version_loop_by_alias_check): Force
12963 cond_expr to simple gimple operand.
12965 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
12967 PR middle-end/46932
12968 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
12970 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
12974 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
12975 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
12976 TARGET_GASISR_PROLOGUES.
12977 * config/avr/avr.c (avr_option_override): Same.
12978 (avr_pass_pre_proep::execute): Same.
12980 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
12983 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
12984 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
12986 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
12988 * config/i386/i386.md (*load_tp_<mode>): Redefine as
12989 define_insn_and_split. Split to a memory load from 0 in
12990 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
12991 using PTR mode iterator.
12992 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
12993 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
12994 (*add_tp_<mode>): Redefine as define_insn_and_split.
12995 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
12996 address space. Merge with *add_tp_x32 using PTR mode iterator.
12997 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
12998 Split to an add with a memory load from 0 in
12999 DEFAULT_TLS_SEG_REG address space.
13001 2017-08-12 Andrew Pinski <apinski@cavium.com>
13003 * config/aarch64/aarch64-option-extensions.def (rdma):
13004 Fix feature string to what Linux prints out in /proc/cpuinfo.
13006 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
13009 * dwarf2out.c (modified_type_die): For C typedef types that have
13010 an ultimate origin, process the ultimate origin instead of the
13012 (gen_typedef_die): Assert that input DECLs have no ultimate
13014 (gen_type_die_with_usage): For typedef variants that have an
13015 ultimate origin, just call gen_decl_die on the original DECL.
13016 (process_scope_var): Avoid creating DIEs for local typedefs and
13017 concrete static variables.
13019 2017-08-12 Alan Modra <amodra@gmail.com>
13023 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
13024 match gnu-user.h startfile.
13025 (ENDFILE_LINUX_SPEC): Similarly.
13027 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
13030 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
13032 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
13034 2017-08-11 Tamar Christina <tamar.christina@arm.com>
13035 * config/aarch64/aarch64.md (mov<mode>): Change.
13036 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
13037 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
13038 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
13040 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
13042 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
13043 for storage order barriers.
13045 2017-08-11 Martin Liska <mliska@suse.cz>
13048 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
13049 variables of void type.
13051 2017-08-11 Martin Liska <mliska@suse.cz>
13053 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
13054 TARGET_SUPPORTS_ALIASES.
13055 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
13056 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
13057 (optimize_weakref): Likewise.
13058 * symtab.c (symtab_node::noninterposable_alias): Likewise.
13059 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
13060 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
13062 2017-08-11 Martin Liska <mliska@suse.cz>
13065 * config/i386/i386.c (make_resolver_func): Do complete
13066 refactoring of the function.
13068 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13071 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
13072 * config/i386/i386.c (ix86_stack_protect_guard): Use
13073 ix86_stack_protect_guard_symbol_str to generate varible declaration.
13074 * doc/invoke.texi (x86 Options): Document
13075 -mstack-protector-guard-symbol= option.
13077 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13079 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
13080 * config/i386/i386.c (ix86_split_stack_guard): New function.
13081 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
13082 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
13083 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
13084 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
13085 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
13086 (split_stack_space_check): Call ix86_split_stack_guard.
13088 2017-08-10 Martin Sebor <msebor@redhat.com>
13090 * print-tree.c (print_node): Print location using the established
13092 Replace spaces with colons.
13093 (debug_raw, debug): Ditto.
13095 2017-08-10 Martin Sebor <msebor@redhat.com>
13098 * pretty-print.c (pp_format): Correct the handling of %s precision.
13100 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
13103 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13105 (ix86_finalize_stack_frame_flags): This. Also clear
13106 frame_pointer_needed if -fno-omit-frame-pointer is used without
13108 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13109 with ix86_finalize_stack_frame_flags.
13110 (ix86_expand_epilogue): Likewise.
13111 (ix86_expand_split_stack_prologue): Likewise.
13112 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
13114 2017-08-10 Martin Liska <mliska@suse.cz>
13117 * c-attribs.c (handle_target_attribute):
13118 Report warning for an empty string argument of target attribute.
13120 2017-08-09 Jakub Jelinek <jakub@redhat.com>
13123 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
13125 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
13126 or DECL_NONLOCAL labels.
13127 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
13128 or DECL_NONLOCAL labels here.
13130 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
13132 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
13133 to indicate when early gimple folding has been disabled.
13134 (rs6000_gimple_fold_builtin): Add debug content.
13135 (rs6000_invalid_builtin): Fix whitespace.
13136 (rs6000_expand_builtin): Fix whitespace.
13137 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
13139 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
13142 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
13143 SAVE_MULTIPLE if not all the registers that saves, should be saved.
13145 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
13147 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
13148 (qdf24xx): Likewise.
13149 * config/aarch64/aarch64.md: Include falkor.md.
13150 * config/aarch64/falkor.md: New.
13152 2017-08-09 Marek Polacek <polacek@redhat.com>
13155 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
13156 * diagnostic.c (emit_diagnostic): Add a comment.
13157 (emit_diagnostic_valist): New function.
13159 2017-08-09 Marek Polacek <polacek@redhat.com>
13162 * input.c (make_location): New overload.
13163 * input.h (make_location): Declare.
13165 2017-08-08 Alan Modra <amodra@gmail.com>
13166 H.J. Lu <hongjiu.lu@intel.com>
13169 * gcc.c (NO_PIE_SPEC): Delete.
13170 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
13172 (LINK_PIE_SPEC): ..to here.
13173 (LINK_COMMAND_SPEC): Support -no-pie.
13174 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
13175 chain of crtbegin*.o selection, update for PIE_SPEC changes and
13177 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
13178 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
13179 (ENDFILE_CRTEND_SPEC): Similarly.
13181 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
13184 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
13185 (mstack-protector-guard-offset=): Ditto.
13186 * config/i386/i386.c (ix86_option_override): Handle
13187 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
13189 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
13190 ix86_stack_protect_guard_offset variables.
13191 (TARGET_STACK_PROTECT_GUARD): Always define.
13192 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
13193 and -mstack-protector-guard-offset= options.
13195 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13197 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
13198 boundary case for the last candidate.
13200 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13202 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
13203 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
13205 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13207 PR middle-end/19706
13208 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
13209 * config/aarch64/aarch64-builtins.c
13210 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
13211 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
13212 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
13214 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13215 Andrew Pinski <pinskia@gmail.com>
13217 PR middle-end/19706
13218 * internal-fn.def (XORSIGN): New.
13219 * optabs.def (xorsign_optab): New.
13220 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
13221 (convert_expand_mult_copysign): New.
13222 (pass_optimize_widening_mul::execute): Call
13223 convert_expand_mult_copysign.
13225 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13227 PR tree-optimization/81354
13228 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
13229 Insert on edges rather than explicitly creating landing pads.
13230 (analyze_candidates_and_replace): Commit edge inserts.
13232 2017-08-08 Richard Biener <rguenther@suse.de>
13234 PR middle-end/81719
13235 * tree-ssa-loop-niter.c: Include tree-dfa.h.
13236 (expand_simple_operations): Also look through ADDR_EXPRs with
13237 MEM_REF bases treating them as POINTER_PLUS_EXPR.
13239 2017-08-08 Richard Biener <rguenther@suse.de>
13241 PR tree-optimization/81723
13242 * tree-vect-slp.c (struct bst_traits): New hash traits.
13243 (bst_fail): New global.
13244 (vect_build_slp_tree_2): New worker, split out from ...
13245 (vect_build_slp_tree): ... this now wrapping it with using
13246 bst_fail set to cache SLP tree build fails. Properly handle
13248 (vect_analyze_slp_instance): Allocate and free bst_fail.
13250 2017-08-08 Martin Liska <mliska@suse.cz>
13253 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
13254 LABEL_DECLs that can be from a different function.
13256 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13258 PR tree-optimization/81744
13259 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
13260 loop's number of iterations.
13262 2017-08-08 Martin Liska <mliska@suse.cz>
13264 * asan.c: Include header files.
13265 * attribs.c (build_decl_attribute_variant): New function moved
13267 (build_type_attribute_qual_variant): Likewise.
13268 (cmp_attrib_identifiers): Likewise.
13269 (simple_cst_list_equal): Likewise.
13270 (omp_declare_simd_clauses_equal): Likewise.
13271 (attribute_value_equal): Likewise.
13272 (comp_type_attributes): Likewise.
13273 (build_type_attribute_variant): Likewise.
13274 (lookup_ident_attribute): Likewise.
13275 (remove_attribute): Likewise.
13276 (merge_attributes): Likewise.
13277 (merge_type_attributes): Likewise.
13278 (merge_decl_attributes): Likewise.
13279 (merge_dllimport_decl_attributes): Likewise.
13280 (handle_dll_attribute): Likewise.
13281 (attribute_list_equal): Likewise.
13282 (attribute_list_contained): Likewise.
13283 * attribs.h (lookup_attribute): New function moved from tree.[ch].
13284 (lookup_attribute_by_prefix): Likewise.
13285 * bb-reorder.c: Include header files.
13286 * builtins.c: Likewise.
13287 * calls.c: Likewise.
13288 * cfgexpand.c: Likewise.
13289 * cgraph.c: Likewise.
13290 * cgraphunit.c: Likewise.
13291 * convert.c: Likewise.
13292 * dwarf2out.c: Likewise.
13293 * final.c: Likewise.
13294 * fold-const.c: Likewise.
13295 * function.c: Likewise.
13296 * gimple-expr.c: Likewise.
13297 * gimple-fold.c: Likewise.
13298 * gimple-pretty-print.c: Likewise.
13299 * gimple.c: Likewise.
13300 * gimplify.c: Likewise.
13301 * hsa-common.c: Likewise.
13302 * hsa-gen.c: Likewise.
13303 * internal-fn.c: Likewise.
13304 * ipa-chkp.c: Likewise.
13305 * ipa-cp.c: Likewise.
13306 * ipa-devirt.c: Likewise.
13307 * ipa-fnsummary.c: Likewise.
13308 * ipa-inline.c: Likewise.
13309 * ipa-visibility.c: Likewise.
13311 * lto-cgraph.c: Likewise.
13312 * omp-expand.c: Likewise.
13313 * omp-general.c: Likewise.
13314 * omp-low.c: Likewise.
13315 * omp-offload.c: Likewise.
13316 * omp-simd-clone.c: Likewise.
13317 * opts-global.c: Likewise.
13318 * passes.c: Likewise.
13319 * predict.c: Likewise.
13320 * sancov.c: Likewise.
13321 * sanopt.c: Likewise.
13322 * symtab.c: Likewise.
13323 * toplev.c: Likewise.
13324 * trans-mem.c: Likewise.
13325 * tree-chkp.c: Likewise.
13326 * tree-eh.c: Likewise.
13327 * tree-into-ssa.c: Likewise.
13328 * tree-object-size.c: Likewise.
13329 * tree-parloops.c: Likewise.
13330 * tree-profile.c: Likewise.
13331 * tree-ssa-ccp.c: Likewise.
13332 * tree-ssa-live.c: Likewise.
13333 * tree-ssa-loop.c: Likewise.
13334 * tree-ssa-sccvn.c: Likewise.
13335 * tree-ssa-structalias.c: Likewise.
13336 * tree-ssa.c: Likewise.
13337 * tree-streamer-in.c: Likewise.
13338 * tree-vectorizer.c: Likewise.
13339 * tree-vrp.c: Likewise.
13340 * tsan.c: Likewise.
13341 * ubsan.c: Likewise.
13342 * varasm.c: Likewise.
13343 * varpool.c: Likewise.
13344 * tree.c: Remove functions moved to attribs.[ch].
13345 * tree.h: Likewise.
13346 * config/aarch64/aarch64.c: Add attrs.h header file.
13347 * config/alpha/alpha.c: Likewise.
13348 * config/arc/arc.c: Likewise.
13349 * config/arm/arm.c: Likewise.
13350 * config/avr/avr.c: Likewise.
13351 * config/bfin/bfin.c: Likewise.
13352 * config/c6x/c6x.c: Likewise.
13353 * config/cr16/cr16.c: Likewise.
13354 * config/cris/cris.c: Likewise.
13355 * config/darwin.c: Likewise.
13356 * config/epiphany/epiphany.c: Likewise.
13357 * config/fr30/fr30.c: Likewise.
13358 * config/frv/frv.c: Likewise.
13359 * config/ft32/ft32.c: Likewise.
13360 * config/h8300/h8300.c: Likewise.
13361 * config/i386/winnt.c: Likewise.
13362 * config/ia64/ia64.c: Likewise.
13363 * config/iq2000/iq2000.c: Likewise.
13364 * config/lm32/lm32.c: Likewise.
13365 * config/m32c/m32c.c: Likewise.
13366 * config/m32r/m32r.c: Likewise.
13367 * config/m68k/m68k.c: Likewise.
13368 * config/mcore/mcore.c: Likewise.
13369 * config/microblaze/microblaze.c: Likewise.
13370 * config/mips/mips.c: Likewise.
13371 * config/mmix/mmix.c: Likewise.
13372 * config/mn10300/mn10300.c: Likewise.
13373 * config/moxie/moxie.c: Likewise.
13374 * config/msp430/msp430.c: Likewise.
13375 * config/nds32/nds32-isr.c: Likewise.
13376 * config/nds32/nds32.c: Likewise.
13377 * config/nios2/nios2.c: Likewise.
13378 * config/nvptx/nvptx.c: Likewise.
13379 * config/pa/pa.c: Likewise.
13380 * config/pdp11/pdp11.c: Likewise.
13381 * config/powerpcspe/powerpcspe.c: Likewise.
13382 * config/riscv/riscv.c: Likewise.
13383 * config/rl78/rl78.c: Likewise.
13384 * config/rx/rx.c: Likewise.
13385 * config/s390/s390.c: Likewise.
13386 * config/sh/sh.c: Likewise.
13387 * config/sol2.c: Likewise.
13388 * config/sparc/sparc.c: Likewise.
13389 * config/spu/spu.c: Likewise.
13390 * config/stormy16/stormy16.c: Likewise.
13391 * config/tilegx/tilegx.c: Likewise.
13392 * config/tilepro/tilepro.c: Likewise.
13393 * config/v850/v850.c: Likewise.
13394 * config/vax/vax.c: Likewise.
13395 * config/visium/visium.c: Likewise.
13396 * config/xtensa/xtensa.c: Likewise.
13398 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13401 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
13402 constraints since the -mupper-regs-* switches have been
13404 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
13405 into a vector from a double word element that was extracted from
13406 another vector, and eliminate extra XXPERMDI instructions.
13407 (vsx_concat_<mode>_2): Likewise.
13408 (vsx_concat_<mode>_3): Likewise.
13409 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
13410 concat to allow optimizing inserts from previous extracts.
13412 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
13414 * config/i386/i386.c (ix86_stack_protect_guard): Generate
13415 memory reference to a SSP offset in TLS address space.
13416 (ix86_print_operand) <case '@'>: Remove.
13417 (ix86_print_operand_punct_valid_p): Remove '@' code.
13418 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
13419 UNSPEC_SP_TLS_TEST.
13420 (stack_tls_protect_set_<mode>): Remove.
13421 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
13422 (stack_tls_protect_test_<mode>): Remove.
13423 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
13425 2017-08-07 Olivier Hainque <hainque@adacore.com>
13428 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
13430 2017-08-07 Douglas Rupp <rupp@adacore.com>
13432 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
13433 variable was referenced as multidir in command.
13435 2017-08-07 Jakub Jelinek <jakub@redhat.com>
13438 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
13441 2017-08-07 Martin Liska <mliska@suse.cz>
13443 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
13444 * config/rl78/rl78.c: Add include of attribs.h.
13445 * config/sh/sh.c: Likewise.
13446 * config/v850/v850.c: Likewise.
13448 2017-08-07 Tom de Vries <tom@codesourcery.com>
13450 PR middle-end/78266
13451 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
13453 2017-08-07 Martin Liska <mliska@suse.cz>
13455 * config/mips/mips.c: Include attribs.h.
13457 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
13460 * doc/invoke.texi: Document change in behvaior for -Ofast for
13463 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
13465 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
13467 (aarch64_pop_regs): Likewise.
13468 (aarch64_gen_load_pair): Likewise.
13469 (aarch64_save_callee_saves): Likewise.
13470 (aarch64_restore_callee_saves): Likewise.
13472 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13474 * config/i386/i386.c: Revert the last change.
13476 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13479 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13481 (ix86_finalize_stack_frame_flags): This. Also clear
13482 frame_pointer_needed if -fno-omit-frame-pointer is used without
13484 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13485 with ix86_finalize_stack_frame_flags.
13486 (ix86_expand_epilogue): Likewise.
13487 (ix86_expand_split_stack_prologue): Likewise.
13489 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13492 * config/i386/i386.c (get_builtin_code_for_version): Set priority
13493 to P_AES for Westmere.
13495 2017-08-07 Jonathan Yong <10walls@gmail.com>
13497 * config/i386/mingw.opt (fset-stack-executable): Removed.
13498 * config/i386/cygming.opt (fset-stack-executable): Moved
13500 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
13502 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
13504 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
13506 2017-08-07 Marek Polacek <polacek@redhat.com>
13508 PR middle-end/81737
13509 * fold-const.c (fold_indirect_ref_1): Check type_domain.
13511 2017-08-07 Martin Liska <mliska@suse.cz>
13513 * attribs.h (canonicalize_attr_name): New function.
13514 (cmp_attribs): Move from c-format.c and adjusted.
13515 (is_attribute_p): Moved from tree.h.
13516 * tree-inline.c: Add new includes.
13517 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
13518 (private_is_attribute_p): Remove.
13519 (private_lookup_attribute): Likewise.
13520 (private_lookup_attribute_by_prefix): Simplify.
13521 (remove_attribute): Use is_attribute_p.
13522 * tree.h: Remove removed declarations.
13524 2017-08-07 Jakub Jelinek <jakub@redhat.com>
13526 PR middle-end/81698
13527 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
13528 instead of computing it in the function. Formatting fix.
13529 (expand_case): Don't rely on default_edge being the first edge,
13530 clear it if removing it, pass default_edge to
13531 emit_case_dispatch_table.
13532 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
13535 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
13537 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
13538 insn in the function, emit NOP after the insn.
13540 2017-08-06 Tom de Vries <tom@codesourcery.com>
13542 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
13545 2017-08-06 Tom de Vries <tom@codesourcery.com>
13547 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
13550 2017-08-04 Yury Gribov <tetra2005@gmail.com>
13552 PR tree-optimization/57371
13553 * match.pd: New pattern.
13555 2017-08-04 Marek Polacek <polacek@redhat.com>
13557 PR middle-end/81695
13558 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
13559 perform the computation in offset_int.
13561 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13563 PR tree-optimization/81136
13564 * tree-vectorizer.h: Include tree-hash-traits.h.
13565 (vec_base_alignments): New typedef.
13566 (vec_info): Add a base_alignments field.
13567 (vect_record_base_alignments): Declare.
13568 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
13570 (DR_IS_CONDITIONAL_IN_STMT): New macro.
13571 (create_data_ref): Add an is_conditional_in_stmt argument.
13572 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
13573 the is_conditional_in_stmt field.
13574 (data_ref_loc): Add an is_conditional_in_stmt field.
13575 (get_references_in_stmt): Set the is_conditional_in_stmt field.
13576 (find_data_references_in_stmt): Update call to create_data_ref.
13577 (graphite_find_data_references_in_stmt): Likewise.
13578 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
13579 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
13580 (vect_record_base_alignment): New function.
13581 (vect_record_base_alignments): Likewise.
13582 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
13583 for nested statements even if we fail to compute a misalignment.
13584 Use pooled base alignments for unconditional references.
13585 (vect_find_same_alignment_drs): Compare base addresses instead
13587 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
13588 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
13590 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13592 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
13593 Add an explicit name for the enum. Use auto_vec for slp_instances
13594 and grouped_stores.
13595 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
13597 (_bb_vec_info): Add a constructor and destructor.
13598 (vinfo_for_stmt): Return NULL for uids of -1 as well.
13599 (destroy_loop_vec_info): Delete.
13600 (vect_destroy_datarefs): Likewise.
13601 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
13602 (vec_info::vec_info): New function.
13603 (vec_info::~vec_info): Likewise.
13604 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
13605 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
13606 destroy_loop_vec_info.
13607 * tree-vect-loop.c (new_loop_vec_info): Replace with...
13608 (_loop_vec_info::_loop_vec_info): ...this.
13609 (destroy_loop_vec_info): Replace with...
13610 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
13611 the stmt_vec_infos. Leave handling of vec_info information to its
13612 destructor. Remove explicit vector releases.
13613 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
13614 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
13615 * tree-vect-slp.c (new_bb_vec_info): Replace with...
13616 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
13617 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
13618 (destroy_bb_vec_info): Replace with...
13619 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
13620 information to its destructor.
13621 (vect_slp_analyze_bb_1): Use new and delete instead of
13622 new_bb_vec_info and destroy_bb_vec_info.
13623 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
13626 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13628 * tree-data-ref.h (subscript): Add access_fn field.
13629 (data_dependence_relation): Add could_be_independent_p.
13630 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
13631 (same_access_functions): Move to tree-data-ref.c.
13632 * tree-data-ref.c (ref_contains_union_access_p): New function.
13633 (access_fn_component_p): Likewise.
13634 (access_fn_components_comparable_p): Likewise.
13635 (dr_analyze_indices): Add a reference to access_fn_component_p.
13636 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
13638 (constant_access_functions): Likewise.
13639 (add_other_self_distances): Likewise.
13640 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
13641 (initialize_data_dependence_relation): Use XCNEW and remove
13642 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
13643 of access functions that have the same type. Allow the
13644 subsequence to end with different bases in some circumstances.
13645 Record the chosen access functions in SUB_ACCESS_FN.
13646 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
13647 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
13648 (subscript_dependence_tester_1): Likewise dra and drb.
13649 (build_classic_dist_vector): Update calls accordingly.
13650 (subscript_dependence_tester): Likewise.
13651 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
13652 DDR_COULD_BE_INDEPENDENT_P.
13653 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
13654 comp_alias_ddrs instead of may_alias_ddrs.
13655 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
13657 (vect_analyze_data_ref_dependence): Use it if
13658 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
13659 distance vectors if that fails.
13660 (dependence_distance_ge_vf): New function.
13661 (vect_prune_runtime_alias_test_list): Use it. Don't clear
13662 LOOP_VINFO_MAY_ALIAS_DDRS.
13664 2017-08-04 Richard Biener <rguenther@suse.de>
13666 PR middle-end/81705
13667 * fold-const.c (fold_binary_loc): Properly restrict
13668 minus_var0 && minus_var1 case when associating undefined overflow
13671 2017-08-04 Tom de Vries <tom@codesourcery.com>
13673 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
13675 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13677 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13678 Don't start diagnostic messages with a capital letter.
13679 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13681 (rs6000_invalid_builtin): Likewise.
13682 (rs6000_trampoline_init): Likewise.
13684 2017-08-03 Jakub Jelinek <jakub@redhat.com>
13687 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
13688 after setting changeable df flags.
13690 2017-08-03 Richard Biener <rguenther@suse.de>
13692 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
13693 up if the use is in USE - X.
13695 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
13697 * toplev.c (dumpfile.h): New include.
13698 (internal_error_reentered): New static function. Use it...
13699 (internal_error_function): ...here to handle reentered internal_error.
13701 2017-08-03 Richard Biener <rguenther@suse.de>
13703 PR middle-end/81148
13704 * fold-const.c (split_tree): Add minus_var and minus_con
13705 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
13706 here but always use minus_*.
13707 (associate_trees): Assert we never associate with MINUS_EXPR
13708 and NULL first operand. Do not recurse for PLUS_EXPR operands
13709 when associating as MINUS_EXPR either.
13710 (fold_binary_loc): Track minus_var and minus_con.
13712 2017-08-03 Tom de Vries <tom@codesourcery.com>
13715 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
13716 ACCEL_COMPILER, apply finish_options on
13717 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
13719 2017-08-03 Tom de Vries <tom@codesourcery.com>
13722 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
13723 function_entry_patch_area_size > 0.
13725 2017-08-03 Jakub Jelinek <jakub@redhat.com>
13728 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
13729 instead of 10??LU, perform unit multiplication in wide_int,
13730 don't change alloc_object_size_limit if the limit is larger
13733 PR tree-optimization/81655
13734 PR tree-optimization/81588
13735 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
13736 the case when ranges[i].low and high are 1 for unsigned type with
13739 PR middle-end/81052
13740 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
13741 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
13743 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13745 * tree-vrp.h: Add include guard.
13747 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
13750 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
13751 (ud2): New insn pattern.
13752 * config/i386/i386.c (ix86_expand_epilogue):
13753 For naked functions, generate ud2 instead of trap insn.
13755 2017-08-02 Marek Polacek <polacek@redhat.com>
13758 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
13760 2017-08-02 Tom de Vries <tom@codesourcery.com>
13761 Cesar Philippidis <cesar@codesourcery.com>
13763 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
13764 Add missing edge probabilities.
13766 2017-08-02 Tamar Christina <tamar.christina@arm.com>
13768 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
13769 Correct endianness.
13771 2017-08-02 Jakub Jelinek <jakub@redhat.com>
13773 PR middle-end/79499
13774 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
13775 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
13776 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
13777 prologue_seq sequences - if any.
13779 2017-08-02 Richard Biener <rguenther@suse.de>
13781 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
13782 via vectors if supported, integer extracts via punning if supported
13783 or otherwise vector extracts.
13785 2017-08-02 Richard Biener <rguenther@suse.de>
13787 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
13789 (bitmap_insert_into_set): ... this.
13791 2017-08-02 Richard Biener <rguenther@suse.de>
13793 PR tree-optimization/81633
13795 2015-08-17 Alan Hayward <alan.hayward@arm.com>
13797 PR tree-optimization/71752
13798 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
13800 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
13802 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
13803 (machine_function::call_ms2sysv_pad_out): Remove field.
13804 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
13805 (ix86_compute_frame_layout): Likewise.
13807 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
13810 * config/i386/i386.c (ix86_set_func_type): Disallow naked
13811 attribute with interrupt attribute.
13813 2017-08-01 Andrew Pinski <apinski@cavium.com>
13815 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
13816 BIT_INSERT_EXPR's operand 1
13817 to see if the types precision matches.
13819 2017-08-01 Martin Liska <mliska@suse.cz>
13821 PR middle-end/70140
13822 * builtins.c (expand_builtin_memcpy_args): Remove.
13823 (expand_builtin_memcpy): Call newly added function
13824 expand_builtin_memory_copy_args.
13825 (expand_builtin_memcpy_with_bounds): Likewise.
13826 (expand_builtin_mempcpy): Remove last argument.
13827 (expand_builtin_mempcpy_with_bounds): Likewise.
13828 (expand_builtin_memory_copy_args): New function created from
13829 expand_builtin_mempcpy_args with small modifications.
13830 (expand_builtin_mempcpy_args): Remove.
13831 (expand_builtin_stpcpy): Remove unused argument.
13832 (expand_builtin): Likewise.
13833 (expand_builtin_with_bounds): Likewise.
13835 2017-08-01 Martin Liska <mliska@suse.cz>
13838 Make mempcpy more optimal (PR middle-end/70140).
13840 2017-08-01 Jakub Jelinek <jakub@redhat.com>
13843 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
13844 __builtin_vec_cmpne verify both arguments are compatible vectors
13845 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
13846 verify arg1_type is a pointer or array type. For __builtin_vec_st,
13847 move computation of aligned to after checking the argument types.
13851 * config/rs6000/vsx.md (vextract_fp_from_shorth,
13852 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
13855 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
13856 Doug Rupp <rupp@adacore.com>
13857 Olivier Hainque <hainque@adacore.com>
13859 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
13860 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
13862 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
13863 for TARGET_OS_CPP_BUILTIN.
13864 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
13865 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
13867 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
13868 passing required abi options to the assembler for EABI configurations.
13869 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
13870 of .text.hot and .text.unlikely sections for kernel modules when
13871 using ARM style exceptions.
13872 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
13873 options. Add EXTRA_CC1_SPEC.
13874 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
13876 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
13878 (ARM_TARGET2_DWARF_FORMAT): Define.
13879 * config/arm/t-vxworks: Adjust multilib control to removal of the
13880 Diab command line options.
13882 2017-08-01 Martin Liska <mliska@suse.cz>
13884 PR gcov-profile/81561
13885 * gcov.c (unblock): Make unblocking safe as we need to preserve
13886 index correspondence of blocks and block_lists.
13888 2017-08-01 Richard Biener <rguenther@suse.de>
13890 PR tree-optimization/81181
13891 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
13892 (compute_antic): ... end of iteration here.
13894 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
13896 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
13897 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
13898 (ftree-slp-vectorize): Likewise.
13899 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
13900 can no longer be set independent of flag_tree_loop_vectorize.
13901 * omp-general.c (emp_max_vf): Likewise.
13902 * opts.c (enable_fdo_optimizations): Remove references to
13903 flag_tree_vectorize, these are now implicit.
13904 (common_handle_option): Remove handling for OPT_ftree_vectorize,
13905 and leave it for the options machinery.
13907 2017-08-01 Martin Liska <mliska@suse.cz>
13909 PR middle-end/70140
13910 * builtins.c (expand_builtin_memcpy_args): Remove.
13911 (expand_builtin_memcpy): Call newly added function
13912 expand_builtin_memory_copy_args.
13913 (expand_builtin_memcpy_with_bounds): Likewise.
13914 (expand_builtin_mempcpy): Remove last argument.
13915 (expand_builtin_mempcpy_with_bounds): Likewise.
13916 (expand_builtin_memory_copy_args): New function created from
13917 expand_builtin_mempcpy_args with small modifications.
13918 (expand_builtin_mempcpy_args): Remove.
13919 (expand_builtin_stpcpy): Remove unused argument.
13920 (expand_builtin): Likewise.
13921 (expand_builtin_with_bounds): Likewise.
13923 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
13926 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
13927 print "ds:" only for immediates in generic address space.
13929 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
13932 * config/i386/i386.c (ix86_funciton_naked): New prototype.
13933 (ix86_function_ok_for_sibcall): Return false for naked functions.
13935 2017-08-01 Richard Biener <rguenther@suse.de>
13937 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
13938 (compute_antic): Seed worklist with exit block predecessors.
13939 * cfganal.c (dfs_find_deadend): For a cycle return the source
13940 of the edge closing it.
13942 2017-08-01 Tamar Christina <tamar.christina@arm.com>
13944 * config/aarch64/aarch64.c
13945 (aarch64_can_const_movi_rtx_p): Move 0 check.
13947 2017-08-01 Bin Cheng <bin.cheng@arm.com>
13949 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
13950 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
13952 * match.pd: Ditto in address comparison pattern.
13954 2017-08-01 Bin Cheng <bin.cheng@arm.com>
13956 PR tree-optimization/81627
13957 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
13958 closed ssa form for store-store chain.
13960 2017-08-01 Bin Cheng <bin.cheng@arm.com>
13962 PR tree-optimization/81620
13963 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
13964 for store-store chain.
13966 2017-08-01 Jakub Jelinek <jakub@redhat.com>
13968 PR tree-optimization/81588
13969 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
13970 ranges[i].in_p, invert comparison code ccode. For >/>=,
13971 swap rhs1 and rhs2 and comparison code unconditionally,
13972 for </<= don't do that. Don't swap rhs1/rhs2 again if
13973 ranges[i].in_p, instead invert comparison code ccode if
13974 opcode or oe->rank is BIT_IOR_EXPR.
13977 * optabs.def (vec_extract_optab, vec_init_optab): Change from
13978 a direct optab to conversion optab.
13979 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
13980 with GET_MODE_INNER as last argument instead of optab_handler.
13981 * expmed.c (extract_bit_field_1): Likewise. Use vector from
13982 vector extraction if possible and optab is available.
13983 * expr.c (store_constructor): Use convert_optab_handler instead
13984 of optab_handler. Use vector initialization from smaller
13985 vectors if possible and optab is available.
13986 * tree-vect-stmts.c (vectorizable_load): Likewise.
13987 * doc/md.texi (vec_extract, vec_init): Document that the optabs
13988 now have two modes.
13989 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
13990 of vec_init from half-sized vectors with the same element mode.
13991 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
13992 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
13993 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
13994 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
13995 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
13996 after mode in gen_vec_extract* calls.
13997 (vec_extract<mode>): Renamed to ...
13998 (vec_extract<mode><ssescalarmodelower>): ... this.
13999 (vec_extract<mode><ssehalfvecmodelower>): New expander.
14000 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
14001 element mode after mode in gen_vec_init* calls.
14002 (VEC_INIT_HALF_MODE): New mode iterator.
14003 (vec_init<mode>): Renamed to ...
14004 (vec_init<mode><ssescalarmodelower>): ... this.
14005 (vec_init<mode><ssehalfvecmodelower>): New expander.
14006 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
14007 (vec_extractv2sfsf): ... this.
14008 (vec_initv2sf): Renamed to ...
14009 (vec_initv2sfsf): ... this.
14010 (vec_extractv2si): Renamed to ...
14011 (vec_extractv2sisi): ... this.
14012 (vec_initv2si): Renamed to ...
14013 (vec_initv2sisi): ... this.
14014 (vec_extractv4hi): Renamed to ...
14015 (vec_extractv4hihi): ... this.
14016 (vec_initv4hi): Renamed to ...
14017 (vec_initv4hihi): ... this.
14018 (vec_extractv8qi): Renamed to ...
14019 (vec_extractv8qiqi): ... this.
14020 (vec_initv8qi): Renamed to ...
14021 (vec_initv8qiqi): ... this.
14022 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
14023 (vec_init<mode>): Renamed to ...
14024 (vec_init<mode><VEC_base_l>): ... this.
14025 (vec_extract<mode>): Renamed to ...
14026 (vec_extract<mode><VEC_base_l>): ... this.
14027 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
14028 (vec_initv2sfsf): ... this.
14029 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
14030 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14031 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
14032 element mode after mode in gen_vec_init* calls.
14033 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
14034 (vec_init<mode><Vel>): ... this.
14035 (vec_extract<mode>): Renamed to ...
14036 (vec_extract<mode><Vel>): ... this.
14037 * config/aarch64/iterators.md (Vel): New mode attribute.
14038 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
14039 Add element mode after mode in gen_vec_extract* calls.
14040 * config/s390/vector.md (non_vec_l): New mode attribute.
14041 (vec_extract<mode>): Renamed to ...
14042 (vec_extract<mode><non_vec_l>): ... this.
14043 (vec_init<mode>): Renamed to ...
14044 (vec_init<mode><non_vec_l>): ... this.
14045 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
14046 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
14048 * config/arm/iterators.md (V_elem_l): New mode attribute.
14049 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
14050 (vec_extract<mode><V_elem_l>): ... this.
14051 (vec_extractv2di): Renamed to ...
14052 (vec_extractv2didi): ... this.
14053 (vec_init<mode>): Renamed to ...
14054 (vec_init<mode><V_elem_l>): ... this.
14055 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
14056 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
14057 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
14058 Add element mode after gen_vec_extract* calls.
14059 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
14060 (vec_init<mode><unitmode>): ... this.
14061 (vec_extract<mode>): Renamed to ...
14062 (vec_extract<mode><unitmode>): ... this.
14063 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
14064 (vec_init<mode><unitmode>): ... this.
14065 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
14066 (vec_initv2sfsf): ... this.
14067 (vec_extractv2sf): Renamed to ...
14068 (vec_extractv2sfsf): ... this.
14069 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
14070 Add element mode after gen_vec_extract* calls.
14071 * config/mips/mips.md (unitmode): New mode iterator.
14072 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
14073 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
14074 * config/spu/spu.md (inner_l): New mode attribute.
14075 (vec_init<mode>): Renamed to ...
14076 (vec_init<mode><inner_l>): ... this.
14077 (vec_extract<mode>): Renamed to ...
14078 (vec_extract<mode><inner_l>): ... this.
14079 * config/sparc/sparc.md (veltmode): New mode iterator.
14080 (vec_init<VMALL:mode>): Renamed to ...
14081 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
14082 * config/ia64/vect.md (vec_initv2si): Renamed to ...
14083 (vec_initv2sisi): ... this.
14084 (vec_initv2sf): Renamed to ...
14085 (vec_initv2sfsf): ... this.
14086 (vec_extractv2sf): Renamed to ...
14087 (vec_extractv2sfsf): ... this.
14088 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
14089 (vec_init<mode>): Renamed to ...
14090 (vec_init<mode><VEC_base_l>): ... this.
14091 (vec_extract<mode>): Renamed to ...
14092 (vec_extract<mode><VEC_base_l>): ... this.
14093 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
14094 (vec_initv2sfsf): ... this.
14095 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
14096 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14097 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
14098 gen_vec_init* calls.
14100 2017-08-01 Richard Biener <rguenther@suse.de>
14102 PR tree-optimization/81297
14103 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
14104 TREE_OVERFLOW from INTEGER_CSTs.
14106 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
14108 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
14110 2017-07-31 Carl Love <cel@us.ibm.com>
14112 * config/rs6000/rs6000-c: Add support for built-in functions
14113 vector signed char vec_xl_be (signed long long, signed char *);
14114 vector unsigned char vec_xl_be (signed long long, unsigned char *);
14115 vector signed int vec_xl_be (signed long long, signed int *);
14116 vector unsigned int vec_xl_be (signed long long, unsigned int *);
14117 vector signed long long vec_xl_be (signed long long, signed long long *);
14118 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
14119 vector signed short vec_xl_be (signed long long, signed short *);
14120 vector unsigned short vec_xl_be (signed long long, unsigned short *);
14121 vector double vec_xl_be (signed long long, double *);
14122 vector float vec_xl_be (signed long long, float *);
14123 * config/rs6000/altivec.h (vec_xl_be): Add #define.
14124 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
14125 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
14127 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
14128 (altivec_expand_builtin): Add switch statement to call
14129 altivec_expand_xl_be for each builtin.
14130 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
14131 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
14132 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
14133 __builtin_vsx_le_be_v16qi.
14134 * doc/extend.texi: Update the built-in documentation file for the
14135 new built-in functions.
14137 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14140 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
14142 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
14144 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14146 * config.gcc: Add z14.
14147 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
14148 CPU model numbers for z13s and z14.
14149 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
14151 * config/s390/s390-opts.h (enum processor_type): Rename
14152 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14153 * config/s390/s390.c (processor_table): Add field for CPU name to
14154 be passed to Binutils.
14155 (s390_asm_output_machine_for_arch): Use the new field in
14156 processor_table for Binutils.
14157 (s390_expand_builtin): Replace arch12 with z14.
14158 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14159 (s390_get_sched_attrmask): Likewise.
14160 (s390_get_unit_mask): Likewise.
14161 * config/s390/s390.opt: Add z14 to processor_type enum.
14163 2017-07-31 Martin Jambor <mjambor@suse.cz>
14166 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
14167 regardless of optimization level.
14169 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
14170 Martin Liska <mliska@suse.cz>
14172 * predict.def: Remove old comment and adjust probability.
14173 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
14174 PREDICT statements.
14176 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14179 * config/i386/i386.c (ix86_function_naked): New function.
14180 (ix86_can_use_return_insn_p): Return false for naked functions.
14181 (ix86_expand_prologue): Skip prologue for naked functions.
14182 (ix86_expand_epilogue): Skip epilogue for naked functions
14183 and emit trap instruction.
14184 (ix86_warn_func_return): New function.
14185 (ix86_attribute_table): Add "naked" attribute specification.
14186 (TARGET_WARN_FUNC_RETURN): Define.
14187 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
14189 2017-07-31 Martin Liska <mliska@suse.cz>
14191 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
14192 (dump_gimple_bb_header): Always dump BB info.
14193 (pp_cfg_jump): Do not append info about BB when dumping a jump.
14195 2017-07-31 Martin Liska <mliska@suse.cz>
14198 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
14199 also with current_function_decl non-null equality.
14201 2017-07-31 Jakub Jelinek <jakub@redhat.com>
14204 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
14205 change type to the element type, instead add eltype variable and
14206 use it where we are interested in the element type.
14208 PR tree-optimization/81603
14209 * ipa-polymorphic-call.c
14210 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
14211 offset arithmetic in offset_int, bail out if the resulting bit offset
14212 doesn't fit into shwi.
14214 2017-07-31 Martin Liska <mliska@suse.cz>
14216 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
14217 (gimplify_save_expr): Fix comment.
14219 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
14222 * config/i386/i386.c (ix86_function_arg): Update arguments for
14224 (ix86_compute_frame_layout): Set the initial stack offset to
14225 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
14226 INCOMING_FRAME_SP_OFFSET.
14227 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
14228 stack before exception handler returns.
14229 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
14230 the 'ERROR_CODE' for exception handler.
14232 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
14234 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
14235 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
14236 (ASM_OUTPUT_REG_POP): Ditto.
14237 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
14238 instead of asm_fprintf to output pure string.
14240 2017-07-29 Jakub Jelinek <jakub@redhat.com>
14242 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
14243 to imported_module_or_decl hook.
14244 (debug_nothing_tree_tree_tree_bool): Remove.
14245 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
14246 * debug.c (do_nothing_debug_hooks): Use
14247 debug_nothing_tree_tree_tree_bool_bool instead of
14248 debug_nothing_tree_tree_tree_bool.
14249 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14250 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
14251 * sdbout.c (sdb_debug_hooks): Likewise.
14252 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
14253 (gen_namespace_die): Add DW_AT_export_symbols attribute if
14255 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
14256 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
14257 attribute, don't add anything.
14259 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14261 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
14262 (fold_build2_stat_loc): Likewise.
14263 (fold_build3_stat_loc): Likewise.
14264 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
14265 (fold_build1_loc): Remove macro.
14266 (fold_build2_loc): Likewise.
14267 (fold_build3_loc): Likewise.
14269 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14271 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
14272 (gimple_build_debug_bind_source_stat): Likewise.
14273 * gimple.h (gimple_build_debug_bind): Remove macro.
14274 (gimple_build_debug_bind_source): Likewise.
14276 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14278 * bitmap.c (bitmap_alloc): Adjust.
14279 (bitmap_gc_alloc): Likewise.
14280 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
14282 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14284 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
14285 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
14286 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
14287 (bitmap_gc_alloc_stat): Likewise.
14288 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
14290 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14292 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
14293 * rtl.h (shallow_copy_rtx): Remove macro.
14295 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14297 * emit-rtl.c (gen_raw_REG): Adjust.
14298 * gengenrtl.c (gendef): Likewise.
14299 * rtl.c (rtx_alloc_stat): Remove _stat from name.
14300 * rtl.h (rtx_alloc): Remove macro.
14302 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14304 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
14305 (build_tree_list_stat): Likewise.
14306 * tree.h (build_tree_list): Remove macro.
14307 (build_tree_list_vec): Likewise.
14309 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14311 * tree.c (make_vector_stat): Remove _stat from name.
14312 (build_vector_stat): Likewise.
14313 * tree.h (make_vector_stat): Remove macro.
14314 (build_vector_stat): Likewise.
14316 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14318 * tree.h (build_var_debug_value): Remove prototype.
14320 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14322 * tree.c (tree_cons_stat): Remove _stat from name.
14323 * tree.h (tree_cons): Remove macro.
14325 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14327 * tree.c (build_vl_exp_stat): Remove _stat from name.
14328 * tree.h (build_vl_exp): Remove macro.
14330 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14332 * tree.c (build_decl_stat): Remove _stat from name.
14333 * tree.h (build_decl): Remove macro.
14335 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14337 * gimple.c (gimple_build_with_ops_stat): Adjust.
14338 (gimple_alloc_stat): Remove _stat from name.
14339 * gimple.h (gimple_alloc): Remove macro.
14341 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14343 * tree.c (make_tree_vec_stat): Remove _stat from name.
14344 (grow_tree_vec_stat): Likewise.
14345 * tree.h (make_tree_vec_stat): Adjust prototype.
14346 (grow_tree_vec_stat): Likewise.
14347 (make_tree_vec): Remove macro.
14348 (grow_tree_vec): Likewise.
14350 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14352 * fold-const.c (fold_build1_stat_loc): Adjust.
14353 (fold_build2_stat_loc): Likewise.
14354 (fold_build3_stat_loc): Likewise.
14355 * tree.c (build0_stat): Remove _stat from name.
14356 (build1_stat): Likewise.
14357 (build2_stat): Likewise.
14358 (build3_stat): Likewise.
14359 (build4_stat): Likewise.
14360 (build5_stat): Likewise.
14361 * tree.h (build1_loc): Remove macro, and rename _stat function
14363 (build2_loc): Likewise.
14364 (build3_loc): Likewise.
14365 (build4_loc): Likewise.
14366 (build5_loc): Likewise.
14368 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14370 * tree.c (make_int_cst_stat): Remove _stat from name.
14371 * tree.h (make_int_cst_stat): Adjust prototype.
14372 (make_int_cst): Remove macro.
14374 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14376 * tree.c (make_tre_binfo_stat): Remove _stat from name.
14377 * tree.h (make_tree_binfo_stat): Adjust prototype.
14378 (make_tree_binfo): Remove.
14380 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14382 * tree.c (copy_node_stat): Rename to copy_node.
14383 (build_distinct_type_copy): Adjust.
14384 * tree.h (copy_node_stat): Adjust prototype.
14385 (copy_node): Remove macro.
14387 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14389 * tree.c (make_node_stat): rename to make_node.
14390 (build_tree_list_stat): Adjust.
14391 (build0_stat): Likewise.
14392 (build2_stat): Likewise.
14393 (build3_stat): Likewise.
14394 (build4_stat): Likewise.
14395 (build5_stat): Likewise.
14396 (build_decl_stat): Likewise.
14397 * tree.h (make_node_stat): Adjust prototype.
14398 (make_node): remove macro.
14400 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
14402 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
14403 (PPC_FEATURE2_SCV): Likewise.
14404 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
14406 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14408 * config/aarch64/aarch64.c
14409 (aarch64_internal_mov_immediate): Add new special pattern.
14410 * config/aarch64/aarch64.md (*movdi_aarch64):
14411 Add reg/32bit const mov case.
14413 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14414 Richard Sandiford <richard.sandiford@linaro.org>
14416 * config/aarch64/aarch64.md (mov<mode>): Generalize.
14417 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14418 Add integer and movi cases.
14419 (movi-split-hf-df-sf split, fp16): New.
14420 (enabled): Added TARGET_FP_F16INST.
14421 * config/aarch64/iterators.md (GPF_HF): New.
14422 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
14424 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14426 * config/aarch64/aarch64.c
14427 (aarch64_simd_container_mode): Add prototype.
14428 (aarch64_expand_mov_immediate): Add HI support.
14429 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
14430 (aarch64_can_const_movi_rtx_p): New.
14431 (aarch64_preferred_reload_class):
14432 Remove restrictions of using FP registers for certain SIMD operations.
14433 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
14434 (aarch64_valid_floating_const): Add integer move validation.
14435 (aarch64_simd_imm_scalar_p): Remove.
14436 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
14437 (aarch64_legitimate_constant_p): Expand list of supported cases.
14438 * config/aarch64/aarch64-protos.h
14439 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
14440 (aarch64_reinterpret_float_as_int): New.
14441 (aarch64_simd_imm_scalar_p): Remove.
14442 * config/aarch64/constraints.md (Uvi): New.
14443 (Dd): Split into Ds and new Dd.
14444 * config/aarch64/aarch64.md (*movsi_aarch64):
14446 (*movdi_aarch64): Add SIMD mov case.
14448 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14450 * tree-predcom.c: (struct chain): Handle store-store chain in which
14451 stores for elimination only store loop invariant values.
14452 (execute_pred_commoning_chain): Ditto.
14453 (prepare_initializers_chain_store_elim): Ditto.
14454 (prepare_finalizers): Ditto.
14455 (is_inv_store_elimination_chain): New function.
14456 (initialize_root_vars_store_elim_1): New function.
14458 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14460 * tree-predcom.c: Revise general description of the pass.
14461 (enum chain_type): New enum type for store elimination.
14462 (struct chain): New field supporting store elimination.
14463 (struct component): Ditto.
14464 (dump_chain): Dump store-stores chain.
14465 (release_chain): Release resources.
14466 (split_data_refs_to_components): Compute and create component
14467 contains only stores for elimination.
14468 (get_chain_last_ref_at): New function.
14469 (make_invariant_chain): Initialization.
14470 (make_rooted_chain): Specify chain type in parameter and record it.
14471 (add_looparound_copies): Skip for store-stores chain.
14472 (determine_roots_comp): Compute type of chain and pass it to
14474 (initialize_root_vars_store_elim_2): New function.
14475 (finalize_eliminated_stores): New function.
14476 (remove_stmt): Handle store for elimination.
14477 (execute_pred_commoning_chain): Execute predictive commoning on
14478 store-store chains.
14479 (determine_unroll_factor): Skip unroll for store-stores chain.
14480 (prepare_initializers_chain_store_elim): New function.
14481 (prepare_initializers_chain): Hanlde store-store chain.
14482 (prepare_finalizers_chain, prepare_finalizers): New function.
14483 (tree_predictive_commoning_loop): Return integer value indicating
14484 if loop is unrolled or lcssa form is corrupted.
14485 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
14487 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14489 * tree-predcom.c (initialize_root): Delete.
14490 (execute_pred_commoning_chain): Initialize root vars and replace
14491 reference of non-combined chain directly, rather than call above
14494 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14496 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
14497 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
14499 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14501 * tree-predcom.c (struct chain): New field init_seq.
14502 (release_chain): Release init_seq.
14503 (prepare_initializers_chain): Record intialization stmts in above
14505 (insert_init_seqs): New function.
14506 (tree_predictive_commoning_loop): Call insert_init_seqs.
14508 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14510 * tree-predcom.c (determine_roots_comp): Skip trivial components.
14512 2017-07-28 Richard Biener <rguenther@suse.de>
14514 * match.pd: Remove superfluous :c.
14515 * genmatch.c (simplify::id): Add member.
14516 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
14518 (current_id): New global.
14519 (dt_node::parent): Move from ...
14520 (dt_operand::parent): ... here. Add for_id member.
14521 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
14522 (decision_tree::find_node): Relax order requirement when
14523 merging DT_TRUE nodes to ones inbetween the current simplify
14524 and the one we try to merge with. Add diagnostic whenever
14525 we need to enforce pattern order by not merging.
14526 (decision_tree::insert): Set current_id.
14527 (decision_tree::print_node): Dump parent node and for_id.
14528 (parser::last_id): Add member.
14529 (parser::push_simplify): Assign unique id.
14530 (parser::parser): Initialize last_id.
14532 2017-07-28 Martin Liska <mliska@suse.cz>
14535 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
14536 gimple_build_debug_bind.
14538 2017-07-28 Richard Biener <rguenther@suse.de>
14540 PR tree-optimization/81502
14541 * match.pd: Add pattern combining BIT_INSERT_EXPR with
14543 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
14545 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
14546 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
14547 for BIT_FIELD_REF args.
14548 * fold-const.c (make_bit_field_ref): Likewise.
14549 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
14551 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14554 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
14555 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
14556 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
14557 Or it into SANITIZER_UNDEFINED.
14558 * ubsan.c: Include gimple-fold.h and varasm.h.
14559 (ubsan_expand_ptr_ifn): New function.
14560 (instrument_pointer_overflow): New function.
14561 (maybe_instrument_pointer_overflow): New function.
14562 (instrument_object_size): Formatting fix.
14563 (pass_ubsan::execute): Call instrument_pointer_overflow
14564 and maybe_instrument_pointer_overflow.
14565 * internal-fn.c (expand_UBSAN_PTR): New function.
14566 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
14567 * sanitizer.def (__ubsan_handle_pointer_overflow,
14568 __ubsan_handle_pointer_overflow_abort): New builtins.
14569 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
14570 * internal-fn.def (UBSAN_PTR): New internal function.
14571 * opts.c (sanitizer_opts): Add pointer-overflow.
14572 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
14573 * fold-const.c (build_range_check): Compute pointer range check in
14574 integral type if pointer arithmetics would be needed. Formatting
14577 2017-07-28 Martin Liska <mliska@suse.cz>
14580 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
14581 parameters that are of a variable-length.
14583 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
14585 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
14587 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
14588 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
14589 (CRT_CALL_STATIC_FUNCTION): Likewise.
14590 (ASM_DEFAULT_SPEC): New define.
14591 (ASM_SPEC32): Likewise.
14592 (ASM_SPEC64): Likewise.
14593 (ASM_SPEC_COMMON): Likewise.
14594 (ASM_SPEC): Likewise.
14595 (INVALID_64BIT): Likewise.
14596 (LINK_OS_DEFAULT_SPEC): Likewise.
14597 (LINK_OS_SPEC32): Likewise.
14598 (LINK_OS_SPEC64): Likewise.
14599 (POWERPC_LINUX): Likewise.
14600 (PTRDIFF_TYPE): Likewise.
14601 (RESTORE_FP_PREFIX): Likewise.
14602 (RESTORE_FP_SUFFIX): Likewise.
14603 (SAVE_FP_PREFIX): Likewise.
14604 (SAVE_FP_SUFFIX): Likewise.
14605 (SIZE_TYPE): Likewise.
14606 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
14607 (TARGET_64BIT): Likewise.
14608 (TARGET_64BIT): Likewise.
14609 (TARGET_AIX): Likewise.
14610 (WCHAR_TYPE_SIZE): Likewise.
14611 (WCHAR_TYPE): Undefine.
14612 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
14613 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
14614 (CPP_OS_RTEMS_SPEC): Delete.
14615 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
14616 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
14618 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
14620 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14622 PR tree-optimization/81578
14623 * tree-parloops.c (build_new_reduction): Bail out if
14624 reduction_code isn't one of the standard OpenMP reductions.
14625 Move the details printing after that decision.
14627 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14629 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
14630 related to reload_in_progress.
14631 (splat_input_operand): Likewise.
14632 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
14634 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
14636 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
14637 (TARGET_INSTANTIATE_DECLS): Likewise.
14638 (legitimate_indexed_address_p): Delete reload_in_progress code.
14639 (rs6000_debug_legitimate_address_p): Likewise.
14640 (rs6000_eliminate_indexed_memrefs): Likewise.
14641 (rs6000_emit_le_vsx_store): Likewise.
14642 (rs6000_emit_move_si_sf_subreg): Likewise.
14643 (rs6000_emit_move): Likewise.
14644 (register_to_reg_type): Likewise.
14645 (rs6000_pre_atomic_barrier): Likewise.
14646 (rs6000_machopic_legitimize_pic_address): Likewise.
14647 (rs6000_allocate_stack_temp): Likewise.
14648 (rs6000_address_for_fpconvert): Likewise.
14649 (rs6000_address_for_altivec): Likewise.
14650 (rs6000_secondary_memory_needed_rtx): Delete function.
14651 (rs6000_check_sdmode): Likewise.
14652 (rs6000_alloc_sdmode_stack_slot): Likewise.
14653 (rs6000_instantiate_decls): Likewise.
14654 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
14655 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
14656 Delete reload_in_progress.
14657 (*vec_reload_and_plus_<mptrsize>): Likewise.
14658 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
14659 (vsx_div_v2di): Likewise.
14660 (vsx_udiv_v2di): Likewise.
14662 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14664 * config/rs6000/rs6000.opt (mlra): Replace with stub.
14665 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
14666 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
14667 (rs6000_debug_reg_global): Delete print of LRA status.
14668 (rs6000_option_override_internal): Delete dead LRA related code.
14669 (rs6000_lra_p): Delete function.
14670 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
14672 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
14674 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
14675 * config/riscv/rtems.h: New file.
14677 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14678 Sudakshina Das <sudi.das@arm.com>
14680 * config/aarch64/aarch64.md
14681 (define_split for and<mode>3nr_compare): Move
14682 non aarch64_logical_operand to a register.
14683 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
14684 register immediate operand to a register.
14685 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
14687 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14689 PR middle-end/81564
14690 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
14692 2017-07-27 Richard Biener <rguenther@suse.de>
14694 PR tree-optimization/81573
14695 PR tree-optimization/81494
14696 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
14697 multi defuse cycle case.
14699 2017-07-27 Richard Biener <rguenther@suse.de>
14701 PR tree-optimization/81571
14702 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
14705 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
14707 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
14708 earlier and only if MASK_FPU is set. Adjust formatting.
14710 2017-07-27 Martin Liska <mliska@suse.cz>
14712 * opt-functions.awk: Add validation of value of Init.
14713 * optc-gen.awk: Pass new argument.
14715 2017-07-27 Martin Liska <mliska@suse.cz>
14717 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
14718 Fix wrong condition.
14720 2017-07-27 Martin Liska <mliska@suse.cz>
14722 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
14723 BBs and edges seen by autoFDO.
14725 2017-07-27 Richard Biener <rguenther@suse.de>
14727 PR tree-optimization/81502
14728 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
14729 with incompatible but same sized type.
14730 (execute_update_addresses_taken): Likewise.
14732 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
14734 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
14735 flag_tree_loop_vectorize rather than flag_tree_vectorize.
14737 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14740 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
14741 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
14742 Change s_operand to memory_operand.
14744 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
14746 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
14747 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
14748 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
14749 Emit instructions rather than returning an expression. Handle TFmode
14750 and KFmode by casting to TImode.
14751 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
14752 (rs6000_emit_le_vsx_store): Likewise.
14753 * config/rs6000/vsx.md (VSX_TI): New iterator.
14754 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
14755 (*vsx_le_undo_permute_<mode>): Likewise.
14756 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
14757 emit the split sequence.
14758 (*vsx_le_perm_store_<mode>): Likewise.
14760 2017-07-27 Jakub Jelinek <jakub@redhat.com>
14762 PR tree-optimization/81555
14763 PR tree-optimization/81556
14764 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
14765 if true, force CHANGED for the recursive invocation.
14766 (reassociate_bb): Remember original length of ops array, pass
14767 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
14769 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
14770 attributes for noipa attribute. For naked attribute use
14771 lookup_attribute first before lookup_attribute_spec.
14772 * final.c (rest_of_handle_final): Disable IPA RA for functions with
14774 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
14775 for functions with noipa attribute.
14776 (cgraph_externally_visible_p): Return true for functions with noipa
14778 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
14779 for functions with noipa attribute.
14780 * doc/extend.texi: Document noipa function attribute.
14781 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
14782 also for functions with noipa attribute.
14783 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
14785 2017-07-26 Andrew Pinski <apinski@cavium.com>
14787 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
14788 vec_unalign_load_cost and vec_unalign_store_cost.
14790 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14792 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
14793 -mvsx-small-integer option.
14794 (ISA_3_0_MASKS_IEEE): Likewise.
14795 (OTHER_VSX_VECTOR_MASKS): Likewise.
14796 (POWERPC_MASKS): Likewise.
14797 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
14798 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
14799 code, only testing for DImode being allowed in non-VSX floating
14801 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
14802 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
14804 (rs6000_option_override_internal): Delete -mvsx-small-integer.
14805 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
14806 TARGET_P8_VECTOR test.
14807 (rs6000_secondary_reload_simple_move): Likewise.
14808 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
14809 since TARGET_P9_VECTOR was already tested.
14810 (rs6000_opt_masks): Remove -mvsx-small-integer.
14811 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
14812 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14814 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
14815 test for TARGET_VEXTRACTUB was used, and that uses
14817 (p9 extract splitter): Likewise.
14818 (vsx_extract_<mode>_di_p9): Likewise.
14819 (vsx_extract_<mode>_store_p9): Likewise.
14820 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
14821 for TARGET_P9_VECTOR was used. Delete code that is now dead with
14822 the elimination of TARGET_VSX_SMALL_INTEGER.
14823 (vsx_extract_<mode>_p8): Likewise.
14824 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
14825 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
14826 (vsx_set_<mode>_p9): Likewise.
14827 (vsx_set_v4sf_p9): Likewise.
14828 (vsx_set_v4sf_p9_zero): Likewise.
14829 (vsx_insert_extract_v4sf_p9): Likewise.
14830 (vsx_insert_extract_v4sf_p9_2): Likewise.
14831 * config/rs6000/rs6000.md (sign extend splitter): Change
14832 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
14833 (floatsi<mode>2_lfiwax_mem): Likewise.
14834 (floatunssi<mode>2_lfiwzx_mem): Likewise.
14835 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
14836 since a test for TARGET_P9_VECTOR was used.
14837 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14838 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
14839 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14840 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
14841 TARGET_P8_VECTOR test.
14842 (fix_trunc<mode>si2_stfiwx): Likewise.
14843 (fix_trunc<mode>si2_internal): Likewise.
14844 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
14845 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14847 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
14848 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
14849 TARGET_P8_VECTOR test.
14850 (fixuns_trunc<mode>si2_stfiwx): Likewise.
14851 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
14852 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14854 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
14855 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
14856 since a test for TARGET_P9_VECTOR was used.
14857 (splitter for loading small constants): Likewise.
14859 2017-07-26 Andrew Pinski <apinski@cavium.com>
14861 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
14864 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
14867 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
14868 (fp_valid_at): Likewise.
14870 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
14872 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
14873 (qdf24xx_addrcost_table): Likewise.
14874 (cortexa57_tunings): Update to use generic_branch_cost.
14875 (cortexa72_tunings): Likewise.
14876 (cortexa73_tunings): Likewise.
14877 (qdf24xx_tunings): Likewise.
14879 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
14881 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
14882 (thunderx2t99_branch_cost): Likewise.
14883 (cortexa35_tunings): Update to use generic_branch_cost.
14884 (cortexa53_tunings): Likewise.
14885 (cortexa57_tunings): Likewise.
14886 (cortexa72_tunings): Likewise.
14887 (cortexa73_tunings): Likewise.
14888 (thunderx2t99_tunings): Likewise.
14890 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
14892 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
14893 (sparc_option_override): Honour MASK_FSMULD.
14894 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
14895 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
14896 * config/sparc/sparc.opt (mfsmuld): New option.
14897 * doc/invoke.texi (mfsmuld): Document option.
14899 2017-07-26 Marek Polacek <polacek@redhat.com>
14901 PR middle-end/70992
14902 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
14904 2017-07-26 Richard Biener <rguenther@suse.de>
14906 * gimple-match-head.c (do_valueize): Return OP if valueize
14908 (get_def): New helper to get at the def stmt of a SSA name
14909 if valueize allows.
14910 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
14911 do_valueize to get at the def stmt.
14912 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
14914 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
14916 PR middle-end/46932
14917 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
14919 2017-07-26 Martin Liska <mliska@suse.cz>
14922 * function.c (expand_function_start): Make expansion of
14923 nonlocal_goto_save_area after parm_birth_insn.
14925 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
14927 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
14928 from all CPU target flags enable members.
14930 2017-07-26 Richard Biener <rguenther@suse.de>
14932 * genmatch.c (dt_simplify::gen): Make iterator vars const.
14933 (decision_tree::gen): Make 'type' const.
14934 (write_predicate): Likewise.
14936 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
14938 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
14939 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
14940 (rs6000_option_override_internal): Likewise.
14941 (rs6000_expand_vector_set): Likewise.
14942 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
14943 (TARGET_UPPER_REGS_SF): Likewise.
14944 (TARGET_UPPER_REGS_DI): Likewise.
14945 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
14946 (TARGET_DIRECT_MOVE_64BIT): Likewise.
14947 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
14948 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14949 (Splitters for DI constants in Altivec registers): Likewise.
14950 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
14951 (vsx_set_v4sf_p9): Likewise.
14952 (vsx_set_v4sf_p9_zero): Likewise.
14953 (vsx_insert_extract_v4sf_p9): Likewise.
14954 (vsx_insert_extract_v4sf_p9_2): Likewise.
14956 2017-07-25 Carl Love <cel@us.ibm.com>
14958 * doc/extend.texi: Update the built-in documentation file for the
14959 existing built-in functions
14960 vector signed char vec_cnttz (vector signed char);
14961 vector unsigned char vec_cnttz (vector unsigned char);
14962 vector signed short vec_cnttz (vector signed short);
14963 vector unsigned short vec_cnttz (vector unsigned short);
14964 vector signed int vec_cnttz (vector signed int);
14965 vector unsigned int vec_cnttz (vector unsigned int);
14966 vector signed long long vec_cnttz (vector signed long long);
14967 vector unsigned long long vec_cnttz (vector unsigned long long);
14969 2017-07-25 Andrew Pinski <apinski@cavium.com>
14971 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
14972 accesses where the use is for the first operand of a BIT_INSERT.
14974 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
14977 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
14978 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
14980 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
14982 * config/i386/gstabs.h: Delete.
14983 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
14985 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
14987 * config/i386/i386.c (ix86_decompose_address): Do not check for
14988 register RTX when looking at index_reg or base_reg.
14989 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
14991 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
14993 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
14994 to update EH info here.
14996 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
14998 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
15000 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15002 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
15004 2017-07-25 Torsten Duwe <duwe@suse.de>
15006 * common.opt: Introduce -fpatchable-function-entry
15007 command line option, and its variables function_entry_patch_area_size
15008 and function_entry_patch_area_start.
15009 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
15010 including a two-value parser.
15011 * target.def (print_patchable_function_entry): New target hook.
15012 * targhooks.h (default_print_patchable_function_entry): New function.
15013 * targhooks.c (default_print_patchable_function_entry): Likewise.
15014 * toplev.c (process_options): Switch off IPA-RA if
15015 patchable function entries are being generated.
15016 * varasm.c (assemble_start_function): Look at the
15017 patchable-function-entry command line switch and current
15018 function attributes and maybe generate NOP instructions by
15019 calling the print_patchable_function_entry hook.
15020 * doc/extend.texi: Document patchable_function_entry attribute.
15021 * doc/invoke.texi: Document -fpatchable_function_entry
15022 command line option.
15023 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
15025 * doc/tm.texi: Re-generate.
15027 2017-07-25 Jakub Jelinek <jakub@redhat.com>
15030 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
15031 TARGET_AVX512DQ rather than TARGET_AVX512BW.
15033 2017-07-25 Tamar Christina <tamar.christina@arm.com>
15035 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
15037 2017-07-25 Richard Biener <rguenther@suse.de>
15039 PR tree-optimization/81455
15040 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
15041 not walk in cycles when looking for guards.
15043 2017-07-25 Richard Biener <rguenther@suse.de>
15045 PR tree-optimization/81529
15046 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
15047 when optimizing backedge uses.
15049 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
15051 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
15053 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
15054 to dl_section_ref. On AIX, append an expression to subtract
15055 the size of the section length to dl_section_ref.
15057 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
15059 * configure.ac: If any of the config.* scripts fail, exit 1.
15060 * configure: Regenerate.
15062 2017-07-25 Richard Biener <rguenther@suse.de>
15064 PR middle-end/81546
15065 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
15066 of immediate uses, be more verbose on errors.
15068 2017-07-25 Richard Biener <rguenther@suse.de>
15070 PR tree-optimization/81510
15071 * tree-vect-loop.c (vect_is_simple_reduction): When the
15072 reduction stmt is not inside the loop bail out.
15074 2017-07-25 Richard Biener <rguenther@suse.de>
15076 PR tree-optimization/81303
15077 * tree-vect-loop-manip.c (vect_loop_versioning): Build
15078 profitability check against LOOP_VINFO_NITERSM1.
15080 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15082 * domwalk.c (cmp_bb_postorder): Simplify.
15083 (sort_bbs_postorder): New function. Use it...
15084 (dom_walker::walk): ...here to optimize common cases.
15086 2017-07-25 Martin Liska <mliska@suse.cz>
15089 * ipa-visibility.c (function_and_variable_visibility): Make the
15090 redirection just on target that supports aliasing.
15091 Fix GNU coding style.
15093 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
15096 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
15098 * config/arm/rtems.h: Likewise.
15099 * config/bfin/rtems.h: Likewise.
15100 * config/i386/rtemself.h: Likewise.
15101 * config/lm32/rtems.h: Likewise.
15102 * config/m32c/rtems.h: Likewise.
15103 * config/m68k/rtemself.h: Likewise.
15104 * config/microblaze/rtems.h: Likewise.
15105 * config/mips/rtems.h: Likewise.
15106 * config/moxie/rtems.h: Likewise.
15107 * config/nios2/rtems.h: Likewise.
15108 * config/powerpcspe/rtems.h: Likewise.
15109 * config/rs6000/rtems.h: Likewise.
15110 * config/rtems.h: Likewise.
15111 * config/sh/rtems.h: Likewise.
15112 * config/sh/rtemself.h: Likewise.
15113 * config/sparc/rtemself.h: Likewise.
15115 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15118 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
15119 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
15120 * tree-ssa-structalias.c (alias_get_name): Same.
15122 2017-07-25 Bin Cheng <bin.cheng@arm.com>
15125 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
15126 instructions if no du chain is found.
15128 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15130 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
15132 2017-07-25 Richard Biener <rguenther@suse.de>
15134 PR middle-end/81505
15135 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
15138 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
15140 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
15141 upper-regs options.
15142 (ISA_2_7_MASKS_SERVER): Likewise.
15143 (ISA_3_0_MASKS_IEEE): Likewise.
15144 (OTHER_P8_VECTOR_MASKS): Likewise.
15145 (OTHER_VSX_VECTOR_MASKS): Likewise.
15146 (POWERPC_MASKS): Likewise.
15147 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
15148 duplicate list of options.
15149 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
15150 explicit -mupper-regs options.
15151 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
15152 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
15153 alias for -mupper-regs-df.
15154 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
15155 (rs6000_init_hard_regno_mode_ok): Likewise.
15156 (rs6000_option_override_internal): Likewise.
15157 (rs6000_opt_masks): Likewise.
15158 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
15159 options in terms of whether -mvsx or -mpower8-vector was used.
15160 (TARGET_UPPER_REGS_DI): Likewise.
15161 (TARGET_UPPER_REGS_SF): Likewise.
15162 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
15163 -mupper-regs-* options.
15165 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15167 * passes.c (emergency_dump_function): Print some empty lines and a
15168 header before the RTL dump.
15170 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15172 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
15174 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
15177 * config/aarch64/aarch64.c (aarch64_classify_symbol):
15178 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
15180 2017-07-24 Carl Love <cel@us.ibm.com>
15182 * config/rs6000/rs6000-c.c: Add support for built-in functions
15183 vector float vec_extract_fp32_from_shorth (vector unsigned short);
15184 vector float vec_extract_fp32_from_shortl (vector unsigned short);
15185 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
15186 vec_extract_fp_from_shortl): Add defines for the two builtins.
15187 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
15188 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
15190 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
15191 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
15192 * doc/extend.texi: Update the built-in documentation file for the
15193 new built-in function.
15195 2017-07-24 Jakub Jelinek <jakub@redhat.com>
15198 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
15200 * doc/generic.texi: Likewise.
15201 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
15202 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
15204 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
15206 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
15207 (aarch64_mls_elt_merge<mode>): Likewise.
15209 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
15211 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
15212 having __cxa_atexit.
15214 2017-07-23 Michael Collison <michael.collison@arm.com>
15216 * config/arm/arm.c (arm_option_override): Deprecate
15217 use of -mstructure-size-boundary.
15218 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
15219 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
15221 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15224 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
15225 Reduce cost estimate for direct moves.
15227 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
15230 * config/i386/i386.c (ix86_option_override_internal): Disable
15231 BMI, BMI2 and TBM instructions for -m16.
15233 2017-07-21 Carl Love <cel@us.ibm.com>
15235 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15236 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15237 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15238 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15239 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15240 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15241 VMULOSW): New enum "unspec" values.
15242 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15243 altivec_vmulosw): New patterns.
15244 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15245 VMULOSW): Add definitions.
15247 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
15249 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
15250 (qdf24xx): Likewise.
15251 * config/aarch64/aarch64-options-extensions.def (rdma); New.
15252 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
15253 (AARCH64_FL_V8_1): Renumber.
15254 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
15255 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
15256 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
15257 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
15258 rdma to feature modifiers list.
15260 2017-07-21 Yury Gribov <tetra2005@gmail.com>
15262 PR middle-end/56727
15263 * ipa-visibility (function_and_variable_visibility): Convert
15264 recursive PLT call to direct call if appropriate.
15266 2017-07-21 Andrew Pinski <apinski@cavium.com>
15268 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
15269 operand 1 to see if the types precision matches.
15270 * fold-const.c (operand_equal_p): Likewise.
15272 2017-07-21 Richard Biener <rguenther@suse.de>
15274 PR tree-optimization/81303
15275 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
15276 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
15277 (vect_peeling_hash_get_lowest_cost): Adjust.
15278 (vect_enhance_data_refs_alignment): Likewise. Use
15279 vect_get_peeling_costs_all_drs to compute the penalty for no
15280 peeling to match up costs.
15282 2017-07-21 Richard Biener <rguenther@suse.de>
15284 PR tree-optimization/81500
15285 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
15286 we didn't identify a reduction path.
15288 2017-07-21 Tom de Vries <tom@codesourcery.com>
15289 Cesar Philippidis <cesar@codesourcery.com>
15291 PR gcov-profile/81442
15292 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
15295 2017-07-21 Tom de Vries <tom@codesourcery.com>
15298 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
15300 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
15301 nvptx_override_options_after_change.
15303 2017-07-21 Ulrich Drepper <drepper@redhat.com>
15305 * dwarf2out.c (output_file_names): Avoid double testing for
15306 dwarf_version >= 5.
15308 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
15310 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
15312 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15314 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
15316 (try_crossjump_to_edge): Do not punt on partitioned functions.
15318 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15320 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
15321 Put all BBs reachable only via paths crossing cold region to cold
15323 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
15325 2016-07-21 Richard Biener <rguenther@suse.de>
15327 PR tree-optimization/81303
15328 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
15329 into account prologue and epilogue iterations when raising
15330 min_profitable_iters to sth at least covering one vector iteration.
15332 2017-07-21 Tamar Christina <tamar.christina@arm.com>
15334 * config/arm/arm.c (arm_test_cpu_arch_dat):
15337 2017-07-20 Nathan Sidwell <nathan@acm.org>
15339 Remove TYPE_METHODS.
15340 * tree.h (TYPE_METHODS): Delete.
15341 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
15342 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
15343 (dbxout_type_methods): Scan TYPE_FIELDS.
15344 (dbxout_type): Don't check TYPE_METHODS here.
15345 * function.c (use_register_for_decl): Always ignore register for
15346 class types when not optimizing.
15347 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
15348 * tree.c (free_lang_data_in_type): Stitch out member functions and
15349 templates from TYPE_FIELDS.
15350 (build_distinct_type_copy, verify_type_variant,
15351 verify_type): Member fns are on TYPE_FIELDS.
15352 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
15353 * tree-pretty-print.c (dump_generic_node): Likewise.
15355 2017-07-20 Jakub Jelinek <jakub@redhat.com>
15358 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
15359 V2TImode and V4TImode.
15360 (ix86_expand_vector_extract): Likewise.
15361 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
15362 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
15363 (ssescalarmode): Handle V4TImode and V2TImode.
15364 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
15365 (*vec_extractv2ti, *vec_extractv4ti): New insns.
15366 (VEXTRACTI128_MODE): New mode iterator.
15367 (splitter for *vec_extractv?ti first element): New.
15368 (VEC_INIT_MODE): New mode iterator.
15369 (vec_init<mode>): Consolidate 3 expanders into one using
15370 VEC_INIT_MODE mode iterator.
15372 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
15374 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
15375 non_spilled_static_chain_regno_p.
15377 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
15379 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
15381 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
15383 * bb-reorder.c (connect_traces): Allow copying of blocks within
15386 2017-07-20 Richard Biener <rguenther@suse.de>
15388 * gimple.h (gimple_phi_result): Add gphi * overload.
15389 (gimple_phi_result_ptr): Likewise.
15390 (gimple_phi_arg): Likewise. Adjust index assert to only
15391 allow actual argument accesses rather than all slots available
15393 (gimple_phi_arg_def): Add gphi * overload.
15394 * tree-phinodes.c (make_phi_node): Initialize only actual
15396 (resize_phi_node): Clear memory not covered by old node,
15397 do not initialize excess argument slots.
15398 (reserve_phi_args_for_new_edge): Initialize new argument slot
15401 2017-07-20 Bin Cheng <bin.cheng@arm.com>
15403 PR tree-optimization/81388
15405 2016-07-21 Bin Cheng <bin.cheng@arm.com>
15407 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
15408 by removing computation of may_be_zero.
15410 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15411 Tom de Vries <tom@codesourcery.com>
15413 PR middle-end/81030
15414 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
15415 when gimple level profile disagrees with what RTL expander did.
15417 2017-07-20 Richard Biener <rguenther@suse.de>
15419 PR tree-optimization/61171
15420 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
15421 (vect_analyze_stmt): Add slp instance parameter.
15422 (vectorizable_reduction): Likewise.
15423 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
15424 (vect_is_simple_reduction): Deal with chains not detected
15425 as SLP reduction chain, specifically not properly associated
15426 chains containing a mix of plus/minus.
15427 (get_reduction_op): Remove.
15428 (get_initial_defs_for_reduction): Simplify, pass in whether
15429 this is a reduction chain, pass in the SLP node for the PHIs.
15430 (vect_create_epilog_for_reduction): Get the SLP instance as
15432 (vectorizable_reduction): Get the SLP instance as arg.
15433 During analysis remember the SLP node with the PHIs in the
15434 instance. Simplify getting at the vectorized reduction PHIs.
15435 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
15436 through SLP instance.
15437 (vect_slp_analyze_operations): Likewise.
15438 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
15439 (vect_transform_stmt): Likewise.
15441 2017-07-20 Tom de Vries <tom@codesourcery.com>
15443 PR tree-optimization/81489
15444 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
15445 read of phi arg location to before loop that modifies phi.
15447 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15449 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
15452 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15454 PR middle-end/81331
15455 * except.c (execute): Fix ordering issue.
15457 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
15459 PR rtl-optimization/81423
15460 * combine.c (make_compound_operation_int): Don't try to optimize
15461 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
15463 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
15465 PR rtl-optimization/81423
15466 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
15467 with a constant that is -1 in the truncated to mode.
15469 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15471 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
15472 (determine_unlikely_bbs): ... here.
15473 * predict.h (propagate_unlikely_bbs_forward): Declare.
15474 * cfgexpand.c (pass_expand::execute): Use it.
15475 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
15477 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
15478 propagate_unlikely_bbs_forward.
15480 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15482 PR middle-end/81331
15483 * except.c (maybe_add_nop_after_section_switch): New function.
15486 2017-07-19 Tom de Vries <tom@codesourcery.com>
15488 * gimple.h (gimple_phi_set_arg): Make assert more strict.
15490 2017-07-19 Tom de Vries <tom@codesourcery.com>
15492 * gimple.h (gimple_phi_arg): Make assert more strict.
15494 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
15496 * config.gcc (powerpc*-*-*): Add mmintrin.h.
15497 * config/rs6000/mmintrin.h: New file.
15498 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
15500 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15502 PR tree-optimization/81346
15503 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
15505 2017-07-19 Tom de Vries <tom@codesourcery.com>
15507 * config/nvptx/nvptx.md (VECIM): Add V2DI.
15509 2017-07-19 Tom de Vries <tom@codesourcery.com>
15511 * config/nvptx/nvptx-modes.def: Add V2DImode.
15512 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
15513 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
15514 (nvptx_output_mov_insn): Handle lack of mov.b128.
15515 (nvptx_print_operand): Handle 'H' and 'L' codes.
15516 (nvptx_vector_mode_supported): Allow V2DImode.
15517 (nvptx_preferred_simd_mode): New function.
15518 (nvptx_data_alignment): New function.
15519 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
15520 nvptx_preferred_simd_mode.
15521 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
15523 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
15525 2017-07-19 Tom de Vries <tom@codesourcery.com>
15527 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
15528 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
15529 (nvptx_vector_mode_supported): New function. Allow V2SImode.
15530 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
15531 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
15532 (mov<VECIM>_insn): New define_insn.
15533 (define_expand "mov<VECIM>): New define_expand.
15535 2017-07-19 Tom de Vries <tom@codesourcery.com>
15537 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
15539 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15541 PR tree-optimization/81346
15542 * fold-const.h (fold_div_compare, range_check_type): Declare.
15543 * fold-const.c (range_check_type): New function.
15544 (build_range_check): Use range_check_type.
15545 (fold_div_compare): No longer static, rewritten into
15546 a match.pd helper function.
15547 (fold_comparison): Don't call fold_div_compare here.
15548 * match.pd (X / C1 op C2): New optimization using fold_div_compare
15549 as helper function.
15551 2017-07-19 Nathan Sidwell <nathan@acm.org>
15553 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
15554 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
15555 * tree.c (find_decls_types_r, verify_type): Use
15556 TYPE_{MIN,MAX}_VALUE_RAW.
15557 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
15558 (hash_tree): Likewise.
15559 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
15561 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
15564 2017-07-18 Tom de Vries <tom@codesourcery.com>
15566 PR middle-end/81464
15567 * omp-expand.c (expand_omp_for_static_chunk): Handle
15568 equal-argument loop exit phi.
15570 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
15573 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
15574 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
15575 operand 2 predicate.
15576 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
15577 operand 2 predicate.
15578 (ror,rol -> rorx splitters): Use const_int_operand as
15579 operand 2 predicate.
15581 2017-06-18 Richard Biener <rguenther@suse.de>
15583 PR tree-optimization/81410
15584 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
15585 the gap in the ! slp_perm SLP case after each group.
15587 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15589 PR middle-end/81463
15590 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
15593 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15595 PR middle-end/81462
15596 * predict.c (set_even_probabilities): Cleanup; do not affect
15597 probabilities that are already known.
15598 (combine_predictions_for_bb): Call even when count is set.
15600 2017-07-18 Nathan Sidwell <nathan@acm.org>
15602 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
15605 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15608 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
15609 optimization for loop niter analysis.
15611 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
15614 * config/avr/avr.c (avr_optimize_casesi): Don't use
15615 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
15617 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
15619 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
15620 body_cost_vec from _vect_peel_extended_info.
15621 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
15622 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
15625 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15627 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
15629 2017-07-18 Richard Biener <rguenther@suse.de>
15631 PR tree-optimization/80620
15632 PR tree-optimization/81403
15633 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
15634 info when re-using a VN table entry.
15636 2017-07-18 Richard Biener <rguenther@suse.de>
15638 PR tree-optimization/81418
15639 * tree-vect-loop.c (vectorizable_reduction): Properly compute
15640 vectype_in. Verify that with lane-reducing reduction operations
15641 we have a single def-use cycle.
15643 2017-07-17 Carl Love <cel@us.ibm.com>
15645 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
15647 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15648 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15649 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15650 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15651 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15652 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15653 VMULOSW): New enum "unspec" values.
15654 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
15655 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
15656 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15657 altivec_vmulosw): New patterns.
15658 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15659 VMULOSW): Add definitions.
15660 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
15662 * config/alpha/alpha.c: Include predict.h.
15664 2017-07-17 Yury Gribov <tetra2005@gmail.com>
15666 * tree-vrp.c (compare_assert_loc): Fix comparison function
15667 to return predictable results.
15669 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15671 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
15673 (subdi3): Likewise.
15674 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
15675 * doc/invoke.texi (mexpand-adddi): Update text.
15677 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15679 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
15680 that also clobbers the CC register. The old expand code is moved
15682 (*arc_clzsi2): ... here.
15683 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
15684 the CC register. The old expand code is moved to ...
15685 (arc_ctzsi2): ... here.
15687 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15689 * config/arc/arc.opt (mindexed-loads): Use initial value
15690 TARGET_INDEXED_LOADS_DEFAULT.
15691 (mauto-modify-reg): Use initial value
15692 TARGET_AUTO_MODIFY_REG_DEFAULT.
15693 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
15694 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
15695 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
15696 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
15698 2017-07-17 Martin Liska <mliska@suse.cz>
15701 * opts.c (finish_options): Do not allow -fgnu-tm
15702 w/ -fsanitize={kernel-,}address. Say sorry.
15704 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15707 * tree-loop-distribution.c (classify_partition): Only assert on
15708 numer of iterations.
15709 (merge_dep_scc_partitions): Delete prameter. Update function call.
15710 (distribute_loop): Remove code handling loop with unknown niters.
15711 (pass_loop_distribution::execute): Skip loop with unknown niters.
15713 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15716 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
15717 function sort_partitions_by_post_order.
15719 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15721 PR tree-optimization/81374
15722 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
15723 the max index of basic blocks, rather than number of basic blocks.
15725 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15727 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
15729 (arc_legitimate_pic_operand_p): Likewise.
15730 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
15732 (arc_needs_pcl_p): Likewise.
15733 (arc_legitimate_pc_offset_p): Likewise.
15734 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
15735 function is also used in constrains.md.
15736 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
15737 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
15738 PLUS. Only return true/false in known cases, otherwise assert.
15739 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
15740 is already called in arc_legitimate_constant_p.
15741 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
15743 (LEGITIMATE_PIC_OPERAND_P): Use
15744 arc_raw_symbolic_reference_mentioned_p function.
15745 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
15750 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15751 Andrew Burgess <andrew.burgess@embecosm.com>
15753 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
15754 (arc_return_address_register): New function.
15755 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
15756 (arc_handle_fndecl_attribute): Add naked attribute.
15757 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
15758 (TARGET_WARN_FUNC_RETURN): Likewise.
15759 (arc_allocate_stack_slots_for_args): New function.
15760 (arc_warn_func_return): Likewise.
15761 (machine_function): Change type fn_type.
15762 (arc_compute_function_type): Consider new naked function type,
15763 change function return type.
15764 (arc_must_save_register): Adapt to handle new
15765 arc_compute_function_type's return type.
15766 (arc_expand_prologue): Likewise.
15767 (arc_expand_epilogue): Likewise.
15768 (arc_return_address_regs): Delete.
15769 (arc_return_address_register): New function.
15770 (arc_epilogue_uses): Use above function.
15771 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
15772 (arc_function_type): Change encoding, add naked type.
15773 (ARC_INTERRUPT_P): Change to handle the new encoding.
15774 (ARC_FAST_INTERRUPT_P): Likewise.
15775 (ARC_NORMAL_P): Define.
15776 (ARC_NAKED_P): Likewise.
15777 (arc_compute_function_type): Delete prototype.
15778 * config/arc/arc.md (in_ret_delay_slot): Use
15779 arc_return_address_register function.
15780 (simple_return): Likewise.
15781 (p_return_i): Likewise.
15783 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15785 PR tree-optimization/81428
15786 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
15787 can't be built for those types.
15789 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
15791 Remove stuff dead since r239246.
15793 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
15794 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
15795 (avr_inform_devices): Remove dead stuff.
15797 2017-07-17 Tamar Christina <tamar.christina@arm.com>
15799 * config/arm/arm_neon.h: Fix softp typo.
15801 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15803 PR tree-optimization/81365
15804 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
15805 aggregate moves onto bb predecessor edges, make sure there are no
15806 loads that could alias the lhs in between the start of bb and the
15809 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
15812 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
15813 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
15814 [LSHIFTRT, outer_code = TRUNCATE]: Same.
15816 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15818 PR tree-optimization/81396
15819 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
15820 (init_symbolic_number): Initialize it to 1.
15821 (perform_symbolic_merge): Add n_ops from both operands into the new
15823 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
15824 without base_addr as useless if they need more than one operation.
15825 (bswap_replace): Handle !bswap case for NULL base_addr.
15827 2017-07-17 Tom de Vries <tom@codesourcery.com>
15830 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
15833 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
15835 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
15836 conditional builtin define __FIX_LEON3FT_B2BST.
15838 2017-07-17 Daniel Cederman <cederman@gaisler.com>
15840 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
15841 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
15844 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
15846 PR rtl-optimization/81424
15847 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
15848 to remove potential trapping from operands if -fnon-call-exceptions.
15850 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15852 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
15853 profile_proability for scalling.
15854 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
15856 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15858 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
15860 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15862 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
15863 fixpoint arithmetics.
15865 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15867 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
15868 fixpoint arithmetics.
15870 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15872 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
15873 fixpoint arithmetics.
15875 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15877 * profile-count.h (profile_probability::from_reg_br_prob_note,
15878 profile_probability::to_reg_br_prob_note): New functions.
15879 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
15880 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
15881 * predict.c (probability_reliable_p): Update.
15882 (edge_probability_reliable_p): Update.
15883 (br_prob_note_reliable_p): Update.
15884 (invert_br_probabilities): Update.
15885 (add_reg_br_prob_note): New function.
15886 (combine_predictions_for_insn): Update.
15887 * asan.c (asan_clear_shadow): Update.
15888 * cfgbuild.c (compute_outgoing_frequencies): Update.
15889 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
15890 (update_br_prob_note): Update.
15891 (rtl_verify_edges): Update.
15892 (purge_dead_edges): Update.
15893 (fixup_reorder_chain): Update.
15894 * emit-rtl.c (try_split): Update.
15895 * ifcvt.c (cond_exec_process_insns): Update.
15896 (cond_exec_process_if_block): Update.
15897 (dead_or_predicable): Update.
15898 * internal-fn.c (expand_addsub_overflow): Update.
15899 (expand_neg_overflow): Update.
15900 (expand_mul_overflow): Update.
15901 * loop-doloop.c (doloop_modify): Update.
15902 * loop-unroll.c (compare_and_jump_seq): Update.
15903 * optabs.c (emit_cmp_and_jump_insn_1): Update.
15904 * predict.h: Update.
15905 * reorg.c (mostly_true_jump): Update.
15907 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
15908 * config/alpha/alpha.c (emit_unlikely_jump): Update.
15909 * config/arc/arc.c: (emit_unlikely_jump): Update.
15910 * config/arm/arm.c: (emit_unlikely_jump): Update.
15911 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
15912 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
15913 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
15914 (ix86_print_operand): Update.
15915 (ix86_split_fp_branch): Update.
15916 (predict_jump): Update.
15917 * config/ia64/ia64.c (ia64_print_operand): Update.
15918 * config/mmix/mmix.c (mmix_print_operand): Update.
15919 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
15920 (rs6000_expand_split_stack_prologue): Update.
15921 * config/rs6000/rs6000.c: Update.
15922 * config/s390/s390.c (s390_expand_vec_strlen): Update.
15923 (s390_expand_vec_movstr): Update.
15924 (s390_expand_cs_tdsi): Update.
15925 (s390_expand_split_stack_prologue): Update.
15926 * config/sh/sh.c (sh_print_operand): Update.
15927 (expand_cbranchsi4): Update.
15928 (expand_cbranchdi4): Update.
15929 * config/sparc/sparc.c (output_v9branch): Update.
15930 * config/spu/spu.c (get_branch_target): Update.
15931 (ea_load_store_inline): Update.
15932 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
15933 * config/tilepro/tilepro.c: Update.
15935 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
15937 * gimplify.c (mostly_copy_tree_r): Revert latest change.
15938 (gimplify_save_expr): Likewise.
15940 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15942 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
15944 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15946 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
15948 * timevar.def (TV_IPA_FNSUMMARY): Define.
15950 2017-07-16 Daniel Cederman <cederman@gaisler.com>
15952 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
15953 to back store errata sensitive sequence from being generated.
15954 (sqrtdf2_fix): Likewise.
15956 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15958 * tree-ssa-threadupdate.c (compute_path_counts,
15959 update_joiner_offpath_counts): Use profile_probability.
15961 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
15964 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15966 * config/arm/arm-c.c (arm_cpu_builtins): Define
15967 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
15969 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
15971 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15972 array entries to represent __ieee128 versions of the
15973 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
15974 scalar_extract_sig, and scalar_insert_exp built-in functions.
15975 (altivec_resolve_overloaded_builtin): Add special case handling
15976 for the __builtin_scalar_insert_exp function, as represented by
15977 the P9V_BUILTIN_VEC_VSIEDP constant.
15978 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
15979 exponent support for __ieee128 argument.
15980 (VSESQP): Add scalar extract signature support for __ieee128
15982 (VSTDCNQP): Add scalar test negative support for __ieee128
15984 (VSIEQP): Add scalar insert exponent support for __int128 argument
15985 with __ieee128 result.
15986 (VSIEQPF): Add scalar insert exponent support for __ieee128
15987 argument with __ieee128 result.
15988 (VSTDCQP): Add scalar test data class support for __ieee128
15990 (VSTDCNQP): Add overload support for scalar test negative with
15991 __ieee128 argument.
15992 (VSTDCQP): Add overload support for scalar test data class
15993 __ieee128 argument.
15994 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
15995 UNSPEC_VSX_SXSIGDP.
15996 (UNSPEC_VSX_SIEXPQP): New constant.
15997 (xsxexpqp): New insn for VSX scalar extract exponent quad
15999 (xsxsigqp): New insn for VSX scalar extract significand quad
16001 (xsiexpqpf): New insn for VSX scalar insert exponent quad
16002 precision with floating point argument.
16003 (xststdcqp): New expand for VSX scalar test data class quad
16005 (xststdcnegqp): New expand for VSX scalar test negative quad
16007 (xststdcqp): New insn to match expansions for VSX scalar test data
16008 class quad precision and VSX scalar test negative quad precision.
16009 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
16010 special case operand checking to enforce that second operand of
16011 VSX scalar test data class with quad precision argument is a 7-bit
16013 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
16014 prototypes and descriptions of __ieee128 versions of
16015 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
16016 scalar_test_data_class, and scalar_test_neg built-in functions.
16018 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16020 PR tree-optimization/81162
16021 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
16022 replace a negate with an add.
16024 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
16026 * doc/invoke.texi (arm/-mcpu): Document +crypto.
16028 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16030 * config/arm/arm-c.c (arm_cpu_builtins): Define
16031 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
16033 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16035 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
16036 (armv8-r): Set ARM Cortex-R52 as default CPU.
16037 * config/arm/arm-tables.opt: Regenerate.
16038 * config/arm/arm-tune.md: Regenerate.
16039 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
16041 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
16042 extension for -mcpu=cortex-r52.
16044 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16046 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
16047 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
16048 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
16049 (fp-armv8): Define it as FP_ARMv8 only.
16050 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
16051 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
16053 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
16054 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
16055 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
16056 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
16057 than TARGET_FPU_ARMV8.
16058 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
16059 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
16060 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
16062 * config/arm/neon.md (neon_vrint): Likewise.
16063 (neon_vcvt): Likewise.
16064 (neon_<fmaxmin_op><mode>): Likewise.
16065 (<fmaxmin><mode>3): Likewise.
16066 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
16067 * config/arm/predicates.md (arm_cond_move_operator): Check against
16068 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
16070 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
16072 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
16073 to top of function.
16075 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16077 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
16078 loop in comment with memset.
16080 2017-07-14 Martin Liska <mliska@suse.cz>
16082 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
16083 * dwarf2out.c (is_java): Remove the function.
16084 (output_pubname): Remove usage of the function.
16085 (lower_bound_default): Remove usage of DW_LANG_Java.
16086 (gen_compile_unit_die): Likewise.
16087 * gcc.c: Remove compiler defaults for .java and .zip files.
16088 * gimple-expr.c (remove_suffix): Change as there's no longer
16089 extension than 4-letter one.
16090 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
16091 (gimplify_save_expr): Likewise.
16092 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
16093 as it's possible even for other languages than Java.
16094 * langhooks.h (struct lang_hooks): Remove Java from a comment.
16095 * lto-opts.c (lto_write_options): Remove reference to Java.
16096 * opts.c (strip_off_ending): Update file extension handling.
16097 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
16098 * tree-eh.c (lower_resx): Likewise.
16099 * tree.c (free_lang_data_in_type): Remove dead code.
16100 (find_decls_types_r): Likewise.
16101 (build_common_builtin_nodes): Remove Java from a comment.
16102 (verify_type): Remove dead code.
16103 * varasm.c (assemble_external): Remove Java from a comment.
16105 2017-07-14 Martin Liska <mliska@suse.cz>
16107 * opts.c (finish_options): Add quotes.
16108 (common_handle_option): Likewise.
16110 2017-07-14 Martin Liska <mliska@suse.cz>
16112 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
16113 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
16114 Remove N_SO_PASCAL.
16115 * dwarf2out.c (lower_bound_default): Do not handle
16117 (gen_compile_unit_die): Likewise.
16118 * gcc.c: Remove default extension binding for GNU Pascal.
16119 * stmt.c: Remove Pascal language from a comment.
16120 * xcoffout.c: Likewise.
16122 2017-07-13 David Malcolm <dmalcolm@redhat.com>
16125 * diagnostic-show-locus.c (fixit_cmp): New function.
16126 (layout::layout): Sort m_fixit_hints.
16127 (column_range::column_range): Assert that the values are valid.
16128 (struct char_span): New struct.
16129 (correction::overwrite): New method.
16130 (struct source_line): New struct.
16131 (line_corrections::add_hint): Add assertions. Reimplement memcpy
16132 calls in terms of classes source_line and char_span, and
16133 correction::overwrite.
16134 (selftest::test_overlapped_fixit_printing_2): New function.
16135 (selftest::diagnostic_show_locus_c_tests): Call it.
16137 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
16139 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
16140 early if there is no lhs.
16142 2017-07-13 Martin Liska <mliska@suse.cz>
16144 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
16145 (gen_reference_type_die): Likewise.
16146 * stor-layout.c: Remove Pascal-related comment.
16148 2017-07-13 Martin Liska <mliska@suse.cz>
16150 * opts.c (finish_options): Add quotes to error messages.
16151 (parse_sanitizer_options): Likewise.
16153 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
16155 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
16157 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
16159 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
16161 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
16163 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
16165 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
16167 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
16170 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
16171 provides the hardware capability bits, define the macro
16172 __BUILTIN_CPU_SUPPORTS__.
16173 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
16174 if GLIBC does not provide the hardware capability bits. Add a
16175 gcc_unreachable call if the built-in cpu function is neither
16176 __builtin_cpu_is nor __builtin_cpu_supports.
16177 (rs6000_get_function_versions_dispatcher): Change the warning
16178 that an old GLIBC is used which does not export the capability
16179 bits to be an error.
16180 * doc/extend.texi (target_clones attribute): Document the
16181 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
16182 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
16183 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
16184 the macros defined by GCC if the newer GLIBC is available.
16186 2017-07-12 Jeff Law <law@redhat.com>
16188 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
16189 remaining includes slightly.
16190 * config/riscv/riscv-builtins.c: Include profile-count.h.
16192 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16195 * config/avr/avr.c (avr_set_current_function): In diagnostic
16196 messages: Quote keywords and (parts of) identifiers.
16197 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
16200 2017-07-12 Carl Love <cel@us.ibm.com>
16202 * config/rs6000/rs6000-c.c: Add support for built-in functions
16203 vector bool char vec_revb (vector bool char);
16204 vector bool short vec_revb (vector short char);
16205 vector bool int vec_revb (vector bool int);
16206 vector bool long long vec_revb (vector bool long long);
16207 * doc/extend.texi: Update the built-in documentation file for the
16208 new built-in functions.
16210 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16212 * config/s390/s390.md: Remove movcc splitter.
16214 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16216 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
16217 load/store on condition.
16219 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16222 * config/avr/avr.c (avr_encode_section_info)
16223 [progmem && !TREE_READONLY]: Error if progmem object needs
16226 2017-07-11 Michael Collison <michael.collison@arm.com>
16228 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
16231 2017-07-11 Carl Love <cel@us.ibm.com>
16233 * config/rs6000/rs6000-c.c: Add support for builtins
16234 vector unsigned int vec_parity_lsbb (vector signed int);
16235 vector unsigned int vec_parity_lsbb (vector unsigned int);
16236 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
16237 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
16238 vector unsigned long long vec_parity_lsbb (vector signed long long);
16239 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
16240 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
16241 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
16242 * doc/extend.texi: Update the built-in documentation file for the
16243 new built-in functions.
16245 2017-07-11 David Malcolm <dmalcolm@redhat.com>
16247 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
16248 (layout::m_primary_loc): New field.
16249 (layout::layout): Initialize new field. Move location filtering
16250 logic from here to...
16251 (layout::maybe_add_location_range): ...this new method. Add
16252 support for filtering to just the lines already specified by other
16254 (layout::will_show_line_p): New method.
16255 (gcc_rich_location::add_location_if_nearby): New method.
16256 (selftest::test_add_location_if_nearby): New test function.
16257 (selftest::diagnostic_show_locus_c_tests): Call it.
16258 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
16261 2017-07-11 Tom de Vries <tom@codesourcery.com>
16263 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
16264 (bb_first_real_insn): New function.
16265 (nvptx_single): Add extra initialization of broadcasted condition
16268 2017-07-11 Nathan Sidwell <nathan@acm.org>
16270 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
16272 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
16274 * doc/extend.texi (AVR Function Attributes): Remove weblink to
16275 Binutils doc as TEXI will mess them up.
16276 * doc/invoke.texi (AVR Options): Same here.
16278 2017-07-11 Daniel Cederman <cederman@gaisler.com>
16280 * config/sparc/sparc.opt (mfix-ut700): New option.
16281 (mfix-gr712rc): Likewise.
16282 (sparc_fix_b2bst): New variable.
16283 * doc/invoke.texi (SPARC options): Document them.
16284 (ARM options): Fix warnings.
16285 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
16286 instructions to prevent sequences that can trigger the store-store
16287 errata for certain LEON3FT processors.
16288 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
16289 (sparc_option_override): Set sparc_fix_b2bst appropriately.
16290 * config/sparc/sparc.md (fix_b2bst): New attribute.
16291 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
16293 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
16296 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
16298 (*rsqrtsf2_sse): Ditto.
16300 (div<mode>3): Macroize insn from divdf3 and divsf3
16301 using MODEF mode iterator.
16303 2017-07-10 Martin Sebor <msebor@redhat.com>
16305 PR tree-optimization/80397
16306 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
16307 instead of testing for equality to INTEGER_TYPE.
16309 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
16311 * config.gcc: Remove uclibc from arc target spec.
16313 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
16315 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
16317 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16320 * lto-wrapper.c (remove_option): New function.
16321 (merge_and_complain): Merge PIC/PIE options more realistically.
16323 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16325 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
16329 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
16330 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
16331 * config.in: Regenerate.
16332 * configure: Regenerate.
16333 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
16334 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
16335 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
16336 (TARGET_GASISR_PROLOGUES): ...target mask.
16337 * common/config/avr/avr-common.c
16338 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
16339 Set -mgas-isr-prologues.
16340 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
16341 INSERT_PASS_BEFORE for it.
16342 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
16343 * config/avr/avr.c (avr_option_override)
16344 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
16345 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
16346 (avr_attribute_table) <no_gccisr>: Add new function attribute.
16347 (avr_set_current_function) <is_no_gccisr>: Init machine field.
16348 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
16350 (make_avr_pass_pre_proep): New function.
16351 (emit_push_sfr) <treg>: Add argument to function and use it
16352 instead of TMP_REG.
16353 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
16354 and set machine->gasisr.yes.
16355 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
16356 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
16357 __gcc_isr.n_pushed to .L__stack_usage.
16358 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
16359 (avr_asm_final_postscan_insn): ...this new static function.
16360 * config/avr/avr.h (machine_function)
16361 <is_no_gccisr, use_L__stack_usage>: New fields.
16362 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
16363 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
16364 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
16365 (gasisr, *gasisr): New expander and insn.
16366 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
16367 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
16368 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
16370 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
16372 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
16375 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16377 Move jump-tables out of .text again.
16380 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
16381 (ASM_OUTPUT_ADDR_VEC): New function.
16382 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
16383 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
16384 INSN_ADDRESSes as asm comment.
16385 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
16386 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
16387 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
16388 * config/avr/avr.md (*tablejump): Adjust comment.
16389 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
16390 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
16392 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
16393 (avr_output_addr_vec): New proto.
16394 (avr_log_t) <insn_addresses>: New field.
16396 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
16399 * config/i386/i386.c (ix86_function_arg_advance): Set
16400 outgoing_args_on_stack to true if there are outgoing arguments
16402 (ix86_function_arg): Likewise.
16403 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
16404 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
16405 * config/i386/i386.h (machine_function): Add
16406 outgoing_args_on_stack.
16408 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
16410 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
16411 supporting pthreds.
16412 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
16414 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
16416 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
16417 (REAL_H): Remove $(MACHMODE_H).
16418 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
16420 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
16421 $(MACHMODE_H) and double-int.h.
16422 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
16424 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
16427 2017-07-07 Andrew Pinski <apinski@cavium.com>
16429 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
16430 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
16432 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16434 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
16435 Add warning if GCC was not configured to link against a GLIBC that
16436 exports the hardware capability bits.
16437 (make_resolver_func): Make resolver function private and not a
16438 COMDAT function. Create the name with clone_function_name instead
16439 of make_unique_name.
16442 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
16443 correct operand in doing the split.
16445 2017-07-07 Carl Love <cel@us.ibm.com>
16447 * config/rs6000/rs6000-c: Add support for built-in function
16448 vector unsigned short vec_pack_to_short_fp32 (vector float,
16450 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
16451 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
16452 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
16453 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
16454 (convert_4f32_8i16): Add define_expand.
16455 * doc/extend.texi: Update the built-in documentation file for the
16456 new built-in function.
16458 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16460 * config/sparc/m8.md: New file.
16461 * config/sparc/sparc.md: Include m8.md.
16463 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16465 * config/sparc/sparc.opt: New option -mvis4b.
16466 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
16467 (sparc_option_override): Handle VIS4B.
16468 (enum sparc_builtins): Define
16469 SPARC_BUILTIN_DICTUNPACK{8,16,32},
16470 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
16471 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
16472 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
16473 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
16474 (check_constant_argument): New function.
16475 (sparc_vis_init_builtins): Define builtins
16476 __builtin_vis_dictunpack{8,16,32},
16477 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
16478 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
16479 __builtin_vis_fpcmpde{8,16,32}shl and
16480 __builtin_vis_fpcmpur{8,16,32}shl.
16481 (sparc_expand_builtin): Check that the constant operands to
16482 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
16483 constant and in range.
16484 * config/sparc/sparc-c.c (sparc_target_macros): Handle
16486 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
16487 (SPARC_IMM5_P): Likewise.
16488 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
16489 (enabled): Handle vis4b.
16490 (UNSPEC_DICTUNPACK): New unspec.
16491 (UNSPEC_FPCMPSHL): Likewise.
16492 (UNSPEC_FPUCMPSHL): Likewise.
16493 (UNSPEC_FPCMPDESHL): Likewise.
16494 (UNSPEC_FPCMPURSHL): Likewise.
16495 (cpu_feature): New CPU feature `vis4b'.
16496 (dictunpack{8,16,32}): New insns.
16497 (FPCSMODE): New mode iterator.
16498 (fpcscond): New code iterator.
16499 (fpcsucond): Likewise.
16500 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
16501 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
16502 (fpcmpde{8,16,32}{si,di}shl): Likewise.
16503 (fpcmpur{8,16,32}{si,di}shl): Likewise.
16504 * config/sparc/constraints.md: Define constraints `q' for unsigned
16505 2-bit integer constants and `t' for unsigned 5-bit integer
16507 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
16509 (imm5_operand_dictunpack16): Likewise.
16510 (imm5_operand_dictunpack32): Likewise.
16511 (imm2_operand): Likewise.
16512 * doc/invoke.texi (SPARC Options): Document -mvis4b.
16513 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
16514 ditunpack* and fpcmp*shl builtins.
16516 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16518 * config.gcc: Handle m8 in --with-{cpu,tune} options.
16519 * config.in: Add HAVE_AS_SPARC6 define.
16520 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
16522 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
16524 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
16525 (CPP_CPU_SPEC): Handle m8.
16526 (ASM_CPU_SPEC): Likewise.
16527 * config/sparc/sparc-opts.h (enum processor_type): Add
16529 * config/sparc/sparc.c (m8_costs): New struct.
16530 (sparc_option_override): Handle TARGET_CPU_m8.
16531 (sparc32_initialize_trampoline): Likewise.
16532 (sparc64_initialize_trampoline): Likewise.
16533 (sparc_issue_rate): Likewise.
16534 (sparc_register_move_cost): Likewise.
16535 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
16536 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
16537 (ASM_CPU64_DEFAULT_SPEC): Likewise.
16538 (CPP_CPU_SPEC): Handle M8.
16539 (ASM_CPU_SPEC): Likewise.
16540 (AS_M8_FLAG): Define.
16541 * config/sparc/sparc.md: Add m8 to the cpu attribute.
16542 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
16543 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
16545 * configure: Regenerate.
16546 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
16549 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16551 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
16553 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
16554 ("*movdi_insn_sp32"): Do not set v3pipe.
16555 ("*movsi_insn"): Likewise.
16556 ("*movdi_insn_sp64"): Likewise.
16557 ("*movsf_insn"): Likewise.
16558 ("*movdf_insn_sp32"): Likewise.
16559 ("*movdf_insn_sp64"): Likewise.
16560 ("*zero_extendsidi2_insn_sp64"): Likewise.
16561 ("*sign_extendsidi2_insn"): Likewise.
16562 ("*mov<VM32:mode>_insn"): Likewise.
16563 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16564 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16565 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16566 ("<vlop:code><VL:mode>3"): Likewise.
16567 ("*not_<vlop:code><VL:mode>3"): Likewise.
16568 ("*nand<VL:mode>_vis"): Likewise.
16569 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
16570 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
16571 ("one_cmpl<VL:mode>2"): Likewise.
16572 ("faligndata<VM64:mode>_vis"): Likewise.
16573 ("alignaddrsi_vis"): Likewise.
16574 ("alignaddrdi_vis"): Likweise.
16575 ("alignaddrlsi_vis"): Likewise.
16576 ("alignaddrldi_vis"): Likewise.
16577 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16578 ("bmaskdi_vis"): Likewise.
16579 ("bmasksi_vis"): Likewise.
16580 ("bshuffle<VM64:mode>_vis"): Likewise.
16581 ("cmask8<P:mode>_vis"): Likewise.
16582 ("cmask16<P:mode>_vis"): Likewise.
16583 ("cmask32<P:mode>_vis"): Likewise.
16584 ("pdistn<P:mode>_vis"): Likewise.
16585 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16587 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16589 * config/sparc/sparc.md ("subtype"): New insn attribute.
16590 ("*wrgsr_sp64"): Set insn subtype.
16591 ("*rdgsr_sp64"): Likewise.
16592 ("alignaddrsi_vis"): Likewise.
16593 ("alignaddrdi_vis"): Likewise.
16594 ("alignaddrlsi_vis"): Likewise.
16595 ("alignaddrldi_vis"): Likewise.
16596 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16597 ("fexpand_vis"): Likewise.
16598 ("fpmerge_vis"): Likewise.
16599 ("faligndata<VM64:mode>_vis"): Likewise.
16600 ("bshuffle<VM64:mode>_vis"): Likewise.
16601 ("cmask8<P:mode>_vis"): Likewise.
16602 ("cmask16<P:mode>_vis"): Likewise.
16603 ("cmask32<P:mode>_vis"): Likewise.
16604 ("fchksm16_vis"): Likewise.
16605 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
16606 ("fmean16_vis"): Likewise.
16607 ("fp<plusminus_insn>64_vis"): Likewise.
16608 ("<plusminus_insn>v8qi3"): Likewise.
16609 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16610 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
16611 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
16612 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
16613 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
16614 ("*movqi_insn"): Likewise.
16615 ("*movhi_insn"): Likewise.
16616 ("*movsi_insn"): Likewise.
16617 ("movsi_pic_gotdata_op"): Likewise.
16618 ("*movdi_insn_sp32"): Likewise.
16619 ("*movdi_insn_sp64"): Likewise.
16620 ("movdi_pic_gotdata_op"): Likewise.
16621 ("*movsf_insn"): Likewise.
16622 ("*movdf_insn_sp32"): Likewise.
16623 ("*movdf_insn_sp64"): Likewise.
16624 ("*zero_extendhisi2_insn"): Likewise.
16625 ("*zero_extendqihi2_insn"): Likewise.
16626 ("*zero_extendqisi2_insn"): Likewise.
16627 ("*zero_extendqidi2_insn"): Likewise.
16628 ("*zero_extendhidi2_insn"): Likewise.
16629 ("*zero_extendsidi2_insn_sp64"): Likewise.
16630 ("ldfsr"): Likewise.
16631 ("prefetch_64"): Likewise.
16632 ("prefetch_32"): Likewise.
16633 ("tie_ld32"): Likewise.
16634 ("tie_ld64"): Likewise.
16635 ("*tldo_ldub_sp32"): Likewise.
16636 ("*tldo_ldub1_sp32"): Likewise.
16637 ("*tldo_ldub2_sp32"): Likewise.
16638 ("*tldo_ldub_sp64"): Likewise.
16639 ("*tldo_ldub1_sp64"): Likewise.
16640 ("*tldo_ldub2_sp64"): Likewise.
16641 ("*tldo_ldub3_sp64"): Likewise.
16642 ("*tldo_lduh_sp32"): Likewise.
16643 ("*tldo_lduh1_sp32"): Likewise.
16644 ("*tldo_lduh_sp64"): Likewise.
16645 ("*tldo_lduh1_sp64"): Likewise.
16646 ("*tldo_lduh2_sp64"): Likewise.
16647 ("*tldo_lduw_sp32"): Likewise.
16648 ("*tldo_lduw_sp64"): Likewise.
16649 ("*tldo_lduw1_sp64"): Likewise.
16650 ("*tldo_ldx_sp64"): Likewise.
16651 ("*mov<VM32:mode>_insn"): Likewise.
16652 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16653 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16655 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16657 * config/sparc/sparc.md ("type"): New insn type viscmp.
16658 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
16660 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
16661 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
16662 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16663 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
16665 ("n7_vis_logical_11cycle"): Likewise.
16666 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
16667 * config/sparc/niagara2.md ("niag3_vis": Likewise.
16668 * config/sparc/niagara.md ("niag_vis"): Likewise.
16669 * config/sparc/ultra3.md ("us3_fga"): Likewise.
16670 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
16672 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16674 * config/sparc/sparc.md: New instruction type `bmask'.
16675 (bmaskdi_vis): Use the `bmask' type.
16676 (bmasksi_vis): Likewise.
16677 * config/sparc/ultra3.md (us3_array): Likewise.
16678 * config/sparc/niagara7.md (n7_array): Likewise.
16679 * config/sparc/niagara4.md (n4_array): Likewise.
16680 * config/sparc/niagara2.md (niag2_vis): Likewise.
16681 (niag3_vis): Likewise.
16682 * config/sparc/niagara.md (niag_vis): Likewise.
16684 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16686 * ipa-comdats.c: Remove optimize check from gate.
16687 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
16688 for functions not optimized.
16689 (ipa_fn_summary_read): Skip optimize check.
16690 (ipa_fn_summary_write): Likewise.
16691 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
16693 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
16695 (can_inline_edge_p): Check flag_pcc_struct_return for match.
16696 (check_callers): Give up on caller which is not optimized.
16697 (inline_small_functions): Likewise.
16698 (ipa_inline): Do not give up when not optimizing.
16699 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
16700 away unoptimizes cdtors.
16701 (whole_program_function_and_variable_visibility): Do
16702 ipa_discover_readonly_nonaddressable_vars in LTO mode.
16703 * ipa.c (process_references): Do not check optimize.
16704 (symbol_table::remove_unreachable_nodes): Update optimize check.
16705 (set_writeonly_bit): Update optimize check.
16706 (pass_ipa_cdtor_merge::gate): Do not check optimize.
16707 (pass_ipa_single_use::gate): Remove.
16709 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16711 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
16712 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
16713 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
16714 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
16715 permute_load, permute_store, adjust_extract, adjust_splat,
16716 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
16717 replace_swap_with_copy, dump_swap_insn_table,
16718 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
16719 recombine_lvx_pattern, recombine_stvx_pattern,
16720 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
16721 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
16722 to file rs6000-p8swap.c.
16723 * config/rs6000/rs6000-p8swap.c: New file.
16724 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
16725 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
16726 and rs6000*-*-* targets.
16728 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16730 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
16732 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16734 * lto-wrapper.c (merge_and_complain): Do not merge
16735 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
16736 fsigned_zeros, ftrapping_math, fwrapv.
16737 (append_compiler_options): Do not track these options.
16738 (append_linker_options): Likewie
16740 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16742 * cgraphunit.c (cgraph_node::finalize_function): When
16743 !flag_toplevel_reorde set no_reorder flag.
16744 (varpool_node::finalize_decl): Likewise.
16745 (symbol_table::compile): Drop no toplevel reorder path.
16747 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16749 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
16750 edges; zero probability is not better than uninitialized.
16752 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
16754 * asan.h (asan_sanitize_allocas_p): Declare.
16755 * asan.c (asan_sanitize_allocas_p): New function.
16756 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
16757 (handle_builtin_alloca): Likewise.
16758 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
16759 if !asan_sanitize_allocas_p.
16760 * params.def (asan-instrument-allocas): Add new option.
16761 * params.h (ASAN_PROTECT_ALLOCAS): Define.
16762 * opts.c (common_handle_option): Disable allocas sanitization for
16765 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
16767 * asan.c: Include gimple-fold.h.
16768 (get_last_alloca_addr): New function.
16769 (handle_builtin_stackrestore): Likewise.
16770 (handle_builtin_alloca): Likewise.
16771 (asan_emit_allocas_unpoison): Likewise.
16772 (get_mem_refs_of_builtin_call): Add new parameter, remove const
16773 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
16774 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
16775 (instrument_builtin_call): Pass gimple iterator to
16776 get_mem_refs_of_builtin_call.
16777 (last_alloca_addr): New global.
16778 * asan.h (asan_emit_allocas_unpoison): Declare.
16779 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
16780 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
16781 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
16782 if function calls alloca.
16783 * gimple-fold.c (replace_call_with_value): Remove static keyword.
16784 * gimple-fold.h (replace_call_with_value): Declare.
16785 * internal-fn.c: Include asan.h.
16786 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
16787 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
16789 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16791 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
16792 (C_SELFTEST_FLAGS): New.
16793 (CPP_SELFTEST_FLAGS): New.
16794 (SELFTEST_DEPS): New, from deps of s-selftest.
16795 (C_SELFTEST_DEPS): New, from deps of s-selftest.
16796 (CPP_SELFTEST_DEPS): New.
16797 (selftest): Add dependency on s-selftest-c++.
16798 (s-selftest): Rename to...
16799 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
16800 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
16801 than SELFTEST_FLAGS.
16802 (selftest-gdb): Rename to...
16803 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
16805 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
16806 (selftest-valgrind): Rename to...
16807 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
16809 (selftest-valgrind): Reintroduce as an alias for
16810 selftest-c-valgrind.
16811 (s-selftest-c++): New.
16812 (selftest-c++-gdb): New.
16813 (selftest-c++-valgrind): New.
16815 2017-07-06 Olivier Hainque <hainque@adacore.com>
16817 * gcc.c (process_command): When deciding if undefined variables
16818 should be ignored when processing specs, accept "gcc -v" as well.
16820 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16822 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
16823 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
16825 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
16827 * config/arm/arm-cpus.in (armv8-r): Add new entry.
16828 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
16829 * config/arm/arm-tables.opt: Regenerate.
16830 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
16832 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
16834 2017-07-06 Carl Love <cel@us.ibm.com>
16836 * ChangeLog: Clean up from mid air collision
16838 2017-07-06 Carl Love <cel@us.ibm.com>
16840 * config/rs6000/rs6000-c.c: Add support for built-in functions
16841 vector signed int vec_subc (vector signed int, vector signed int);
16842 vector signed __int128 vec_subc (vector signed __int128,
16843 vector signed __int128);
16844 vector unsigned __int128 vec_subc (vector unsigned __int128,
16845 vector unsigned __int128);
16846 vector signed int vec_sube (vector signed int, vector signed int,
16847 vector signed int);
16848 vector unsigned int vec_sube (vector unsigned int,
16849 vector unsigned int,
16850 vector unsigned int);
16851 vector signed __int128 vec_sube (vector signed __int128,
16852 vector signed __int128,
16853 vector signed__int128);
16854 vector unsigned __int128 vec_sube (vector unsigned __int128,
16855 vector unsigned __int128,
16856 vector unsigned __int128);
16857 vector signed int vec_subec (vector signed int, vector signed int,
16858 vector signed int);
16859 vector unsigned int vec_subec (vector unsigned int,
16860 vector unsigned int,
16861 vector unsigned int);
16862 vector signed __int128 vec_subec (vector signed __int128,
16863 vector signed __int128,
16864 vector signed__int128);
16865 vector unsigned __int128 vec_subec (vector unsigned __int128,
16866 vector unsigned __int128,
16867 vector unsigned __int128);
16868 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
16869 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
16870 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
16871 BU_ALTIVEC_OVERLOAD_X definitions.
16872 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
16873 * doc/extend.texi: Update the built-in documentation file for the new
16874 built-in functions.
16876 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16879 * diagnostic-show-locus.c (layout::layout): Use start and finish
16880 spelling location for the start and finish of each range.
16881 * genmatch.c (linemap_client_expand_location_to_spelling_point):
16882 Add unused aspect param.
16883 * input.c (expand_location_1): Add "aspect" param, and use it
16884 to access the correct part of the location.
16885 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
16887 (expand_location_to_spelling_point): Likewise.
16888 (linemap_client_expand_location_to_spelling_point): Add "aspect"
16889 param, and pass it to expand_location_1.
16891 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
16893 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
16894 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
16895 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
16896 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
16897 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
16898 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
16899 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
16900 _mm_maskz_getmant_ss): New intrinsics.
16901 (__builtin_ia32_getexpss128_mask): Changed to ...
16902 __builtin_ia32_getexpss128_round ... this.
16903 (__builtin_ia32_getexpsd128_mask): Changed to ...
16904 __builtin_ia32_getexpsd128_round ... this.
16905 * config/i386/i386-builtin-types.def
16906 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
16907 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
16908 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
16909 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
16910 __builtin_ia32_getmantss_mask_round): New builtins.
16911 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
16912 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
16913 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
16914 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
16915 * config/i386/sse.md
16916 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
16917 avx512f_sgetexp<mode><mask_scalar_name>
16918 <round_saeonly_scalar_name> ... this.
16919 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
16920 %0, %1, %2<round_saeonly_op3>}): Changed to ...
16921 vgetexp<ssescalarmodesuffix>
16922 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16923 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
16924 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
16925 avx512f_vgetmant<mode><mask_scalar_name>
16926 <round_saeonly_scalar_name> ... this.
16927 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
16928 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
16929 vgetmant<ssescalarmodesuffix>
16930 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
16931 %0<mask_scalar_operand4>, %1, %2
16932 <round_saeonly_scalar_mask_op4>, %3} ... this.
16933 * config/i386/subst.md (mask_scalar_operand4,
16934 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
16935 round_saeonly_scalar_nimm_predicate): New subst attributes.
16937 2017-07-06 Julia Koval <julia.koval@intel.com>
16939 * config/i386/i386.c (ix86_erase_embedded_rounding):
16940 Remove code for old rounding pattern.
16942 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
16944 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
16946 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
16948 * doc/sourcebuild.texi (Test Directives, Variants of
16949 dg-require-support): Add documentation for dg-require-stack-check.
16951 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
16953 * config/i386/subst.md (mask_scalar, round_scalar,
16954 round_saeonly_scalar): New meta-templates.
16955 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
16956 round_scalar_mask_operand3, round_scalar_mask_op3,
16957 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
16958 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
16959 round_saeonly_scalar_constraint,
16960 round_saeonly_scalar_prefix): New subst attribute.
16961 * config/i386/sse.md
16962 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
16963 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
16964 <round_scalar_name> ... this.
16965 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
16966 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
16967 <round_scalar_name> ... this.
16968 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
16969 <sse>_vm<code><mode>3<mask_scalar_name>
16970 <round_saeonly_scalar_name> ... this.
16971 (v<plusminus_mnemonic><ssescalarmodesuffix>
16972 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
16973 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
16974 v<plusminus_mnemonic><ssescalarmodesuffix>
16975 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16976 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
16977 (v<multdiv_mnemonic><ssescalarmodesuffix>
16978 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
16979 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
16980 v<multdiv_mnemonic><ssescalarmodesuffix>
16981 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16982 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
16983 (v<maxmin_float><ssescalarmodesuffix>
16984 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
16985 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
16986 v<maxmin_float><ssescalarmodesuffix>
16987 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16988 %0<mask_scalar_operand3>, %1, %<iptr>2
16989 <round_saeonly_scalar_mask_op3>} ... this.
16991 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
16993 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
16994 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
16996 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
16997 Alan Hayward <alan.hayward@arm.com>
16998 David Sherwood <david.sherwood@arm.com>
17000 * combine.c (simplify_if_then_else): Remove "enum" before
17002 * compare-elim.c (can_eliminate_compare): Likewise.
17003 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
17005 (aarch64_lookup_simd_builtin_type): Likewise.
17006 (aarch64_simd_builtin_type): Likewise.
17007 (aarch64_init_simd_builtin_types): Likewise.
17008 (aarch64_simd_expand_args): Likewise.
17009 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
17011 (aarch64_reverse_mask): Likewise.
17012 (aarch64_simd_emit_reg_reg_move): Likewise.
17013 (aarch64_gen_adjusted_ldpstp): Likewise.
17014 (aarch64_ccmp_mode_to_code): Likewise.
17015 (aarch64_operands_ok_for_ldpstp): Likewise.
17016 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17017 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
17019 (aarch64_min_divisions_for_recip_mul): Likewise.
17020 (aarch64_reassociation_width): Likewise.
17021 (aarch64_get_condition_code_1): Likewise.
17022 (aarch64_simd_emit_reg_reg_move): Likewise.
17023 (aarch64_simd_attr_length_rglist): Likewise.
17024 (aarch64_reverse_mask): Likewise.
17025 (aarch64_operands_ok_for_ldpstp): Likewise.
17026 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17027 (aarch64_gen_adjusted_ldpstp): Likewise.
17028 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
17030 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
17031 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
17032 (arm_lookup_simd_builtin_type): Likewise.
17033 (arm_simd_builtin_type): Likewise.
17034 (arm_init_simd_builtin_types): Likewise.
17035 (arm_expand_builtin_args): Likewise.
17036 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
17037 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
17038 (ft32_setup_incoming_varargs): Likewise.
17039 (ft32_function_arg): Likewise.
17040 (ft32_function_arg_advance): Likewise.
17041 (ft32_pass_by_reference): Likewise.
17042 (ft32_arg_partial_bytes): Likewise.
17043 (ft32_valid_pointer_mode): Likewise.
17044 (ft32_addr_space_pointer_mode): Likewise.
17045 (ft32_addr_space_legitimate_address_p): Likewise.
17046 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
17048 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
17049 (ix86_emit_outlined_ms2sysv_restore): Likewise.
17050 (iamcu_alignment): Likewise.
17051 (canonicalize_vector_int_perm): Likewise.
17052 (ix86_noce_conversion_profitable_p): Likewise.
17053 (ix86_mpx_bound_mode): Likewise.
17054 (ix86_operands_ok_for_move_multiple): Likewise.
17055 * config/microblaze/microblaze-protos.h
17056 (microblaze_expand_conditional_branch_reg): Likewise.
17057 * config/microblaze/microblaze.c
17058 (microblaze_expand_conditional_branch_reg): Likewise.
17059 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
17061 (rs6000_reassociation_width): Likewise.
17062 (rs6000_invalid_binary_op): Likewise.
17063 (fusion_p9_p): Likewise.
17064 (emit_fusion_p9_load): Likewise.
17065 (emit_fusion_p9_store): Likewise.
17066 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
17068 (riscv_hard_regno_mode_ok_p): Likewise.
17069 (riscv_address_insns): Likewise.
17070 (riscv_split_symbol): Likewise.
17071 (riscv_legitimize_move): Likewise.
17072 (riscv_function_value): Likewise.
17073 (riscv_hard_regno_nregs): Likewise.
17074 (riscv_expand_builtin): Likewise.
17075 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
17076 (riscv_build_integer): Likewise.
17077 (riscv_split_integer): Likewise.
17078 (riscv_legitimate_constant_p): Likewise.
17079 (riscv_cannot_force_const_mem): Likewise.
17080 (riscv_regno_mode_ok_for_base_p): Likewise.
17081 (riscv_valid_base_register_p): Likewise.
17082 (riscv_valid_offset_p): Likewise.
17083 (riscv_valid_lo_sum_p): Likewise.
17084 (riscv_classify_address): Likewise.
17085 (riscv_legitimate_address_p): Likewise.
17086 (riscv_address_insns): Likewise.
17087 (riscv_load_store_insns): Likewise.
17088 (riscv_force_binary): Likewise.
17089 (riscv_split_symbol): Likewise.
17090 (riscv_force_address): Likewise.
17091 (riscv_legitimize_address): Likewise.
17092 (riscv_move_integer): Likewise.
17093 (riscv_legitimize_const_move): Likewise.
17094 (riscv_legitimize_move): Likewise.
17095 (riscv_address_cost): Likewise.
17096 (riscv_subword): Likewise.
17097 (riscv_output_move): Likewise.
17098 (riscv_canonicalize_int_order_test): Likewise.
17099 (riscv_emit_int_order_test): Likewise.
17100 (riscv_function_arg_boundary): Likewise.
17101 (riscv_pass_mode_in_fpr_p): Likewise.
17102 (riscv_pass_fpr_single): Likewise.
17103 (riscv_pass_fpr_pair): Likewise.
17104 (riscv_get_arg_info): Likewise.
17105 (riscv_function_arg): Likewise.
17106 (riscv_function_arg_advance): Likewise.
17107 (riscv_arg_partial_bytes): Likewise.
17108 (riscv_function_value): Likewise.
17109 (riscv_pass_by_reference): Likewise.
17110 (riscv_setup_incoming_varargs): Likewise.
17111 (riscv_print_operand): Likewise.
17112 (riscv_elf_select_rtx_section): Likewise.
17113 (riscv_save_restore_reg): Likewise.
17114 (riscv_for_each_saved_reg): Likewise.
17115 (riscv_register_move_cost): Likewise.
17116 (riscv_hard_regno_mode_ok_p): Likewise.
17117 (riscv_hard_regno_nregs): Likewise.
17118 (riscv_class_max_nregs): Likewise.
17119 (riscv_memory_move_cost): Likewise.
17120 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
17121 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
17122 (rl78_addr_space_address_mode): Likewise.
17123 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17125 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17126 (rs6000_reassociation_width): Likewise.
17127 (rs6000_invalid_binary_op): Likewise.
17128 (fusion_p9_p): Likewise.
17129 (emit_fusion_p9_load): Likewise.
17130 (emit_fusion_p9_store): Likewise.
17131 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
17132 (ok_for_simple_move_operands): Likewise.
17133 (ok_for_simple_move_strict_operands): Likewise.
17134 (ok_for_simple_arith_logic_operands): Likewise.
17135 (visium_legitimize_reload_address): Likewise.
17136 (visium_select_cc_mode): Likewise.
17137 (output_cbranch): Likewise.
17138 (visium_split_double_move): Likewise.
17139 (visium_expand_copysign): Likewise.
17140 (visium_expand_int_cstore): Likewise.
17141 (visium_expand_fp_cstore): Likewise.
17142 * config/visium/visium.c (visium_pass_by_reference): Likewise.
17143 (visium_function_arg): Likewise.
17144 (visium_function_arg_advance): Likewise.
17145 (visium_libcall_value): Likewise.
17146 (visium_setup_incoming_varargs): Likewise.
17147 (visium_legitimate_constant_p): Likewise.
17148 (visium_legitimate_address_p): Likewise.
17149 (visium_legitimize_address): Likewise.
17150 (visium_secondary_reload): Likewise.
17151 (visium_register_move_cost): Likewise.
17152 (visium_memory_move_cost): Likewise.
17153 (prepare_move_operands): Likewise.
17154 (ok_for_simple_move_operands): Likewise.
17155 (ok_for_simple_move_strict_operands): Likewise.
17156 (ok_for_simple_arith_logic_operands): Likewise.
17157 (visium_function_value_1): Likewise.
17158 (rtx_ok_for_offset_p): Likewise.
17159 (visium_legitimize_reload_address): Likewise.
17160 (visium_split_double_move): Likewise.
17161 (visium_expand_copysign): Likewise.
17162 (visium_expand_int_cstore): Likewise.
17163 (visium_expand_fp_cstore): Likewise.
17164 (visium_split_cstore): Likewise.
17165 (visium_select_cc_mode): Likewise.
17166 (visium_split_cbranch): Likewise.
17167 (output_cbranch): Likewise.
17168 (visium_print_operand_address): Likewise.
17169 * expmed.c (flip_storage_order): Likewise.
17170 * expmed.h (emit_cstore): Likewise.
17171 (flip_storage_order): Likewise.
17172 * genrecog.c (validate_pattern): Likewise.
17173 * hsa-gen.c (gen_hsa_addr): Likewise.
17174 * internal-fn.c (expand_arith_overflow): Likewise.
17175 * ira-color.c (allocno_copy_cost_saving): Likewise.
17176 * lra-assigns.c (find_hard_regno_for_1): Likewise.
17177 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
17178 (process_invariant_for_inheritance): Likewise.
17179 * lra-eliminations.c (move_plus_up): Likewise.
17180 * omp-low.c (lower_oacc_reductions): Likewise.
17181 * simplify-rtx.c (simplify_subreg): Likewise.
17182 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
17183 (TARGET_CHKP_BOUND_MODE): Likewise..
17184 * targhooks.c (default_chkp_bound_mode): Likewise.
17185 (default_setup_incoming_vararg_bounds): Likewise.
17186 * targhooks.h (default_chkp_bound_mode): Likewise.
17187 (default_setup_incoming_vararg_bounds): Likewise.
17188 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
17189 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
17190 (have_whole_vector_shift): Likewise.
17191 * tree-vect-stmts.c (vectorizable_load): Likewise.
17192 * doc/tm.texi: Regenerate.
17194 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17196 Graceful degrade if Binutils PR21472 is not available.
17199 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
17200 .rodata in flash test fails.
17201 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
17202 * confgure: Regenerate.
17203 * config.in: Regenerate.
17204 * config/avr/avr.c (avr_asm_named_section)
17205 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
17206 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
17207 (avr_asm_init_sections): Same.
17209 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17211 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
17212 (fma<VH:mode>4_intrinsic): Likewise.
17213 (*fmsub<VCVTF:mode>4): Likewise.
17214 (*fmsub<VH:mode>4_intrinsic): Likewise.
17216 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17219 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
17220 Don't depend on "optimize > 0".
17221 (out_movhi_r_mr, out_movqi_mr_r): Same.
17222 (out_movhi_mr_r, out_movqi_r_mr): Same.
17223 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
17224 io_address_operand on "optimize > 0".
17226 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17228 * tree-loop-distribution.c: Add general explanantion on the pass.
17229 (generate_loops_for_partition): Mark distributed loop.
17230 (pg_add_dependence_edges): New parameter. Handle alias data
17231 dependence specially and record it in the parameter if asked.
17232 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
17233 (init_partition_graph_vertices, add_partition_graph_edge): New.
17234 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
17235 (free_partition_graph_vdata, build_partition_graph): New.
17236 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
17237 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
17238 (data_ref_segment_size, latch_dominated_by_data_ref): New.
17239 (compute_alias_check_pairs, version_loop_by_alias_check): New.
17240 (version_for_distribution_p, finalize_partitions): New.
17241 (distribute_loop): Handle alias data dependence specially. Factor
17242 out loop fusion code as functions and call these functions.
17244 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17246 * tree-loop-distribution.c (classify_partition): New parameter and
17247 better handle reduction statement.
17248 (rdg_build_partitions): Revise comment.
17249 (distribute_loop): Compute statements in all partitions and pass it
17250 to classify_partition.
17252 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17254 * tree-loop-distribution.c (enum partition_type): New.
17255 (struct partition): New field type.
17256 (partition_merge_into): Add parameter. Update partition type.
17257 (data_dep_in_cycle_p, update_type_for_merge): New functions.
17258 (build_rdg_partition_for_vertex): Compute partition type.
17259 (rdg_build_partitions): Dump partition type.
17260 (distribute_loop): Update calls to partition_merge_into.
17262 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17264 * tree-loop-distribution.c (struct ddr_hasher): New.
17265 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
17267 (classify_partition): Call get_data_dependence.
17268 (pg_add_dependence_edges): Ditto.
17269 (distribute_loop): Release data dependence hash table.
17271 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17273 * tree-loop-distribution.c (ref_base_address): Delete.
17274 (similar_memory_accesses): Rename ...
17275 (share_memory_accesses): ... to this. Check if partitions access
17276 the same memory reference.
17277 (distribute_loop): Call share_memory_accesses.
17279 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17281 * tree-loop-distribution.c (struct partition): New field recording
17282 its data reference.
17283 (partition_alloc, partition_free): Init and release data refs.
17284 (partition_merge_into): Merge data refs.
17285 (build_rdg_partition_for_vertex): Collect data refs for partition.
17286 (pg_add_dependence_edges): Change parameters from vector to bitmap.
17288 (distribute_loop): Remve data refs from vertice data of partition
17291 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17293 * tree-loop-distribution.c (params.h): Include header file.
17294 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
17295 (datarefs_vec): New global var.
17296 (create_rdg_vertices): Use datarefs_vec directly.
17297 (free_rdg): Don't free data references.
17298 (build_rdg): Update use. Don't free data references.
17299 (distribute_loop): Compute global variable for data references.
17300 Bail out if there are too many data references.
17302 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17304 * tree-loop-distribution.c (loop_nest): New global var.
17305 (build_rdg): Use loop directly, rather than loop nest.
17306 (pg_add_dependence_edges): Remove loop nest parameter. Use global
17308 (distribute_loop): Compute global variable loop nest. Update use.
17310 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17312 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
17313 (partition_merge_into): New parameter. Dump reason for fusion.
17314 (distribute_loop): Update use of partition_merge_into.
17316 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17318 * tree-loop-distribution.c (bb_top_order_index): New.
17319 (bb_top_order_index_size, bb_top_order_cmp): New.
17320 (stmts_from_loop): Use topological order.
17321 (pass_loop_distribution::execute): Compute and release topological
17322 order for basic blocks.
17324 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17326 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
17329 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17331 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
17332 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
17333 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
17334 * internal-fn.def (LOOP_DIST_ALIAS): New.
17335 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
17336 (fold_loop_internal_call): ... this.
17337 (vect_loop_dist_alias_call): New function.
17338 (set_uid_loop_bbs): Call fold_loop_internal_call.
17339 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
17342 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
17345 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
17346 Require dead FLAGS_REG at the beginning of a peephole.
17348 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
17351 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
17352 arguments in the call to __builtin_ia32_sbb_u32.
17353 (_subborrow_u64): Swap _X and _Y arguments in the call to
17354 __builtin_ia32_sbb_u64.
17356 2017-07-04 Jakub Jelinek <jakub@redhat.com>
17359 * tree-vrp.c (compare_assert_loc): Turn into a function template
17360 with stable template parameter. Only test if a->e is NULL,
17361 !a->e == !b->e has been verified already. Use e == NULL or
17362 e != NULL instead of e or ! e tests. If stable is true, don't use
17363 iterative_hash_expr, on the other side allow a or b or both NULL
17364 and sort the NULLs last.
17365 (process_assert_insertions): Sort using compare_assert_loc<false>
17366 instead of compare_assert_loc, later sort using
17367 compare_assert_loc<true> before calling process_assert_insertions_for
17368 in a loop. Use break instead of continue once seen NULL pointer.
17370 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
17372 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17373 Cortex-R7 and Cortex-R8 processors.
17375 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17377 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
17378 uninitialized while src is not.
17380 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
17382 * common/config/arm/arm-common.c: Adjust include path for
17384 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
17385 (arm-cpu.h): Create in build directory. Adjust dependency rules.
17386 (arm-cpu-data.h): Likewise.
17387 (arm-cpu-cdata.h): Likewise.
17388 * config/arm/arm-cpu.h: Delete.
17389 * config/arm/arm-cpu-cdata.h: Delete.
17390 * config/arm/arm-cpu-data.h: Delete.
17392 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
17394 * config/arm/arm-cpus.in (cortex-a55): New.
17395 (cortex-a75): Likewise.
17396 (cortex-a75.cortex-a55): Likewise.
17397 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
17399 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
17400 * config/arm/arm-cpu-cdata.h: Regenerate.
17401 * config/arm/arm-cpu-data.h: Regenerate.
17402 * config/arm/arm-cpu.h: Regenerate.
17403 * config/arm/arm-tables.opt: Regenerate.
17404 * config/arm/arm-tune.md: Regenerate.
17406 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17408 * haifa-sched.c (sched_create_recovery_edges): Update profile.
17410 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17412 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
17415 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
17417 PR tree-optimization/81292
17418 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
17419 full_string_p, also call adjust_related_strinfos if the adjustment
17420 is simple, otherwise invalidate related strinfos.
17422 2017-07-04 Martin Liska <mliska@suse.cz>
17425 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
17426 newly created variable as DECL_IGNORED_P.
17428 2017-07-04 Martin Liska <mliska@suse.cz>
17431 * ipa-inline.c (inline_small_functions):
17432 Use xstrdup_for_dump.
17434 2017-07-04 Tom de Vries <tom@codesourcery.com>
17436 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
17438 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
17441 * config/darwin.c (darwin_function_switched_text_sections):
17444 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
17446 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
17448 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17450 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
17452 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17454 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
17455 min_profitable_iters, and th as inclusive lower bounds.
17456 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
17457 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
17458 for min_profitable_iters and min_profitable_estimate.
17459 (vect_transform_loop): Treat th as an inclusive lower bound.
17460 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
17462 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
17465 * config/darwin.c (darwin_function_switched_text_sections):
17466 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
17467 in two pieces, and suppress the use of buf.
17469 2017-07-03 Nathan Sidwell <nathan@acm.org>
17471 * hash-table.h (hash_table_mod1): Fix indentation.
17473 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17475 PR middle-end/81290
17476 * predict.c (force_edge_cold): Be more careful about propagation
17478 * profile-count.h (profile_probability::guessed,
17479 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
17481 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
17483 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
17485 * doc/invoke.texi (rcpc architecture extension): Document it.
17487 2017-07-03 Richard Biener <rguenther@suse.de>
17489 PR tree-optimization/60510
17490 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
17491 the scalar reduction PHI and use it.
17492 (vectorizable_reduction): Properly guard the single_defuse_cycle
17493 path for non-SLP reduction chains where we cannot use it.
17494 Rework reduc_def/index and vector type deduction. Rework
17495 vector operand gathering during reduction op code-gen.
17496 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
17497 chains dissolve the chain and leave it to non-SLP reduction
17500 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17502 * tree-data-ref.h (dr_alignment): Declare.
17503 * tree-data-ref.c (dr_alignment): New function.
17504 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
17505 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
17507 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
17509 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17511 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
17512 and base_misalignment fields.
17513 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
17514 * tree-data-ref.c: Include builtins.h.
17515 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
17516 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
17517 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
17518 * tree-vect-data-refs.c: Include tree-cfg.h.
17519 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
17520 fields instead of calculating an alignment here.
17521 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
17522 innermost_loop_behavior fields.
17524 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17526 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
17528 (DR_STEP_ALIGNMENT): New macro.
17529 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
17530 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
17531 (create_data_ref): Print it.
17532 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
17533 to tell whether the step preserves vector (mis)alignment.
17534 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17535 Move the check for an integer step and generalise to all INTEGER_CST.
17536 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
17537 Print the outer step alignment.
17539 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17541 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
17542 with offset_alignment.
17543 (DR_ALIGNED_TO): Delete.
17544 (DR_OFFSET_ALIGNMENT): New macro.
17545 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
17546 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
17547 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
17548 (create_data_ref): Likewise.
17549 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17550 (vect_analyze_data_refs): Likewise.
17551 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
17552 creating dummy innermost behavior.
17554 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17556 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
17557 with a "innermost_loop_behavior *" and refeence tree.
17558 * tree-data-ref.c (dr_analyze_innermost): Likewise.
17559 (create_data_ref): Update call accordingly.
17560 * tree-predcom.c (find_looparound_phi): Likewise.
17562 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17564 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
17565 fields with dr_wrt_vec_loop.
17566 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
17567 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
17568 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
17569 (vect_dr_behavior): New function.
17570 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17571 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
17572 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
17573 track whether the step preserves the misalignment.
17574 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17575 Use vect_dr_behavior.
17576 (vect_setup_realignment): Update call accordingly.
17577 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
17578 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
17579 call to vect_create_addr_base_for_vector_ref.
17580 (vect_create_cond_for_align_checks): Likewise.
17581 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
17582 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
17583 (vect_recog_mask_conversion_pattern): Likewise.
17584 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
17585 (new_stmt_vec_info): Remove redundant zeroing.
17587 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17589 * common/config/arm/arm-common.c (arm_be8_option): New function.
17590 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
17591 (ISA_ARMv6): Add isa_bit_be8.
17592 * config/arm/arm.h (arm_be8_option): Add prototype.
17593 (BE8_SPEC_FUNCTION): New define.
17594 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
17595 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
17596 (mlittle-endian): Similarly.
17597 (mbe8, mbe32): New options.
17598 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
17599 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
17601 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17603 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
17605 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17607 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
17608 (cleanup_tree_cfg_bb): Use it.
17609 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
17611 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
17613 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17616 * loop-doloop.c (add_test): Update profile.
17618 2017-07-03 Martin Liska <mliska@suse.cz>
17621 * sanopt.c (rewrite_usage_of_param): New function.
17622 (sanitize_rewrite_addressable_params): Likewise.
17623 (pass_sanopt::execute): Call rewrite_usage_of_param.
17625 2017-07-03 Richard Biener <rguenther@suse.de>
17627 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
17628 back to using VIEW_CONVERT_EXPR.
17630 2017-07-03 Martin Liska <mliska@suse.cz>
17633 * doc/extend.texi: Document when a resolver function is
17634 generated for target_clones.
17636 2017-07-03 Martin Liska <mliska@suse.cz>
17638 * asan.c (asan_emit_stack_protection): Unpoison just red zones
17639 and shadow memory of auto variables which are subject of
17640 use-after-scope sanitization.
17641 (asan_expand_mark_ifn): Add do set only when is_poison.
17643 2016-07-03 Richard Biener <rguenther@suse.de>
17645 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
17647 (vect_force_simple_reduction): Record reduction def -> phi mapping.
17648 (vectorizable_reduction): Perform reduction PHI creation when
17649 visiting a reduction PHI and adjust and simplify code generation
17650 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
17651 (vect_transform_loop): Visit reduction PHIs.
17652 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
17653 defs into the SLP tree.
17654 (vect_build_slp_tree): Reduction defs terminate the recursion.
17655 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
17657 (vect_get_vec_defs_for_stmt_copy): Export.
17658 (vect_get_vec_defs): Likewise.
17659 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
17661 (vect_get_vec_defs_for_stmt_copy): Declare.
17662 (vect_get_vec_defs): Likewise.
17664 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17666 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
17667 parameter with a "loop" parameter and use it instead of the
17668 loop containing DR_STMT. Don't check simple_iv when doing
17669 BB analysis. Describe the two analysis modes in the comment.
17671 2017-07-03 Tom de Vries <tom@codesourcery.com>
17673 PR tree-optimization/69468
17674 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
17675 (find_same_succ_bb): Handle ignore_edge_flags.
17677 2017-07-03 Tom de Vries <tom@codesourcery.com>
17679 PR tree-optimization/81192
17680 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
17682 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
17684 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
17686 2017-07-03 Tom de Vries <tom@codesourcery.com>
17688 PR tree-optimization/81192
17689 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
17690 BB_SAME_SUCC (bb) == NULL.
17692 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17694 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
17697 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17699 * dumpfile.c: Include profile-count.h
17700 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
17702 (insert_cond_bb): Update profile.
17703 * tree-cfg.h (insert_cond_bb): Update prototype.
17704 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
17705 * tree-dump.c: Do not include tree-cfg.
17707 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17709 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
17711 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17713 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
17716 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17718 * tree-complex.c (expand_complex_div_wide): update profile.
17720 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17721 Alan Hayward <alan.hayward@arm.com>
17722 David Sherwood <david.sherwood@arm.com>
17724 * Makefile.in (MACHMODE_H): Remove insn-modes.h
17725 (CORETYPES_H): New define.
17726 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
17727 (insn-modes-inline.h, s-modes-inline-h): New rules.
17728 (generated_files): Add insn-modes-inline.h.
17729 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
17730 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
17731 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
17732 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
17733 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
17734 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
17735 (build/gencodes.o, build/genconditions.o): Likewise.
17736 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
17737 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
17738 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
17739 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
17740 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
17741 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
17742 * coretypes.h: Include everything up to real.h for generators.
17743 Include insn-modes.h first. Include wide-int-print.h after
17744 wide-int.h. Include insn-modes-inline.h and then machmode.h.
17745 * machmode.h: Don't include insn-modes.h here.
17746 * function-tests.c: Remove includes of signop.h, machmode.h,
17747 double-int.h and wide-int.h.
17749 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
17751 * optc-save-gen.awk: Likewise.
17752 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
17753 * godump.c: Remove include of wide-int-print.h.
17754 * pretty-print.h: Likewise.
17755 * wide-int-print.cc: Likewise.
17756 * wide-int.cc: Likewise.
17757 * hash-map-tests.c: Remove include of signop.h.
17758 * hash-set-tests.c: Likewise.
17759 * rtl-tests.c: Likewise.
17760 * mkconfig.sh: Remove include of machmode.h.
17761 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
17763 (emit_insn_modes_inline_h): ...this new function. Emit the code
17764 into an insn-modes-inline.h header file, adding appropriate
17765 include guards and end comments.
17766 (emit_insn_modes_c_header): Remove include of machmode.h.
17767 (emit_min_insn_modes_c_header): Include coretypes.h rather than
17769 (main): Handle -i flag and call emit_insn_modes_inline_h when
17772 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17774 * tree-ssa-strlen.c (strinfo): Rename the length field to
17775 nonzero_chars. Add a full_string_p field.
17776 (compare_nonzero_chars, zero_length_string_p): New functions.
17777 (get_addr_stridx): Add an offset_out parameter.
17778 Use compare_nonzero_chars.
17779 (get_stridx): Update accordingly. Use compare_nonzero_chars.
17780 (new_strinfo): Update after above changes to strinfo.
17781 (set_endptr_and_length): Set full_string_p.
17782 (get_string_length): Update after above changes to strinfo.
17783 (unshare_strinfo): Update call to new_strinfo.
17784 (maybe_invalidate): Likewise.
17785 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
17786 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
17787 as a uhwi instead of an shwi. Update after above changes to
17788 strinfo and new_strinfo.
17789 (zero_length_string): Assert that chainsi contains full strings.
17790 Use zero_length_string_p. Update call to new_strinfo.
17791 (adjust_related_strinfos): Update after above changes to strinfo.
17792 Copy full_string_p from origsi.
17793 (adjust_last_stmt): Use zero_length_string_p.
17794 (handle_builtin_strlen): Update after above changes to strinfo and
17795 new_strinfo. Install the lhs as the string length if the previous
17796 entry didn't describe a full string.
17797 (handle_builtin_strchr): Update after above changes to strinfo
17799 (handle_builtin_strcpy): Likewise.
17800 (handle_builtin_strcat): Likewise.
17801 (handle_builtin_malloc): Likewise.
17802 (handle_pointer_plus): Likewise.
17803 (handle_builtin_memcpy): Likewise. Track nonzero characters
17804 that aren't necessarily followed by a nul terminator.
17805 (handle_char_store): Likewise.
17807 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17809 PR tree-optimization/80769
17810 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
17811 for malloc and calloc. Document the new invariant that all related
17812 strinfos have delayed lengths or none do.
17813 (verify_related_strinfos): Move earlier in file.
17814 (set_endptr_and_length): New function, split out from...
17815 (get_string_length): ...here. Also set the lengths of related
17817 (zero_length_string): Assert that chainsi has known (rather than
17819 (adjust_related_strinfos): Likewise.
17821 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17823 PR tree-optimization/81136
17824 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
17825 assert that two references with the same misalignment have the same
17826 compile-time misalignment if those compile-time misalignments
17829 2017-07-01 Andi Kleen <ak@linux.intel.com>
17831 * print-tree.c (print_node): Print all attributes.
17833 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17835 * cfg.c (scale_bbs_frequencies): New function.
17836 * cfg.h (scale_bbs_frequencies): Declare it.
17837 * cfgloopanal.c (single_likely_exit): Cleanup.
17838 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
17840 (scale_loop_profile): Likewise.
17841 (loop_version): Likewise.
17842 (create_empty_loop_on_edge): Update.
17843 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
17844 scale_loop_frequencies, scale_loop_profile, loopify,
17845 loop_version): Update prototypes.
17846 * modulo-sched.c (sms_schedule): Update.
17847 * predict.c (unlikely_executed_edge_p): Also check probability.
17848 (probably_never_executed_edge_p): Fix typo.
17849 * tree-if-conv.c (version_loop_for_if_conversion): Update.
17850 * tree-parloops.c (gen_parallel_loop): Update.
17851 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
17852 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
17853 * tree-ssa-loop-split.c (split_loop): Update.
17854 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
17855 * tree-vect-loop-manip.c (vect_do_peeling): Update.
17856 (vect_loop_versioning): Update.
17857 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
17859 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17861 * trans-mem.c (split_bb_make_tm_edge): Update profile.
17863 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17865 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
17866 to keep profile consistent.
17868 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17870 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
17871 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
17872 * profile-count.h (max_safe_multiplier): Make unsigned.
17873 (profile_count::guessed_zero): New.
17875 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17877 * bb-reorder.c (fix_up_crossing_landing_pad,
17878 fix_crossing_conditional_branches): Use make_single_succ_edge
17879 to keep profile consistent.
17881 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17883 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
17886 2017-07-01 Jakub Jelinek <jakub@redhat.com>
17889 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
17890 the right scopes, make sure cond_jump isn't preserved between multiple
17891 iterations. Search for fallthru edge whenever there are 3+ edges and
17892 use find_fallthru_edge for it.
17894 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17896 Patch by Alexander Monakov <amonakov@ispras.ru>
17897 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
17898 probabilities consistently.
17900 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17902 * pa.c (pa_expand_compare_and_swap_loop): Update call of
17903 emit_cmp_and_jump_insns.
17905 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17908 * tree-inline.c (expand_call_inline): Combine profile statuses.
17910 2017-06-30 Andrew Pinski <apinski@cavium.com>
17912 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
17913 fold_stmt returned true.
17915 2017-06-30 Nathan Sidwell <nathan@acm.org>
17917 * ggc.h (empty_string): Delete.
17918 * cfgexpand.c (expand_asm_stmt): Use plain "".
17919 * optabs.c (expand_asm_memory_barrier): Likewise.
17920 * stringpool.c (empty_string): Delete.
17921 (digit_vector, digit_string): Delete.
17922 (ggc_alloc_string): Use plain "", don't optimize single digit
17923 strings. Use ggc_alloc_atomic.
17925 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
17927 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
17928 comparison set and one other set, use the cost of the non-comparison
17931 2017-06-30 Nathan Sidwell <nathan@acm.org>
17933 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
17936 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
17938 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
17939 loops. Remove now unneeded calls to gimple_switch_set_label() that
17940 just set removed labels to NULL_TREE.
17942 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
17944 * tree-ssanames.c (set_range_info_raw): Abstract from ...
17945 (set_range_info): ...here. Only call set_range_info_raw if domain
17947 (set_nonzero_bits): Call set_range_info_raw.
17948 * tree-ssanames.h (set_range_info_raw): New.
17950 2017-06-30 Jakub Jelinek <jakub@redhat.com>
17953 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
17954 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
17955 of nonimmediate_operand and <store_mask_constraint> instead of m
17956 for the input operand. For V8FI iterator, always split if input
17957 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
17958 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
17959 <store_mask_predicate> instead of register_operand and
17960 <store_mask_constraint> instead of v for the input operand. Make
17961 sure both operands aren't MEMs for if not <mask_applied>.
17963 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
17965 * lto-wrapper.c (copy_file) Close both file descriptors before
17968 2017-06-30 Martin Liska <mliska@suse.cz>
17971 * multiple_target.c (create_dispatcher_calls): Make ifunc
17972 also for function that don't have calls or are not referenced.
17974 2017-06-30 Richard Biener <rguenther@suse.de>
17976 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
17977 analyze the first scalar stmt. Move vector type computation
17978 for the BB case here from ...
17979 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
17980 live operation processing in the SLP case properly.
17982 2017-06-30 Richard Biener <rguenther@suse.de>
17984 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
17986 2017-06-30 Martin Liska <mliska@suse.cz>
17989 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
17990 before BUILT_IN_UNWIND_RESUME when ASAN is used.
17992 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
17994 * doc/invoke.texi (AArch64): Add missing options and remove redundant
17997 2017-06-30 Richard Biener <rguenther@suse.de>
17999 PR tree-optimization/81249
18000 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
18001 condition reduction result to original scalar type.
18003 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18005 * profile-count.h (enum profile_quality): Fix typos and whitespace
18008 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18010 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
18011 type for branch probabilities.
18013 2017-06-29 Julian Brown <julian@codesourcery.com>
18014 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18016 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
18017 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
18018 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
18019 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
18021 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18023 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
18024 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
18025 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
18026 CC usage from generic code to here.
18027 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
18028 CC usage into the target macros.
18030 2017-06-29 Maya Rashish <coypu@sdf.org>
18032 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
18035 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18037 * arm/arm-builtins.c: Include profile-count.h
18038 * except.c (sjlj_emit_function_enter): Use
18039 profile_probability::unlikely.
18041 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18043 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
18044 and tocrel_offset be pointer args rather than implicitly using
18046 (legitimate_constant_pool_address_p, rs6000_emit_move,
18047 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
18048 tocrel_offset and use in toc_relative_expr_p call.
18049 (print_operand, print_operand_address): Use static tocrel_base_oac
18050 and tocrel_offset_oac.
18051 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
18054 2017-06-29 Maya Rashish <coypu@sdf.org>
18056 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
18058 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
18060 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
18061 objects, take into account only the alignment of 'op0' and 'mode1' if
18064 2017-06-29 Steve Ellcey <sellcey@cavium.com>
18066 * ccmp.c (ccmp_tree_comparison_p): New function.
18067 (ccmp_candidate_p): Update to use above function.
18068 (get_compare_parts): New function.
18069 (expand_ccmp_next): Update to use new functions.
18070 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
18072 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
18073 take mode as argument.
18074 * ccmp.h (expand_ccmp_expr): Add mode as argument.
18075 * expr.c (expand_expr_real_1): Pass mode as argument.
18077 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
18079 * combine.c (combine_instructions): Print insns to dump_file, together
18082 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18084 * asan.c (asan_emit_stack_protection): Update.
18085 (create_cond_insert_point): Update.
18086 * auto-profile.c (afdo_propagate_circuit): Update.
18087 * basic-block.h (struct edge_def): Turn probability to
18088 profile_probability.
18089 (EDGE_FREQUENCY): Update.
18090 * bb-reorder.c (find_traces_1_round): Update.
18091 (better_edge_p): Update.
18092 (sanitize_hot_paths): Update.
18093 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
18094 (make_single_succ_edge): Update.
18095 (check_bb_profile): Update.
18096 (dump_edge_info): Update.
18097 (update_bb_profile_for_threading): Update.
18098 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
18099 probabilitycount to 0.
18100 * cfgbuild.c (compute_outgoing_frequencies): Update.
18101 * cfgcleanup.c (try_forward_edges): Update.
18102 (outgoing_edges_match): Update.
18103 (try_crossjump_to_edge): Update.
18104 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
18105 (expand_gimple_tailcall): Update.
18106 (construct_init_block): Use make_single_succ_edge.
18107 (construct_exit_block): Use make_single_succ_edge.
18108 * cfghooks.c (verify_flow_info): Update.
18109 (redirect_edge_succ_nodup): Update.
18110 (split_edge): Update.
18111 (account_profile_record): Update.
18112 * cfgloopanal.c (single_likely_exit): Update.
18113 * cfgloopmanip.c (scale_loop_profile): Update.
18114 (set_zero_probability): Remove.
18115 (duplicate_loop_to_header_edge): Update.
18116 * cfgloopmanip.h (loop_version): Update prototype.
18117 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
18118 (force_nonfallthru_and_redirect): Update.
18119 (update_br_prob_note): Update.
18120 (rtl_verify_edges): Update.
18121 (purge_dead_edges): Update.
18122 (rtl_lv_add_condition_to_bb): Update.
18123 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
18124 * cgraphunit.c (init_lowered_empty_function): Update.
18125 (cgraph_node::expand_thunk): Update.
18126 * cilk-common.c: Include profile-count.h
18127 * dojump.c (inv): Remove.
18128 (jumpifnot): Update.
18129 (jumpifnot_1): Update.
18130 (do_jump_1): Update.
18132 (do_jump_by_parts_greater_rtx): Update.
18133 (do_compare_rtx_and_jump): Update.
18134 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
18135 do_jump_1. do_compare_rtx_and_jump): Update prototype.
18136 * dwarf2cfi.c: Include profile-count.h
18137 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
18138 (sjlj_emit_dispatch_table): Likewise.
18139 * explow.c: Include profile-count.h
18140 * expmed.c (emit_store_flag_force): Update.
18141 (do_cmp_and_jump): Update.
18142 * expr.c (compare_by_pieces_d::generate): Update.
18143 (compare_by_pieces_d::finish_mode): Update.
18144 (emit_block_move_via_loop): Update.
18145 (store_expr_with_bounds): Update.
18146 (store_constructor): Update.
18147 (expand_expr_real_2): Update.
18148 (expand_expr_real_1): Update.
18149 * expr.h (try_casesi, try_tablejump): Update prototypes.
18150 * gimple-pretty-print.c (dump_probability): Update.
18151 (dump_profile): New.
18152 (dump_gimple_label): Update.
18153 (dump_gimple_bb_header): Update.
18154 * graph.c (draw_cfg_node_succ_edges): Update.
18155 * hsa-gen.c (convert_switch_statements): Update.
18156 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
18157 (find_if_case_1): Update.
18158 (find_if_case_2): Update.
18159 * internal-fn.c (expand_arith_overflow_result_store): Update.
18160 (expand_addsub_overflow): Update.
18161 (expand_neg_overflow): Update.
18162 (expand_mul_overflow): Update.
18163 (expand_vector_ubsan_overflow): Update.
18164 * ipa-cp.c (good_cloning_opportunity_p): Update.
18165 * ipa-split.c (split_function): Use make_single_succ_edge.
18166 * ipa-utils.c (ipa_merge_profiles): Update.
18167 * loop-doloop.c (add_test): Update.
18168 (doloop_modify): Update.
18169 * loop-unroll.c (compare_and_jump_seq): Update.
18170 (unroll_loop_runtime_iterations): Update.
18171 * lra-constraints.c (lra_inheritance): Update.
18172 * lto-streamer-in.c (input_cfg): Update.
18173 * lto-streamer-out.c (output_cfg): Update.
18174 * mcf.c (adjust_cfg_counts): Update.
18175 * modulo-sched.c (sms_schedule): Update.
18176 * omp-expand.c (expand_omp_for_init_counts): Update.
18177 (extract_omp_for_update_vars): Update.
18178 (expand_omp_ordered_sink): Update.
18179 (expand_omp_for_ordered_loops): Update.
18180 (expand_omp_for_generic): Update.
18181 (expand_omp_for_static_nochunk): Update.
18182 (expand_omp_for_static_chunk): Update.
18183 (expand_cilk_for): Update.
18184 (expand_omp_simd): Update.
18185 (expand_omp_taskloop_for_outer): Update.
18186 (expand_omp_taskloop_for_inner): Update.
18187 * omp-simd-clone.c (simd_clone_adjust): Update.
18188 * optabs.c (expand_doubleword_shift): Update.
18189 (expand_abs): Update.
18190 (emit_cmp_and_jump_insn_1): Update.
18191 (expand_compare_and_swap_loop): Update.
18192 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
18193 * predict.c (predictable_edge_p): Update.
18194 (edge_probability_reliable_p): Update.
18195 (set_even_probabilities): Update.
18196 (combine_predictions_for_insn): Update.
18197 (combine_predictions_for_bb): Update.
18198 (propagate_freq): Update.
18199 (estimate_bb_frequencies): Update.
18200 (force_edge_cold): Update.
18201 * profile-count.c (profile_count::dump): Add missing space into dump.
18202 (profile_count::debug): Add newline.
18203 (profile_count::differs_from_p): Explicitly convert to unsigned.
18204 (profile_count::stream_in): Update.
18205 (profile_probability::dump): New member function.
18206 (profile_probability::debug): New member function.
18207 (profile_probability::differs_from_p): New member function.
18208 (profile_probability::differs_lot_from_p): New member function.
18209 (profile_probability::stream_in): New member function.
18210 (profile_probability::stream_out): New member function.
18211 * profile-count.h (profile_count_quality): Rename to ...
18212 (profile_quality): ... this one.
18213 (profile_probability): New.
18214 (profile_count): Update.
18215 * profile.c (compute_branch_probabilities): Update.
18216 * recog.c (peep2_attempt): Update.
18217 * sched-ebb.c (schedule_ebbs): Update.
18218 * sched-rgn.c (find_single_block_region): Update.
18219 (compute_dom_prob_ps): Update.
18220 (schedule_region): Update.
18221 * sel-sched-ir.c (compute_succs_info): Update.
18222 * stmt.c (struct case_node): Update.
18223 (do_jump_if_equal): Update.
18224 (get_outgoing_edge_probs): Update.
18225 (conditional_probability): Update.
18226 (emit_case_dispatch_table): Update.
18227 (expand_case): Update.
18228 (expand_sjlj_dispatch_table): Update.
18229 (emit_case_nodes): Update.
18230 * targhooks.c: Update.
18231 * tracer.c (better_p): Update.
18232 (find_best_successor): Update.
18233 * trans-mem.c (expand_transaction): Update.
18234 * tree-call-cdce.c: Update.
18235 * tree-cfg.c (gimple_split_edge): Upate.
18236 (move_sese_region_to_fn): Upate.
18237 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
18238 * tree-eh.c (lower_resx): Upate.
18239 (cleanup_empty_eh_move_lp): Upate.
18240 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18241 * tree-inline.c (copy_edges_for_bb): Update.
18242 (copy_cfg_body): Update.
18243 * tree-parloops.c (gen_parallel_loop): Update.
18244 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
18245 (gimple_gen_time_profiler): Update.
18246 * tree-ssa-dce.c (remove_dead_stmt): Update.
18247 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
18248 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
18249 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
18250 (unloop_loops): Update.
18251 (try_peel_loop): Update.
18252 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18253 * tree-ssa-loop-split.c (connect_loops): Update.
18254 (split_loop): Update.
18255 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
18256 (hoist_guard): Update.
18257 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
18258 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
18259 (value_replacement): Update.
18260 * tree-ssa-reassoc.c (branch_fixup): Update.
18261 * tree-ssa-tail-merge.c (replace_block_by): Update.
18262 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
18263 (create_edge_and_update_destination_phis): Update.
18264 (compute_path_counts): Update.
18265 (recompute_probabilities): Update.
18266 (update_joiner_offpath_counts): Update.
18267 (freqs_to_counts_path): Update.
18268 (duplicate_thread_path): Update.
18269 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
18270 (struct switch_conv_info): Update.
18271 (gen_inbound_check): Update.
18272 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
18273 (vect_do_peeling): Update.
18274 (vect_loop_versioning): Update.
18275 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18276 (optimize_mask_stores): Update.
18277 * ubsan.c (ubsan_expand_null_ifn): Update.
18278 * value-prof.c (gimple_divmod_fixed_value): Update.
18279 (gimple_divmod_fixed_value_transform): Update.
18280 (gimple_mod_pow2): Update.
18281 (gimple_mod_pow2_value_transform): Update.
18282 (gimple_mod_subtract): Update.
18283 (gimple_mod_subtract_transform): Update.
18284 (gimple_ic): Update.
18285 (gimple_stringop_fixed_value): Update.
18286 (gimple_stringops_transform): Update.
18287 * value-prof.h: Update.
18289 2017-06-29 Carl Love <cel@us.ibm.com>
18291 * config/rs6000/rs6000-c.c: Add support for built-in functions
18292 vector signed int vec_signed (vector float);
18293 vector signed long long vec_signed (vector double);
18294 vector signed int vec_signed2 (vector double, vector double);
18295 vector signed int vec_signede (vector double);
18296 vector signed int vec_signedo (vector double);
18297 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
18298 instruction generator.
18299 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18300 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
18301 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
18303 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
18304 vunsignede_v2df): Add define_expands.
18305 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
18306 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
18307 VEC_UNSIGNEDO): Add definitions.
18308 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18309 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
18310 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
18311 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
18312 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
18313 * config/rs6000/altivec.h (vec_signed, vec_signed2,
18314 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
18315 vec_unsignede, vec_unsignedo): Add builtin defines.
18316 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
18318 * doc/extend.texi: Update the built-in documentation file for the
18319 new built-in functions.
18321 2017-06-29 Richard Biener <rguenther@suse.de>
18323 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
18324 reduction chains to LOOP_VINFO_REDUCTIONS.
18325 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
18326 SLP reductions after processing reduction chains.
18328 2017-06-29 Nathan Sidwell <nathan@acm.org>
18330 * builtins.c (fold_builtin_FUNCTION): Use
18331 lang_hooks.decl_printable_name.
18333 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
18335 PR middle-end/81194
18336 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
18337 with only one label.
18338 * stmt.c (expand_case): Assert NCASES is greater than one.
18340 2017-06-29 Richard Biener <rguenther@suse.de>
18342 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
18344 (group_case_labels): Likewise.
18345 (find_taken_edge): Push sanity checking on val to workers...
18346 (find_taken_edge_cond_expr): ... here
18347 (find_taken_edge_switch_expr): ... and here, handle cases
18348 with just a default label.
18349 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
18350 (group_case_labels): Likewise.
18351 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
18352 group_case_labels does anything cleanup the CFG again.
18354 2017-06-29 Bin Cheng <bin.cheng@arm.com>
18356 PR tree-optimization/81196
18357 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
18358 exit condition comparing two IVs.
18360 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
18362 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
18363 profile to the dummy entry at the end of the list of architectures.
18364 * config/arm/arm-cpu-cdata.h: Regenerated.
18366 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18367 Michael Collison <michael.collison@arm.com>
18370 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
18372 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
18373 (*aarch64_reg_<mode>3_minus_mask): New pattern.
18374 (*aarch64_<optab>_reg_di3_mask2): New pattern.
18375 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
18376 of shift when the shift amount is masked with constant equal to
18377 the size of the mode.
18378 * config/aarch64/predicates.md (subreg_lowpart_operator): New
18381 2017-06-29 Martin Liska <mliska@suse.cz>
18383 * config/i386/i386.opt: Change range from [1,5] to [0,5].
18385 2017-06-29 Yury Gribov <tetra2005@gmail.com>
18388 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
18391 (edge_growth_cache_entry::edge_growth_cache_entry): New
18393 (reset_edge_growth_cache): Update to use constructor.
18395 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18397 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
18398 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
18399 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
18401 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
18403 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
18404 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
18406 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
18408 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
18409 (*-linux-uclibc*): Add t-uclibc tmake_file.
18410 * config/t-musl: New.
18411 * config/t-uclibc: New.
18413 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
18415 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
18417 (gen_comm_data): Emit architectural setting of arch_prof.
18418 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
18420 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
18421 (armv8-m.base, armv8-m.main): Likewise.
18422 * arm-protos.h (arm_build_target): Add profile field.
18423 (arch_option): Likewise.
18424 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
18426 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
18427 arm_active_target.profile.
18429 2017-06-28 Richard Biener <rguenther@suse.de>
18431 PR middle-end/81227
18432 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
18433 TYPE_OVERFLOW_WRAPS.
18434 * match.pd (negate_expr_p): Likewise.
18435 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
18436 fold_build2, not fold_binary.
18438 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
18440 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
18441 Convert memory address to Pmode.
18442 (aarch64_print_operand): Assert MEM operands are always Pmode.
18444 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
18447 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
18448 Remove redundant if.
18449 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
18450 * config/arm/aarch-common-protos.h
18451 (aarch_forward_to_shift_is_not_shifted_re): Remove.
18452 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
18454 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
18457 * multiple_target.c (create_dispatcher_calls): Set the default
18458 clone to be static, not public.
18460 2017-06-28 Richard Biener <rguenther@suse.de>
18462 * tree-vect-loop.c (vectorizable_reduction): Move special
18463 cond reduction IV var creation ...
18464 (vect_create_epilog_for_reduction): ... here. Remove induction_index
18465 parameter. Use STMT_VINFO_VECTYPE.
18466 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
18469 2017-06-28 Martin Liska <mliska@suse.cz>
18472 * ipa-visibility.c (non_local_p): Handle visibility.
18474 2017-06-28 Martin Liska <mliska@suse.cz>
18477 * common.opt: Add IntegerRange to various options.
18478 * opt-functions.awk (integer_range_info): New function.
18479 * optc-gen.awk: Add integer_range_info to cl_options struct.
18480 * opts-common.c (decode_cmdline_option): Handle
18481 CL_ERR_INT_RANGE_ARG.
18482 (cmdline_handle_error): Likewise.
18483 * opts.c (print_filtered_help): Show valid interval in
18484 when --help is provided.
18485 * opts.h (struct cl_option): Add range_min and range_max fields.
18486 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
18488 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
18490 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
18491 (x * C EQ/NE y * C): New transformation.
18493 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
18495 * genmultilib (combination_space): Accept '+' in option names.
18497 2017-06-28 Martin Liska <mliska@suse.cz>
18500 * asan.c (instrument_derefs): Bail out inner references
18501 that are hard register variables.
18503 2017-06-28 Jakub Jelinek <jakub@redhat.com>
18506 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
18507 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
18509 2017-06-28 Richard Biener <rguenther@suse.de>
18511 * tree-vectorizer.h (vect_get_vec_defs): Remove.
18512 (vect_get_slp_defs): Adjust.
18513 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
18515 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
18517 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
18518 get_initial_defs_for_reduction instead of vect_get_vec_defs.
18519 (vectorizable_reduction): Adjust.
18520 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
18522 (vect_get_slp_defs): Likewise.
18523 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
18524 (vectorizable_bswap): Adjust.
18525 (vectorizable_call): Likewise.
18526 (vectorizable_conversion): Likewise.
18527 (vectorizable_assignment): Likewise.
18528 (vectorizable_shift): Likewise.
18529 (vectorizable_operation): Likewise.
18530 (vectorizable_store): Likewise.
18531 (vectorizable_condition): Likewise.
18532 (vectorizable_comparison): Likewise.
18534 2017-06-28 Michael Collison <michael.collison@arm.com>
18537 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
18539 (arm_gen_movmemqi): Removed unused variable 'i'.
18540 Convert 'for' loop into 'while' loop.
18541 (arm_expand_prologue): Remove last unnecessary set of insn.
18542 (thumb_pop): Remove unused variable 'pushed_words'.
18543 (thumb_exit): Remove last unnecessary set of regs_to_pop.
18545 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18547 * config/s390/predicates.md: Use s390_rel_address_ok_p.
18548 * config/s390/s390-protos.h: Add prototype of
18549 s390_rel_address_ok_p.
18550 * config/s390/s390.c (s390_got_symbol): New function.
18551 (s390_rel_address_ok_p): New function.
18552 (legitimize_pic_address): Use s390_rel_address_ok_p.
18553 (s390_load_got): Use s390_got_symbol.
18554 (s390_option_override): Issue error if
18555 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
18556 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
18558 * config/s390/s390.opt: New option mpic-data-is-text-relative.
18560 2017-06-27 Andrew Pinski <apinski@cavium.com>
18562 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
18563 (X * copysign (1.0, X)): New pattern.
18564 (X * copysign (1.0, -X)): New pattern.
18565 (copysign (-1.0, CST)): New pattern.
18567 2017-06-27 Joseph Myers <joseph@codesourcery.com>
18569 * genmultilib (combination_space): Remove variable.
18570 Validate reuse rules against regular expression for any sequence
18571 of multilib options in any order.
18573 2017-06-27 Michael Collison <michael.collison@arm.com>
18575 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
18576 call aarch64_split_simd_combine.
18577 * (aarch64_combine_internal<mode>): Delete pattern.
18578 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
18579 Allow register and subreg operands.
18581 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18583 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
18584 specific need, just fallback on defaults.
18585 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
18587 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18588 Olivier Hainque <hainque@adacore.com>
18590 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
18592 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
18593 targets. Pick a default if no particular attempt applied.
18594 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
18597 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18599 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
18600 (x86_64-wrs-vxworks7): Likewise.
18602 2017-06-27 Marek Polacek <polacek@redhat.com>
18605 * ubsan.c (instrument_null): Check get_base_address's result for null.
18607 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18609 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
18611 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18613 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
18614 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
18615 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
18616 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
18617 New function types.
18618 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
18619 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
18620 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
18621 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
18622 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
18623 BUILT_IN_FEUPDATEENV): New builtins.
18624 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
18625 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
18626 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
18627 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
18629 (builtin_structptr_types): Adjust size.
18630 * tree.c (builtin_structptr_types): Add four entries.
18632 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18633 Olivier Hainque <hainque@adacore.com>
18635 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
18636 (TLS_SYM): New local macro, forcing reference to __tls__ on
18637 link command lines for VxWorks 7 RTPs, triggering initialization
18639 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
18640 OS features TLS support, true for RTPs on VxWorks 7.
18641 * config/vxworks.c (vxworks_override_options): Setup emutls
18644 2017-06-27 Jakub Jelinek <jakub@redhat.com>
18646 * predict.c (test_prediction_value_range): Use -1U instead of -1
18647 to avoid narrowing conversion warning.
18648 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
18649 to avoid narrowing conversion warning.
18650 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
18652 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
18654 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18656 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
18657 64bit configurations.
18658 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
18659 (SIZE_TYPE): Likewise.
18660 * config/vxworks.c (vxworks_emutls_var_fields): Use
18661 long_unsigned_type_node instead of unsigned_type_node as the offset
18662 field type, which is "pointer" mode in emutls.c.
18664 2017-06-27 Jakub Jelinek <jakub@redhat.com>
18667 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
18669 PR middle-end/81207
18670 * gimple-fold.c (replace_call_with_call_and_fold): Handle
18671 gimple_vuse copying separately from gimple_vdef copying.
18673 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18675 * value-prof.c (free_hist): Remove call to memset and the enclosing if
18678 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
18679 Olivier Hainque <hainque@adacore.com>
18681 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
18682 for all vxworks7 targets.
18683 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
18684 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
18685 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
18686 variations for VX6/VX7 and 32/64bits later on in ...
18687 (VXWORKS_LIB_SPEC): Leverage new macros.
18688 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
18689 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
18691 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
18693 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
18694 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
18696 2017-06-26 Carl Love <cel@us.ibm.com>
18698 * config/rs6000/rs6000-c.c: Add support for built-in functions
18699 vector bool char vec_reve (vector bool char);
18700 vector signed char vec_reve (vector signed char);
18701 vector unsigned char vec_reve (vector unsigned char);
18702 vector bool int vec_reve (vector bool int);
18703 vector signed int vec_reve (vector signed int);
18704 vector unsigned int vec_reve (vector unsigned int);
18705 vector bool long long vec_reve (vector bool long long);
18706 vector signed long long vec_reve (vector signed long long);
18707 vector unsigned long long vec_reve (vector unsigned long long);
18708 vector bool short vec_reve (vector bool short);
18709 vector signed short vec_reve (vector signed short);
18710 vector double vec_reve (vector double);
18711 vector float vec_reve (vector float);
18712 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
18713 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
18714 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
18715 (altivec_vreve): New pattern.
18716 * config/rs6000/altivec.h (vec_reve): New define.
18717 * doc/extend.texi (vec_rev): Update the built-in documentation file
18718 for the new built-in functions.
18720 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18722 PR tree-optimization/71815
18723 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
18725 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
18727 (slsr_process_phi): Likewise.
18728 (replace_uncond_cands_and_profitable_phis): Don't replace a
18729 multiply candidate with a stride of 1 (copy or cast).
18730 (phi_incr_cost): Call uses_consumed_by_stmt rather than
18732 (lowest_cost_path): Likewise.
18733 (total_savings): Likewise.
18735 2017-06-26 Richard Biener <rguenther@suse.de>
18738 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
18739 Use def_builtin_pure for all gather builtins.
18741 2017-06-26 Richard Biener <rguenther@suse.de>
18743 PR tree-optimization/81203
18744 * tree-tailcall.c (find_tail_calls): Do not move stmts into
18745 non-dominating BBs.
18747 2017-06-26 Marek Polacek <polacek@redhat.com>
18750 * doc/invoke.texi: Document -Wmultistatement-macros.
18752 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
18754 * doc/sourcebuild.texi (ARM-specific attributes): Document new
18755 arm_neon_ok_no_float_abi effective target.
18757 2017-06-26 Richard Biener <rguenther@suse.de>
18759 PR tree-optimization/80928
18760 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
18761 (copy_bbs): Set BB_DUPLICATED flag early.
18762 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
18764 (execute_on_shrinking_pred): Likewise.
18765 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
18766 BB_DUPLICATED blocks.
18767 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
18768 iterate over all PHIs considering removal of *gsi.
18770 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
18772 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
18775 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18777 * config/rs6000/rs6000-string.c: (expand_block_clear,
18778 do_load_for_compare, select_block_compare_mode,
18779 compute_current_alignment, expand_block_compare,
18780 expand_strncmp_align_check, expand_strn_compare,
18781 expand_block_move, rs6000_output_load_multiple)
18782 Move functions related to string/block move/compare
18783 to a separate file.
18784 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
18785 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
18786 for this function which is now used in two files.
18787 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
18788 * config.gcc: Add rs6000-string.o to extra_objs for
18789 targets powerpc*-*-* and rs6000*-*-*.
18791 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
18794 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
18795 32-bit, since indexed is not valid for DImode.
18796 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
18797 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
18798 (define_peephole2 for Altivec d-form load): Add 32-bit support.
18799 (define_peephole2 for Altivec d-form store): Likewise.
18802 * multiple_target.c (create_dispatcher_calls): Only create the
18803 dispatcher call if the function is the default clone of a
18804 versioned function.
18806 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
18808 PR middle-end/80902
18809 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
18810 a call, force the call to not be a tail call.
18812 2017-06-23 Jeff Law <law@redhat.com>
18814 * doc/contrib.texi: Add entry for Steven Pemberton's work on
18817 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
18819 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
18820 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
18821 handling for early expansion of vector shifts (sl,sr,sra,rl).
18822 (builtin_function_type): Add vector shift right instructions
18823 to the unsigned argument list.
18825 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
18827 rtl-optimizatoin/79286
18828 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
18829 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
18830 trap. PIC register plus a const unspec without offset can never trap.
18832 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
18834 * tree.h (builtin_structptr_type): New type.
18835 (builtin_structptr_types): Declare new array.
18836 * tree.c (builtin_structptr_types): New array.
18837 (free_lang_data, build_common_tree_nodes): Use it.
18839 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
18842 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
18845 2017-06-22 Matt Turner <mattst88@gmail.com>
18847 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
18848 Lake models to skylake case. Assume skylake for unknown
18849 models with clflushopt.
18851 2017-06-22 Jeff Law <law@redhat.com>
18853 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
18854 frame sizes that do not satisfy aarch64_uimm12_shift.
18856 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
18858 * profile-count.h (apply_probability,
18859 apply_scale, probability_in): Fix checks for zero.
18861 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18863 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
18864 * doc/cppdiropts.texi (-I @var{dir}): Document it.
18866 2016-06-22 Richard Biener <rguenther@suse.de>
18868 * tree-vect-loop.c (vect_model_reduction_cost): Handle
18869 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
18870 REDUC_MAX_EXPR support.
18871 (vectorizable_reduction): Likewise.
18872 (vect_create_epilog_for_reduction): Likewise.
18874 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
18876 * match.pd (A / (1 << B) -> A >> B): New.
18877 * generic-match-head.c: Include optabs-tree.h.
18878 * gimple-match-head.c: Likewise.
18879 * optabs-tree.h (target_supports_op_p): New.
18880 * optabs-tree.c (target_supports_op_p): New.
18882 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18884 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
18885 $gcc_cv_ld --help output.
18886 (gcc_cv_ld_demangle): Likewise.
18887 (gcc_cv_ld_eh_frame_hdr): Likewise.
18888 (gcc_cv_ld_pie): Likewise.
18889 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
18890 (gcc_cv_ld_buildid): Likewise.
18891 (gcc_cv_ld_sysroot): Likewise.
18892 (ld_bndplt_support): Likewise.
18893 (ld_pushpopstate_support): Likewise.
18894 * configure: Regenerate.
18895 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
18897 2017-06-21 Jakub Jelinek <jakub@redhat.com>
18900 * config/i386/sse.md (round<mode>2): Renumber match_dup and
18901 operands indexes to avoid gap between operands and match_dups.
18903 2017-06-21 Andrew Pinski <apinski@cavium.com>
18905 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
18906 Increment Arith_shift and Arith_shift_reg by 1.
18907 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
18909 * config/aarch64/aarch64.c (thunderx_tunings): Enable
18910 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
18911 (aarch64_strip_extend): Add new argument and test for it.
18912 (aarch64_cheap_mult_shift_p): New function.
18913 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
18914 add a cost if it is true.
18915 Update calls to aarch64_strip_extend.
18916 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
18918 2017-06-21 Andrew Pinski <apinski@cavium.com>
18920 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
18922 (thunderxt88): Likewise.
18923 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
18924 (thunderx_prefetch_tune): New variable.
18925 (thunderx2t99_prefetch_tune): Update for the correct values.
18926 (thunderxt88_tunings): New variable.
18927 (thunderx_tunings): Use thunderx_prefetch_tune instead of
18928 generic_prefetch_tune.
18929 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
18931 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18933 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
18934 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
18935 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
18936 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
18937 (aarch64_atomic_cas<mode>, GPI): Likewise.
18939 2017-06-21 Martin Liska <mliska@suse.cz>
18941 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
18942 statements on cold and hot labels.
18943 * predict.c (tree_estimate_probability_bb): Remove the
18944 prediction from this place.
18946 2017-06-21 Martin Liska <mliska@suse.cz>
18948 PR tree-optimization/79489
18949 * gimplify.c (maybe_add_early_return_predict_stmt): New
18951 (gimplify_return_expr): Call the function.
18952 * predict.c (tree_estimate_probability_bb): Remove handling
18954 * predict.def: Update comment about early return predictor.
18955 * gimple-predict.h (is_gimple_predict): New function.
18956 * predict.def: Change default value of early return to 66.
18957 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
18959 * passes.def: Put pass_strip_predict_hints to the beginning of
18962 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
18964 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
18965 FUNCTION_DECL declarations.
18966 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
18968 (dwaf2out_decl): Likewise.
18969 * godump.c (go_early_global_decl): Skip call to the real debug hook
18970 for FUNCTION_DECL declarations.
18971 * passes.c (rest_of_decl_compilation): Skip call to the
18972 early_global_decl debug hook for FUNCTION_DECL declarations, unless
18973 -fdump-go-spec is passed.
18975 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
18977 * config/i386/i386.c (struct builtin_isa): New field pure_p.
18978 Reorder for compactness.
18979 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
18980 (def_builtin_pure, def_builtin_pure2): New functions.
18981 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
18983 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
18985 * match.pd (nop_convert): New predicate.
18986 ((A +- CST1) +- CST2): Allow some NOP conversions.
18988 2017-06-21 Jakub Jelinek <jakub@redhat.com>
18991 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
18992 with ctors/dtors if GOVD_SHARED is set.
18994 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
18996 * config/aarch64/aarch64.md (movti_aarch64):
18997 Emit mov rather than orr.
18998 (movtf_aarch64): Likewise.
18999 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
19000 Emit mov rather than orr.
19002 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19004 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
19005 Swap alternatives, make integer dup more expensive.
19007 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19009 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
19010 Return true for non-tls symbols.
19012 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
19014 * config/aarch64/aarch64-cores.def (cortex-a55): New.
19015 (cortex-a75): Likewise.
19016 (cortex-a75.cortex-a55): Likewise.
19017 * config/aarch64/aarch64-tune.md: Regenerate.
19018 * doc/invoke.texi (-mtune): Document new values for -mtune.
19020 2017-06-21 Tom de Vries <tom@codesourcery.com>
19022 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
19023 stack_size feature.
19024 (Effective-Target Keywords, Other attributes): Suggest using
19025 dg-add-options stack_size feature to get stack limit in stack_size
19026 effective target documentation.
19028 2017-06-21 Julian Brown <julian@codesourcery.com>
19029 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19031 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
19032 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
19033 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
19035 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
19036 attribute type list for neon_multiply.
19037 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
19038 attribute type list for neon_multiply.
19039 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
19040 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
19041 attribute type list for neon_multiply.
19042 * config/arm/types.md (crypto_pmull): Add.
19043 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
19044 attribute type list.
19046 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
19048 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
19051 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19053 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
19054 to make sure not to dereference a NULL cost_classes_ptr pointer.
19056 2017-06-20 Carl Love <cel@us.ibm.com>
19058 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19059 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19060 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
19061 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
19062 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
19063 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
19064 VMULOSW): New enum "unspec" values.
19065 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
19066 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
19067 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
19068 altivec_vmulosw): New patterns.
19069 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
19070 VMULOSW): Add definitions.
19072 2017-06-20 Julia Koval <julia.koval@intel.com>
19074 * config/i386/i386.c: Fix rounding expand for new pattern.
19075 * config/i386/subst.md: Fix pattern (parallel -> unspec).
19077 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19079 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
19080 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
19082 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19084 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
19087 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19089 * config/aarch64/aarch64-cores.def: Rearrange to sort by
19090 architecture, then by implementer ID.
19091 * config/aarch64/aarch64-tune.md: Regenerate.
19093 2017-06-20 Richard Biener <rguenther@suse.de>
19095 PR middle-end/81097
19096 * fold-const.c (split_tree): Fold to type before negating.
19098 2017-06-20 David Malcolm <dmalcolm@redhat.com>
19100 * diagnostic-show-locus.c
19101 (selftest::test_fixit_deletion_affecting_newline): New function.
19102 (selftest::diagnostic_show_locus_c_tests): Call it.
19104 2017-06-20 Andreas Schwab <schwab@suse.de>
19107 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
19110 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19112 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
19113 __ARM_FEATURE_COPROC according to support.
19115 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19117 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
19118 Rewritten to avoid overflow for > 32-bit pointers.
19121 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
19122 by removing enum keyword.
19123 (ubsan_type_descriptor): Likewise. Formatting fix.
19126 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
19127 splitter): Require TARGET_SSE2 in the condition.
19129 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
19132 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
19133 for doing vector set of SFmode on ISA 3.0.
19134 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
19135 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
19137 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
19138 SFmode value into a V4SF variable that was extracted from another
19139 V4SF variable without converting the element to double precision
19140 and back to single precision vector format.
19141 (vsx_insert_extract_v4sf_p9_2): Likewise.
19143 2017-06-19 Jakub Jelinek <jakub@redhat.com>
19145 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
19146 in UWHI to avoid undefined overflow.
19149 * ubsan.h (enum ubsan_encode_value_phase): New.
19150 (ubsan_encode_value): Change second argument to
19151 enum ubsan_encode_value_phase with default value of
19152 UBSAN_ENCODE_VALUE_GENERIC.
19153 * ubsan.c (ubsan_encode_value): Change second argument to
19154 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
19155 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
19156 create_tmp_var_raw instead of create_tmp_var and use a
19158 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
19159 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
19160 ubsan_encode_value callers.
19163 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
19164 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
19165 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
19167 2017-06-19 Richard Biener <rguenther@suse.de>
19169 PR middle-end/81118
19170 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
19171 estimates if we changed anything.
19173 2017-06-19 Richard Biener <rguenther@suse.de>
19175 PR tree-optimization/80887
19176 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
19177 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
19178 simplified lookups, then reset mprts_hook.
19179 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
19181 (try_to_simplify): Likewise.
19183 2017-06-19 Martin Liska <mliska@suse.cz>
19186 * gimplify.c (gimplify_switch_expr):
19187 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
19189 2017-06-19 Martin Liska <mliska@suse.cz>
19191 * doc/install.texi: Document that PGO runs in 4 stages.
19193 2017-06-19 Martin Liska <mliska@suse.cz>
19196 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
19197 to dispatcher function name.
19198 * multiple_target.c (replace_function_decl): New function.
19199 (create_dispatcher_calls): Redirect both edges and references.
19201 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
19203 * profile-count.c (profile_count::dump): Dump quality.
19204 (profile_count::differs_from_p): Update for unsigned val.
19205 * profile-count.h (profile_count_quality): New enum.
19206 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
19208 2017-06-19 Richard Biener <rguenther@suse.de>
19210 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
19211 struct function as arg.
19212 (estimate_numbers_of_iterations): Export overload with loop arg.
19213 (free_numbers_of_iterations_estimates_loop): Use an overload of
19214 free_numbers_of_iterations_estimates instead.
19215 * tree-cfg.c (remove_bb): Adjust.
19216 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
19217 * tree-parloops.c (gen_parallel_loop): Likewise.
19218 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
19220 (tree_unroll_loops_completely): Likewise.
19221 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
19222 Use an overload instead and export.
19223 (estimated_loop_iterations): Adjust.
19224 (max_loop_iterations): Likewise.
19225 (likely_max_loop_iterations): Likewise.
19226 (estimate_numbers_of_iterations): Take struct function as arg
19228 (loop_exits_before_overflow): Adjust.
19229 (free_numbers_of_iterations_estimates_loop): Use an overload.
19230 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
19231 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
19233 2017-06-19 Richard Biener <rguenther@suse.de>
19236 * ipa-prop.c (find_constructor_constant_at_offset): Handle
19237 RANGE_EXPR conservatively.
19239 2017-06-16 Carl Love <cel@us.ibm.com>
19241 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19242 definitions for vec_float, vec_float2, vec_floato,
19243 vec_floate built-ins.
19244 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
19245 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
19247 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
19248 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
19249 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
19250 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
19251 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
19252 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
19253 vec_floato): Add builtin defines.
19254 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
19255 Update the built-in documentation file for the new built-in
19258 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19260 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
19261 (mthumb): Mark as the negative of -marm.
19263 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19265 * doc/invoke.texi (ARM Options, -mcpu): Document supported
19267 (ARM Options, -mtune): Document that this accepts the same
19268 extension options as -mcpu.
19269 (ARM Options, -mfpu): Document addition of -mfpu=auto.
19271 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19273 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
19274 permitted extensions.
19276 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19278 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
19279 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
19280 (armv8-m.main): Add option +nodsp.
19281 * config/arm/arm-cpu-cdata.h: Regenerated.
19283 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19285 * config/arm/t-fuchsia: New file.
19286 * config.gcc (arm*-*-fuchsia*): Use it.
19288 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19290 * config/arm/t-symbian: Rewrite for new option infrastructure.
19292 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19294 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
19295 (MULTILIB_REQUIRED): Likewise.
19297 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19299 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
19300 (MULTILIB_RESUE): Likewise.
19301 (MULTILIB_MATCHES): Likewise.
19302 (MULTLIB_REQUIRED): Likewise.
19304 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19306 * config/arm/t-rtems: Rewrite for new option framework.
19308 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19310 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
19311 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
19312 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
19313 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
19314 * config/arm/t-multilib: ... here.
19315 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
19316 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
19317 armv7-a and armv8*-a when A-profile libraries have not been built.
19318 * config/arm/t-rmprofile: Rewrite.
19320 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19322 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
19323 with a backslash. Remove the backslash after substituting unescaped
19325 * doc/fragments.texi (MULTILIB_REUSE): Document it.
19327 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19329 * config.gcc: (arm*-*-*): When building a-profile libraries, force
19330 the driver to pass through the default setting of -mfloat-abi.
19331 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
19333 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
19334 (all_feat_combs): New rule.
19335 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
19337 * config/arm/t-aprofile: Rewrite.
19339 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19341 * config/arm/arm.h (FPUTYPE_AUTO): Define.
19342 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
19343 fpu is not specified by the user/command-line.
19344 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
19345 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
19346 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
19347 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
19348 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
19349 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
19351 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19353 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
19354 * config/arm/t-arm-elf: Rewritten.
19356 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19358 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
19359 have some floating-point instructions.
19360 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
19361 (TARGET_MAYBE_HARD_FLOAT): New macro.
19362 * config/arm/arm-builtins.c (arm_init_builtins): Use
19363 TARGET_MAYBE_HARD_FLOAT.
19364 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
19366 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19368 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
19369 (configargs.h): Include it.
19370 (arm_print_hint_for_fpu_option): New function.
19371 (arm_parse_fpu_option): New function.
19372 (candidate_extension): New class.
19373 (arm_canon_for_multilib): New function.
19374 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
19375 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
19376 (ARCH_CANONICAL_SPECS): New macro.
19377 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
19379 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19381 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
19382 are set after handling multilib fragments. Set target_cpu_default2
19385 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19387 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
19389 (arm*-*-*): Set target_cpu_default2 to a quoted string.
19390 * config/arm/parsecpu.awk (check_cpu): Validate any extension
19392 (check_arch): Likewise.
19393 * config/arm/arm.c (arm_configure_build_target): Handle
19394 TARGET_CPU_DEFAULT being a string constant. Scan any feature
19395 options in the default.
19397 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19399 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
19400 when an option is an alias of another.
19401 * config/arm/parsecpu.awk (optalias): New parser token.
19402 (gen_comm_data): Mark non-alias options as such. Emit entries
19403 for extension aliases.
19404 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
19405 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
19406 (armv6kz, armv6zk, armv6t2): Likewise.
19407 (armv7): Make vfpv3-d16 an alias.
19408 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
19410 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
19411 Sort in canonical order.
19412 (armv8-a): Sort in canonical order.
19413 (armv8.1-a, armv8.2-a): Likewise.
19414 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
19416 (cortex-a9): Sort in canonical order.
19417 * config/arm/arm.c (selftests.h): Include it.
19418 (arm_test_cpu_arch_data): New function.
19419 (arm_run_self_tests): New function.
19420 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
19421 (targetm): Move declaration to the end of the file.
19422 * arm-cpu-cdata.h: Regenerated.
19424 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19426 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
19427 call to target_mode_check describing the type of option passed.
19428 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
19429 (arm_target_thumb_only): Use arm_parse_arch_option_name or
19430 arm_parse_cpu_option_name to match parameters against list of
19432 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
19433 arm_arch_core_flags data structure.
19434 * config/arm/arm-cpu_cdata.h: Regenerated.
19436 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19438 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
19440 (arm_print_hint_for_cpu_option): Likewise.
19441 (arm_print_hint_for_arch_option): Likewise.
19442 (arm_parse_cpu_option_name): Likewise.
19443 (arm_parse_arch_option_name): Likewise.
19444 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
19445 of entries in the all_fpus list.
19446 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
19447 (arm_parse_cpu_option_name): Declare.
19448 (arm_parse_arch_option_name): Declare.
19449 (arm_parse_option_features): Declare.
19450 (arm_intialize_isa): Declare.
19451 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
19453 (gen_comm_data): ... here. Make definitions non-static.
19454 * config/arm/arm-cpu-data.h: Regenerated.
19455 * config/arm/arm-cpu-cdata.h: Regenerated.
19457 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19459 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
19460 (cpu_arch_extension): New structure.
19461 (cpu_arch_option, arch_option, cpu_option): New structures.
19462 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
19463 architecture types.
19464 (gen_data): Generate new format data tables.
19465 * config/arm/arm.c (cpu_tune): New structure.
19466 (cpu_option, processors): Delete.
19467 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
19468 (arm_print_hint_for_cpu_option): ... this and ...
19469 (arm_print_hint_for_arch_option): ... this.
19470 (arm_parse_arch_cpu_name): Delete. Replace with ...
19471 (arm_parse_cpu_option_name): ... this and ...
19472 (arm_parse_arch_option_name): ... this.
19473 (arm_unrecognized_feature): Change type of target parameter to
19475 (arm_parse_arch_cpu_features): Delete. Replace with ...
19476 (arm_parse_option_features): ... this.
19477 (arm_configure_build_target): Rework to use new configuration data
19479 (arm_print_tune_info): Rework for new configuration data tables.
19480 * config/arm/arm-cpu-data.h: Regenerated.
19481 * config/arm/arm-cpu.h: Regenerated.
19483 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19485 * Makefile.in (OBJS): Move sbitmap.o from here ...
19486 (OBJS-libcommon): ... to here.
19488 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19490 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
19491 (ISA_ALL_CRYPTO): New macro.
19492 (ISA_ALL_SIMD): New macro
19493 (ISA_ALL_FP): New macro.
19494 * config/arm/arm.c (fpu_bitlist): Update initializer.
19495 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
19497 (arm9e): Add fpu. Add option for nofp
19498 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
19499 (arm926ej-s, arm1026ej-s): Likewise.
19500 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
19501 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
19502 neon-fp16, neon-vfpv4, nofp and nosimd.
19503 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
19504 (cortex-a8): Add fpu. Add option for nofp.
19505 (cortex-a9): Add fpu. Add options for nosimd and nofp.
19506 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
19507 (cortex-r4f): Add fpu.
19508 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
19509 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
19511 (cortex-r8): Likewise.
19512 (cortex-m4): Add fpu. Add option for nofp.
19513 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
19514 (cortex-a17.cortex-a7): Likewise.
19515 (cortex-a32): Add fpu. Add options for crypto and nofp.
19516 (cortex-a35, cortex-a53): Likewise.
19517 (cortex-a57): Add fpu. Add option for crypto.
19518 (cortex-a72, cortex-a73): Likewise.
19519 (exynos-m1): Likewise.
19520 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
19521 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
19522 (cortex-m33): Add fpu. Add option for nofp.
19523 * config/arm/arm-cpu-cdata.h: Regenerated
19524 * config/arm/arm-cpu-data.h: Regenerated.
19526 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19528 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
19529 (armv5te, armv5tej): Likewise.
19530 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
19531 (armv7): Add options fp and vfpv3-d16.
19532 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
19533 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
19535 (armv7ve): Likewise.
19536 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
19537 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
19538 (armv8-a): Add nocrypto option.
19539 (armv8.1-a, armv8.2-a): Likewise.
19540 (armv8-m.main): add options fp, fp.dp and nofp.
19542 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19544 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
19546 (armv8-a+crc): Delete.
19547 (armv8.1-a): Add options simd, crypto and nofp.
19548 (armv8.2-a): Add options fp16, simd, crypto and nofp.
19549 (armv8.2-a+fp16): Delete.
19550 (armv8-m.main): Add option dsp.
19551 (armv8-m.main+dsp): Delete.
19552 (cortex-a8): Add fpu. Add nofp option.
19553 (cortex-a9): Add fpu. Add nofp and nosimd options.
19554 * config/arm/parsecpu.awk (gen_data): Generate option tables and
19555 link to main cpu and architecture data structures.
19556 (gen_comm_data): Only put isa attributes from the main architecture
19558 (option): New statement for architecture and CPU entries.
19559 * arm.c (struct cpu_option): New structure.
19560 (struct processors): Add entry for options.
19561 (arm_unrecognized_feature): New function.
19562 (arm_parse_arch_cpu_name): Ignore any characters after the first
19564 (arm_parse_arch_cpu_feature): New function.
19565 (arm_configure_build_target): Separate out any CPU and architecture
19566 features and parse separately. Don't error out if -mfpu=auto is
19567 used with only an architecture string.
19568 (arm_print_asm_arch_directives): New function.
19569 (arm_file_start): Call it.
19570 * config/arm/arm-cpu-cdata.h: Regenerated.
19571 * config/arm/arm-cpu-data.h: Likewise.
19572 * config/arm/arm-tables.opt: Likewise.
19574 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19576 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
19577 assembler when it is not -mfpu=auto.
19579 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19581 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
19582 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
19583 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
19584 (ASM_CPU_SPEC): Rewrite.
19585 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
19586 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
19587 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
19588 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
19589 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
19590 copied string is NUL-terminated. Also strip any characters prefixed
19592 (arm_rewrite_selected_arch): New function.
19593 (arm_rewrite_march): New function.
19595 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19597 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
19598 (x_arm_cpu_string, x_arm_tune_string): Likewise.
19599 (march, mcpu, mtune): Convert to string-based options.
19600 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
19601 (arm_parse_arch_cpu_name): New function.
19602 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
19603 identify selected architecture or CPU.
19604 (arm_option_save): New function.
19605 (TARGET_OPTION_SAVE): Redefine.
19606 (arm_option_restore): Restore string options.
19607 (arm_option_print): Print string options.
19609 2017-06-16 Martin Sebor <msebor@redhat.com>
19611 PR tree-optimization/80933
19612 PR tree-optimization/80934
19613 * builtins.c (fold_builtin_3): Do not handle bcmp here.
19614 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
19615 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
19616 (gimple_fold_builtin): Call them.
19618 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19620 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
19621 as unlikely; update profile.
19623 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19625 * predict.c (force_edge_cold): Handle declaring edges impossible
19628 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19630 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
19632 (try_unroll_loop_completely): Fix reporting.
19634 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19636 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
19638 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
19641 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
19642 if given a non-constant argument for an intrinsic which requires a
19645 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19647 * profile.c (compare_freqs): New function.
19648 (branch_prob): Sort edge list.
19649 (find_spanning_tree): Assume that the list is priority sorted.
19651 2017-06-16 Richard Biener <rguenther@suse.de>
19653 PR tree-optimization/81090
19654 * passes.def (pass_record_bounds): Remove.
19655 * tree-pass.h (make_pass_record_bounds): Likewise.
19656 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
19657 make_pass_record_bounds): Likewise.
19658 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
19659 not free niter estimates at the beginning but at the end.
19660 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
19662 2017-06-16 Richard Biener <rguenther@suse.de>
19664 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
19665 initializer to workaround ICE in host GCC 4.8.
19667 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19669 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
19671 (clone_inlined_nodes): Update.
19673 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19675 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
19676 prefetch settings, and enable prefetching by default at -O3.
19678 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19680 * config/aarch64/aarch64.c (aarch64_override_options_internal):
19681 Set flag_prefetch_loop_arrays according to tuning data.
19683 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19685 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
19686 New tune structure.
19687 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
19688 [Unrelated to main purpose of the patch] Place the pointer field last
19689 to enable type checking errors when tune structure are wrongly merged.
19690 * config/aarch64/aarch64.c (generic_prefetch_tune,)
19691 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
19692 (thunderx2t99_prefetch_tune): New tune constants.
19693 (tune_params *_tunings): Update all tunings (no functional change).
19694 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
19695 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
19696 from tunings structures.
19698 2017-06-16 Jakub Jelinek <jakub@redhat.com>
19701 * ubsan.c (instrument_null): Add T argument, use it instead
19702 of computing it based on IS_LHS.
19703 (instrument_object_size): Likewise.
19704 (pass_ubsan::execute): Adjust instrument_null and
19705 instrument_object_size callers to pass gimple_get_lhs or
19706 gimple_assign_rhs1 result to it. Use instrument_null instead of
19707 calling get_base_address and instrument_mem_ref. Handle
19708 aggregate call arguments for object-size sanitization.
19710 2017-06-16 Yury Gribov <tetra2005@gmail.com>
19712 PR tree-optimization/81089
19713 * tree-vrp.c (is_masked_range_test): Validate operands of
19716 2017-06-15 Martin Sebor <msebor@redhat.com>
19719 * dumpfile.c (dump_register): Avoid calling memset to initialize
19720 a class with a default ctor.
19721 * gcc.c (struct compiler): Remove const qualification.
19722 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
19723 * hash-table.h: Ditto.
19724 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
19726 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
19727 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
19729 * params.h (struct param_info): Make struct members non-const.
19730 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
19731 with default initialization.
19732 * vec.h (vec_copy_construct, vec_default_construct): New helper
19734 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
19735 with vec_copy_construct.
19736 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
19737 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
19738 * doc/invoke.texi (-Wclass-memaccess): Document.
19740 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19742 * emit-rtl.h (is_leaf): Update comment about local
19743 register allocator.
19745 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19748 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
19749 for a variable to have a section before checking if the section has a
19751 Set section to.persistent if persistent attribute is set.
19752 Warn if .persistent attribute is used on an automatic variable.
19754 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
19756 PR rtl-optimization/80474
19757 * reorg.c (update_block): Do not ignore instructions in a delay slot.
19759 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
19761 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
19764 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
19766 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
19767 (casesi): Emit bounds checking as RTL.
19768 (casesi_internal_mips16_<mode>): Remove bounds checking.
19770 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
19772 * config/xtensa/xtensa.c (xtensa_option_override): Append
19773 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
19774 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
19775 xtensa_doloop_hooks): Define unconditionally.
19776 (xtensa_reorg_loops): Only call reorg_loops in the presence of
19778 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
19779 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
19780 for it in xtensa_option_override.
19781 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
19782 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
19784 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
19786 * doc/cppopts.texi: Document '-' special value to -MF.
19788 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
19790 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
19791 (cortex_a53_fconst): Likewise.
19792 (cortex_a53_fpmul): Likewise.
19793 (cortex_a53_f_load_64): Likewise.
19794 (cortex_a53_f_load_many): Likewise.
19795 (cortex_a53_advsimd_alu): Likewise.
19796 (cortex_a53_advsimd_alu_q): Likewise.
19797 (cortex_a53_advsimd_mul): Likewise.
19798 (cortex_a53_advsimd_mul_q): Likewise.
19799 (fpmac bypass): Add new bypass for fpmac-fpmac case.
19800 Add missing fmul, r2f_cvt and fconst cases.
19802 2017-06-14 Richard Biener <rguenther@suse.de>
19804 PR middle-end/81088
19805 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
19807 (fold_binary_loc): When associating do not treat pre-existing
19808 TREE_OVERFLOW on literal constants as a reason to allow
19809 TREE_OVERFLOW on associated literal constants.
19811 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19813 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
19814 (MASK_FEATURES): New macro.
19815 * config/sparc/sparc.c (sparc_option_override): Remove the special
19816 handling of -mfpu and generalize it to all MASK_FEATURES switches.
19818 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19820 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
19821 a division of 0 if non-call exceptions are enabled.
19823 2017-06-14 Andrew Pinski <apinski@cavium.com>
19824 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19827 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
19828 Improve vector initialization code gen for only variable case.
19830 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19832 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
19834 2017-06-14 Richard Biener <rguenther@suse.de>
19836 PR tree-optimization/81083
19837 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
19840 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19842 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
19843 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
19844 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
19845 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
19846 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
19847 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
19849 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19851 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
19852 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
19854 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19856 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
19858 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19860 * config/rs6000/t-rtems: Don't handle SPE.
19862 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19864 * config/rs6000/t-linux: Don't handle SPE.
19866 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19868 * config/rs6000/eabispe.h: Delete file.
19870 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19872 * config/rs6000/t-spe: Delete file.
19874 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19876 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
19877 (rs6000_legitimate_offset_address_p): Return false for anything in
19878 V2SImode or V2SFmode.
19880 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19882 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
19883 except V2SF and V2SI. Rearrange the vector modes, and add comments.
19884 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
19886 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
19887 (rs6000_legitimate_offset_address_p): Ditto.
19888 (rs6000_emit_move): Ditto.
19889 (rs6000_init_builtins): Remove V4HI_type_node.
19891 2017-06-13 Martin Liska <mliska@suse.cz>
19894 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
19895 (gate_asan): Likewise.
19896 * asan.h (asan_no_sanitize_address_p): Remove the function.
19897 (sanitize_flags_p): New function.
19898 * builtins.def: Fix coding style.
19899 * common.opt: Use renamed enum value.
19900 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
19901 * doc/extend.texi: Document no_sanitize attribute.
19902 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
19903 to SANITIZE_UNDEFINED_NONDEFAULT.
19904 * gcc.c (sanitize_spec_function): Use the renamed enum value.
19905 * gimple-fold.c (optimize_atomic_compare_exchange_p):
19906 Use sanitize_flags_p.
19907 * gimplify.c (gimplify_function_tree): Likewise.
19908 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
19909 * opts.c (parse_no_sanitize_attribute): New function.
19910 (common_handle_option): Use renamed enum value.
19911 * opts.h (parse_no_sanitize_attribute): Declare.
19912 * tree.c (sanitize_flags_p): New function.
19913 * tree.h: Declared here.
19914 * tsan.c: Use sanitize_flags_p.
19915 * ubsan.c (ubsan_expand_null_ifn): Likewise.
19916 (instrument_mem_ref): Likewise.
19917 (instrument_bool_enum_load): Likewise.
19918 (do_ubsan_in_current_function): Remove the function.
19919 (pass_ubsan::execute): Use sanitize_flags_p.
19920 * ubsan.h: Remove do_ubsan_in_current_function
19921 * tree-cfg.c (print_no_sanitize_attr_value): New function.
19922 (dump_function_to_file): Use it here.
19924 2017-06-13 Martin Jambor <mjambor@suse.cz>
19926 PR tree-optimization/80803
19927 PR tree-optimization/81063
19928 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
19929 (propagate_subaccesses_across_link): Enqueue subtree whenever
19930 necessary instead of relying on the caller.
19932 2017-06-13 Martin Jambor <mjambor@suse.cz>
19934 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
19935 that have a first_link.
19936 (sort_and_splice_var_accesses): Do not check first_link before
19938 (subtree_mark_written_and_enqueue): Likewise.
19939 (propagate_all_subaccesses): Likewise and do not stop at first
19940 parent with a first_link.
19942 2017-06-13 Martin Jambor <mjambor@suse.cz>
19944 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
19947 2017-06-13 Yury Gribov <tetra2005@gmail.com>
19949 * match.pd: New pattern.
19951 2017-06-13 Yury Gribov <tetra2005@gmail.com>
19953 * tree-vrp.c (is_masked_range_test): New function.
19954 (register_edge_assert_for): Determine ranges for
19957 2017-06-13 Yury Gribov <tetra2005@gmail.com>
19959 PR tree-optimization/67328
19960 * fold-const.c (maskable_range_p): New function.
19961 (build_range_check): Generate bittests if possible.
19963 2017-06-13 Martin Liska <mliska@suse.cz>
19965 * gimple-pretty-print.c (dump_probability): Add new argument.
19966 (dump_edge_probability): Dump both probability and count.
19967 (dump_gimple_label): Likewise.
19968 (dump_gimple_bb_header): Likewise.
19970 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
19973 * config/avr/avr-devices.c: Fix indentation.
19974 * config/avr/gen-avr-mmcu-specs.c: Dito.
19976 2017-06-13 Richard Biener <rguenther@suse.de>
19978 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
19979 instead get vector type from stmt_info.
19980 (vectorizable_reduction): Adjust. Remove dead code.
19982 2017-06-13 Richard Biener <rguenther@suse.de>
19984 PR middle-end/81065
19985 * fold-const.c (extract_muldiv_1): Remove bogus distribution
19986 case of C * (x * C2 + C3).
19987 (fold_addr_of_array_ref_difference): Properly fold index difference.
19989 2017-06-12 David S. Miller <davem@davemloft.net>
19992 * config/sparc/sparc.md (return expander): Emit frame blockage if
19993 function uses alloca.
19995 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
19997 * combine.c (make_field_assignment): Check len rather than the mode
19998 precision when calling force_to_mode.
20000 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
20002 Support multilibs and devices that see flash in RAM address range.
20005 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
20006 (avr_mcu_t) <flash_pm_offset>: New field.
20007 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
20008 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
20009 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
20010 (AVR_TINY_PM_OFFSET): Remove macro.
20011 * config/avr/avr.opt (-mshort-calls): New option.
20012 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
20013 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
20014 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
20015 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
20016 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
20017 instead of avr_arch->have_jmp_call.
20018 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
20019 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
20020 avr_arch->flash_pm_offset to define.
20021 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
20022 new field flash_pm_offset. Add entry for avrxmega3.
20023 (avr_texinfo): Add entry for avrxmega3.
20024 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
20025 attiny212, attiny214,
20026 attiny412, attiny414, attiny416, attiny417,
20027 attiny814, attiny816, attiny817,
20028 attiny1614, attiny1616, attiny1617,
20029 attiny3214, attiny3216, attiny3217.
20030 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
20031 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
20032 (avr_print_operand_address) [AVR_TINY]: Same.
20033 (avr_asm_init_sections) <readonly_data_section>: Only patch
20034 callback if avr_arch->flash_pm_offset = 0.
20035 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
20036 for rodata if avr_arch->flash_pm_offset != 0.
20037 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
20038 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
20039 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
20040 (m_options): Append opt_rcall.
20041 (m_dirnames): Append dir_rcall.
20042 * config/avr/t-multilib: Regenerate.
20044 * configure.ac [target=avr]: Check whether avrxmega3 default
20045 linker description file works as needed.
20046 * configure: Regenerate.
20047 * doc/avr-mmcu.texi: Regenerate.
20048 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
20049 <__AVR_ARCH__>: Document avrxmega3 and 103.
20050 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
20051 <__AVR_SHORT_CALLS__>: Document it.
20052 <__AVR_PM_BASE_ADDRESS__>: Document it.
20053 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
20054 (AVR Variable Attributes) <progmem>: Document this is
20055 not needed for avrxmega3.
20056 (AVR Named Address Spaces) <__flash>: Dito.
20058 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
20060 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
20062 2017-06-12 Doug Rupp <rupp@adacore.com>
20064 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
20065 Append vxworks-stdint.h to the tm_file list.
20066 * config/vxworks-stdint.h: New file.
20068 2017-06-12 Martin Liska <mliska@suse.cz>
20070 PR tree-optimization/81041
20071 * tree-profile.c (gimple_gen_ic_func_profiler):
20072 Create an extra BB in profile-generate
20073 (gimple_gen_time_profiler): Likewise.
20075 2017-06-12 Jakub Jelinek <jakub@redhat.com>
20077 PR tree-optimization/81003
20078 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
20079 (update_range_test): Use it instead of force_gimple_operand_gsi.
20081 2017-06-12 Richard Biener <rguenther@suse.de>
20083 PR tree-optimization/81053
20084 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
20085 with backedge value not defined in loop. Simplify def stmt
20088 2017-06-11 Tom de Vries <tom@codesourcery.com>
20091 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
20093 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
20094 nvptx_cannot_force_const_mem.
20096 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20098 * opts.c (finish_options): Move test for flag_split_stack after
20099 it has been initialized.
20101 2017-06-11 Jason Merrill <jason@redhat.com>
20103 * tree.h (id_equal): New.
20104 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
20105 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
20106 instead of strcmp of IDENTIFIER_POINTER.
20108 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20110 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
20111 (mark_all_inlined_calls_cdtor): Fix formating.
20112 (inline_transform): Rescale profile before inlining.
20114 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20116 * cgraph.h (cgraph_edge::clone): Update prototype.
20117 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
20118 (cgraph_node::create_clone): Update.
20119 (cgraph_node::create_version_clone): Update.
20120 * tree-inline.c (copy_bb): Update.
20121 (expand_call_inline): Update.
20123 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20125 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
20126 factored out from ...
20127 (rs6000_emit_prologue): ... here.
20129 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20131 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
20132 factored out from ...
20133 (rs6000_emit_prologue): ... here.
20135 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20137 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
20139 (handle_missing_profiles): Fix computation of tp_first_run.
20140 (counts_to_freqs): Do not touch freqs when count is 0.
20142 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20144 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
20147 2017-06-10 Tom de Vries <tom@codesourcery.com>
20149 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
20150 attributes): Document signal effective target.
20152 2017-06-10 Tom de Vries <tom@codesourcery.com>
20154 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
20155 Document effective target stack_size.
20157 2017-06-09 David Malcolm <dmalcolm@redhat.com>
20159 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
20160 to the edit_context if they can be auto-applied.
20162 2017-06-9 Ian Lance Taylor <iant@golang.org>
20164 * opts.c (finish_options): If -fsplit-stack, disable implicit
20165 -forder-blocks-and-partition.
20166 * doc/invoke.texi (Optimize Options): Document that when using
20167 -fsplit-stack -forder-blocks-and-partition is not implicitly
20170 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20172 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
20173 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
20174 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
20175 * builtins.def (abort, trap, unreachable): Declare cold.
20176 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
20177 * tree-core.h (ECF_COLD): New.
20178 * tree.c (set_call_expr_flags): Handle ECF_COLD.
20179 (build_common_builtin_nodes): Mark unreachable and abort as cold.
20181 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20183 * predict.c (unlikely_executed_stmt_p): Cleanup.
20185 2017-06-09 Richard Biener <rguenther@suse.de>
20187 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
20188 model if the ref is always written to.
20190 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20192 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
20194 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20196 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
20199 2017-06-09 Tom de Vries <tom@codesourcery.com>
20202 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
20203 "target cannot support label values" when encountering LABEL_REF.
20205 2017-06-09 Martin Liska <mliska@suse.cz>
20207 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
20208 (gimple_gen_ic_func_profiler): Emit direct comparison
20209 of __gcov_indirect_call_callee with NULL.
20210 (gimple_gen_time_profiler): Change probability from
20211 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
20213 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20215 * profile.c (edge_gcov_counts): Turn to pointer.
20216 (compute_branch_probabilities, compute_branch_probabilities): Update.
20217 (branch_prob): Do not clear edge_gcov_count.
20218 * profile.h (edge_gcov_counts): Turn to pointer.
20219 (edge_gcov_count): Update.
20221 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20223 * gimple.h (gimple_check_failed): Mark cold.
20225 2017-06-09 Richard Biener <rguenther@suse.de>
20227 PR tree-optimization/66623
20228 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
20229 refactor check_reduction into two parts, properly computing
20230 whether we have to check reduction validity for outer loop
20233 2017-06-09 Richard Biener <rguenther@suse.de>
20235 PR tree-optimization/79483
20236 * graphite-scop-detection.c (order): New global.
20237 (get_order): Compute bb to order mapping that satisfies code
20238 generation constraints.
20239 (cmp_pbbs): New helper.
20240 (build_scops): Start domwalk at entry block, sort generated
20243 2017-06-09 Richard Biener <rguenther@suse.de>
20245 PR middle-end/81007
20246 * ipa-polymorphic-call.c
20247 (ipa_polymorphic_call_context::restrict_to_inner_class):
20248 Skip FIELD_DECLs with error_mark_node type.
20249 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
20252 2017-06-09 Martin Liska <mliska@suse.cz>
20254 * predict.c (struct branch_predictor): New struct.
20255 (test_prediction_value_range): New test.
20256 (predict_c_tests): New function.
20257 * selftest-run-tests.c (selftest::run_tests): Run the function.
20258 * selftest.h: Declare new tests.
20260 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
20263 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
20264 gen_add3_insn did not fail.
20265 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
20266 r0, construct that number in a temporary reg and add that reg to r0.
20267 If asked to put the result in r0 as well, fail.
20269 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
20271 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
20272 for early expansion of vec_eqv.
20274 2017-06-08 Jakub Jelinek <jakub@redhat.com>
20276 PR middle-end/81005
20277 * ubsan.c (instrument_null): Avoid pointless code temporary.
20278 (pass_ubsan::execute): Instrument aggregate arguments of calls.
20280 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
20284 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
20287 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
20290 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20292 * predict.c (unlikely_executed_edge_p): Move ahead.
20293 (probably_never_executed_edge_p): Use it.
20295 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20297 PR middle-end/79988
20298 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
20299 gimple_call_builtin_p call.
20301 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20303 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
20304 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
20305 rtl_check_failed_type2, rtl_check_failed_code1,
20306 rtl_check_failed_code2, rtl_check_failed_code_mode,
20307 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
20308 rtvec_check_failed_bounds, rtl_check_failed_flag,
20309 _fatal_insn_not_found, _fatal_insn): Likewise.
20310 * tree.h (tree_contains_struct_check_failed,
20311 tree_check_failed, tree_not_check_failed,
20312 tree_class_check_failed, tree_range_check_failed,
20313 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
20314 tree_vec_elt_check_failed, phi_node_elt_check_failed,
20315 tree_operand_check_failed, omp_clause_check_failed,
20316 omp_clause_operand_check_failed, omp_clause_range_check_failed):
20319 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20321 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
20322 flag_branch_probabilities.
20323 * ipa-inline.c (edge_badness): Likewise.
20324 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
20325 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
20326 * predict.c (maybe_hot_frequency_p): Likewise.
20327 (probably_never_executed): Likewise.
20328 * sched-ebb.c (schedule_ebbs): Likewise.
20329 * sched-rgn.c (find_single_block_region): Likewise.
20330 * tracer.c (tail_duplicate): Likewise.
20332 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20334 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
20335 longer requires x_flag_profile_use.
20337 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20339 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
20340 instead of flag_reorder_blocks_and_partition.
20341 * dbxout.c (dbxout_function_end): Likewise.
20342 * dwarf2out.c (gen_subprogram_die): Likewise.
20343 * haifa-sched.c (sched_create_recovery_edges): Likewise.
20344 * hw-doloop.c (reorg_loops): Likewise.
20345 * varasm.c (assemble_start_function,
20346 assemble_end_function): Likewise.
20347 (decide_function_section): Do not check for
20348 flag_reorder_blocks_and_partition.
20350 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20352 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20354 (chkp_get_hard_register_fake_addr_expr): Ditto.
20355 (chkp_build_addr_expr): Add check for hard reg case.
20356 (chkp_parse_array_and_component_ref): Ditto.
20357 (chkp_find_bounds_1): Ditto.
20358 (chkp_process_stmt): Don't generate bounds store for
20361 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20363 * predict.c (maybe_hot_bb_p): Do not check profile status.
20364 (maybe_hot_edge_p): Likewise.
20365 (probably_never_executed): Check for zero counts even if profile
20367 (unlikely_executed_edge_p): New function.
20368 (unlikely_executed_stmt_p): New function.
20369 (unlikely_executed_bb_p): New function.
20370 (set_even_probabilities): Use unlikely predicates.
20371 (combine_predictions_for_bb): Likewise.
20372 (predict_paths_for_bb): Likewise.
20373 (predict_paths_leading_to_edge): Likewise.
20374 (determine_unlikely_bbs): New function.
20375 (estimate_bb_frequencies): Use it.
20376 (compute_function_frequency): Use zero counts even if profile is
20378 * profile-count.h: Fix typo.
20380 2017-08-08 Julia Koval <julia.koval@intel.com>
20382 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
20383 _mm512_mask_cvtsepi16_storeu_epi8,
20384 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
20385 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
20386 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
20387 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
20388 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
20389 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
20390 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
20391 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
20392 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
20393 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
20394 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
20395 __builtin_ia32_pmovuswb256mem_mask,
20396 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
20397 __builtin_ia32_pmovwb512mem_mask): New builtins.
20399 2017-08-08 Julia Koval <julia.koval@intel.com>
20401 PR target/73350,80862
20402 * config/i386/subst.md (round): Fix round pattern.
20403 * config/i386/i386.c (ix86_erase_embedded_rounding):
20404 Fix erasing rounding for the fixed pattern.
20406 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20408 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
20410 2017-06-08 Martin Liska <mliska@suse.cz>
20412 PR gcov-profile/80911
20413 * gcov.c (block_info::block_info): New constructor.
20415 2017-06-07 Carl Love <cel@us.ibm.com>
20417 * config/rs6000/rs6000-c: The return type of the following
20418 built-in functions was implemented as int not long long. Fix sign
20419 of return value for the unsigned version of vec_mulo and vec_mule.
20420 vector unsigned long long vec_bperm (vector unsigned long long,
20421 vector unsigned char)
20422 vector signed long long vec_mule (vector signed int,
20424 vector unsigned long long vec_mule (vector unsigned int,
20425 vector unsigned int)
20426 vector signed long long vec_mulo (vector signed int,
20428 vector unsigned long long vec_mulo (vector unsigned int,
20429 vector unsigned int)
20430 * doc/extend.texi: Fix the documentation for the built-in
20433 2017-06-07 Carl Love <cel@us.ibm.com>
20436 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
20439 2017-06-07 Carl Love <cel@us.ibm.com>
20441 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
20442 support, Generate doublehv for signed int/float for BE case only.
20444 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
20446 * doc/invoke.texi (mcx16): Rewrite.
20448 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20450 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
20451 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
20452 *mov<mode>_softfloat, and an anonymous splitter): Use
20453 nonimmediate_operand instead of rs6000_nonimmediate_operand.
20455 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20457 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
20459 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
20460 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
20461 (rs6000_debug_reg_global): Adjust.
20462 (rs6000_init_hard_regno_mode_ok): Adjust.
20463 (rs6000_dbx_register_number): Adjust.
20464 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
20465 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
20466 Remove SPE_ACC and SPEFSCR.
20467 (REG_ALLOC_ORDER): Ditto.
20468 (FRAME_POINTER_REGNUM): Change to 111.
20469 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
20470 (REG_CLASS_NAMES): Ditto.
20471 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
20472 (REGISTER_NAMES): Ditto.
20473 (ADDITIONAL_REG_NAMES): Ditto.
20474 (rs6000_reg_names): Ditto.
20475 * config/rs6000/rs6000.md: Renumber some register number
20478 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20480 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
20482 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
20483 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
20485 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
20486 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
20487 Delete the SPE high registers.
20488 (REG_ALLOC_ORDER): Ditto.
20489 (enum reg_class): Remove SPE_HIGH_REGS.
20490 (REG_CLASS_NAMES): Ditto.
20491 (REG_CLASS_CONTENTS): Delete the SPE high registers.
20492 (REGISTER_NAMES): Ditto.
20493 (rs6000_reg_names): Ditto.
20494 * doc/tm.texi.in: Remove SPE as example.
20495 * doc/tm.texi: Regenerate.
20497 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20499 * config/rs6000/8540.md (ppc8540_brinc): Delete.
20500 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
20501 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
20502 * config/rs6000/rs6000.md (type): Remove "brinc".
20504 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20506 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
20507 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
20508 * config/rs6000/linuxspe.h: Delete file.
20509 * config/rs6000/rs6000.md: Don't include spe.md.
20510 * config/rs6000/spe.h: Delete file.
20511 * config/rs6000/spe.md: Delete file.
20512 * config/rs6000/t-rs6000: Remove spe.md.
20514 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20516 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
20517 (reg_or_none500mem_operand): Delete.
20518 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
20519 instead of reg_or_none500mem_operand.
20521 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20523 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
20524 handling of SPE flags.
20525 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
20527 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20529 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
20531 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
20532 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
20533 paired_divv2sf3): Similar.
20534 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
20535 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
20536 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
20538 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
20539 Rename the paired_* instruction patterns.
20540 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
20542 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
20543 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
20544 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
20545 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
20546 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
20547 PAIRED_VECTOR_MODE.
20548 (struct machine_function): Delete field spe_insn_chain_scanned_p.
20549 (spe_func_has_64bit_regs_p): Delete.
20550 (spe_expand_predicate_builtin): Delete.
20551 (spe_expand_evsel_builtin): Delete.
20552 (TARGET_DWARF_REGISTER_SPAN): Do not define.
20553 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
20554 (invalid_e500_subreg): Delete.
20555 (rs6000_legitimize_address): Always force_reg op2 as well, for
20556 paired single memory accesses.
20557 (rs6000_member_type_forces_blk): Delete.
20558 (rs6000_spe_function_arg): Delete.
20559 (rs6000_expand_unop_builtin): Delete SPE handling.
20560 (rs6000_expand_binop_builtin): Ditto.
20561 (spe_expand_stv_builtin): Delete.
20562 (bdesc_2arg_spe): Delete.
20563 (spe_expand_builtin): Delete.
20564 (spe_expand_predicate_builtin): Delete.
20565 (spe_expand_evsel_builtin): Delete.
20566 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
20567 (spe_init_builtins): Delete.
20568 (spe_func_has_64bit_regs_p): Delete.
20569 (savres_routine_name): Delete "info" parameter. Adjust callers.
20570 (rs6000_emit_stack_reset): Ditto.
20571 (rs6000_dwarf_register_span): Delete.
20572 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
20573 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
20574 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
20576 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
20578 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
20579 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
20580 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
20581 mulv2sf3, divv2sf3): Delete expanders.
20583 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20585 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
20587 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20589 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
20590 * config/rs6000/rs6000.c: Ditto.
20592 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20594 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
20595 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
20596 comparison_operator.
20598 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20600 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
20601 * config/rs6000/rs6000.opt: Ditto.
20602 * config/rs6000/t-rtems: Ditto.
20604 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20606 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
20607 TARGET_E500_SINGLE by 0, simplify.
20608 * config/rs6000/rs6000.c: Ditto.
20609 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
20610 (spe_build_register_parallel): Delete.
20611 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
20612 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
20613 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
20614 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
20615 (E500_CONVERT): Delete.
20616 * config/rs6000/spe.md: Remove many patterns and all define_constants.
20618 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20620 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
20621 * config/rs6000/dfp.md: Ditto.
20622 (negdd2, *negdd2_fpr): Merge.
20623 (absdd2, *absdd2_fpr): Merge.
20624 (negtd2, *negtd2_fpr): Merge.
20625 (abstd2, *abstd2_fpr): Merge.
20626 * config/rs6000/e500.h: Delete file.
20627 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
20628 TARGET_FPRS by 1 and simplify.
20629 * config/rs6000/rs6000-c.c: Ditto.
20630 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
20631 TARGET_DF_SPE by 0.
20632 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
20634 * config/rs6000/rs6000.md: Ditto.
20635 (floatdidf2, *floatdidf2_fpr): Merge.
20636 (move_from_CR_gt_bit): Delete.
20637 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
20638 (E500_CR_IOR_COMPARE): Delete.
20639 (All patterns that require !TARGET_FPRS): Delete.
20640 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
20642 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20644 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
20646 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20648 * graphds.c (add_edge): Intitialize edge's attached data.
20649 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
20650 pointer parameter. Call pointed function on each edge during
20651 graph traversing. Skip traversing the edge when the function
20653 (graphds_dfs, graphds_scc): Ditto.
20654 (for_each_edge): New parameter. Pass the new parameter to callback
20656 * graphds.h (skip_edge_callback): New function pointer type.
20657 (graphds_dfs, graphds_scc): New function pointer parameter.
20658 (graphds_edge_callback, for_each_edge): New parameter.
20660 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20662 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
20663 out code checking if runtime alias check is possible to below ...
20664 Call the new function.
20665 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
20666 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
20668 2017-06-07 Marek Polacek <polacek@redhat.com>
20671 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
20672 TYPE_OVERFLOW_WRAPS check.
20674 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20676 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
20677 if versioning is required.
20678 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
20679 peeling with the check for versioning.
20681 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20683 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
20684 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
20685 Set true to new parameter if new ssa variable is defined.
20686 (vect_gen_vector_loop_niters): Refactor. Set range information
20687 for the new vector loop bound variable.
20688 (vect_do_peeling): Ditto.
20690 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20692 * tree-affine.c (ssa.h): Include header file.
20693 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
20694 has wrapping overflow behavior.
20696 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20698 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
20700 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20702 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
20703 (tree_to_aff_combination): ... here.
20705 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20707 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
20708 reg_pressure model function.
20709 (ivopts_global_cost_for_size): Delete.
20710 (determine_set_costs, iv_ca_recount_cost): Call new model function
20711 ivopts_estimate_reg_pressure.
20713 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20715 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
20716 expensive than udiv. Remove floating point cases from mod.
20718 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20720 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
20721 Increase idiv cost.
20723 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20725 * config/aarch64/aarch64.md
20726 (copysignsf3): Fix mask generation.
20728 2017-06-07 Jakub Jelinek <jakub@redhat.com>
20730 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
20732 (class dump_manager): Add register_dumps method.
20733 * dumpfile.c: Include langhooks.h.
20734 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
20735 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
20736 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
20737 (dump_manager::dump_register): Start with 512 entries instead of 32.
20738 (dump_manager::register_dumps): New method.
20739 * toplev.c (general_init): Instead of invoking register_dumps
20740 langhook, invoke register_dumps method on the dump manager.
20741 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
20744 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
20746 * doc/md.texi: Clarify the restrictions on a define_insn condition.
20747 Say that # requires an associated define_split to exist, and that
20748 the define_split must be suitable for use after register allocation.
20750 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20752 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
20753 (compute_outgoing_frequencies): Also initialize zero counts.
20754 (find_many_sub_basic_blocks): Do not produce uninitialized profile
20755 around loops; preserve more of profile when nothing changes.
20757 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
20759 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
20761 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
20762 * config/arm/arm-cpu-cdata.h: Regenerate.
20763 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
20764 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
20765 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
20766 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
20767 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
20769 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
20771 * config/arm/t-rmprofile: Likewise.
20772 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
20774 2017-06-06 David S. Miller <davem@davemloft.net>
20777 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
20778 blockage if function uses alloca.
20780 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20782 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
20783 New "uid" fields to hold pretty-print IDs of group and ref.
20784 Memory references are now identified as <group_id>:<ref_id>
20785 instead of using [random] addresses.
20786 (dump_mem_details): Simplify, no functional change.
20787 (dump_mem_ref): Simplify and make output more concise.
20788 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
20789 (find_or_create_group): Initialize group uid.
20790 (record_ref): Initialize ref uid. Improve debug output.
20791 (prune_group_by_reuse, should_issue_prefetch_p,)
20792 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
20793 (mark_nontemporal_store, determine_loop_nest_reuse):
20794 Improve debug output.
20796 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20798 * dbgcnt.def (prefetch): New debug counter.
20799 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
20800 (schedule_prefetches): Stop issueing prefetches if debug counter
20803 2017-06-06 Tom de Vries <tom@codesourcery.com>
20805 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
20806 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
20808 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20810 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
20811 Use aarch64_reg_or_zero predicate for operand 4.
20812 (aarch64_compare_and_swap<mode> define_insn_and_split):
20813 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
20814 (aarch64_store_exclusive<mode>): Likewise for operand 2.
20816 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
20818 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
20819 (arm_compute_save_core_reg_mask): This.
20820 (thumb1_compute_save_reg_mask): Rename into ...
20821 (thumb1_compute_save_core_reg_mask): This.
20822 (arm_compute_save_reg0_reg12_mask): Adapt comment.
20823 (arm_compute_frame_layout): Likewise.
20825 2017-06-06 Richard Biener <rguenther@suse.de>
20827 PR tree-optimization/80974
20828 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
20829 keep or clear leaders SSA info.
20831 2017-06-06 Tom de Vries <tom@codesourcery.com>
20833 * config/nvptx/nvptx.c (split_mode_p): New function.
20834 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
20836 2017-06-06 Tom de Vries <tom@codesourcery.com>
20838 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
20840 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20843 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
20846 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20848 * shrink-wrap.c (handle_simple_exit): Update profile.
20849 (try_shrink_wrapping): Upate profile.
20851 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20853 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
20854 (tree_guess_outgoing_edge_probabilities): New.
20855 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
20856 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
20858 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20860 * ipa-split.c (split_function): Initialize return bb profile.
20862 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20864 * profile.c (compute_branch_probabilities): Also initialize
20865 EXIT_BLOCK profile.
20867 2017-06-06 Richard Biener <rguenther@suse.de>
20869 PR tree-optimization/80928
20870 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
20871 (vect_analyze_loop_operations): Properly guard analysis for
20873 (vect_transform_loop): Likewise.
20874 (vect_analyze_loop_2): Also reset SLP type on PHIs.
20875 (vect_model_induction_cost): Do not cost for pure SLP.
20876 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
20877 of induction in inner loop vectorization.
20878 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
20879 (vect_get_and_check_slp_defs): Handle vect_induction_def.
20880 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
20882 (vect_analyze_slp_cost_1): Cost induction.
20883 (vect_detect_hybrid_slp_stmts): Handle PHIs.
20884 (vect_get_slp_vect_defs): Likewise.
20885 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
20886 (vect_transform_stmt): Handle SLP reductions.
20887 * tree-vectorizer.h (vectorizable_induction): Adjust.
20889 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
20891 * config/rs6000/rs6000.c (make_resolver_func): Update
20892 init_lowered_empty_function call.
20894 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
20896 * doc/invoke.texi: Document the -fprofile-abs-path option.
20897 * common.opt (fprofile-abs-path): New option.
20898 * gcov-io.h (gcov_write_filename): Declare.
20899 * gcov-io.c (gcov_write_filename): New function.
20900 * coverage.c (coverage_begin_function): Use gcov_write_filename.
20901 * profile.c (output_location): Likewise.
20903 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20905 * shring-wrap.c: Revert accidental commit.
20907 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
20909 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
20911 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20913 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
20915 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
20916 profile in callgraph edge.
20917 * profile-count.h (apply_probability): If THIS is 0, then result is 0
20918 (apply_scale): Likewise.
20919 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
20920 Also scale profile when inlining function with zero profile.
20921 (initialize_cfun): Update exit block profile even when it is zero.
20922 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
20923 when profile is read.
20925 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
20927 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
20928 (CLONE_*): New constants to define the processors we can generate
20929 code for with the target_clone attribute.
20930 (rs6000_clone_map): New array to identify which clone processors
20931 the current program is running on.
20932 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
20933 target_clone attribute.
20934 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
20935 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
20936 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
20937 (cpu_expand_builtin): Add support for target_clone attribute.
20938 (rs6000_valid_attribute_p): Allow "default" attribute.
20939 (get_decl_name): New debug function to simplify printing the
20940 current function name in debugging statements.
20941 (rs6000_clone_priority): New functions to support the target_clone
20942 attribute, and be able to generate code to switch between ISA 2.05
20943 through ISA 3.0 (power6 through power9).
20944 (rs6000_compare_version_priority): Likewise.
20945 (rs6000_get_function_versions_dispatcher): Likewise.
20946 (make_resolver_func): Likewise.
20947 (add_condition_to_bb): Likewise.
20948 (dispatch_function_versions): Likewise.
20949 (rs6000_generate_version_dispatcher_body): Likewise.
20950 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
20951 (fusion_gpr_load_p): Fix a spacing issue.
20952 * doc/extend.texi (Common Function Attributes): Document that the
20953 PowerPC supports the target_clone attribute.
20955 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
20957 * config/arm/arm.h: explain F symbol found in description of ARM
20958 register allocation in its legend.
20960 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20962 * config/mips/frame-header-opt.c: Include profile-count.h.
20963 * config/riscv/riscv.c: Include profile-count.h
20965 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20967 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
20969 (sm_set_flag_if_changed): Add bbs field.
20970 (execute_sm_if_changed_flag_set): Pass BBS.
20971 (execute_sm): Update.
20973 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20975 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
20978 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20980 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
20981 (peephole2): New peephole2 to emit the above.
20982 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
20984 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20986 * config/aarch64/aarch64.c (define_peephole2 above
20987 *sub_<shift>_<mode>): New peephole.
20989 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
20991 * config/i386/i386.c (make_resolver_func): Update.
20992 * Makefile.in: Add profile-count.h and profile-count.o
20993 * auto-profile.c (afdo_indirect_call): Update to new API.
20994 (afdo_set_bb_count): Update.
20995 (afdo_propagate_edge): Update.
20996 (afdo_propagate_circuit): Update.
20997 (afdo_calculate_branch_prob): Update.
20998 (afdo_annotate_cfg): Update.
20999 * basic-block.h: Include profile-count.h
21000 (struct edge_def): Turn count to profile_count.
21001 (struct basic_block_def): Likewie.
21002 (REG_BR_PROB_BASE): Move to profile-count.h
21003 (RDIV): Move to profile-count.h
21004 * bb-reorder.c (max_entry_count): Turn to profile_count.
21005 (find_traces): Update.
21006 (rotate_loop):Update.
21007 (connect_traces):Update.
21008 (sanitize_hot_paths):Update.
21009 * bt-load.c (migrate_btr_defs): Update.
21010 * cfg.c (RDIV): Remove.
21011 (init_flow): Use alloc_block.
21012 (alloc_block): Uninitialize count.
21013 (unchecked_make_edge): Uninitialize count.
21014 (check_bb_profile): Update.
21015 (dump_edge_info): Update.
21016 (dump_bb_info): Update.
21017 (update_bb_profile_for_threading): Update.
21018 (scale_bbs_frequencies_int): Update.
21019 (scale_bbs_frequencies_gcov_type): Update.
21020 (scale_bbs_frequencies_profile_count): New.
21021 * cfg.h (update_bb_profile_for_threading): Update.
21022 (scale_bbs_frequencies_profile_count): Declare.
21023 * cfgbuild.c (compute_outgoing_frequencies): Update.
21024 (find_many_sub_basic_blocks): Update.
21025 * cfgcleanup.c (try_forward_edges): Update.
21026 (try_crossjump_to_edge): Update.
21027 * cfgexpand.c (expand_gimple_tailcall): Update.
21028 (construct_exit_block): Update.
21029 * cfghooks.c (verify_flow_info): Update.
21030 (dump_bb_for_graph): Update.
21031 (split_edge): Update.
21032 (make_forwarder_block): Update.
21033 (duplicate_block): Update.
21034 (account_profile_record): Update.
21035 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
21036 (get_estimated_loop_iterations): Update.
21037 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
21038 (single_likely_exit): Update.
21039 * cfgloopmanip.c (scale_loop_profile): Update.
21041 (set_zero_probability): Update.
21042 (lv_adjust_loop_entry_edge): Update.
21043 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
21044 (purge_dead_edges): Update.
21045 (rtl_account_profile_record): Update.
21046 * cgraph.c (cgraph_node::create): Uninitialize count.
21047 (symbol_table::create_edge): Uninitialize count.
21048 (cgraph_update_edges_for_call_stmt_node): Update.
21049 (cgraph_edge::dump_edge_flags): Update.
21050 (cgraph_node::dump): Update.
21051 (cgraph_edge::maybe_hot_p): Update.
21052 * cgraph.h: Include profile-count.h
21053 (create_clone), create_edge, create_indirect_edge): Update.
21054 (cgraph_node): Turn count to profile_count.
21055 (cgraph_edge0: Likewise.
21056 (make_speculative, clone): Update.
21057 (create_edge): Update.
21058 (init_lowered_empty_function): Update.
21059 * cgraphclones.c (cgraph_edge::clone): Update.
21060 (duplicate_thunk_for_node): Update.
21061 (cgraph_node::create_clone): Update.
21062 * cgraphunit.c (cgraph_node::analyze): Update.
21063 (cgraph_node::expand_thunk): Update.
21064 * final.c (dump_basic_block_info): Update.
21065 * gimple-streamer-in.c (input_bb): Update.
21066 * gimple-streamer-out.c (output_bb): Update.
21067 * graphite.c (print_global_statistics): Update.
21068 (print_graphite_scop_statistics): Update.
21069 * hsa-brig.c: Include basic-block.h.
21070 * hsa-dump.c: Include basic-block.h.
21071 * hsa-gen.c (T sum_slice): Update.
21072 (convert_switch_statements):Update.
21073 * hsa-regalloc.c: Include basic-block.h.
21074 * ipa-chkp.c (chkp_produce_thunks): Update.
21075 * ipa-cp.c (struct caller_statistics): Update.
21076 (init_caller_stats): Update.
21077 (gather_caller_stats): Update.
21078 (ipcp_cloning_candidate_p): Update.
21079 (good_cloning_opportunity_p): Update.
21080 (get_info_about_necessary_edges): Update.
21081 (dump_profile_updates): Update.
21082 (update_profiling_info): Update.
21083 (update_specialized_profile): Update.
21084 (perhaps_add_new_callers): Update.
21085 (decide_about_value): Update.
21086 (ipa_cp_c_finalize): Update.
21087 * ipa-devirt.c (struct odr_type_warn_count): Update.
21088 (struct decl_warn_count): Update.
21089 (struct final_warning_record): Update.
21090 (possible_polymorphic_call_targets): Update.
21091 (ipa_devirt): Update.
21092 * ipa-fnsummary.c (redirect_to_unreachable): Update.
21093 * ipa-icf.c (sem_function::merge): Update.
21094 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
21095 * ipa-inline.c (compute_uninlined_call_time): Update.
21096 (compute_inlined_call_time): Update.
21097 (want_inline_small_function_p): Update.
21098 (want_inline_self_recursive_call_p): Update.
21099 (edge_badness): Update.
21100 (lookup_recursive_calls): Update.
21101 (recursive_inlining): Update.
21102 (inline_small_functions): Update.
21103 (dump_overall_stats): Update.
21104 (dump_inline_stats): Update.
21105 * ipa-profile.c (ipa_profile_generate_summary): Update.
21106 (ipa_propagate_frequency): Update.
21107 (ipa_profile): Update.
21108 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
21109 * ipa-utils.c (ipa_merge_profiles): Update.
21110 * loop-doloop.c (doloop_modify): Update.
21111 * loop-unroll.c (report_unroll): Update.
21112 (unroll_loop_runtime_iterations): Update.
21113 * lto-cgraph.c (lto_output_edge): Update.
21114 (lto_output_node): Update.
21115 (input_node): Update.
21116 (input_edge): Update.
21117 (merge_profile_summaries): Update.
21118 * lto-streamer-in.c (input_cfg): Update.
21119 * lto-streamer-out.c (output_cfg): Update.
21120 * mcf.c (create_fixup_graph): Update.
21121 (adjust_cfg_counts): Update.
21122 (sum_edge_counts): Update.
21123 * modulo-sched.c (sms_schedule): Update.
21124 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
21125 * predict.c (maybe_hot_count_p): Update.
21126 (probably_never_executed): Update.
21127 (dump_prediction): Update.
21128 (combine_predictions_for_bb): Update.
21129 (propagate_freq): Update.
21130 (handle_missing_profiles): Update.
21131 (counts_to_freqs): Update.
21132 (rebuild_frequencies): Update.
21133 (force_edge_cold): Update.
21134 * predict.h: Include profile-count.h
21135 (maybe_hot_count_p, counts_to_freqs): UPdate.
21136 * print-rtl-function.c: Do not include cfg.h
21137 * print-rtl.c: Include basic-block.h
21138 * profile-count.c: New file.
21139 * profile-count.h: New file.
21140 * profile.c (is_edge_inconsistent): Update.
21141 (correct_negative_edge_counts): Update.
21142 (is_inconsistent): Update.
21143 (set_bb_counts): Update.
21144 (read_profile_edge_counts): Update.
21145 (compute_frequency_overlap): Update.
21146 (compute_branch_probabilities): Update; Initialize and deinitialize
21148 (branch_prob): Update.
21149 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
21150 (edge_gcov_count): New.
21151 (bb_gcov_count): New.
21152 * shrink-wrap.c (try_shrink_wrapping): Update.
21153 * tracer.c (better_p): Update.
21154 * trans-mem.c (expand_transaction): Update.
21155 (ipa_tm_insert_irr_call): Update.
21156 (ipa_tm_insert_gettmclone_call): Update.
21157 * tree-call-cdce.c: Update.
21158 * tree-cfg.c (gimple_duplicate_sese_region): Update.
21159 (gimple_duplicate_sese_tail): Update.
21160 (gimple_account_profile_record): Update.
21161 (execute_fixup_cfg): Update.
21162 * tree-inline.c (copy_bb): Update.
21163 (copy_edges_for_bb): Update.
21164 (initialize_cfun): Update.
21165 (freqs_to_counts): Update.
21166 (copy_cfg_body): Update.
21167 (expand_call_inline): Update.
21168 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
21169 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
21170 (try_unroll_loop_completely): Update.
21171 (try_peel_loop): Update.
21172 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
21173 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
21174 * tree-ssa-loop-split.c (connect_loops): Update.
21175 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
21176 * tree-ssa-reassoc.c (branch_fixup): Update.
21177 * tree-ssa-tail-merge.c (replace_block_by): Update.
21178 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
21179 (compute_path_counts): Update.
21180 (update_profile): Update.
21181 (recompute_probabilities): Update.
21182 (update_joiner_offpath_counts): Update.
21183 (estimated_freqs_path): Update.
21184 (freqs_to_counts_path): Update.
21185 (clear_counts_path): Update.
21186 (ssa_fix_duplicate_block_edges): Update.
21187 (duplicate_thread_path): Update.
21188 * tree-switch-conversion.c (case_bit_test_cmp): Update.
21189 (struct switch_conv_info): Update.
21190 * tree-tailcall.c (decrease_profile): Update.
21191 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
21192 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
21193 * value-prof.c (check_counter): Update.
21194 (gimple_divmod_fixed_value): Update.
21195 (gimple_mod_pow2): Update.
21196 (gimple_mod_subtract): Update.
21197 (gimple_ic_transform): Update.
21198 (gimple_stringop_fixed_value): Update.
21199 * value-prof.h (gimple_ic): Update.
21201 2017-06-02 Carl Love <cel@us.ibm.com>
21203 * config/rs6000/rs6000-c: Add support for built-in functions
21204 vector double vec_doublee (vector signed int);
21205 vector double vec_doublee (vector unsigned int);
21206 vector double vec_doublee (vector float);
21207 vector double vec_doubleh (vector signed int);
21208 vector double vec_doubleh (vector unsigned int);
21209 vector double vec_doubleh (vector float);
21210 vector double vec_doublel (vector signed int);
21211 vector double vec_doublel (vector unsigned int);
21212 vector double vec_doublel (vector float);
21213 vector double vec_doubleo (vector signed int);
21214 vector double vec_doubleo (vector unsigned int);
21215 vector double vec_doubleo (vector float);.
21216 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
21217 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
21219 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
21220 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
21221 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
21223 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
21224 vec_doublel, vec_doubleh.
21225 * doc/extend.texi: Update the built-in documentation file for the
21226 new built-in functions.
21228 2017-06-02 David Malcolm <dmalcolm@redhat.com>
21231 * ipa-inline-analysis.c (free_growth_caches): Set
21232 edge_removal_hook_holder to NULL after removing it.
21234 2017-06-02 Sudakshina Das <sudi.das@arm.com>
21236 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
21237 comparision with zero.
21239 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21240 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21241 for early expansion of vec_min and vec_max builtins.
21242 (builtin_function_type): Add min/max unsigned variants to those
21243 identified as having unsigned arguments.
21245 2017-06-02 Olivier Hainque <hainque@adacore.com>
21247 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
21249 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21251 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
21252 Use VALL_F16 iterator rather than VALL.
21254 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21256 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
21257 Emit CBNZ inside loop when doing a strong exchange and comparing
21258 against zero. Generate the CC flags after the loop.
21260 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
21262 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
21263 (dl_section_ref): New.
21264 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
21265 On AIX, append an expression to subtract the size of the
21266 section length to dl_section_ref.
21268 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21270 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21271 for early expansion of vector absolute builtins.
21273 2017-06-02 Richard Biener <rguenther@suse.de>
21275 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
21276 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
21278 2017-06-02 Richard Biener <rguenther@suse.de>
21280 PR tree-optimization/80948
21281 * tree-tailcall.c (find_tail_calls): Track stmts to move in
21282 stmt order as well.
21284 2017-06-02 Richard Biener <rguenther@suse.de>
21286 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
21288 * tree-vect-stmts.c (process_use): Do not mark backedge defs
21289 for inductions as relevant.
21291 2017-06-02 Richard Biener <rguenther@suse.de>
21293 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
21294 (vectorizable_induction): ... this. Remove dead code.
21296 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
21298 * builtins. (expand_builtin_alloca): Remove second parameter and
21299 infer its value from the first parameter instead.
21300 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
21302 2017-06-02 Jakub Jelinek <jakub@redhat.com>
21304 PR rtl-optimization/80903
21305 * loop-doloop.c (add_test): Unshare sequence.
21307 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
21309 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
21311 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
21313 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
21315 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
21316 xlogue_layout::get_instance, logue_layout::xlogue_layout,
21317 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
21318 (xlogue_layout::get_stub_rtx): Make static.
21319 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
21320 (xlogue_layout::compute_stub_managed_regs): Rename to...
21321 (xlogue_layout::count_stub_managed_regs): ...this.
21322 (xlogue_layout::is_stub_managed_reg): New function.
21323 (xlogue_layout::m_stub_names): Rename to...
21324 (xlogue_layout::s_stub_names): ...this, make static.
21325 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
21326 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
21327 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
21328 xlogue_layout::s_stub_names): Instantiate statics.
21329 (stub_managed_regs): Remove.
21330 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
21331 (disable_call_ms2sysv_xlogues): Rename to...
21332 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
21333 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
21335 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
21336 change after reload_completed.
21337 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
21339 (ix86_expand_prologue): Likewise.
21340 (ix86_expand_epilogue): Likewise.
21341 (ix86_expand_split_stack_prologue): Likewise.
21342 (ix86_compute_frame_layout): Remove frame parameter ...
21343 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
21344 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
21346 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
21347 (ix86_frame): Move from here ...
21348 * config/i386/i386.h (ix86_frame): ... to here.
21349 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
21350 complete ix86_frame data structure instead. Remove some_ld_name.
21352 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
21354 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
21355 symbols that hold a DECL_VALUE_EXPR.
21357 2017-06-01 Martin Jambor <mjambor@suse.cz>
21359 PR tree-optimization/80898
21360 * tree-sra.c (process_subtree_disqualification): Removed.
21361 (disqualify_candidate): Do not acll
21362 process_subtree_disqualification.
21363 (subtree_mark_written_and_enqueue): New function.
21364 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
21365 RHS has been disqualified and re-queue LHS if necessary. Apart
21366 from that, ignore disqualified RHS.
21368 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21370 * config/s390/s390.c (s390_emit_epilogue): Disable early return
21371 address fetch for z10 or later.
21373 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21375 * config/arc/arc.md (tst_movb): Add guard when splitting.
21377 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21379 * config/arc/arc.c (arc_can_eliminate): Test against
21380 arc_frame_pointer_needed.
21382 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21384 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
21385 to prevent store reordering.
21386 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
21387 (type): Add block type.
21388 (stack_tie): Define special instruction to be used in
21391 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21393 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
21394 constraint. It is not valid for the pattern.
21395 (noncommutative_binary_comparison): Likewise.
21397 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21399 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
21402 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21404 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
21405 be used by the reg-alloc.
21407 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21409 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
21410 reg-alloc when having mul64 or mul32x16 instructions.
21411 (mulsidi3): Likewise.
21412 (umulsidi3): Likewise.
21413 (mulsi32x16): New pattern.
21414 (mulsi64): Likewise.
21415 (mulsidi64): Likewise.
21416 (umulsidi64): Likewise.
21417 (MUL32x16_REG): Define.
21418 (mul64_600): Use MUL32x16_REG.
21419 (mac64_600): Likewise.
21420 (umul64_600): Likewise.
21421 (umac64_600): Likewise.
21423 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21425 * config/arc/arc.md (mulsi3_700): Make it commutative.
21427 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
21429 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
21431 (*sign_extendsidi2_insn): Likewise for movstosw.
21433 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
21435 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
21436 the type of the input discriminant value. Convert the
21437 discriminant value of signedness vary.
21439 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
21441 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
21442 Add to -Wall section.
21444 2017-06-01 Richard Biener <rguenther@suse.de>
21446 PR middle-end/66313
21447 * fold-const.c (fold_plusminus_mult_expr): If the factored
21448 factor may be zero use a wrapping type for the inner operation.
21449 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
21450 and handle moved defs.
21451 (process_assignment): Properly guard the unary op case. Return a
21452 tri-state indicating that moving the stmt before the call may allow
21453 to continue. Pass through to_move.
21454 (find_tail_calls): Handle moving unrelated defs before
21457 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
21460 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
21461 splitter result in the canonical way.
21463 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
21465 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
21466 also for 32bit target. Update insn attributes.
21467 (zero-extendsidi2 splitter): Allow all registers for operand 1.
21469 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
21471 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
21472 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
21473 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
21474 (_mm_maskz_min_ss): New intrinsics.
21476 2017-05-31 Martin Liska <mliska@suse.cz>
21478 * tree-vect-loop.c (vect_create_epilog_for_reduction):
21479 Change comment style to one we normally use.
21480 (vectorizable_reduction): Likewise.
21481 (vectorizable_induction): Likewise.
21482 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
21483 (vectorizable_call): Likewise.
21484 (vectorizable_simd_clone_call): Likewise.
21485 (vectorizable_conversion): Likewise.
21486 (vectorizable_assignment): Likewise.
21487 (vectorizable_shift): Likewise.
21488 (vectorizable_operation): Likewise.
21489 (vectorizable_store): Likewise.
21490 (vectorizable_load): Likewise.
21491 * tree-vectorizer.h: Likewise.
21493 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
21495 * passes.c (emergency_dump_function): New.
21496 * tree-pass.h (emergency_dump_function): Declare.
21497 * plugin.c (plugins_internal_error_function): Remove.
21498 * plugin.h (plugins_internal_error_function): Remove declaration.
21499 * toplev.c (internal_error_function): New static function. Use it...
21500 (general_init): ...here.
21502 2017-05-31 Graham Markall <graham.markall@embecosm.com>
21504 * config/arc/arc.c (arc_print_operand): Handle constant operands.
21505 (arc_rtx_costs): Add costs for new patterns.
21506 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
21507 * config/arc/predicates.md: Add _1_2_3_operand predicate.
21509 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21511 * tree-ssa-strlen.c (get_next_strinfo): New function.
21512 (get_stridx_plus_constant): Use it.
21513 (zero_length_string): Likewise.
21514 (adjust_related_strinfos): Likewise.
21515 (adjust_last_stmt): Likewise.
21517 2017-05-31 Richard Biener <rguenther@suse.de>
21520 * config/i386/i386.c (ix86_expand_builtin): Remove assert
21521 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
21523 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21525 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
21526 loop_vinfo argument and use of dependence distance vectors.
21527 Check instead whether the two references differ only in their
21528 initial value and assume that they have the same alignment if the
21529 difference is a multiple of the vector alignment.
21530 (vect_analyze_data_refs_alignment): Update call accordingly.
21532 2017-05-31 Martin Liska <mliska@suse.cz>
21535 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
21537 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21539 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
21540 (create_intersect_range_checks): Move from ...
21541 * tree-data-ref.c (create_intersect_range_checks_index)
21542 (create_intersect_range_checks): ... to here.
21543 (create_runtime_alias_checks): New function factored from ...
21544 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
21545 here. Call above function.
21546 * tree-data-ref.h (create_runtime_alias_checks): New function.
21548 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21550 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
21551 segment length for dr_b and compute it in wide_int.
21553 2017-05-31 Richard Biener <rguenther@suse.de>
21555 PR tree-optimization/80906
21556 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
21557 and pass through iv_map.
21558 (copy_bb_and_scalar_dependences): Adjust.
21559 (translate_pending_phi_nodes): Likewise.
21560 (copy_loop_close_phi_args): Handle code-generating IVs instead
21563 2017-05-30 David Malcolm <dmalcolm@redhat.com>
21565 * diagnostic-color.c (color_dict): Add "type-diff".
21566 (parse_gcc_colors): Update comment.
21567 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
21568 -fdiagnostics-show-template-tree and -fno-elide-type.
21569 (GCC_COLORS): Add type-diff to example.
21571 (-fdiagnostics-show-template-tree): New.
21572 (-fno-elide-type): New.
21573 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
21574 the pp_format_decoder callback. Call any m_format_postprocessor's
21576 (pretty_printer::pretty_printer): Initialize
21577 m_format_postprocessor.
21578 (pretty_printer::~pretty_printer): Delete any
21579 m_format_postprocessor.
21580 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
21581 (class format_postprocessor): New class.
21582 (struct pretty_printer::format_decoder): Document the new parameters.
21583 (struct pretty_printer::m_format_postprocessor): New field.
21584 * tree-diagnostic.c (default_tree_printer): Update for new
21585 bool and const char ** params.
21586 * tree-diagnostic.h (default_tree_printer): Likewise.
21588 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
21590 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
21591 (lwa_operand): Delete rs6000_gen_cell_microcode test.
21592 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
21593 rs6000_gen_cell_microcode code.
21594 (rs6000_final_prescan_insn): Delete.
21595 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
21596 "warn-cell-microcode" entries.
21597 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
21598 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
21599 throughout. Change cc_reg_not_micro_cr0_operand to
21600 cc_reg_not_cr0_operand throughout.
21601 (*extendhi<mode>2_noload): Delete.
21602 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
21603 (mwarn-cell-microcode): Delete.
21604 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
21605 -mgen-cell-microcode and -mwarn-cell-microcode.
21607 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
21610 * config/i386/constraints.md (Yd): New constraint.
21612 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
21613 and (?Yd, r) alternatives. Update insn attributes.
21614 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
21615 and (?*Yd, r) alternatives. Update insn attributes.
21616 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
21618 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
21620 * gimplify.c (gimplify_modify_expr): Don't create a
21621 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
21624 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
21626 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
21628 2017-05-30 Richard Biener <rguenther@suse.de>
21630 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
21631 and reduc_def fields.
21632 (STMT_VINFO_REDUC_TYPE): New define.
21633 (STMT_VINFO_REDUC_DEF): Likewise.
21634 (vect_force_simple_reduction): Adjust prototype.
21635 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
21636 (vect_is_simple_reduction): Remove check_reduction argument.
21637 (vect_force_simple_reduction): Adjust and set
21638 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21639 (vectorizable_reduction): Do not re-do reduction analysis
21640 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21641 * tree-parloops.c (gather_scalar_reductions): Adjust.
21643 2017-05-30 Richard Biener <rguenther@suse.de>
21645 PR middle-end/80901
21646 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
21649 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
21651 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
21652 Introduce unknown_misalignment parameter and remove vf.
21653 (vect_peeling_hash_get_lowest_cost):
21654 Pass unknown_misalignment parameter.
21655 (vect_enhance_data_refs_alignment):
21656 Fix unsupportable data ref treatment.
21658 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21660 * tree-vect-data-refs.c (vect_get_data_access_cost):
21661 Workaround for SLP handling.
21662 (vect_enhance_data_refs_alignment):
21663 Compute costs for doing no peeling at all, compare to the best
21664 peeling costs so far and avoid peeling if cheaper.
21666 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21668 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
21669 Return peeling info and set costs to zero for unlimited cost
21671 (vect_enhance_data_refs_alignment): Also inspect all datarefs
21672 with unknown misalignment. Compute and costs for unknown
21673 misalignment, compare them to the costs for known misalignment
21674 and choose the cheapest for peeling.
21676 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21678 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
21679 (vect_get_peeling_costs_all_drs): Create function.
21680 (vect_peeling_hash_get_lowest_cost):
21681 Use vect_get_peeling_costs_all_drs.
21682 (vect_peeling_supportable): Create function.
21683 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
21685 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21687 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
21688 DR_HAS_NEGATIVE_STEP.
21689 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
21690 (vect_enhance_data_refs_alignment): Use.
21691 (vect_duplicate_ssa_name_ptr_info): Use.
21692 * tree-vectorizer.h (dr_misalignment): Use.
21693 (known_alignment_for_access_p): Use.
21695 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21698 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
21700 (has_section_name): New function.
21702 2017-05-30 Martin Liska <mliska@suse.cz>
21705 * auto-profile.c (get_function_decl_from_block): Fix
21708 2017-05-30 Richard Biener <rguenther@suse.de>
21710 PR middle-end/80876
21711 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
21713 2017-05-30 Martin Liska <mliska@suse.cz>
21715 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
21716 * dumpfile.h (struct dump_file_info): Remove ctors.
21718 2017-05-30 Martin Liska <mliska@suse.cz>
21720 * predict.def: Fix GNU coding style.
21722 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
21724 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
21725 Mark 'to' argument with ATTRIBUTE_UNUSED.
21727 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
21729 * config/xtensa/xtensa.c (xtensa_emit_call): Use
21730 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
21731 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
21734 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
21736 * doc/install.texi (Options specification): Restore entry of
21737 --enable-sjlj-exceptions.
21739 2017-05-27 Michael Eager <eager@eagercon.com>
21742 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
21744 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
21746 * config/microblaze/microblaze.h
21747 (FIXED_REGISTERS): Update in macro.
21748 (CALL_USED_REGISTERS): Update in macro.
21750 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
21752 * doc/install.texi: Add links to macOS binary distributions.
21754 2017-05-27 Jakub Jelinek <jakub@redhat.com>
21758 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21760 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
21762 2017-05-26 Martin Liska <mliska@suse.cz>
21764 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
21766 2017-05-26 Martin Liska <mliska@suse.cz>
21768 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
21769 always leading ';; '.
21770 (dump_bb_info): Likewise.
21771 (brief_dump_cfg): Likewise.
21772 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
21773 * dumpfile.c: Remove usage of TDF_VERBOSE.
21774 * dumpfile.h (enum dump_kind): Likewise.
21775 (dump_gimple_bb_header): Do not use TDF_COMMENT.
21776 * print-tree.c (debug_verbose): Remove.
21777 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
21778 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
21779 * tree-diagnostic.c (default_tree_printer): Replace
21780 TDF_DIAGNOSTIC with TDF_SLIM.
21782 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21784 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
21785 in parameter loop, rather than loop_vinfo.
21786 (create_intersect_range_checks): Ditto.
21787 (vect_create_cond_for_alias_checks): Update call to above functions.
21789 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21791 PR tree-optimization/80815
21792 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
21793 for merging runtime alias checks. Handle negative DR_STEPs.
21795 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21797 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
21799 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
21800 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
21801 out code pruning runtime alias checks.
21802 * tree-data-ref.c (prune_runtime_alias_test_list): New function
21803 factored out from above.
21804 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
21806 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
21808 (prune_runtime_alias_test_list): New decalaration.
21810 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21812 * tree-vect-data-refs.c (compare_tree): Rename and move ...
21813 * tree-data-ref.c (data_ref_compare_tree): ... to here.
21814 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
21815 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
21816 (operator==, comp_dr_with_seg_len_pair): Ditto.
21817 (vect_prune_runtime_alias_test_list): Ditto.
21819 2017-05-26 Martin Liska <mliska@suse.cz>
21822 * params.def: Bound partial-inlining-entry-probability param.
21824 2017-05-26 Marek Polacek <polacek@redhat.com>
21827 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
21830 2017-05-26 Richard Biener <rguenther@suse.de>
21832 PR tree-optimization/80842
21833 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
21836 2017-05-26 Richard Biener <rguenther@suse.de>
21838 PR tree-optimization/80844
21839 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
21841 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
21843 * doc/md.texi (Machine Constraints): Update x86 family
21844 machine constraints section to match 'config/i386/constraints.md'.
21846 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
21848 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
21850 2017-05-25 Nathan Sidwell <nathan@acm.org>
21852 * doc/invoke.texi (--enable-languages): Update documentation.
21854 2017-05-25 Martin Liska <mliska@suse.cz>
21856 * dumpfile.c: Add TDF_FOLDING.
21857 * dumpfile.h (enum dump_kind): Likewise.
21858 * genmatch.c (dt_simplify::gen_1): Use it.
21860 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21862 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
21864 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21866 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
21867 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
21869 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21871 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
21872 * match.pd (X == C): Rewrite it here.
21873 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
21874 with_certain_nonzero_bits2): New predicates.
21875 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
21877 2017-05-24 Nathan Sidwell <nathan@acm.org>
21879 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
21882 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
21885 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
21887 * config/powerpcspe: New port. Files are copied from the rs6000
21888 port, with "rs6000" in filenames replaced by "powerpcspe".
21890 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
21892 PR rtl-optimization/80754
21893 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
21895 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
21897 * config/sparc/sparc.md (length): Return the correct value for -mflat
21898 sibcalls to match output_sibcall.
21900 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
21904 * config/rs6000/rs6000.c (struct machine_function): Add new field
21906 (rs6000_get_separate_components): Init that field, use it.
21907 (rs6000_components_for_bb): Use the field.
21909 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
21911 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
21913 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
21915 PR middle-end/80823
21916 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
21918 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21921 * config/s390/s390.c (s390_check_qrst_address): Check incoming
21922 address against address_operand predicate.
21923 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
21925 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
21927 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
21928 parameters passed indirectly.
21930 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
21932 * config/i386/i386.md (*movdi_internal): Remove SSE4
21933 alternative 18 (?r, *v). Update insn attributes.
21934 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
21935 Update insn attributes.
21936 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
21937 Update insn attributes.
21938 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
21939 alternative 1 (r, v). Remove isa attribute.
21940 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
21941 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
21942 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
21944 2017-05-23 Tom de Vries <tom@codesourcery.com>
21946 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
21949 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21951 * cgraphunit.c (symbol_table::process_new_functions): Update.
21952 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
21953 (inline_generate_summary): Rename to ...
21954 (ipa_fn_summary_generate): ... this one.
21955 (inline_read_summary): Rename to ...
21956 (ipa_fn_summary_read): ... this one.
21957 (inline_write_summary): Rename to ...
21958 (ipa_fn_summary_write): ... this one.
21959 (inline_free_summary): Rename to ...
21960 (ipa_free_fn_summary): ... this one.
21961 (pass_data_local_fn_summary, pass_local_fn_summary,
21962 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
21963 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
21964 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
21965 make_pass_ipa_fn_summary): New.
21966 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
21967 inline_write_summary, inline_free_summary): Remove.
21968 (ipa_free_fn_summary) : New.
21969 * ipa-inline.c (ipa_inline): Update.
21970 (pass_ipa_inline): Do not generate summaries.
21971 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
21973 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
21974 and add pass_ipa_fn_summary.
21975 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
21977 (make_pass_inline_parameters): Remove.
21979 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
21981 * omp-low.c (struct omp_context): Remove "default_kind" member.
21984 * omp-offload.c (execute_oacc_device_lower): Remove the
21985 parallelism dimensions function attributes for unparallelized
21986 OpenACC kernels constructs.
21988 2017-05-23 Martin Liska <mliska@suse.cz>
21990 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
21992 (cgraph_edge::make_speculative): Likewise.
21993 (cgraph_edge::resolve_speculation): Likewise.
21994 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
21995 (cgraph_node::dump): Likewise.
21996 * cgraph.h: Likewise.
21997 * cgraphunit.c (analyze_functions): Likewise.
21998 (symbol_table::compile): Likewise.
21999 * ipa-cp.c (print_all_lattices): Likewise.
22000 (determine_versionability): Likewise.
22001 (initialize_node_lattices): Likewise.
22002 (ipcp_verify_propagated_values): Likewise.
22003 (estimate_local_effects): Likewise.
22004 (update_profiling_info): Likewise.
22005 (create_specialized_node): Likewise.
22006 (perhaps_add_new_callers): Likewise.
22007 (decide_about_value): Likewise.
22008 (decide_whether_version_node): Likewise.
22009 (identify_dead_nodes): Likewise.
22010 (ipcp_store_bits_results): Likewise.
22011 * ipa-devirt.c (dump_targets): Likewise.
22012 (ipa_devirt): Likewise.
22013 * ipa-icf.c (sem_item::dump): Likewise.
22014 (sem_function::equals): Likewise.
22015 (sem_variable::equals): Likewise.
22016 (sem_item_optimizer::read_section): Likewise.
22017 (sem_item_optimizer::execute): Likewise.
22018 (congruence_class::dump): Likewise.
22019 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
22020 (dump_inline_summary): Likewise.
22021 (estimate_node_size_and_time): Likewise.
22022 (inline_analyze_function): Likewise.
22023 * ipa-inline-transform.c (inline_call): Likewise.
22024 * ipa-inline.c (report_inline_failed_reason): Likewise.
22025 (want_early_inline_function_p): Likewise.
22026 (edge_badness): Likewise.
22027 (update_edge_key): Likewise.
22028 (inline_small_functions): Likewise.
22029 * ipa-profile.c (ipa_profile): Likewise.
22030 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
22031 (ipa_make_edge_direct_to_target): Likewise.
22032 (remove_described_reference): Likewise.
22033 (ipa_impossible_devirt_target): Likewise.
22034 (propagate_controlled_uses): Likewise.
22035 (ipa_print_node_params): Likewise.
22036 (ipcp_transform_function): Likewise.
22037 * ipa-pure-const.c (pure_const_read_summary): Likewise.
22038 (propagate_pure_const): Likewise.
22039 * ipa-reference.c (generate_summary): Likewise.
22040 (read_write_all_from_decl): Likewise.
22041 (propagate): Likewise.
22042 (ipa_reference_read_optimization_summary): Likewise.
22043 * ipa-utils.c (ipa_merge_profiles): Likewise.
22044 * ipa.c (walk_polymorphic_call_targets): Likewise.
22045 (symbol_table::remove_unreachable_nodes): Likewise.
22046 (ipa_single_use): Likewise.
22047 * passes.c (execute_todo): Likewise.
22048 * predict.c (drop_profile): Likewise.
22049 * symtab.c (symtab_node::get_dump_name): New function.
22050 (symtab_node::dump_name): Likewise.
22051 (symtab_node::dump_asm_name): Likewise.
22052 (symtab_node::dump_references): Likewise.
22053 (symtab_node::dump_referring): Likewise.
22054 (symtab_node::dump_base): Likewise.
22055 (symtab_node::debug_symtab): Likewise.
22056 * tree-sra.c (convert_callers_for_node): Likewise.
22057 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
22058 * value-prof.c (init_node_map): Likewise.
22060 2017-05-23 Martin Liska <mliska@suse.cz>
22062 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
22063 and symtab_node::debug_symtab to symbol_table::debug.
22064 * cgraphunit.c (analyze_functions): Use the renamed function.
22065 (symbol_table::compile): Likewise.
22066 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
22067 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
22068 * passes.c (execute_todo): Likewise.
22069 * symtab.c (symbol_table::dump): New function.
22070 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
22072 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22074 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
22075 that nonconst implies exec.
22077 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22079 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22080 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22081 (inline_edge_summary_vec): Turn into ...
22082 (ipa_call_summaries): ... this one.
22083 (redirect_to_unreachable, edge_set_predicate,
22084 evaluate_properties_for_edge, inline_summary_alloc,
22085 reset_ipa_call_summary, reset_inline_summary,
22086 inline_summary_t::duplicate): Update.
22087 (inline_edge_duplication_hook): Turn to ...
22088 (ipa_call_summary_t::duplicate): ... this one.
22089 (inline_edge_removal_hook): Turn to ...
22090 (ipa_call_summary_t::remove): ... this one.
22091 (dump_inline_edge_summary): Turn to ...
22092 (dump_ipa_call_summary): ... this one.
22093 (estimate_function_body_sizes): Update.
22094 (inline_update_callee_summaries): Update.
22095 (remap_edge_change_prob): Update.
22096 (remap_edge_summaries): Update.
22097 (inline_merge_summary): Update.
22098 (do_estimate_edge_time): Update.
22099 (inline_generate_summary): Update.
22100 (inline_read_section): Update.
22101 (inline_read_summary): Update.
22102 (inline_free_summary): Update.
22103 * ipa-inline.c (can_inline_edge_p): Update.
22104 (compute_inlined_call_time): Update.
22105 (want_inline_small_function_p): Update.
22106 (edge_badness): Update.
22107 (early_inliner): Update.
22108 * ipa-inline.h (inline_edge_summary): Turn to ...
22109 (ipa_call_summary): ... this one.
22110 (ipa_call_summary_t): New class.
22111 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22112 (ipa_call_summaries): New.
22113 (inline_edge_summary): Remove.
22114 (estimate_edge_growth): Update.
22115 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22116 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22117 * ipa-split.c (execute_split_functions): Update.
22118 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22120 2017-05-23 Tom de Vries <tom@codesourcery.com>
22122 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22123 attributes): Document rdrand effective target.
22125 2017-05-23 Tom de Vries <tom@codesourcery.com>
22127 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22128 attributes): Sort alphabetically.
22130 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
22132 * config/avr/genmultilib.awk: Use gsub instead of gensub.
22134 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
22137 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
22138 V2DF/V2DI splat into two separate patterns, one that handles
22139 registers, and the other that only handles memory. Drop support
22140 for splatting from a GPR on ISA 2.07 and then splitting the
22141 splat into direct move and splat.
22142 (vsx_splat_<mode>_reg): Likewise.
22143 (vsx_splat_<mode>_mem): Likewise.
22145 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
22147 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
22149 2017-05-22 Jakub Jelinek <jakub@redhat.com>
22151 PR middle-end/80809
22152 * omp-low.c (finish_taskreg_remap): New function.
22153 (finish_taskreg_scan): If unit size of ctx->record_type
22154 is non-constant, unshare the size expression and replace
22155 decls in it with possible outer var refs.
22157 PR middle-end/80809
22158 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
22159 GOVD_SHARED rather than GOVD_PRIVATE with it.
22160 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
22161 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
22163 PR middle-end/80853
22164 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
22165 as last argument to build_outer_var_ref for pointer bases of array
22166 section reductions.
22168 2017-05-19 Martin Sebor <msebor@redhat.com>
22170 * print-tree.c (print_node): Print DECL_READ_P flag.
22172 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22174 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
22175 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
22176 * cgraph.c: Likewise.
22177 * cgraphunit.c: Likewise.
22178 * gengtype.c: Likewise.
22179 * ipa-cp.c: Likewise.
22180 * ipa-devirt.c: Likewise.
22181 * ipa-icf.c: Likewise.
22182 * ipa-predicate.c: Likewise.
22183 * ipa-profile.c: Likewise.
22184 * ipa-prop.c: Likewise.
22185 * ipa-split.c: Likewise.
22187 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
22188 edge_predicate_pool, dump_inline_hints,
22189 inline_summary::account_size_time, redirect_to_unreachable,
22190 edge_set_predicate, set_hint_predicate,
22191 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
22192 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
22193 inline_summary_t::remove, remap_hint_predicate_after_duplication,
22194 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
22195 ipa_call_summary_t::remove, initialize_growth_caches,
22196 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
22197 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
22198 mark_modified, unmodified_parm_1, unmodified_parm,
22199 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
22200 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
22201 compute_bb_predicates, will_be_nonconstant_expr_predicate,
22202 will_be_nonconstant_predicate, record_modified_bb_info,
22203 get_minimal_bb, record_modified, param_change_prob,
22204 phi_result_unknown_predicate, predicate_for_phi_result,
22205 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
22206 estimate_function_body_sizes, compute_inline_parameters,
22207 compute_inline_parameters_for_curren, pass_data_inline_parameters,
22208 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
22209 inline_update_callee_summaries, remap_edge_change_prob,
22210 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
22211 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
22212 inline_analyze_function, inline_summary_t::insert,
22213 inline_generate_summary, read_ipa_call_summary, inline_read_section,
22214 inline_read_summary, write_ipa_call_summary, inline_write_summary,
22215 inline_free_summary): Move to ipa-fnsummary.h
22216 (predicate_t): Remove.
22217 * ipa-fnsummary.c: New file.
22218 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
22219 (enum inline_hints_vals, inline_hints, agg_position_info,
22220 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
22221 inline_summaries, ipa_call_summary, ipa_call_summary_t,
22222 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
22223 dump_inline_summary, dump_inline_hints, inline_generate_summary,
22224 inline_read_summary, inline_write_summary, inline_free_summary,
22225 inline_analyze_function, initialize_inline_failed,
22226 inline_merge_summary, inline_update_overall_summary,
22227 compute_inline_parameters): Move to ipa-fnsummary.h
22228 * ipa-fnsummary.h: New file.
22229 * ipa-inline-transform.h: Include ipa-inline.h.
22230 * ipa-inline.c: LIkewise.
22232 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22234 * ipa-inline.c (edge_badness): Use inlined_time instead of
22235 inline_summaries->get.
22237 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22239 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
22241 2017-05-22 Nathan Sidwell <nathan@acm.org>
22243 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
22244 (fdump-lang): Document 'raw' option.
22245 * dumpfile.h (TDI_tu): Delete.
22246 * dumpfile.c (dump_files): Remove translation-unit.
22247 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
22249 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
22251 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
22252 command option from $(AWK) call.
22253 * config/avr/genmultilib.awk: Simplify and rewrite so that it
22254 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
22255 [FORMAT]: Remove handling of variable.
22256 * config/avr/t-multilib: Regenerate.
22258 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22260 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
22262 (dump_inline_summary): Do not print self_time.
22263 (estimate_function_body_sizes): Do not set self_time.
22264 (compute_inline_parameters): Likewise.
22265 (inline_read_section, inline_write_summary): Do not stream self_time.
22266 * ipa-inline.h (inline_summary): Drop self_time.
22268 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22270 * ipa-inline-analysis.c (account_size_time): Rename to ...
22271 (inline_summary::account_size_time): ... this one.
22272 (reset_ipa_call_summary): Turn to ...
22273 (ipa_call_summary::reset): ... this one.
22274 (reset_inline_summary): Turn to ...
22275 (inline_summary::reset): ... this one.
22276 (inline_summary_t::remove): Update.
22277 (inline_summary_t::duplicate): Update.
22278 (ipa_call_summary_t::remove): Update.
22279 (dump_inline_summary): Update.
22280 (estimate_function_body_sizes): Update.
22281 (compute_inline_parameters): Update.
22282 (estimate_node_size_and_time): Update.
22283 (inline_merge_summary): Update.
22284 (inline_update_overall_summary): Update.
22285 (inline_read_section): Update.
22286 (inline_write_summary): Update.
22287 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
22288 add account_size_time and reset member functions.
22289 (ipa_call_summary): Add reset function.
22290 * ipa-predicate.h (predicate::operator &): Constify.
22292 2017-05-22 Richard Biener <rguenther@suse.de>
22294 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
22296 2017-05-19 Jason Merrill <jason@redhat.com>
22298 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
22300 2017-05-19 Marek Polacek <polacek@redhat.com>
22303 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
22304 TYPE_OVERFLOW_WRAPS checks.
22306 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
22308 * tree-core.h (enum omp_clause_default_kind): Add
22309 "OMP_CLAUSE_DEFAULT_PRESENT".
22310 * tree-pretty-print.c (dump_omp_clause): Handle it.
22311 * gimplify.c (enum gimplify_omp_var_data): Add
22312 "GOVD_MAP_FORCE_PRESENT".
22313 (gimplify_adjust_omp_clauses_1): Map it to
22314 "GOMP_MAP_FORCE_PRESENT".
22315 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
22317 * gimplify.c (oacc_default_clause): Clarify.
22319 2017-05-19 Nathan Sidwell <nathan@acm.org>
22321 LANG_HOOK_REGISTER_DUMPS
22322 * toplev.c (general_init): Call register dump lang hook.
22323 * doc/invoke.texi: Document -fdump-lang option family.
22324 * dumpfile.c (dump_files): Remove class dump here.
22325 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
22326 * dumpfile.h (tree_dump_index): Remove TDI_class.
22327 * langhooks-def.h (lhd_register_dumps): Declare.
22328 (LANG_HOOKS_REGISTER_DUMPS): Define.
22329 (LANG_HOOKS_INITIALIZER): Add it.
22330 * langhooks.c (lhd_register_dumps): Define.
22331 * langhooks.h (struct lang_hooks): Add register_dumps.
22333 2017-05-19 Nathan Sidwell <nathan@acm.org>
22335 * context.h (context::set_passes): New.
22336 * context.c (context::context): Do not create pass manager.
22337 * toplev.c (general_init): Create pass manager here.
22339 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
22341 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
22342 use this splitter if two add or or instructions would also work for
22343 the constant we want to generate.
22345 2017-05-19 Richard Biener <rguenther@suse.de>
22348 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
22349 predicate evaluation.
22351 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
22353 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
22355 * ipa-inline.c (want_inline_small_function_p): Do not cast to
22358 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
22360 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22361 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22362 (inline_edge_summary_vec): Turn into ...
22363 (ipa_call_summaries): ... this one.
22364 (redirect_to_unreachable, edge_set_predicate,
22365 evaluate_properties_for_edge, inline_summary_alloc,
22366 reset_ipa_call_summary, reset_inline_summary,
22367 inline_summary_t::duplicate): Update.
22368 (inline_edge_duplication_hook): Turn to ...
22369 (ipa_call_summary_t::duplicate): ... this one.
22370 (inline_edge_removal_hook): Turn to ...
22371 (ipa_call_summary_t::remove): ... this one.
22372 (dump_inline_edge_summary): Turn to ...
22373 (dump_ipa_call_summary): ... this one.
22374 (estimate_function_body_sizes): Update.
22375 (inline_update_callee_summaries): Update.
22376 (remap_edge_change_prob): Update.
22377 (remap_edge_summaries): Update.
22378 (inline_merge_summary): Update.
22379 (do_estimate_edge_time): Update.
22380 (inline_generate_summary): Update.
22381 (inline_read_section): Update.
22382 (inline_read_summary): Update.
22383 (inline_free_summary): Update.
22384 * ipa-inline.c (can_inline_edge_p): Update.
22385 (compute_inlined_call_time): Update.
22386 (want_inline_small_function_p): Update.
22387 (edge_badness): Update.
22388 (early_inliner): Update.
22389 * ipa-inline.h (inline_edge_summary): Turn to ...
22390 (ipa_call_summary): ... this one.
22391 (ipa_call_summary_t): New class.
22392 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22393 (ipa_call_summaries): New.
22394 (inline_edge_summary): Remove.
22395 (estimate_edge_growth): Update.
22396 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22397 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22398 * ipa-split.c (execute_split_functions): Update.
22399 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22401 2017-05-19 Richard Biener <rguenther@suse.de>
22403 PR middle-end/80764
22404 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
22406 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
22408 * config/rs6000/rs6000.c (struct machine_function): Add field
22409 fpr_is_wrapped_separately.
22410 (rs6000_get_separate_components): Use 64 components. Handle the
22411 new FPR components.
22412 (rs6000_components_for_bb): Handle the FPR components.
22413 (rs6000_emit_prologue_components): Handle the FPR components.
22414 (rs6000_emit_epilogue_components): Handle the FPR components.
22415 (rs6000_set_handled_components): Handle the FPR components.
22416 (rs6000_emit_prologue): Don't output prologue code for those FPRs
22417 that are already separately shrink-wrapped.
22418 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
22419 that are already separately shrink-wrapped.
22421 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
22424 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
22427 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
22428 (define_peephole2 for Altivec d-form load): Add peepholes to catch
22429 cases where the register allocator uses a move and an offsettable
22430 memory operation to/from a FPR register on ISA 2.06/2.07.
22431 (define_peephole2 for Altivec d-form store): Likewise.
22433 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
22436 * config/i386/mmx.md (*mov<mode>_internal): Enable
22437 alternatives 11, 12, 13 and 14 also for 32bit targets.
22438 Remove alternatives 15, 16, 17 and 18.
22439 * config/i386/sse.md (vec_concatv2di): Change
22440 alternative (!x, *y) to (x, ?!*Yn).
22442 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
22444 * dumpfile.h (enum dump_kind): Remove stray comma.
22446 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
22448 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
22449 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
22450 predicate::num_conditions
22451 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
22452 (CHANGED): turn into predicate::changed.
22453 (agg_position_info): Move to ipa-predicate.h
22454 (add_condition, predicate::add_clause, predicate::operator &=,
22455 predicate::or_with, predicate::evaluate, predicate::probability,
22456 dump_condition, dump_clause, predicate::dump,
22457 predicate::remap_after_duplication, predicate::remap_after_inlining,
22458 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
22459 (evaluate_conditions_for_known_args): Update.
22460 (set_cond_stmt_execution_predicate): Update.
22461 * ipa-inline.h: Include ipa-predicate.h
22462 (condition, inline_param_summary, conditions, agg_position_info,
22463 predicate): Move to ipa-predicate.h
22464 * ipa-predicate.c: New file.
22465 * ipa-predicate.h: New file.
22467 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
22469 * final.c (leaf_function_p): Check we are not in a sequence.
22471 2017-05-18 Martin Liska <mliska@suse.cz>
22473 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
22474 * dumpfile.c (dump_register): Use new enum dump_kind.
22475 (get_dump_file_name): Likewise.
22476 (dump_enable_all): Likewise.
22477 (dump_switch_p_1): Likewise.
22478 (enable_rtl_dump_file): Remove usage of TDF_RTL.
22479 * dumpfile.h (enum dump_kind): New enum type.
22480 (struct dump_file_info): Create constructor and
22481 format fields and comments.
22482 * passes.c (pass_manager::register_one_dump_file):
22484 * statistics.c (statistics_early_init): Likewise.
22485 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
22486 TDF_TREE with TDF_SLIM.
22487 (gather_memory_references_ref): Likewise.
22489 2017-05-18 Martin Liska <mliska@suse.cz>
22491 * vec.h (struct vnull): Use it.
22493 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
22495 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
22496 (true_predicate, false_predicate, true_predicate_p,
22497 false_predicate_p): Remove.
22498 (single_cond_predicate, not_inlined_predicate): Turn to member function
22500 (add_condition): Update.
22501 (add_clause): Turn to...
22502 (predicate::add_clause): ... this one; update; allow passing NULL
22504 (and_predicates): Turn to ...
22505 (predicate::operator &=): ... this one.
22506 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
22507 (or_predicates): Turn to ...
22508 (predicate::or_with): ... this one.
22509 (evaluate_predicate): Turn to ...
22510 (predicate::evaluate): ... this one.
22511 (predicate_probability): Turn to ...
22512 (predicate::probability): ... this one.
22513 (dump_condition): Update.
22514 (dump_predicate): Turn to ...
22515 (predicate::dump): ... this one.
22516 (account_size_time): Update.
22517 (edge_set_predicate): Update.
22518 (set_hint_predicate): UPdate.
22519 (evaluate_conditions_for_known_args): Update.
22520 (evaluate_properties_for_edge): Update.
22521 (remap_predicate_after_duplication): Turn to...
22522 (predicate::remap_after_duplication): ... this one.
22523 (remap_hint_predicate_after_duplication): Update.
22524 (inline_summary_t::duplicate): UPdate.
22525 (dump_inline_edge_summary): Update.
22526 (dump_inline_summary): Update.
22527 (set_cond_stmt_execution_predicate): Update.
22528 (set_switch_stmt_execution_predicate): Update.
22529 (compute_bb_predicates): Update.
22530 (will_be_nonconstant_expr_predicate): Update.
22531 (will_be_nonconstant_predicate): Update.
22532 (phi_result_unknown_predicate): Update.
22533 (predicate_for_phi_result): Update.
22534 (array_index_predicate): Update.
22535 (estimate_function_body_sizes): Update.
22536 (estimate_node_size_and_time): Update.
22537 (estimate_ipcp_clone_size_and_time): Update.
22538 (remap_predicate): Rename to ...
22539 (predicate::remap_after_inlining): ... this one.
22540 (remap_hint_predicate): Update.
22541 (inline_merge_summary): Update.
22542 (inline_update_overall_summary): Update.
22543 (estimate_size_after_inlining): Update.
22544 (read_predicate): Rename to ...
22545 (predicate::stream_in): ... this one.
22546 (read_inline_edge_summary): Update.
22547 (write_predicate): Rename to ...
22548 (predicate::stream_out): ... this one.
22549 (write_inline_edge_summary): Update.
22550 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
22551 (clause_t): Turn to uint32_t
22552 (predicate): Turn to class; implement constructor and operators
22554 (size_time_entry): Update.
22555 (inline_summary): Update.
22556 (inline_edge_summary): Update.
22558 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
22560 * fold-const.c (fold_binary_loc): Move transformation...
22561 * match.pd (C - X CMP X): ... here.
22563 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
22565 * config/sparc/sparc.c (sparc_option_override): Set function
22566 alignment for -mcpu=niagara7 to 64 to match the I$ line.
22567 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
22569 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
22571 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
22573 2017-05-18 Marek Polacek <polacek@redhat.com>
22576 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
22577 (pass_ubsan::execute): Call gimple_assign_single_p instead of
22578 gimple_assign_load_p.
22580 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
22582 PR middle-end/80692
22583 * real.c (do_compare): Give decimal_do_compare preference over
22584 comparing just the signs.
22586 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
22588 * doc/md.texi (Canonicalization of Instructions): Describe the
22589 canonical form of instructions that inherently set a condition
22592 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
22594 PR middle-end/80775
22595 * tree-cfg.c: Move deletion of unreachable case statements to after
22596 the merging of consecutive case labels.
22598 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
22600 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
22601 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
22602 restoring of callee-saved registers.
22604 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
22606 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
22607 * config/visium/visium.c (single_set_and_flags): Likewise.
22608 * config/visium/visium.md (Substitutions): Likewise.
22610 2017-05-17 Martin Liska <mliska@suse.cz>
22612 * cfg.c: Introduce dump_flags_t type and
22613 use it instead of int type.
22615 * cfghooks.c: Likewise.
22616 * cfghooks.h (struct cfg_hooks): Likewise.
22617 * cfgrtl.c: Likewise.
22618 * cfgrtl.h: Likewise.
22619 * cgraph.c (cgraph_node::get_body): Likewise.
22620 * coretypes.h: Likewise.
22621 * domwalk.c: Likewise.
22622 * domwalk.h: Likewise.
22623 * dumpfile.c (struct dump_option_value_info): Likewise.
22624 (dump_enable_all): Likewise.
22625 (dump_switch_p_1): Likewise.
22626 (opt_info_switch_p): Likewise.
22627 * dumpfile.h (enum tree_dump_index): Likewise.
22628 (struct dump_file_info): Likewise.
22629 * genemit.c: Likewise.
22630 * generic-match-head.c: Likewise.
22631 * gengtype.c (open_base_files): Likewise.
22632 * gimple-pretty-print.c: Likewise.
22633 * gimple-pretty-print.h: Likewise.
22634 * graph.c (print_graph_cfg): Likewise.
22635 * graphite-scop-detection.c (dot_all_sese): Likewise.
22636 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
22637 * loop-unroll.c (report_unroll): Likewise.
22638 * passes.c (pass_manager::register_one_dump_file): Likewise.
22639 * print-tree.c: Likewise.
22640 * statistics.c: Likewise.
22641 * tree-cfg.c: Likewise.
22642 * tree-cfg.h: Likewise.
22643 * tree-dfa.c: Likewise.
22644 * tree-dfa.h: Likewise.
22645 * tree-dump.c (dump_function): Likewise.
22646 * tree-dump.h (struct dump_info): Likewise.
22647 * tree-pretty-print.c: Likewise.
22648 * tree-pretty-print.h: Likewise.
22649 * tree-ssa-live.c: Likewise.
22650 * tree-ssa-live.h: Likewise.
22651 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
22652 * tree-vect-loop.c: Likewise.
22653 * tree-vect-slp.c: Likewise.
22655 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
22656 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22658 PR tree-optimization/80457
22659 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
22660 of all arguments to a statement as scalar_to_vec operations.
22661 (vectorizable_call): Adjust call to vect_model_simple_cost for
22663 (vectorizable_conversion): Likewise.
22664 (vectorizable_assignment): Likewise.
22665 (vectorizable_shift): Likewise.
22666 (vectorizable_operation): Likewise.
22667 (vectorizable_comparison): Likewise.
22668 (vect_is_simple_cond): Record the def types for operands.
22669 (vectorizable_condition): Likewise, call vect_model_simple_cost.
22670 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
22671 for statement argument count.
22673 2017-05-16 Carl Love <cel@us.ibm.com>
22675 * config/rs6000/rs6000-c: Add support for built-in functions
22676 vector unsigned long long vec_bperm (vector unsigned long long,
22677 vector unsigned char)
22678 vector signed long long vec_mule (vector signed int,
22680 vector unsigned long long vec_mule (vector unsigned int,
22681 vector unsigned int)
22682 vector signed long long vec_mulo (vector signed int,
22684 vector unsigned long long vec_mulo (vector unsigned int,
22685 vector unsigned int)
22686 vector signed char vec_sldw (vector signed char,
22687 vector signed char,
22689 vector unsigned char vec_sldw (vector unsigned char,
22690 vector unsigned char,
22692 vector signed short vec_sldw (vector signed short,
22693 vector signed short,
22695 vector unsigned short vec_sldw (vector unsigned short,
22696 vector unsigned short,
22698 vector signed int vec_sldw (vector signed int,
22701 vector unsigned int vec_sldw (vector unsigned int,
22702 vector unsigned int,
22704 vector signed long long vec_sldw (vector signed long long,
22705 vector signed long long,
22707 vector unsigned long long vec_sldw (vector unsigned long long,
22708 vector unsigned long long,
22710 * config/rs6000/rs6000-c: Add support for built-in functions
22711 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
22712 * config/rs6000/altivec.h: Add defintion for vec_sldw.
22713 * doc/extend.texi: Update the built-in documentation for the
22714 new built-in functions.
22716 2017-05-16 Marek Polacek <polacek@redhat.com>
22720 * tree.c (save_expr): Don't fold the expression.
22722 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
22724 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
22725 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
22726 and (?*y,m). Update insn attributes.
22728 2017-05-16 Martin Liska <mliska@suse.cz>
22730 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
22731 flags argument of print_gimple_stmt, print_gimple_expr,
22732 print_generic_stmt and print_generic_expr.
22733 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
22734 * coretypes.h: Likewise.
22735 * except.c (dump_eh_tree): Likewise.
22736 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
22737 * gimple-pretty-print.h: Likewise.
22738 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
22739 (backprop::push_to_worklist): Likewise.
22740 (backprop::pop_from_worklist): Likewise.
22741 (backprop::process_use): Likewise.
22742 (backprop::intersect_uses): Likewise.
22743 (note_replacement): Likewise.
22744 * gimple-ssa-store-merging.c
22745 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
22746 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
22747 (pass_store_merging::execute): Likewise.
22748 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
22749 (ssa_base_cand_dump_callback): Likewise.
22750 (dump_incr_vec): Likewise.
22751 (replace_refs): Likewise.
22752 (replace_mult_candidate): Likewise.
22753 (create_add_on_incoming_edge): Likewise.
22754 (create_phi_basis): Likewise.
22755 (insert_initializers): Likewise.
22756 (all_phi_incrs_profitable): Likewise.
22757 (introduce_cast_before_cand): Likewise.
22758 (replace_one_candidate): Likewise.
22759 * gimplify.c (gimplify_expr): Likewise.
22760 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
22761 (set_rename): Likewise.
22762 (rename_uses): Likewise.
22763 (copy_loop_phi_nodes): Likewise.
22764 (add_close_phis_to_merge_points): Likewise.
22765 (copy_loop_close_phi_args): Likewise.
22766 (copy_cond_phi_args): Likewise.
22767 (graphite_copy_stmts_from_block): Likewise.
22768 (translate_pending_phi_nodes): Likewise.
22769 * graphite-poly.c (print_pdr): Likewise.
22770 (dump_gbb_cases): Likewise.
22771 (dump_gbb_conditions): Likewise.
22772 (print_scop_params): Likewise.
22773 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
22774 (build_cross_bb_scalars_use): Likewise.
22775 (gather_bbs::before_dom_children): Likewise.
22776 * hsa-dump.c (dump_hsa_immed): Likewise.
22777 * ipa-cp.c (print_ipcp_constant_value): Likewise.
22778 (get_replacement_map): Likewise.
22779 * ipa-inline-analysis.c (dump_condition): Likewise.
22780 (estimate_function_body_sizes): Likewise.
22781 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
22782 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
22783 * ipa-prop.c (ipa_dump_param): Likewise.
22784 (ipa_print_node_jump_functions_for_edge): Likewise.
22785 (ipa_modify_call_arguments): Likewise.
22786 (ipa_modify_expr): Likewise.
22787 (ipa_dump_param_adjustments): Likewise.
22788 (ipa_dump_agg_replacement_values): Likewise.
22789 (ipcp_modif_dom_walker::before_dom_children): Likewise.
22790 * ipa-pure-const.c (check_stmt): Likewise.
22791 (pass_nothrow::execute): Likewise.
22792 * ipa-split.c (execute_split_functions): Likewise.
22793 * omp-offload.c (dump_oacc_loop_part): Likewise.
22794 (dump_oacc_loop): Likewise.
22795 * trans-mem.c (tm_log_emit): Likewise.
22796 (tm_memopt_accumulate_memops): Likewise.
22797 (dump_tm_memopt_set): Likewise.
22798 (dump_tm_memopt_transform): Likewise.
22799 * tree-cfg.c (gimple_verify_flow_info): Likewise.
22800 (print_loop): Likewise.
22801 * tree-chkp-opt.c (chkp_print_addr): Likewise.
22802 (chkp_gather_checks_info): Likewise.
22803 (chkp_get_check_result): Likewise.
22804 (chkp_remove_check_if_pass): Likewise.
22805 (chkp_use_outer_bounds_if_possible): Likewise.
22806 (chkp_reduce_bounds_lifetime): Likewise.
22807 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
22808 (chkp_mark_completed_bounds): Likewise.
22809 (chkp_register_incomplete_bounds): Likewise.
22810 (chkp_mark_invalid_bounds): Likewise.
22811 (chkp_maybe_copy_and_register_bounds): Likewise.
22812 (chkp_build_returned_bound): Likewise.
22813 (chkp_get_bound_for_parm): Likewise.
22814 (chkp_build_bndldx): Likewise.
22815 (chkp_get_bounds_by_definition): Likewise.
22816 (chkp_generate_extern_var_bounds): Likewise.
22817 (chkp_get_bounds_for_decl_addr): Likewise.
22818 * tree-chrec.c (chrec_apply): Likewise.
22819 * tree-data-ref.c (dump_data_reference): Likewise.
22820 (dump_subscript): Likewise.
22821 (dump_data_dependence_relation): Likewise.
22822 (analyze_overlapping_iterations): Likewise.
22823 * tree-inline.c (expand_call_inline): Likewise.
22824 (tree_function_versioning): Likewise.
22825 * tree-into-ssa.c (dump_defs_stack): Likewise.
22826 (dump_currdefs): Likewise.
22827 (dump_names_replaced_by): Likewise.
22828 (dump_update_ssa): Likewise.
22829 (update_ssa): Likewise.
22830 * tree-object-size.c (pass_object_sizes::execute): Likewise.
22831 * tree-parloops.c (build_new_reduction): Likewise.
22832 (try_create_reduction_list): Likewise.
22833 (ref_conflicts_with_region): Likewise.
22834 (oacc_entry_exit_ok_1): Likewise.
22835 (oacc_entry_exit_single_gang): Likewise.
22836 * tree-pretty-print.h: Likewise.
22837 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
22838 (get_scalar_evolution): Likewise.
22839 (add_to_evolution): Likewise.
22840 (get_loop_exit_condition): Likewise.
22841 (analyze_evolution_in_loop): Likewise.
22842 (analyze_initial_condition): Likewise.
22843 (analyze_scalar_evolution): Likewise.
22844 (instantiate_scev): Likewise.
22845 (number_of_latch_executions): Likewise.
22846 (gather_chrec_stats): Likewise.
22847 (final_value_replacement_loop): Likewise.
22848 (scev_const_prop): Likewise.
22849 * tree-sra.c (dump_access): Likewise.
22850 (disqualify_candidate): Likewise.
22851 (create_access): Likewise.
22852 (reject): Likewise.
22853 (maybe_add_sra_candidate): Likewise.
22854 (create_access_replacement): Likewise.
22855 (analyze_access_subtree): Likewise.
22856 (analyze_all_variable_accesses): Likewise.
22857 (sra_modify_assign): Likewise.
22858 (initialize_constant_pool_replacements): Likewise.
22859 (find_param_candidates): Likewise.
22860 (decide_one_param_reduction): Likewise.
22861 (replace_removed_params_ssa_names): Likewise.
22862 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
22863 * tree-ssa-copy.c (dump_copy_of): Likewise.
22864 (copy_prop_visit_cond_stmt): Likewise.
22865 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
22866 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
22867 (record_equivalences_from_stmt): Likewise.
22868 * tree-ssa-dse.c (compute_trims): Likewise.
22869 (delete_dead_call): Likewise.
22870 (delete_dead_assignment): Likewise.
22871 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
22872 (forward_propagate_into_cond): Likewise.
22873 (pass_forwprop::execute): Likewise.
22874 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
22875 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
22877 (move_computations_worker): Likewise.
22878 (execute_sm): Likewise.
22879 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
22880 (remove_exits_and_undefined_stmts): Likewise.
22881 (remove_redundant_iv_tests): Likewise.
22882 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
22883 (adjust_iv_update_pos): Likewise.
22884 * tree-ssa-math-opts.c (bswap_replace): Likewise.
22885 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
22886 (value_replacement): Likewise.
22887 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
22888 * tree-ssa-pre.c (print_pre_expr): Likewise.
22889 (get_representative_for): Likewise.
22890 (create_expression_by_pieces): Likewise.
22891 (insert_into_preds_of_block): Likewise.
22892 (eliminate_insert): Likewise.
22893 (eliminate_dom_walker::before_dom_children): Likewise.
22894 (eliminate): Likewise.
22895 (remove_dead_inserted_code): Likewise.
22896 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
22897 * tree-ssa-reassoc.c (get_rank): Likewise.
22898 (eliminate_duplicate_pair): Likewise.
22899 (eliminate_plus_minus_pair): Likewise.
22900 (eliminate_not_pairs): Likewise.
22901 (undistribute_ops_list): Likewise.
22902 (eliminate_redundant_comparison): Likewise.
22903 (update_range_test): Likewise.
22904 (optimize_range_tests_var_bound): Likewise.
22905 (optimize_vec_cond_expr): Likewise.
22906 (rewrite_expr_tree): Likewise.
22907 (rewrite_expr_tree_parallel): Likewise.
22908 (linearize_expr): Likewise.
22909 (break_up_subtract): Likewise.
22910 (linearize_expr_tree): Likewise.
22911 (attempt_builtin_powi): Likewise.
22912 (attempt_builtin_copysign): Likewise.
22913 (transform_stmt_to_copy): Likewise.
22914 (transform_stmt_to_multiply): Likewise.
22915 (dump_ops_vector): Likewise.
22916 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
22917 (print_scc): Likewise.
22918 (set_ssa_val_to): Likewise.
22919 (visit_reference_op_store): Likewise.
22920 (visit_use): Likewise.
22921 (sccvn_dom_walker::before_dom_children): Likewise.
22922 (run_scc_vn): Likewise.
22923 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
22925 (expr_hash_elt::print): Likewise.
22926 (const_and_copies::pop_to_marker): Likewise.
22927 (const_and_copies::record_const_or_copy_raw): Likewise.
22928 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
22929 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
22930 (dump_predicates): Likewise.
22931 (find_uninit_use): Likewise.
22932 (warn_uninitialized_phi): Likewise.
22933 (pass_late_warn_uninitialized::execute): Likewise.
22934 * tree-ssa.c (verify_vssa): Likewise.
22935 (verify_ssa): Likewise.
22936 (maybe_optimize_var): Likewise.
22937 * tree-vrp.c (dump_value_range): Likewise.
22938 (dump_all_value_ranges): Likewise.
22939 (dump_asserts_for): Likewise.
22940 (register_edge_assert_for_2): Likewise.
22941 (vrp_visit_cond_stmt): Likewise.
22942 (vrp_visit_switch_stmt): Likewise.
22943 (vrp_visit_stmt): Likewise.
22944 (vrp_visit_phi_node): Likewise.
22945 (simplify_cond_using_ranges_1): Likewise.
22946 (fold_predicate_in): Likewise.
22947 (evrp_dom_walker::before_dom_children): Likewise.
22948 (evrp_dom_walker::push_value_range): Likewise.
22949 (evrp_dom_walker::pop_value_range): Likewise.
22950 (execute_early_vrp): Likewise.
22952 2017-05-16 Richard Biener <rguenther@suse.de>
22954 * dwarf2out.c (loc_list_from_tree_1): Do not create
22955 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
22957 2017-05-16 Richard Biener <rguenther@suse.de>
22959 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
22961 (note_variable_value_in_expr): If we resolved the decl ref
22962 do not push to the stack.
22964 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
22966 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
22967 operations in fast-math mode.
22968 (vaddq_f16): Likewise.
22969 (vmul_f16): Likewise.
22970 (vmulq_f16): Likewise.
22971 (vsub_f16): Likewise.
22972 (vsubq_f16): Likewise.
22973 * config/arm/neon.md (add<mode>3): New.
22975 (fma:<VH:mode>3): New. Also remove outdated comment.
22978 2017-05-16 Martin Liska <mliska@suse.cz>
22982 * ipa-devirt.c (warn_types_mismatch): Fix typo.
22983 (odr_types_equivalent_p): Likewise.
22985 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
22987 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
22989 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
22992 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
22993 non-interunit SSE move alternatives with '?'.
22994 (zero-extendsidi peephole2): New peephole to skip intermediate
22995 general register in SSE zero-extend sequence.
22997 2017-05-15 Jeff Law <law@redhat.com>
22999 * reorg.c (relax_delay_slots): Create a new variable to hold
23000 the temporary target rather than clobbering TARGET_LABEL.
23002 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
23003 missing argument to extract_bit_field call.
23004 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
23006 2017-05-15 Martin Liska <mliska@suse.cz>
23009 * gcc.c (process_command): Do not allow empty argument of -o option.
23011 2017-05-15 Renlin Li <renlin.li@arm.com>
23013 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
23014 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
23015 * config/aarch64/constraints.md (Usf): Add long call check.
23016 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
23017 (call_value): Likewise.
23018 (sibcall): Likewise.
23019 (sibcall_value): Likewise.
23021 (call_value_insn): New.
23022 (sibcall_insn): Update rtx pattern.
23023 (sibcall_value_insn): Likewise.
23024 (call_internal): Remove.
23025 (call_value_internal): Likewise.
23026 (sibcall_internal): Likewise.
23027 (sibcall_value_internal): Likewise.
23028 (call_reg): Likewise.
23029 (call_symbol): Likewise.
23030 (call_value_reg): Likewise.
23031 (call_value_symbol): Likewise.
23033 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
23036 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
23038 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
23040 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
23041 compatible with CCGOCmode and with CCZmode.
23043 2017-05-14 Martin Sebor <msebor@redhat.com>
23045 PR middle-end/77671
23046 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
23047 (gimple_fold_builtin_snprintf): Same.
23048 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
23049 (gimple_fold_builtin_snprintf): Same.
23050 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
23051 of character types.
23052 (is_call_safe): New function.
23053 (try_substitute_return_value): Call it.
23054 (try_simplify_call): New function.
23055 (pass_sprintf_length::handle_gimple_call): Call it.
23057 2017-05-14 Martin Sebor <msebor@redhat.com>
23059 PR middle-end/80669
23060 * builtins.c (expand_builtin_stpncpy): Simplify.
23062 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
23064 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
23065 * config/i386/i386.h
23066 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
23067 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
23068 (struct machine_function): Add new members call_ms2sysv,
23069 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
23070 (struct machine_frame_state): New fields sp_realigned and
23071 sp_realigned_offset.
23072 * config/i386/i386.c
23073 (enum xlogue_stub): New enum.
23074 (enum xlogue_stub_sets): New enum.
23075 (class xlogue_layout): New class.
23076 (struct ix86_frame): New fields stack_realign_allocate_offset,
23077 stack_realign_offset and outlined_save_offset. Modify comments to
23078 detail stack layout when using out-of-line stubs.
23079 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
23080 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
23081 -mcall-ms2sysv-xlogues.
23082 (stub_managed_regs): New static variable.
23083 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
23084 registers managed by out-of-line stub.
23085 (disable_call_ms2sysv_xlogues): New function.
23086 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
23087 m->call_ms2sysv when appropriate and compute frame layout for
23089 (sp_valid_at, fp_valid_at): New inline functions.
23090 (choose_basereg): New function.
23091 (choose_baseaddr): Add align parameter, use choose_basereg and modify
23093 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
23094 Use align parameter of choose_baseaddr to generated aligned SSE movs
23096 (pro_epilogue_adjust_stack): Modify to track
23097 machine_frame_state::sp_realigned.
23098 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
23099 (ix86_nsaved_sseregs): Likewise.
23100 (ix86_emit_save_regs): Likewise.
23101 (ix86_emit_save_regs_using_mov): Likewise.
23102 (ix86_emit_save_sse_regs_using_mov): Likewise.
23103 (get_scratch_register_on_entry): Likewise.
23104 (gen_frame_set): New function.
23105 (gen_frame_load): Likewise.
23106 (gen_frame_store): Likewise.
23107 (emit_outlined_ms2sysv_save): Likewise.
23108 (emit_outlined_ms2sysv_restore): Likewise.
23109 (ix86_expand_prologue): Modify stack re-alignment code and call
23110 emit_outlined_ms2sysv_save when appropriate.
23111 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
23112 parameter rtx_insn *insn, which allows the function to be used to only
23113 generate the notes.
23114 (ix86_expand_epilogue): Modify validity checks of frame and stack
23115 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
23116 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
23117 * config/i386/predicates.md
23118 (save_multiple): New predicate.
23119 (restore_multiple): Likewise.
23120 * config/i386/sse.md
23121 (save_multiple<mode>): New pattern.
23122 (save_multiple_realign<mode>): Likewise.
23123 (restore_multiple<mode>): Likewise.
23124 (restore_multiple_and_return<mode>): Likewise.
23125 (restore_multiple_leave_return<mode>): Likewise.
23126 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
23128 2017-05-14 Julia Koval <julia.koval@intel.com>
23130 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
23131 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
23132 (__builtin_ia32_xsetbv): New builtins.
23133 * config/i386/i386.c (ix86_expand_special_args_builtin):
23135 (ix86_expand_builtin): Special expand for new intrinsics.
23136 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
23137 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
23138 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
23140 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23142 * cfganal.c (inverted_post_order_compute): Change argument type
23144 * cfganal.h (inverted_post_order_compute): Adjust prototype.
23145 * df-core.c (rest_of_handle_df_initialize): Adjust.
23146 (rest_of_handle_df_finish): Likewise.
23147 (df_analyze_1): Likewise.
23148 (df_analyze): Likewise.
23149 (loop_inverted_post_order_compute): Change argument to be a vec *.
23150 (df_analyze_loop): Adjust.
23151 (df_get_n_blocks): Likewise.
23152 (df_get_postorder): Likewise.
23153 * df.h (struct df_d): Change field to be a vec.
23154 * lcm.c (compute_laterin): Adjust.
23155 (compute_available): Likewise.
23156 * lra-lives.c (lra_create_live_ranges_1): Likewise.
23157 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
23158 * tree-ssa-pre.c (compute_antic): Likewise.
23160 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23162 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
23163 (depth_first_search::depth_first_search): Change structure init
23164 function to this constructor.
23165 (depth_first_search::add_bb): Rename function to this member.
23166 (depth_first_search::execute): Likewise.
23167 (flow_dfs_compute_reverse_finish): Adjust.
23169 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23171 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
23172 (longest_simple_path): Likewise.
23173 * shrink-wrap.c (spread_components): Likewise.
23174 (disqualify_problematic_components): Likewise.
23175 (emit_common_heads_for_components): Likewise.
23176 (emit_common_tails_for_components): Likewise.
23177 (insert_prologue_epilogue_for_components): Likewise.
23179 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23181 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
23184 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23186 * df-core.c (df_set_blocks): Start using auto_bitmap.
23187 (df_compact_blocks): Likewise.
23188 * df-problems.c (df_rd_confluence_n): Likewise.
23189 * df-scan.c (df_insn_rescan_all): Likewise.
23190 (df_process_deferred_rescans): Likewise.
23191 (df_update_entry_block_defs): Likewise.
23192 (df_update_exit_block_uses): Likewise.
23193 (df_entry_block_bitmap_verify): Likewise.
23194 (df_exit_block_bitmap_verify): Likewise.
23195 (df_scan_verify): Likewise.
23196 * lra-constraints.c (lra_constraints): Likewise.
23197 (undo_optional_reloads): Likewise.
23198 (lra_undo_inheritance): Likewise.
23199 * lra-remat.c (calculate_gen_cands): Likewise.
23200 (do_remat): Likewise.
23201 * lra-spills.c (assign_spill_hard_regs): Likewise.
23202 (spill_pseudos): Likewise.
23203 * tree-ssa-pre.c (bitmap_set_and): Likewise.
23204 (bitmap_set_subtract_values): Likewise.
23206 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23208 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
23209 management with auto_bitmap.
23210 (fix_inter_tick): Likewise.
23211 (fix_recovery_deps): Likewise.
23212 * ira.c (add_store_equivs): Likewise.
23213 (find_moveable_pseudos): Likewise.
23214 (split_live_ranges_for_shrink_wrap): Likewise.
23215 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
23216 (rtx_reuse_manager::seen_def_p): Likewise.
23217 (rtx_reuse_manager::set_seen_def): Likewise.
23218 * print-rtl.h (class rtx_reuse_manager): Likewise.
23220 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23222 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
23224 (migrate_btr_def): Likewise.
23225 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
23226 * df-core.c (loop_post_order_compute): Likewise.
23227 (loop_inverted_post_order_compute): Likewise.
23228 * hsa-common.h: Likewise.
23229 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
23230 * init-regs.c (initialize_uninitialized_regs): Likewise.
23231 * ipa-inline.c (resolve_noninline_speculation): Likewise.
23232 (inline_small_functions): Likewise.
23233 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
23234 * ira.c (combine_and_move_insns): Likewise.
23235 (build_insn_chain): Likewise.
23236 * loop-invariant.c (find_invariants): Likewise.
23237 * lower-subreg.c (propagate_pseudo_copies): Likewise.
23238 * predict.c (tree_predict_by_opcode): Likewise.
23239 (predict_paths_leading_to): Likewise.
23240 (predict_paths_leading_to_edge): Likewise.
23241 (estimate_loops_at_level): Likewise.
23242 (estimate_loops): Likewise.
23243 * shrink-wrap.c (try_shrink_wrapping): Likewise.
23244 (spread_components): Likewise.
23245 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
23246 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
23247 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
23248 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
23249 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
23250 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
23251 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
23252 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
23253 (mark_threaded_blocks): Likewise.
23254 (thread_through_all_blocks): Likewise.
23255 * tree-ssa.c (verify_ssa): Likewise.
23256 (execute_update_addresses_taken): Likewise.
23257 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
23259 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23261 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
23263 (post_order_compute): Likewise.
23264 (inverted_post_order_compute): Likewise.
23265 (pre_and_rev_post_order_compute_fn): Likewise.
23267 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23269 * genrecog.c (int_set::int_set): Explicitly construct our
23270 auto_vec base class.
23271 * vec.h (auto_vec::auto_vec): New constructor.
23273 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23275 * bitmap.h (class auto_bitmap): New constructor taking
23276 bitmap_obstack * argument.
23278 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23280 * bitmap.h (class auto_bitmap): Change type of m_bits to
23281 bitmap_head, and adjust ctor / dtor and member operators.
23283 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
23285 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
23286 when returned register mode doesn't match original mode.
23288 2017-05-12 Jeff Law <law@redhat.com>
23289 Jakub Jelinek <jakub@redhat.com>
23291 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
23292 we look for cc setter after the compare-elim changes.
23293 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
23294 within the vector to match what compare-elim now expects.
23295 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
23296 (xorsi3_flags, one_cmplsi2_flags): Likewise.
23298 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
23299 after the compare-elim changes.
23300 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
23301 the vector to match what compare-elim now expects.
23302 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
23303 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
23304 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
23305 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
23306 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
23308 * config/visium/visium.c (single_set_and_flags): Fix where
23309 we look for cc setter after the compare-elim changes.
23310 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
23311 with the vector to match what compare-elim now expects.
23312 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
23313 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
23314 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
23315 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
23316 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
23317 (neg<mode>2_insn_set_overflow): Likewise.
23319 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
23321 PR middle-end/79794
23322 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
23323 maybe_expand_insn call, set ops[0].target. If still set after call,
23324 set alt_rtl. Add extra arg to recursive calls.
23325 (extract_bit_field): Add alt_rtl argument. Pass to
23327 * expmed.h (extract_bit_field): Fix prototype.
23328 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
23329 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
23330 to extract_bit_field_calls.
23331 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
23332 Pass alt_rtl to extract_bit_field calls.
23333 * calls.c (store_unaligned_arguments_into_psuedos)
23334 load_register_parameters): Pass extra NULL to extract_bit_field calls.
23335 * optabs.c (maybe_legitimize_operand): Clear op->target when call
23337 * optabs.h (struct expand_operand): Add target bitfield.
23339 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
23341 * compare-elim.c (try_eliminate_compare): Canonicalize
23342 operation with embedded compare to
23343 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
23344 (set (reg) (operation)].
23346 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
23348 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
23351 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
23352 cost of adding a carry flag for ADC instruction.
23353 [case MINUS]: Ignore the cost of subtracting a carry flag
23354 for SBB instruction.
23356 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
23358 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
23360 * config/rs6000/bmiintrin.h: New file.
23361 * config/rs6000/bmi2intrin.h: New file.
23362 * config/rs6000/x86intrin.h: New file.
23364 2017-05-12 Jeff Law <law@redhat.com>
23366 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
23369 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
23371 PR middle-end/80707
23372 * tree-cfg.c: Remove cfg edges of unreachable case statements.
23374 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23376 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
23377 early expansion of vector divide builtins.
23378 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
23379 builtins identified as having unsigned arguments.
23381 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23383 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
23384 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
23385 expansion of vector logical operations (and, andc, or, xor,
23388 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23390 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
23391 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
23393 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23395 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
23396 early GIMPLE expansion of vector multiplies.
23398 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23400 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
23401 TARGET_HAVE_MOVT conditional.
23402 (movt splitter): Likewise.
23404 2017-05-12 Richard Biener <rguenther@suse.de>
23406 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
23407 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
23408 Fold all stmts not inplace.
23410 2017-05-12 Richard Biener <rguenther@suse.de>
23412 PR tree-optimization/80713
23413 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
23414 inserted_exprs bit for not removed stmts.
23416 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
23418 PR middle-end/69921
23419 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
23420 parallelized" attribute for parallelized OpenACC kernels.
23421 * omp-offload.c (execute_oacc_device_lower): Use it.
23423 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
23424 Set "oacc kernels" attribute.
23425 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
23426 parameter. Adjust all users.
23427 (oacc_fn_attrib_kernels_p): Remove function.
23428 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
23429 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
23430 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
23431 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
23432 assert "oacc kernels" attribute is set.
23434 2017-05-11 Carl Love <cel@us.ibm.com>
23436 * config/rs6000/rs6000-c: Add support for built-in functions
23437 vector unsigned char vec_popcnt (vector signed char)
23438 vector unsigned char vec_popcnt (vector unsigned char)
23439 vector unsigned short vec_popcnt (vector signed short)
23440 vector unsigned short vec_popcnt (vector unsigned short)
23441 vector unsigned int vec_popcnt (vector signed int)
23442 vector unsigned int vec_popcnt (vector unsigned int)
23443 vector unsigned long long vec_popcnt (vector signed long long)
23444 vector unsigned long long vec_popcnt (vector unsigned long long)
23445 vector signed long long vec_slo (vector signed long long,
23446 vector signed char)
23447 vector signed long long vec_slo (vector signed long long,
23448 vector unsigned char)
23449 vector unsigned long long vec_slo (vector unsigned long long,
23450 vector signed char)
23451 vector unsigned long long vec_slo (vector unsigned long long,
23452 vector unsigned char)
23453 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
23454 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
23455 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
23456 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
23457 * doc/extend.texi: Update the built-in documentation file for the
23458 new built-in functions.
23460 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
23462 * attribs.h (sorted_attr_string): Move machine independent
23463 functions for target clone support from the i386 port to common
23464 code. Rename ix86_function_versions to common_function_versions.
23465 Rename make_name to make_unique_name.
23466 (common_function_versions): Likewise.
23467 (make_unique_name): Likewise.
23468 (make_dispatcher_decl): Likewise.
23469 (is_function_default_version): Likewise.
23470 * attribs.c (attr_strcmp): Likewise.
23471 (sorted_attr_string): Likewise.
23472 (common_function_versions): Likewise.
23473 (make_unique_name): Likewise.
23474 (make_dispatcher_decl): Likewise.
23475 (is_function_default_version): Likewise.
23476 * config/i386/i386.c (attr_strcmp): Likewise.
23477 (sorted_attr_string): Likewise.
23478 (ix86_function_versions): Likewise.
23479 (make_name): Likewise.
23480 (make_dispatcher_decl): Likewise.
23481 (is_function_default_version): Likewise.
23482 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
23484 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23487 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23488 Account for direct move costs for vec_construct of integer
23491 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
23494 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
23495 (UNSPEC_STX_ATOMIC): Ditto.
23496 (loaddi_via_sse): New insn.
23497 (storedi_via_sse): Ditto.
23498 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
23499 Update corresponding peephole2 patterns.
23500 (atomic_storedi_fpu): Ditto.
23502 2017-05-11 Julia Koval <julia.koval@intel.com>
23504 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
23505 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
23507 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
23508 (__builtin_ia32_rsqrt14ss_mask): New builtins.
23509 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
23511 2017-05-11 Nathan Sidwell <nathan@acm.org>
23513 * graphite-poly.c: Include dumpfile.h.
23515 * dumpfle.h (dump_function): Declare here ...
23516 * tree-dump.h (dump_function): ... not here.
23517 * dumpfile.c: #include tree-cfg.h.
23518 (dump_function): Move here from ...
23519 * tree-dump.c (dump_function): ... here.
23520 * gimplify.c: #include splay-tree.h, not tree-dump.h.
23521 * graphite-poly.c: Don't include tree-dump.h.
23522 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
23523 * print-tree.c: Likewise.
23524 * stor-layout.c: Likewise.
23525 * tree-nested.c: Likewise.
23527 * dumpfile.c (dump_start): Use TDF_FLAGS.
23528 (dump_enable_all): Fix TDF_KIND check thinko.
23530 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
23532 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23533 array entries to represent two legal parameterizations of the
23534 overloaded __builtin_cmpb function, as represented by the
23535 P6_OV_BUILTIN_CMPB constant.
23536 (altivec_resolve_overloaded_builtin): Add special case handling
23537 for the __builtin_cmpb function, as represented by the
23538 P6_OV_BUILTIN_CMPB constant.
23539 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
23540 (BU_P6_64BIT_2): New macro.
23541 (BU_P6_OVERLOAD_2): New macro
23542 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
23543 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
23544 (CMPB): Add overload support to represent both 32-bit and 64-bit
23545 compare-bytes function.
23546 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
23547 support for TARGET_CMPB.
23548 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
23549 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23550 documentation of the __builtin_cmpb overloaded built-in function.
23552 2017-05-11 Richard Biener <rguenther@suse.de>
23554 PR tree-optimization/80705
23555 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
23556 bases are not vectorizable.
23558 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23560 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
23561 when counting register pressure.
23563 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23565 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
23566 (iv_ca_more_deps): Renamed to ...
23567 (iv_ca_compare_deps): ... this.
23568 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
23570 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23572 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
23574 (determine_group_iv_costs): ... here.
23575 (find_inv_vars_cb): Record inv var if it's not recorded before.
23577 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23579 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
23580 (get_shiftadd_cost): Ditto.
23582 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23584 * tree-ssa-address.c: Include header file.
23585 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
23587 (add_to_parts): Refactor.
23588 (addr_to_parts): New parameter. Update use of move_hint_to_base.
23589 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
23592 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23594 PR tree-optimization/53090
23595 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
23597 (extract_cond_operands): Detect condition with IV on both sides
23598 and return COMP_IV_EXPR_2.
23599 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
23600 (rewrite_use_compare): Simplify by removing call to function
23601 extract_cond_operands.
23603 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23605 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
23606 (extract_cond_operands): Detect condition comparing against non-
23607 invariant bound and return appropriate enum value.
23608 (find_interesting_uses_cond): Update use of extract_cond_operands.
23609 Handle its return value accordingly.
23610 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
23612 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23614 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
23615 nonlinear iv_use computation in loop invariant sensitive way.
23617 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23619 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
23620 (find_iv_candidates): Call relate_compare_use_with_all_cands.
23622 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23624 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
23625 (dump_cand): Support iv_cand.inv_exprs.
23626 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
23628 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
23631 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23633 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
23635 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
23636 as local function. Include necessary header files.
23637 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
23639 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23641 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
23643 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23645 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
23646 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
23647 RSHIFT_EXPR and BIT_NOT_EXPR.
23649 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23651 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
23652 (adjust_setup_cost): New parameter supporting round up adjustment.
23653 (struct address_cost_data): Delete.
23654 (force_expr_to_var_cost): Don't bound cost with spill_cost.
23655 (split_address_cost, ptr_difference_cost): Delete.
23656 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
23657 (struct ainc_cost_data): New struct.
23658 (get_address_cost_ainc): New function.
23659 (get_address_cost, get_computation_cost): Reimplement.
23660 (determine_group_iv_cost_address): Record inv_expr for all uses of
23662 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
23663 (iv_ca_has_deps): Reimplemented to ...
23664 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
23666 (iv_ca_extend): Call iv_ca_more_deps.
23668 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23670 * tree-ssa-address.c (struct mem_address): Move to header file.
23671 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
23672 * tree-ssa-address.h (struct mem_address): Move from C file.
23673 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
23675 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23677 * tree-affine.h (aff_combination_type): New interface.
23678 (aff_combination_zero_p): Remove static.
23679 (aff_combination_const_p): New interface.
23680 (aff_combination_singleton_var_p): New interfaces.
23682 2017-05-11 Richard Biener <rguenther@suse.de>
23684 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
23685 Skip unreachable blocks and destinations.
23686 (eliminate): Move stmt removal and fixup ...
23687 (fini_eliminate): ... here. Skip inserted exprs.
23688 (pass_pre::execute): Move fini_pre after fini_eliminate.
23689 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
23690 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
23691 PRE to get rid of dead code that has invalid SSA form and
23692 split critical edges again.
23694 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23696 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
23698 2017-05-11 Richard Biener <rguenther@suse.de>
23700 * passes.c (execute_function_todo): Verify loops if they are
23701 said to be up-to-date.
23702 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
23703 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
23705 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
23708 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
23709 handle calling assemble_external ourself.
23712 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
23713 modes with zero size. Enhance comment.
23715 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23717 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
23718 built-ins for vec_xl and vec_xst with short and char pointer
23721 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
23723 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
23724 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
23725 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
23726 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
23727 (_mm_maskz_min_round_ss): New intrinsics.
23728 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
23729 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
23730 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
23731 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
23732 (__builtin_ia32_minss_mask_round): New builtins.
23733 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23734 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23735 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
23737 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
23738 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
23740 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
23743 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
23745 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
23746 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
23747 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
23748 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
23749 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
23750 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
23751 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
23752 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23753 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
23754 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
23755 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
23756 (__builtin_ia32_mulss_mask_round): New builtins.
23757 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23758 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23759 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
23761 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
23762 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
23764 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
23767 2017-05-10 Julia Koval <julia.koval@intel.com>
23769 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
23770 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
23771 (_mm256_setr_m128i): New intrinsics.
23773 2017-05-10 Julia Koval <julia.koval@intel.com>
23775 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
23776 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
23777 (_mm_maskz_rcp14_ss): New intrinsics.
23778 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
23779 (__builtin_ia32_rcp14ss_mask): New builtins.
23780 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
23782 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
23784 PR tree-optimization/51513
23785 * tree-cfg.c (gimple_seq_unreachable_p): New function.
23786 (assert_unreachable_fallthru_edge_p): Use it.
23787 (group_case_labels_stmt): Likewise.
23788 * tree-cfg.h: Prototype it.
23789 * stmt.c: Include cfghooks.h and tree-cfg.h.
23790 (emit_case_dispatch_table) <gap_label>: New local variable.
23791 Use it to fill dispatch table gaps.
23792 Test for default_label before updating probabilities.
23793 (expand_case) <default_label>: Remove unneeded initialization.
23794 Test for unreachable default case statement and remove its edge.
23795 Set default_label accordingly.
23796 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
23798 2017-05-10 Carl Love <cel@us.ibm.com>
23800 * config/rs6000/rs6000-c: Add support for built-in functions
23801 vector signed char vec_neg (vector signed char)
23802 vector signed short int vec_neg (vector short int)
23803 vector signed int vec_neg (vector signed int)
23804 vector signed long long vec_neg (vector signed long long)
23805 vector float vec_neg (vector float)
23806 vector double vec_neg (vector double)
23807 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
23809 * config/rs6000/altivec.h: Add define for vec_neg
23810 * doc/extend.texi: Update the built-in documentation for the
23811 new built-in functions.
23813 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23815 PR tree-optimization/77644
23816 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
23818 2017-05-10 Nathan Sidwell <nathan@acm.org>
23820 * dumpfile.h (TDI_lang_all): New.
23821 (TDF_KIND): New. Renumber others
23822 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
23824 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
23826 (get_dump_file_name): Adjust suffix generation.
23827 (dump_enable_all): Use TDF_KIND.
23828 * doc/invoke.texi (-fdump-lang-all): Document.
23830 * dumpfile.h: Tabify.
23832 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
23835 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
23836 Move member access before delete.
23838 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
23840 * tree-inline.c (expand_call_inline): Split block at stmt
23843 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
23846 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
23847 are now unused after splitting mov{sf,sd}_hardfloat.
23848 (f32_lr2): Likewise.
23849 (f32_lm): Likewise.
23850 (f32_lm2): Likewise.
23851 (f32_li): Likewise.
23852 (f32_li2): Likewise.
23853 (f32_lv): Likewise.
23854 (f32_sr): Likewise.
23855 (f32_sr2): Likewise.
23856 (f32_sm): Likewise.
23857 (f32_sm2): Likewise.
23858 (f32_si): Likewise.
23859 (f32_si2): Likewise.
23860 (f32_sv): Likewise.
23861 (f32_dm): Likewise.
23862 (f32_vsx): Likewise.
23863 (f32_av): Likewise.
23864 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
23865 For movsf, order stores so the VSX stores occur before the GPR
23866 store which encourages the register allocator to use a traditional
23867 FPR instead of a GPR. For movsd, order the stores so that the GPR
23868 store comes before the VSX stores to allow the power6 to work.
23869 This is due to the power6 not having a 32-bit integer store
23870 instruction from a FPR.
23871 (movsf_hardfloat): Likewise.
23872 (movsd_hardfloat): Likewise.
23874 2017-05-09 Martin Sebor <msebor@redhat.com>
23876 PR translation/80280
23877 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
23880 PR translation/80280
23881 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
23882 data member added in r247778.
23883 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
23885 2017-05-09 Nathan Sidwell <nathan@acm.org>
23887 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
23889 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
23892 2017-05-09 Marek Polacek <polacek@redhat.com>
23894 * doc/invoke.texi: Fix typo.
23896 2017-05-09 Richard Biener <rguenther@suse.de>
23898 * tree-vrp.c (vrp_val_is_max): Adjust comment.
23899 (vrp_val_is_min): Likewise.
23900 (set_value_range_to_value): Likewise.
23901 (set_value_range_to_nonnegative): Likewise.
23902 (gimple_assign_nonzero_p): Likewise.
23903 (gimple_stmt_nonzero_p): Likewise.
23904 (vrp_int_const_binop): Likewise. Remove unreachable case.
23905 (adjust_range_with_scev): Adjust comments.
23906 (compare_range_with_value): Likewise.
23907 (extract_range_from_phi_node): Likewise.
23908 (test_for_singularity): Likewise.
23910 2017-05-09 Richard Biener <rguenther@suse.de>
23912 * tree-vrp.c (get_single_symbol): Add assert that we don't
23913 get overflowed constants as invariant part.
23914 (compare_values_warnv): Add comment before the TREE_NO_WARNING
23915 checks. Use wi::cmp instead of recursing for integer constants.
23916 (compare_values): Just ignore whether we assumed undefined
23917 overflow instead of failing the compare.
23918 (extract_range_for_var_from_comparison_expr): Add comment before the
23919 TREE_NO_WARNING sets.
23920 (test_for_singularity): Likewise.
23921 (extract_range_from_comparison): Do not disable optimization
23922 when we assumed undefined overflow.
23923 (extract_range_basic): Remove init of unused var.
23925 2017-05-09 Richard Biener <rguenther@suse.de>
23927 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
23928 (extract_range_from_multiplicative_op_1): Adjust.
23929 (extract_range_from_binary_expr_1): Use int_const_binop.
23931 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
23934 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
23935 rs6000_store_data_bypass_p in seven define_bypass directives and
23936 in several comments.
23937 * config/rs6000/rs6000-protos.h: Add prototype for
23938 rs6000_store_data_bypass_p function.
23939 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
23940 function implements slightly different (rs6000-specific) semantics
23941 than store_data_bypass_p, returning false rather than aborting
23942 with assertion error when arguments do not satisfy the
23943 requirements of store data bypass.
23944 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
23945 rs6000_store_data_bypass_p.
23947 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
23949 * config/xtensa/xtensa-protos.h
23950 (xtensa_initial_elimination_offset): New declaration.
23951 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
23952 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
23953 macro definition, add case for FRAME_POINTER_REGNUM when
23954 FRAME_GROWS_DOWNWARD.
23955 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
23956 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
23957 xtensa_initial_elimination_offset.
23959 2017-05-08 Nathan Sidwell <nathan@acm.org>
23961 * doc/invoke.texi: Alphabetize -fdump options.
23963 2017-05-08 Martin Sebor <msebor@redhat.com>
23965 PR translation/80280
23966 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
23968 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
23970 * target.def (compute_frame_layout): New optional target hook.
23971 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
23972 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
23973 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
23975 * reload1.c (verify_initial_elim_offsets): Likewise.
23976 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
23977 (use_simple_return_p): Call arm_compute_frame_layout if needed.
23978 (arm_get_frame_offsets): Split up into this ...
23979 (arm_compute_frame_layout): ... and this function.
23981 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
23983 * config/aarch64/constraints.md (Usa): New constraint.
23984 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
23986 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
23988 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
23989 with_multilib_list after it has been checked.
23991 2017-05-08 Richard Biener <rguenther@suse.de>
23993 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
23994 (bitmap_set_subtract_values): Likewise.
23996 2017-05-08 Richard Biener <rguenther@suse.de>
23998 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
23999 (gimple_assign_nonzero): ... this and remove strict_overflow_p
24001 (gimple_stmt_nonzero_warnv_p): Rename to ...
24002 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
24004 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
24005 (extract_range_basic): Adjust, do not disable propagation on
24006 strict overflow sensitive simplification.
24007 (vrp_visit_cond_stmt): Likewise.
24009 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
24011 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
24012 body size unconditionally.
24014 2017-05-07 Jeff Law <law@redhat.com>
24017 2017-05-06 Jeff Law <law@redhat.com>
24018 PR tree-optimization/78496
24019 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24022 PR tree-optimization/78496
24023 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24024 (simplify_stmt_using_ranges): Call it.
24025 (vrp_dom_walker::before_dom_children): Extract equivalences
24026 from an ASSERT_EXPR with an equality comparison against a
24029 2017-05-06 Jeff Law <law@redhat.com>
24031 PR tree-optimization/78496
24032 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24035 PR tree-optimization/78496
24036 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24037 (simplify_stmt_using_ranges): Call it.
24038 (vrp_dom_walker::before_dom_children): Extract equivalences
24039 from an ASSERT_EXPR with an equality comparison against a
24042 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24044 * lra-constraints.c (lra_copy_reg_equiv): New function.
24045 (split_reg): Use it to copy equivalence information from the
24046 original register to the spill register.
24048 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24050 PR rtl-optimization/75964
24051 * simplify-rtx.c (simplify_const_relational_operation): Remove
24052 invalid handling of comparisons of integer ABS.
24054 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
24056 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
24057 initialize to zero.
24058 (init_regs): Remove declaration.
24059 (function_arg_advance_32): Initialize error_p as boolean variable.
24061 2017-05-05 Nathan Sidwell <nathan@acm.org>
24063 * store-motion.c (remove_reachable_equiv_notes): Reformat long
24064 lines. Use for (;;).
24066 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24068 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
24069 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
24070 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
24071 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
24072 VF=2 that require versioning.
24074 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24076 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
24079 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24081 * diagnostic.h (diagnostic_override_option_index): Convert from
24082 macro to inline function.
24084 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24086 * diagnostic.c (last_module_changed_p): New function.
24087 (set_last_module): New function.
24088 (diagnostic_report_current_module): Convert macro usage to
24089 the above functions.
24090 * diagnostic.h (diagnostic_context::last_module): Strengthen
24091 from const line_map * to const line_map_ordinary *.
24092 (diagnostic_last_module_changed): Delete macro.
24093 (diagnostic_set_last_module): Delete macro.
24095 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24097 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
24098 with diagnostic_report_diagnostic.
24099 (diagnostic_n_impl_richloc): Likewise.
24100 * diagnostic.h (report_diagnostic): Delete macro.
24101 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
24102 with diagnostic_report_diagnostic.
24103 * substring-locations.c (format_warning_va): Likewise.
24105 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24107 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
24108 save/restor of format_spec. Move option-printing code to...
24109 (print_option_information): ...this new function, and
24110 reimplement by simply printing to the pretty_printer,
24111 rather than appending to the format string.
24113 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24115 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
24116 handling logic into...
24117 (update_effective_level_from_pragmas): ...this new function.
24119 2017-05-04 Andrew Waterman <andrew@sifive.com>
24121 * config/riscv/riscv.opt (mstrict-align): New option.
24122 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
24123 (SLOW_UNALIGNED_ACCESS): Define.
24124 (riscv_slow_unaligned_access): Declare.
24125 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
24127 (riscv_slow_unaligned_access): New variable.
24128 (rocket_tune_info): Set slow_unaligned_access to true.
24129 (optimize_size_tune_info): Set slow_unaligned_access to false.
24130 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
24131 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
24132 (riscv_option_override): Set riscv_slow_unaligned_access.
24133 * doc/invoke.texi: Add -mstrict-align to RISC-V.
24135 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
24137 * config/riscv/riscv.md: Unify indentation.
24139 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
24144 * config/rs6000/rs6000.md (u code attribute): Add FIX and
24146 (extendsi<mode>2): Add support for doing sign extension via
24147 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
24148 don't have ISA 3.0 instructions.
24149 (extendsi<mode>2 splitter): Likewise.
24150 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
24151 generate the normal insns since SImode can now go in vector
24152 registers. Disallow the special UNSPECs needed for previous
24153 machines to hide SImode being used. Add new insns
24154 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
24155 (fix_trunc<mode>si2_stfiwx): Likewise.
24156 (fix_trunc<mode>si2_internal): Likewise.
24157 (fixuns_trunc<mode>si2): Likewise.
24158 (fixuns_trunc<mode>si2_stfiwx): Likewise.
24159 (fctiw<u>z_<mode>_smallint): Likewise.
24160 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
24161 of floating point to 32-bit integer from doing a direct move to
24162 the GPR registers to do a store.
24163 (fctiwz_<mode>): Break long line.
24165 2017-05-05 Bin Cheng <bin.cheng@arm.com>
24167 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
24168 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
24169 (addr_list, addr_offset_valid_p): New.
24170 (split_address_groups): Check offset validity with above function.
24171 (gt-tree-ssa-loop-ivopts.h): Include header file.
24173 2017-05-05 Nathan Sidwell <nathan@acm.org>
24175 * config.gcc (arm*-*-*): Add missing 'fi'.
24177 2017-05-05 Steve Ellcey <sellcey@cavium.com>
24179 * doc/invoke.texi (-fopt-info): Explicitly say order of options
24180 included in -fopt-info does not matter.
24181 * doc/optinfo.texi (-fopt-info): Fix description of default
24182 behavour. Explicitly say order of options included in -fopt-info
24185 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
24187 * config.gcc: Allow combinations of aprofile and rmprofile values for
24188 --with-multilib-list.
24189 * config/arm/t-multilib: New file.
24190 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
24191 variables. Remove setting of ISA and floating-point ABI in
24192 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
24193 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
24194 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
24195 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
24197 * config/arm/t-rmprofile: Likewise except for the matches changes.
24198 * doc/install.texi (--with-multilib-list): Document the combination of
24199 aprofile and rmprofile values and warn about pitfalls in doing that.
24201 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24203 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
24204 (movdi_aarch64): Likewise.
24206 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24208 PR tree-optimization/80632
24209 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
24211 (build_arrays): Initialize it for virtual phis.
24212 (fix_phi_nodes): Use it for virtual phis.
24214 PR tree-optimization/80558
24215 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
24216 [x, y] op z into [x op, y op z] for op & or | if conditions
24219 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
24220 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24223 * config/arm/arm.md (use_literal_pool): Remove.
24224 (64-bit immediate split): No longer takes cost into consideration
24225 if arm_disable_literal_pool is enabled.
24226 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
24227 used when arm_disable_literal_pool is enabled.
24228 (arm_max_const_double_inline_cost): Remove use of
24229 arm_disable_literal_pool.
24230 (push_minipool_fix): Add assert.
24231 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
24232 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
24233 (no_literal_pool_sf_immediate): New.
24235 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24237 PR tree-optimization/80613
24238 * tree-ssa-dce.c (propagate_necessity): Remove cases for
24239 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
24241 2017-05-05 Richard Biener <rguenther@suse.de>
24243 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
24245 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
24247 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
24248 of this flag from insn conditions due to removal from r247495.
24250 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24252 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
24254 (arm_early_store_addr_dep_ptr): Likewise.
24255 * config/arm/aarch-common-protos.h
24256 (arm_early_load_addr_dep_ptr): Add prototype.
24257 (arm_early_store_addr_dep_ptr): Likewise.
24258 * config/arm/cortex-a53.md: Add new bypasses.
24260 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24262 * tree.c (next_type_uid): Change type to unsigned.
24263 (type_hash_canon): Decrement back next_type_uid if
24264 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
24265 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
24268 2017-05-04 Martin Sebor <msebor@redhat.com>
24270 * builtins.c: Fix a trivial typo in a comment.
24272 PR middle-end/79234
24273 * builtins.c (check_sizes): Adjust to handle reading past the end.
24274 Avoid printing excessive upper bound of ranges. Use %E to print
24275 tree nodes instead of converting them to %wu.
24276 (expand_builtin_memchr): New function.
24277 (compute_dest_size): Rename...
24278 (compute_objsize): ...to this.
24279 (expand_builtin_memcpy): Adjust.
24280 (expand_builtin_mempcpy): Adjust.
24281 (expand_builtin_strcat): Adjust.
24282 (expand_builtin_strcpy): Adjust.
24283 (check_strncat_sizes): Adjust.
24284 (expand_builtin_strncat): Adjust.
24285 (expand_builtin_strncpy): Adjust and simplify.
24286 (expand_builtin_memset): Adjust.
24287 (expand_builtin_bzero): Adjust.
24288 (expand_builtin_memcmp): Adjust.
24289 (expand_builtin): Handle memcmp.
24290 (maybe_emit_chk_warning): Check strncat just once.
24292 2017-05-04 Martin Sebor <msebor@redhat.com>
24294 PR preprocessor/79214
24295 PR middle-end/79222
24296 PR middle-end/79223
24297 * builtins.c (check_sizes): Add inlining context and issue
24298 warnings even when -Wno-system-headers is set.
24299 (check_strncat_sizes): Same.
24300 (expand_builtin_strncat): Same.
24301 (expand_builtin_memmove): New function.
24302 (expand_builtin_stpncpy): Same.
24303 (expand_builtin): Handle memmove and stpncpy.
24305 2017-05-04 Bin Cheng <bin.cheng@arm.com>
24307 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
24308 which is not used any more.
24310 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24312 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
24314 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24316 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
24317 (cortexa53_tunings): Likewise.
24318 (cortexa57_tunings): Likewise.
24319 (cortexa72_tunings): Likewise.
24320 (cortexa73_tunings): Likewise.
24322 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24324 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
24325 Set loop alignment to 8.
24327 2017-05-04 Martin Sebor <msebor@redhat.com>
24329 PR translation/80280
24330 * builtins.c (expand_builtin_object_size): Add missing quoting to
24331 %D and like directives.
24332 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
24333 (hsa_type_for_tree_type): Same.
24334 (verify_function_arguments): Same.
24335 * symtab.c (symbol_table::change_decl_assembler_name): Same.
24336 * varasm.c (get_section): Same.
24339 2017-05-04 Martin Sebor <msebor@redhat.com>
24341 PR translation/80280
24342 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
24344 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24346 * config/aarch64/aarch64.c (generic_addrcost_table):
24347 Change HI/TI mode setting.
24349 2017-05-04 Martin Jambor <mjambor@suse.cz>
24351 PR tree-optimization/80622
24352 * tree-sra.c (comes_initialized_p): New function.
24353 (build_accesses_from_assign): Only set write lazily when
24354 comes_initialized_p is false.
24355 (analyze_access_subtree): Use comes_initialized_p.
24356 (propagate_subaccesses_across_link): Assert !comes_initialized_p
24357 instead of testing for PARM_DECL.
24359 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24361 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
24362 constraint on operand 0 to allow more general addressing modes.
24363 Adjust output template.
24364 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
24366 * config/aarch64/aarch64-protos.h
24367 (aarch64_address_valid_for_prefetch_p): Declare prototype.
24368 * config/aarch64/constraints.md (Dp): New address constraint.
24369 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
24372 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
24374 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
24375 update use of estimate_ipcp_clone_size_and_time.
24376 (estimate_local_effects): Update use of
24377 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
24378 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
24379 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
24380 Return nonspecialized time.
24382 2017-05-04 Richard Biener <rguenther@suse.de>
24384 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
24385 for the last VUSE which def dominates the PHI. Directly call
24387 (get_continuation_for_phi_1): Remove.
24389 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
24391 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
24392 to explain the use of truncating division. Cap the number of
24393 iterations to the maximum given by nb_iterations_upper_bound,
24396 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24398 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
24399 * configure: Regenerate.
24400 * config.in: Regenerate.
24401 * config/i386/driver-mingw32.c: new file.
24402 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
24403 * config.host: Link driver-mingw32.o on MinGW host.
24404 * doc/install.texi: Document new --enable-mingw-wildcard configure
24407 2017-05-04 Marek Polacek <polacek@redhat.com>
24409 PR tree-optimization/80612
24410 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
24412 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24413 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
24415 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
24416 (movt splitter): Likewise.
24417 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
24418 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
24419 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
24420 block for Thumb-1 with MOVT.
24421 (thumb2_legitimate_address_p): Move code block ...
24422 (can_avoid_literal_pool_for_label_p): ... into this new function.
24423 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
24425 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
24426 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
24427 "M-profile targets with the MOVT instruction".
24429 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24431 * config/arm/arm-builtins.c (arm_init_builtins): Rename
24432 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
24433 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
24435 2017-05-04 Martin Liska <mliska@suse.cz>
24437 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
24438 variable cond_code.
24440 2017-05-04 Richard Biener <rguenther@suse.de>
24442 * tree.c (array_at_struct_end_p): Handle arrays at struct
24443 end with flexarrays more conservatively. Refactor and treat
24444 arrays of arrays or aggregates more strict. Fix
24445 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
24446 * tree.c (array_at_struct_end_p): Adjust prototype.
24447 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
24448 * gimple-fold.c (get_range_strlen): Likewise.
24449 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
24451 2017-05-04 Richard Biener <rguenther@suse.de>
24453 PR tree-optimization/31130
24454 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
24456 (supports_overflow_infinity): Likewise.
24457 (is_negative_overflow_infinity): Likewise.
24458 (is_positive_overflow_infinity): Likewise.
24459 (is_overflow_infinity): Likewise.
24460 (stmt_overflow_infinity): Likewise.
24461 (overflow_infinity_range_p): Likewise.
24462 (usable_range_p): Remove as always returning true.
24463 (make_overflow_infinity): Remove.
24464 (negative_overflow_infinity): Likewise.
24465 (positive_overflow_infinity): Likewise.
24466 (avoid_overflow_infinity): Likewise.
24467 (set_value_range): Adjust accordingly.
24468 (set_value_range_to_nonnegative): Likewise, remove now unused
24469 overflow_infinity arg.
24470 (vrp_operand_equal_p): Adjust.
24471 (update_value_range): Likewise.
24472 (range_int_cst_singleton_p): Likewise.
24473 (operand_less_p): Likewise.
24474 (compare_values_warnv): Likewise.
24475 (extract_range_for_var_from_comparison_expr): Likewise.
24476 (vrp_int_const_binop): Likewise.
24477 (zero_nonzero_bits_from_vr): Likewise.
24478 (extract_range_from_multiplicative_op_1): Likewise.
24479 (extract_range_from_binary_expr_1): Likewise.
24480 (extract_range_from_unary_expr): Likewise.
24481 (extract_range_from_comparison): Likewise.
24482 (extract_range_basic): Likewise.
24483 (adjust_range_with_scev): Likewise.
24484 (compare_ranges): Likewise.
24485 (compare_range_with_value): Likewise.
24486 (dump_value_range): Likewise.
24487 (test_for_singularity): Likewise, remove strict_overflow_p parameter
24489 (simplify_cond_using_ranges): Adjust.
24491 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
24493 * brig-builtins.def: Added a builtin for class_f64.
24494 * builtin-types.def: Added a builtin type needed by class_f64.
24496 2017-05-03 Jason Merrill <jason@redhat.com>
24498 * timevar.def: Add TV_CONSTEXPR.
24500 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24502 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
24504 2017-05-03 Martin Jambor <mjambor@suse.cz>
24506 * ipa-prop.c (ipa_update_after_lto_read): Removed.
24507 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
24508 * ipa-cp.c (ipcp_propagate_stage): Do not call
24509 ipa_update_after_lto_read.
24510 * ipa-inline.c (ipa_inline): Likewise.
24512 2017-05-03 Martin Jambor <mjambor@suse.cz>
24514 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
24515 tag. Added a default constructor and a destructor.
24516 (ipa_edge_args_sum_t): New class;
24517 (ipa_edge_args_sum): Declare.
24518 (ipa_edge_args_vector): Remove declaration.
24519 (IPA_EDGE_REF): Use ipa_edge_args_sum.
24520 (ipa_free_edge_args_substructures): Remove declaration.
24521 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
24522 (ipa_edge_args_info_available_for_edge_p): Likewise.
24523 * ipa-prop.c (ipa_edge_args_vector): Removed.
24524 (edge_removal_hook_holder): Likewise.
24525 (edge_duplication_hook_holder): Likewise.
24526 (ipa_edge_args_sum): New variable.
24527 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
24528 ipa_edge_args_vector.
24529 (ipa_free_edge_args_substructures): Likewise.
24530 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
24531 ipa_edge_args_vector.
24532 (ipa_edge_removal_hook): Turned into method
24533 ipa_edge_args_sum_t::remove.
24534 (ipa_edge_duplication_hook): Turned into method
24535 ipa_edge_args_sum_t::duplicate.
24536 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
24537 registering edge hooks.
24538 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
24539 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
24540 ipa_edge_args_sum instead of ipa_edge_args_vector.
24541 * ipa-profile.c (ipa_profile): Likewise.
24543 2017-05-03 Martin Jambor <mjambor@suse.cz>
24545 * symbol-summary.h (function_summary): New method exists.
24546 (function_summary::symtab_removal): Deallocate through release.
24547 (call_summary): New class.
24548 (gt_ggc_mx): New overload.
24549 (gt_pch_nx): Likewise.
24550 (gt_pch_nx): Likewise.
24552 2017-05-03 Jeff Law <law@redhat.com>
24554 PR tree-optimization/78496
24555 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
24556 from simplify_cond_using_ranges. Split off code to walk
24557 backwards through casts into ...
24558 (simplify_cond_using_ranges_2): New function.
24559 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
24560 (execute_vrp): After identifying jump threads, call
24561 simplify_cond_using_ranges_2.
24563 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
24566 * ipa-inline.h (inline_summary): Add ctor.
24567 (create_ggc): Do not use ggc_cleared_alloc.
24569 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
24570 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24572 * gcc.c (handle_braces): Support escaping in switch matching
24574 * doc/invoke.texi (Spec Files): Document it.
24575 Remove superfluous @code markup in items.
24577 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24579 * diagnostic-show-locus.c (struct column_range): New struct.
24580 (get_affected_columns): New function.
24581 (get_printed_columns): New function.
24582 (struct correction): New struct.
24583 (correction::ensure_capacity): New function.
24584 (correction::ensure_terminated): New function.
24585 (struct line_corrections): New struct.
24586 (line_corrections::~line_corrections): New dtor.
24587 (line_corrections::add_hint): New function.
24588 (layout::print_trailing_fixits): Reimplement in terms of the new
24590 (selftest::test_overlapped_fixit_printing): New function.
24591 (selftest::diagnostic_show_locus_c_tests): Call it.
24593 2017-05-03 Nathan Sidwell <nathan@acm.org>
24595 Canonicalize canonical type hashing
24596 * tree.h (type_hash_canon_hash): Declare.
24597 * tree.c (type_hash_list, attribute_hash_list): Move into
24598 type_hash_canon_hash.
24599 (build_type_attribute_qual_variant): Break out hash code calc into
24600 type_hash_canon_hash.
24601 (type_hash_canon_hash): New. Generic type hash computation.
24602 (build_range_type_1, build_array_type_1, build_function_type,
24603 build_method_type_directly, build_offset_type, build_complex_type,
24604 make_vector_type): Call it.
24606 2017-05-03 Richard Biener <rguenther@suse.de>
24608 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
24609 When all DRs have unknown misaligned do not always peel
24610 when there is a store but apply the same costing model as if
24611 there were only loads.
24613 2017-05-03 Richard Biener <rguenther@suse.de>
24616 PR tree-optimization/80492
24617 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
24618 compare_base_decls returning dont-know properly.
24620 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
24622 * config/arm/iterators.md (CCSI): New mode iterator.
24623 (arch): New mode attribute.
24624 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
24625 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
24626 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
24627 code iterator for success result mode.
24628 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
24629 the corresponding new insn generators.
24631 2017-05-03 Bin Cheng <bin.cheng@arm.com>
24634 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24635 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24637 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
24639 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
24640 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
24641 (DDR_A): Wrap DDR argument in brackets.
24642 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
24643 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
24644 (DDR_REVERSED_P): Likewise.
24646 2017-05-03 Jakub Jelinek <jakub@redhat.com>
24648 PR tree-optimization/79472
24649 * tree-switch-conversion.c (struct switch_conv_info): Add
24650 contiguous_range and default_case_nonstandard fields.
24651 (collect_switch_conv_info): Compute contiguous_range and
24652 default_case_nonstandard fields, don't clear final_bb if
24653 contiguous_range and only the default case doesn't have the required
24655 (check_all_empty_except_final): Set default_case_nonstandard instead
24656 of failing if contiguous_range and the default case doesn't have empty
24658 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
24659 and only the default case doesn't have the required constants. Skip
24661 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
24662 if default_case_nonstandard.
24663 (build_constructors): Build constant 1 just once. Assert that default
24664 values aren't inserted in between cases if contiguous_range. Skip
24666 (build_arrays): Skip virtual phis.
24667 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
24668 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
24669 Handle virtual phis.
24670 (gen_inbound_check): Handle default_case_nonstandard case.
24671 (process_switch): Adjust check_final_bb caller. Call
24672 gather_default_values with the first non-default case instead of
24673 default case if default_case_nonstandard.
24675 2017-05-02 Nathan Sidwell <nathan@acm.org>
24677 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
24678 check. Fix formatting.
24680 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
24682 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
24683 errors when comparing specialized and unspecialized times.
24685 2017-05-02 David Malcolm <dmalcolm@redhat.com>
24687 * diagnostic-show-locus.c
24688 (layout::should_print_annotation_line_p): Make private.
24689 (layout::print_annotation_line): Make private.
24690 (layout::annotation_line_showed_range_p): Make private.
24691 (layout::show_ruler): Make private.
24692 (layout::print_source_line): Make private. Pass in line and
24693 line_width, rather than calling location_get_source_line. Drop
24695 (layout::print_leading_fixits): New method.
24696 (layout::print_any_fixits): Rename to...
24697 (layout::print_trailing_fixits): ...this, and make private.
24698 Don't print newline fixits.
24699 (diagnostic_show_locus): Move logic for printing one row into...
24700 (layout::print_line): ...this new function. Move the
24701 location_get_source_line call and error-handling from
24702 print_source_line to here. Call print_leading_fixits, and rename
24703 print_any_fixits to print_trailing_fixits.
24704 (selftest::test_fixit_insert_containing_newline): Update now that
24705 newlines are partially supported.
24706 (selftest::test_fixit_insert_containing_newline_2): New test.
24707 (selftest::test_fixit_replace_containing_newline): Update comments.
24708 (selftest::diagnostic_show_locus_c_tests): Call the new test.
24709 * edit-context.c (class added_line): New class.
24710 (class edited_line): Describe newline handling in comment.
24711 (edited_line::actually_edited_p): New method.
24712 (edited_line::print_content): Delete redundant decl.
24713 (edited_line::m_predecessors): New field.
24714 (edited_file::print_content): Call edited_line::print_content.
24715 (edited_file::print_diff): Update to support newlines.
24716 (edited_file::print_diff_hunk): Likewise.
24717 (edited_file::print_run_of_changed_lines): New function.
24718 (edited_file::print_diff_line): Convert to...
24719 (print_diff_line): ...this.
24720 (edited_file::get_effective_line_count): New function.
24721 (edited_line::edited_line): Initialize new field m_predecessors.
24722 (edited_line::~edited_line): Clean up m_predecessors.
24723 (edited_line::apply_fixit): Handle newlines.
24724 (edited_line::get_effective_line_count): New function.
24725 (edited_line::print_content): New function.
24726 (edited_line::print_diff_lines): New function.
24727 (selftest::test_applying_fixits_insert_containing_newline): New
24729 (selftest::test_applying_fixits_replace_containing_newline): New
24731 (selftest::insert_line): New function.
24732 (selftest::test_applying_fixits_multiple_lines): Add example of
24734 (selftest::edit_context_c_tests): Call the new tests.
24736 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24738 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
24739 parameter cand. Update dump information.
24740 (get_computation_cost): Update uses.
24742 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24744 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
24745 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
24746 (get_computation_at, rewrite_use_address): Update use of
24747 get_computation_aff.
24749 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24751 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
24752 (get_computation): Delete.
24753 (get_computation_cost): Implement like get_computation_cost_at.
24754 Use get_computation_at.
24755 (get_computation_cost_at): Delete.
24756 (rewrite_use_nonlinear_expr): Use get_computation_at.
24757 (rewrite_use_compare, remove_unused_ivs): Ditto.
24759 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24761 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
24763 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24765 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
24766 (ivopts_global_cost_for_size): Rename parameter and update uses.
24767 (iv_ca_recount_cost): Update uses.
24768 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
24769 candidates seperately in n_invs and n_cands.
24770 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
24772 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24774 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
24775 (find_inv_vars_cb): New.
24776 (find_depends): Renamed to ...
24777 (find_inv_vars): ... this.
24778 (add_candidate_1, force_var_cost): Call find_inv_vars.
24779 (split_address_cost, determine_group_iv_cost_cond): Ditto.
24781 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24783 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
24784 inv_vars. Add inv_exprs.
24785 (struct iv_cand): Rename depends_on to inv_vars.
24786 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
24787 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
24788 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
24789 (dump_cand): Dump inv_vars.
24790 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
24791 (record_invariant, find_depends, add_candidate_1): Ditto.
24792 (set_group_iv_cost, force_var_cost): Ditto.
24793 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
24794 (get_computation_cost_at, get_computation_cost): Ditto.
24795 (determine_group_iv_cost_generic): Ditto.
24796 (determine_group_iv_cost_address): Ditto.
24797 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
24798 (determine_group_iv_costs): Ditto.
24799 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
24800 (iv_ca_set_remove_invariants): Renamed to ...
24801 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
24802 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
24803 (iv_ca_set_add_invariants): Renamed to ...
24804 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
24805 (iv_ca_set_cp): Use iv_ca_set_add_invs.
24806 (iv_ca_has_deps): Support inv_vars and inv_exprs.
24807 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
24808 (create_new_ivs): Remove useless dump.
24810 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24812 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
24814 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
24815 (iv_ca_set_no_cp, create_new_iv): Ditto.
24817 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24819 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24821 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
24823 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
24824 function tree_check2.
24826 2017-05-02 Martin Liska <mliska@suse.cz>
24828 * doc/gcov.texi: Add missing preposition.
24829 * gcov.c (function_info::function_info): Properly fill up
24830 all member variables.
24832 2017-05-02 Tamar Christina <tamar.christina@arm.com>
24834 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
24836 2017-05-02 Tamar Christina <tamar.christina@arm.com>
24838 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
24840 2017-05-02 Martin Liska <mliska@suse.cz>
24843 * lto-streamer-in.c (lto_read_tree_1): Remove
24844 LTO_STREAMER_DEBUG.
24845 * lto-streamer.c (struct tree_hash_entry): Likewise.
24846 (struct tree_entry_hasher): Likewise.
24847 (tree_entry_hasher::hash): Likewise.
24848 (tree_entry_hasher::equal): Likewise.
24849 (lto_streamer_init): Likewise.
24850 (lto_orig_address_map): Likewise.
24851 (lto_orig_address_get): Likewise.
24852 (lto_orig_address_remove): Likewise.
24853 * lto-streamer.h: Likewise.
24854 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
24855 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
24857 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
24859 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
24860 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
24861 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
24862 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
24863 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
24864 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
24865 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
24866 (mm_maskz_sub_ss): New intrinsics.
24867 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24868 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
24869 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
24870 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
24871 (__builtin_ia32_subss_mask_round): New builtins.
24872 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24873 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24874 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
24876 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
24877 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
24879 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
24882 2017-05-02 Martin Jambor <mjambor@suse.cz>
24884 PR tree-optimization/78687
24885 * tree-sra.c (access): New field parent.
24886 (process_subtree_disqualification): New function.
24887 (disqualify_candidate): Call it.
24888 (build_accesses_from_assign): Reset write flag if creating an
24890 (build_access_subtree): Fill in parent field and also prpagate
24891 down grp_write flag.
24892 (create_artificial_child_access): New parameter set_grp_write, set
24893 grp_write to its value.
24894 (propagate_subaccesses_across_link): Also propagate grp_write flag
24896 (propagate_all_subaccesses): Push the closest parent back to work
24897 queue if add_access_to_work_queue returned true.
24899 2017-05-02 Richard Biener <rguenther@suse.de>
24901 * common.opt (fstrict-overflow): Alias negative to fwrapv.
24902 * doc/invoke.texi (fstrict-overflow): Remove all traces of
24903 -fstrict-overflow documentation.
24904 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
24905 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
24906 flag_strict_overflow.
24907 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
24908 * lto-opts.c (lto_write_options): Do not stream it.
24909 * lto-wrapper.c (merge_and_complain): Do not handle it.
24910 * opts.c (default_options_table): Do not set -fstrict-overflow.
24911 (finish_options): Likewise do not clear it when sanitizing.
24912 * simplify-rtx.c (simplify_const_relational_operation): Do not
24913 test flag_strict_overflow.
24915 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
24917 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
24918 using enabled attribute.
24919 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
24920 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
24921 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
24922 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
24923 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
24924 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
24925 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
24926 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
24927 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
24928 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
24930 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
24932 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
24934 2017-05-02 Richard Biener <rguenther@suse.de>
24936 PR tree-optimization/80591
24938 2017-04-10 Richard Biener <rguenther@suse.de>
24940 * tree-ssa-structalias.c (find_func_aliases): Properly handle
24943 2017-05-02 Richard Biener <rguenther@suse.de>
24945 PR tree-optimization/80549
24946 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
24947 (cleanup_tree_cfg_noloop): Create forwarders to known loop
24948 headers if they do not have a preheader.
24950 2017-05-02 Martin Liska <mliska@suse.cz>
24953 * common.opt: Fix typo.
24954 * doc/invoke.texi: Likewise.
24956 2017-05-01 Jan Beulich <jbeulich@suse.com>
24958 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
24959 swapping, add (x,x,m,x,n) alternative.
24961 2017-05-01 Nathan Sidwell <nathan@acm.org>
24963 * calls.c (combine_pending_stack_adjustment_and_call): Remove
24964 unnecessary unadjusted_alignment check.
24966 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
24969 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
24971 * gimplify.c (gimplify_cilk_detach): New function.
24972 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
24973 * tree-core.h: Document EXPR_CILK_SPAWN.
24974 * tree.h (EXPR_CILK_SPAWN): Define.
24976 2017-05-01 David Malcolm <dmalcolm@redhat.com>
24978 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
24979 to use new fixit_hint representation, using the "replace" logic.
24980 (get_line_span_for_fixit_hint): Likewise.
24981 (layout::print_any_fixits): Likewise.
24982 (selftest::test_one_liner_many_fixits): Rename to...
24983 (selftest::test_one_liner_many_fixits_1): ...this, and update
24984 comment and expected output to reflect that the multiple fix-it
24985 hints are now consolidated into one insertion.
24986 (selftest::test_one_liner_many_fixits_2): New test.
24987 (selftest::test_diagnostic_show_locus_one_liner): Update for
24989 (selftest::test_fixit_consolidation): Update for fix-it API
24991 * diagnostic.c (print_parseable_fixits): Likewise.
24992 * edit-context.c (edited_line::m_line_events): Convert from
24993 auto_vec <line_event *> to auto_vec <line_event>.
24994 (class line_event): Convert from abstract base class to a concrete
24995 class, taking over the role of replace_event.
24996 (class insert_event): Delete.
24997 (class replace_event): Rename to class line_event. Convert to
24999 (edit_context::add_fixits): Reimplement.
25000 (edit_context::apply_insert): Delete.
25001 (edit_context::apply_replace): Rename to...
25002 (edit_context::apply_fixit): ...this. Convert to half-open range.
25003 (edited_file::apply_insert): Delete.
25004 (edited_file::apply_replace): Rename to...
25005 (edited_file::apply_fixit): ...this.
25006 (edited_line::~edited_line): Drop deletion of events.
25007 (edited_line::apply_insert): Delete.
25008 (edited_line::apply_replace): Rename to...
25009 (edited_line::apply_fixit): ...this. Convert to half-open range.
25010 Update for change to type of m_line_events.
25011 * edit-context.h (edit_context::apply_insert): Delete.
25012 (edit_context::apply_replace): Rename to...
25013 (edit_context::apply_fixit): ...this.
25015 2017-05-01 Martin Sebor <msebor@redhat.com>
25017 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
25020 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
25023 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
25024 __get_cpuid_max returns 0.
25025 (__get_cpuid_count): Ditto.
25027 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
25029 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
25030 replacement expression is another instance of one of its arguments.
25032 2017-05-01 Jakub Jelinek <jakub@redhat.com>
25035 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
25036 check for stack push/pop autoinc.
25037 * config/i386/i386.c (ix86_agi_dependent): Return false
25038 if the only reason why modified_in_p returned true is that
25039 addr is SP based and set_insn is a push or pop.
25041 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25043 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
25046 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25049 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
25050 (account_size_time): Use two predicates - exec_pred and
25052 (evaluate_conditions_for_known_args): Compute both clause and
25054 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
25055 (inline_summary_t::duplicate): Update.
25056 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
25058 (compute_inline_parameters): Likewise.
25059 (estimate_edge_size_and_time): Update caluclation of time.
25060 (estimate_node_size_and_time): Compute both time and nonspecialized
25062 (estimate_ipcp_clone_size_and_time): Update.
25063 (inline_merge_summary): Update.
25064 (do_estimate_edge_time): Update.
25065 (do_estimate_edge_size): Update.
25066 (do_estimate_edge_hints): Update.
25067 (inline_read_section, inline_write_summary): Stream both new predicates.
25068 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
25070 (compute_inlined_call_time): Cleanup.
25071 (big_speedup_p): Update.
25072 (edge_badness): Update.
25073 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
25074 (size_time_entry): Replace predicate by exec_predicate and
25075 nonconst_predicate.
25076 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
25077 (estimate_edge_time): Return also nonspec_time.
25078 (reset_edge_growth_cache): Update.
25080 2017-04-29 Jakub Jelinek <jakub@redhat.com>
25082 PR rtl-optimization/80491
25083 * ifcvt.c (noce_process_if_block): When looking for x setter
25084 with missing else_bb, don't check only the insn right before
25085 cond_earliest, but look for the last insn that x is modified in
25086 within the same bb.
25088 PR rtl-optimization/80491
25089 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
25091 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
25093 PR tree-optimization/80487
25094 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
25096 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25098 PR tree-optimization/79697
25099 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
25100 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
25101 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
25103 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
25104 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
25106 2017-04-28 Martin Sebor <msebor@redhat.com>
25108 PR tree-optimization/80523
25109 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
25110 (init_target_to_host_charmap, target_to_host, target_strtol10): New
25112 (maybe_warn, format_directive, parse_directive): Use new functions.
25113 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
25115 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
25117 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
25119 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
25121 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
25122 target_header_dir): Set correctly.
25123 * configure: Regenerated.
25124 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25125 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25126 instead of SYSTEM_HEADER_DIR.
25128 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
25130 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
25131 (estimate_local_effects): Likewise.
25132 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
25133 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
25134 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
25135 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
25136 do_estimate_edge_time, estimate_edge_time): Likewise.
25137 * ipa-inline-analysis.c (estimate_node_size_and_time,
25138 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
25139 (estimate_time_after_inlining): Remove.
25141 2017-04-28 Martin Liska <mliska@suse.cz>
25143 * doc/gcov.texi: Enhance documentation of gcov.
25145 2017-04-28 Martin Liska <mliska@suse.cz>
25147 * doc/gcov.texi: Sort options in alphabetic order.
25148 * doc/gcov-dump.texi: Likewise.
25149 * doc/gcov-tool.texi: Likewise.
25150 * gcov.c (print_usage): Likewise.
25151 * gcov-dump.c (print_usage): Likewise.
25152 * gcov-tool.c (print_merge_usage_message): Likewise.
25153 (print_rewrite_usage_message): Likewise.
25154 (print_overlap_usage_message): Likewise.
25156 2017-04-28 Martin Liska <mliska@suse.cz>
25158 PR gcov-profile/53915
25159 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
25161 2017-04-28 Martin Liska <mliska@suse.cz>
25163 PR gcov-profile/79891
25164 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
25165 is marked by compiler as living on a line.
25166 (get_cycles_count): Remove usage of the union.
25167 (output_intermediate_file): Likewise.
25168 (find_source): Fix GNU coding style.
25169 (accumulate_line_counts): Remove old non-all block mode.
25170 (output_lines): Remove usage of the union.
25171 * profile.c (output_location): Include all BBs, even if
25172 belonging to a same line (and file) as a previous BB.
25174 2017-04-28 Martin Liska <mliska@suse.cz>
25176 * gcov.c (process_args): Handle new argument 'w'.
25177 (read_graph_file): Assign ID to BBs.
25178 (output_branch_count): Display BB # if verbose flag is set.
25179 (output_lines): Likewise for arcs.
25180 (print_usage): Add '--verbose' option help.
25181 * doc/gcov.texi: Document --verbose (-w) option.
25183 2017-04-28 Martin Liska <mliska@suse.cz>
25185 * gcov.c (struct block_location_info): New struct.
25186 (process_file): Fill up the new structure.
25187 (read_graph_file): Replace usage of encoding by the newly added
25189 (add_line_counts): Likewise.
25190 (accumulate_line_counts): Remove usage of the union.
25191 (function_info::function_info): New function.
25192 (function_info::~function_info): Likewise.
25193 (process_file): Call delete instead of release_function.
25194 (release_function): Release the function.
25195 (release_structures): Call delete instead of release_function.
25196 (solve_flow_graph): Replace usage of num_blocks.
25197 (find_exception_blocks): Likewise.
25198 (output_lines): Fix GNU coding style.
25200 2017-04-28 Martin Liska <mliska@suse.cz>
25203 * coverage.c (coverage_remove_note_file): New function.
25204 * coverage.h: Declare the function.
25205 * toplev.c (finalize): Clean if an error has been seen.
25207 2017-04-28 Martin Liska <mliska@suse.cz>
25209 PR gcov-profile/80031
25210 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
25211 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
25212 * gcov.c (read_graph_file): Read just number of blocks.
25213 * profile.c (branch_prob): Do not stream 0 flags per a basic
25216 2017-04-28 Martin Liska <mliska@suse.cz>
25218 * gcov-dump.c (tag_*): Add new argument to declarations.
25219 (dump_gcov_file): Likewise.
25220 (tag_blocks): Add and use new argument depth.
25221 (tag_arcs): Likewise.
25222 (tag_lines): Likewise.
25223 (tag_counters): Likewise.
25224 (tag_summary): Likewise.
25225 (dump_working_sets): Use depth to do a proper indentation.
25227 2017-04-28 Jakub Jelinek <jakub@redhat.com>
25230 * cgraph.h (symtab_node::debug_symtab): No longer inline.
25231 * symtab.c (symtab_node::debug_symtab): Move definition here.
25233 2017-04-28 Richard Biener <rguenther@suse.de>
25235 * lto-streamer.h (LTO_major_version): Bump to 7.
25237 2017-04-28 Richard Biener <rguenther@suse.de>
25239 * tree-vrp.c (assert_info): New struct.
25240 (add_assert_info): New helper.
25241 (register_edge_assert_for_2): Refactor to add asserts to a vector
25243 (register_edge_assert_for_1): Likewise.
25244 (register_edge_assert_for): Likewise.
25245 (finish_register_edge_assert_for): New helper actually registering
25246 asserts where live on edge.
25247 (find_conditional_asserts): Adjust.
25248 (find_switch_asserts): Likewise.
25249 (evrp_dom_walker::try_find_new_range): Generalize.
25250 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
25252 2017-04-27 Marek Polacek <polacek@redhat.com>
25255 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
25256 arg10 and arg11 to itype.
25258 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
25260 * doc/extend.texi (Object Size Checking): Improve grammar.
25262 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
25265 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
25266 that the logic for permitting reciprocal estimates matches that
25269 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25272 * tree.c (type_cache_hasher::equal): Only compare
25273 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
25274 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
25275 non-aggregate element types.
25276 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
25277 about the flag on ARRAY_TYPEs in the comment, formatting fix.
25279 2017-04-27 Richard Biener <rguenther@suse.de>
25281 PR middle-end/80533
25282 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
25283 stripping ARRAY_REFs from MEM_EXPR make sure we're not
25284 keeping a reference to a trailing array.
25286 2017-04-27 Richard Biener <rguenther@suse.de>
25288 PR middle-end/80539
25289 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
25290 being in loop-closed SSA form conservatively.
25291 (chrec_fold_multiply_poly_poly): Likewise.
25293 2017-04-27 Tamar Christina <tamar.christina@arm.com>
25295 PR middle-end/79665
25296 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
25297 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
25299 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25302 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
25303 (aarch64_function_arg_alignment): Return unsigned int again, but still
25304 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
25305 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
25306 Don't emit -Wpsabi note.
25307 (aarch64_function_arg_boundary): Likewise.
25308 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
25311 2017-04-26 Nathan Sidwell <nathan@acm.org>
25313 * tree.h (crc32_unsigned_n): Declare.
25314 (crc32_unsigned, crc32_unsigned): Make inline.
25315 * tree.c (crc32_unsigned_bits): Replace with ...
25316 (crc32_unsigned_n): ... this.
25317 (crc32_unsigned, crc32_byte): Remove.
25318 (crc32_string): Remove unnecessary braces.
25320 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25322 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
25323 * ipa-inline-analysis.c (MAX_TIME): Remove.
25324 (account_size_time): Use sreal for time.
25325 (dump_inline_summary): Update.
25326 (estimate_function_body_sizes): Update.
25327 (estimate_edge_size_and_time): Update.
25328 (estimate_calls_size_and_time): Update.
25329 (estimate_node_size_and_time): Update.
25330 (inline_merge_summary): Update.
25331 (inline_update_overall_summary): Update.
25332 (estimate_time_after_inlining): Update.
25333 (inline_read_section): Update.
25334 (inline_write_summary): Update.
25335 * ipa-inline.c (compute_uninlined_call_time): Update.
25336 (compute_inlined_call_time): Update.
25337 (recursive_inlining): Update.
25338 (inline_small_functions): Update.
25339 (dump_overall_stats): Update.
25340 * ipa-inline.h: Include sreal.h.
25341 (size_time_entry): Turn time to sreal.
25342 (inline_summary): Turn self_time nad time to sreal.
25344 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25346 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
25348 (sreal::stream_out, sreal::stream_in): New.
25349 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
25351 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25353 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
25356 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
25359 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
25361 (dimode_scalar_chain::compute_convert_gain): Ditto.
25362 (dimode_scalar_chain::make_vector_copies): Ditto.
25363 (dimode_scalar_chain::convert_reg): Ditto.
25364 (dimode_scalar_chain::convert_insn): Ditto.
25365 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
25366 (VI248_AVX512BW_1): New mode iterator.
25367 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
25368 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
25371 2017-04-25 Martin Sebor <msebor@redhat.com>
25373 PR tree-optimization/80497
25374 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
25375 constants are representable in HOST_WIDE_INT.
25376 (parse_directive): Ditto.
25378 2017-04-25 Martin Sebor <msebor@redhat.com>
25381 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
25382 (new_zero_array): Adjust signature.
25383 (dom_info::dom_init): Used unsigned rather that size_t.
25384 (dom_info::dom_info): Same.
25386 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25387 Jakub Jelinek <jakub@redhat.com>
25390 * config/arm/arm.c: Include gimple.h.
25391 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
25392 returns negative, increment ncrn only if it returned positive.
25393 (arm_needs_doubleword_align): Return int instead of bool,
25394 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
25395 members, but if there is any such non-FIELD_DECL
25396 > PARM_BOUNDARY aligned decl, return -1 instead of false.
25397 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
25398 returns negative, increment nregs only if it returned positive.
25399 (arm_setup_incoming_varargs): Likewise.
25400 (arm_function_arg_boundary): Emit -Wpsabi note if
25401 arm_needs_doubleword_align returns negative, return
25402 DOUBLEWORD_ALIGNMENT only if it returned positive.
25404 2017-04-25 Marek Polacek <polacek@redhat.com>
25407 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
25408 first argument to type.
25410 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
25413 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
25414 type checks to test for compatibility instead of equality.
25416 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25417 Jakub Jelinek <jakub@redhat.com>
25420 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
25422 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
25423 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
25424 the alignment computation, but return their maximum in warn_alignment.
25425 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
25426 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
25428 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
25429 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
25432 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25434 * config/arc/simdext.md (dmpyh): Fix typo.
25436 2017-04-25 Richard Biener <rguenther@suse.de>
25438 PR tree-optimization/80492
25439 * alias.c (compare_base_decls): Handle registers with asm
25440 specification conservatively.
25441 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
25442 compare_base_decls returning dont-know properly.
25444 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25446 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
25447 (legitimate_offset_address_p): New function.
25448 (arc_legitimate_address_p): Use above function.
25450 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25452 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
25454 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25456 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
25457 ACCH registers whenever they are available.
25459 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25461 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
25462 double regs fix when not used.
25464 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25466 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
25468 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
25469 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
25471 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25473 * config/arc/arc.c (arc_output_addsi): Check for h-register class
25474 when emitting short ADD instructions.
25476 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25478 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
25480 (cmpsi_cc_c_insn): Likewise.
25481 (cbranchsi4_scratch): Compute proper instruction length using
25482 compact_hreg_operand.
25483 * config/arc/predicates.md (compact_hreg_operand): New predicate.
25485 2017-04-25 Richard Biener <rguenther@suse.de>
25487 PR middle-end/80509
25488 * passes.c (pass_manager::pass_manager): Initialize
25489 m_name_to_pass_map.
25491 2017-04-25 Richard Biener <rguenther@suse.de>
25493 PR tree-optimization/79201
25494 * tree-ssa-sink.c (statement_sink_location): Handle calls.
25496 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25499 * config/s390/vector.md: Split MEM->GPR vector moves for
25500 non-s_operand addresses.
25502 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25505 * config/s390/predicates.md (reload_const_wide_int_operand): New
25507 * config/s390/s390.md ("movti"): Remove d/P alternative.
25508 ("movti_bigconst"): New pattern definition.
25510 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25513 * s390-protos.h (s390_expand_cs_hqi): Removed.
25514 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
25515 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
25516 modes as well as CCZ1mode and CCZmode.
25517 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
25518 signature of s390_emit_compare_and_swap.
25519 (s390_expand_cs_hqi): Likewise, make static.
25520 (s390_expand_cs_tdsi): Generate an explicit compare before trying
25521 compare-and-swap, in some cases.
25522 (s390_expand_cs): Wrapper function.
25523 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
25525 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
25526 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
25527 patterns for small and large integers. Forbid symref memory operands.
25528 Move expander to s390.c. Require cc register.
25529 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
25530 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
25531 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
25532 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
25533 symref memory operands. Remove CC mode and call s390_match_ccmode
25535 ("atomic_exchange<mode>"): Allow and implement all integer modes.
25537 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25539 * config/s390/s390.md (define_peephole2): New peephole to help
25540 combining the load-and-test pattern with volatile memory.
25542 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25544 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
25545 with CCZmode for TARGET_Z196.
25547 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25549 PR rtl-optimization/80501
25550 * combine.c (make_compound_operation_int): Set subreg_code to SET
25551 even for AND with mask of the sign bit of mode.
25553 PR rtl-optimization/80500
25554 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
25555 sum's initial value.
25557 2017-04-25 Julian Brown <julian@codesourcery.com>
25558 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25560 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
25562 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
25564 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
25566 2017-04-25 Julian Brown <julian@codesourcery.com>
25567 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25569 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
25570 (thunderx2t99_sha): New Reservation.
25572 2017-04-25 Julian Brown <julian@codesourcery.com>
25573 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25575 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
25576 type for 1-element load.
25578 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
25580 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
25582 2017-04-24 Martin Jambor <mjambor@suse.cz>
25584 PR tree-optimization/80293
25585 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
25586 char arrays not totally scalarizable if it is false.
25587 (analyze_all_variable_accesses): Pass correct value in the new
25588 parameter. Add a statistics counter.
25590 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
25592 PR middle-end/79931
25593 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
25595 2017-04-24 Richard Biener <rguenther@suse.de>
25597 PR tree-optimization/80494
25598 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
25599 out for complex types.
25601 2017-04-24 Richard Biener <rguenther@suse.de>
25603 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
25604 * tree-ssa-sccvn.c (print_scc): Print SCC size.
25605 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
25606 (DFS): Adjust and never fail.
25607 (sccvn_dom_walker::fail): Remove.
25608 (sccvn_dom_walker::before_dom_children): Adjust.
25609 (run_scc_vn): Likewise and never fail.
25610 * tree-ssa-pre.c (pass_pre::execute): Adjust.
25611 (pass_fre::execute): Likewise.
25613 2017-04-24 Richard Biener <rguenther@suse.de>
25615 PR tree-optimization/79725
25616 * tree-ssa-sink.c (statement_sink_location): Return whether
25617 failure reason was zero uses. Move that check later.
25618 (sink_code_in_bb): Deal with zero uses by removing the stmt
25621 2017-04-24 Richard Biener <rguenther@suse.de>
25624 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
25625 pointer-based references.
25627 2017-04-24 Richard Biener <rguenther@suse.de>
25630 * pass_manager.h (pass_manager::operator new): Remove.
25631 (pass_manager::operator delete): Likewise.
25632 * passes.c (pass_manager::operator new): Remove.
25633 (pass_manager::operator delete): Likewise.
25634 (pass_manager::pass_manager): Zero individual pass members.
25636 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
25639 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
25640 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
25641 Check "XEXP (src, 1)" operand here.
25642 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
25643 Check "XEXP (src, 1)" operand here.
25644 (dimode_scalar_chain::make_vector_copies): Detect count register
25645 of a shift instruction. Zero extend count register from QImode
25646 to DImode to satisfy vector shift pattern count operand predicate.
25647 Substitute vector shift count operand with a DImode copy.
25648 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
25651 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
25653 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
25654 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
25655 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
25656 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
25657 (UNSPEC_NOREX_MEM): Remove definition.
25659 2017-04-21 Richard Biener <rguenther@suse.de>
25661 PR tree-optimization/79547
25662 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
25663 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
25664 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
25665 without any constraints.
25667 2017-04-21 Richard Biener <rguenther@suse.de>
25669 PR tree-optimization/78847
25670 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
25672 2017-04-21 Richard Biener <rguenther@suse.de>
25674 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
25675 (build_distinct_type_copy): Likewise.
25676 (build_variant_type_copy): Likewise.
25677 * tree.c (build_qualified_type): Pass down mem-stat info.
25678 (build_distinct_type_copy): Likewise.
25679 (build_variant_type_copy): Likewise.
25681 2017-04-21 Richard Biener <rguenther@suse.de>
25683 PR tree-optimization/80237
25684 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
25686 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
25687 for a simplified result.
25689 2016-04-21 Richard Biener <rguenther@suse.de>
25691 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
25692 sth as strict as a simple_iv but a chrec without symbols and an
25693 operand defined in the loop we are peeling (and not some subloop).
25694 (propagate_constants_for_unrolling): Propagate all constants.
25696 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25699 * config/i386/i386.c (print_reg): Remove assert for disalowed
25700 regno values, call output_operand_lossage instead.
25702 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25705 * config/i386/constraints.md (Yc): New register constraint.
25706 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
25707 Use Yc constraint for alternative 2 of operand 0. Remove
25708 preferred_for_speed attribute.
25710 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
25712 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
25713 lastprivate clauses in SIMT case.
25715 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
25717 * doc/invoke.texi (-Wextra-semi): Document new warning option.
25719 2017-04-20 Richard Biener <rguenther@suse.de>
25721 PR tree-optimization/57796
25722 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
25723 as N scalar stores.
25724 (vect_model_load_cost): Cost gathers as N scalar loads.
25726 2017-04-20 Richard Biener <rguenther@suse.de>
25728 * ggc-page.c (ggc_allocated_p): Rename to ...
25729 (safe_lookup_page_table_entry): ... this and return the lookup
25731 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
25733 2017-04-20 Richard Biener <rguenther@suse.de>
25735 PR tree-optimization/80453
25736 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
25737 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
25738 from the conditions.
25739 (vn_phi_eq): Pass them down.
25740 (vn_phi_lookup): Record them.
25741 (vn_phi_insert): Likewise.
25743 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
25745 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
25746 uninitialized variable warning to avoid buffer overrun.
25748 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
25751 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
25752 is suppressed for '{ 0 }' in C.
25754 2017-04-20 Jakub Jelinek <jakub@redhat.com>
25756 * BASE-VER: Set to 8.0.0.
25758 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
25760 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
25761 priority .init_array and .fini_array section with SECTION_NOTYPE
25764 2017-04-20 Jakub Jelinek <jakub@redhat.com>
25766 PR middle-end/80423
25767 * tree.h (build_array_type): Add typeless_storage default argument.
25768 * tree.c (type_cache_hasher::equal): Also compare
25769 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
25770 (build_array_type): Add typeless_storage argument, set
25771 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
25773 (build_nonshared_array_type): Adjust build_array_type_1 caller.
25774 (build_array_type): Likewise. Add typeless_storage argument.
25776 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
25777 Jakub Jelinek <jakub@redhat.com>
25779 PR tree-optimization/80426
25780 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
25781 operation on symbolic operands, also compute the overflow for the
25782 invariant part when the operation degenerates into a negation.
25784 2017-04-19 Jakub Jelinek <jakub@redhat.com>
25787 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
25788 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
25791 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
25793 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
25796 * config/avr/avr.c (tree.h): Include it.
25797 (cgraph.h): Include it.
25798 (avr_encode_section_info): Don't warn for uninitialized progmem
25799 variable if it's just an alias.
25801 2017-04-19 Richard Biener <rguenther@suse.de>
25804 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
25805 when needed by AutoPGO.
25807 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
25810 * doc/lto.texi: Remove an extra 'that'.
25812 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
25814 PR rtl-optimization/80429
25815 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
25816 are only used in debug insns.
25818 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
25819 Vladimir Makarov <vmakarov@redhat.com>
25821 * config/sparc/predicates.md (input_operand): Add comment. Return
25822 true for any memory operand when LRA is in progress.
25823 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
25825 2017-04-18 Jeff Law <law@redhat.com>
25828 * mips.md ({return,simple_return}_internal): Do not overwrite
25831 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25833 PR tree-optimization/80443
25834 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
25835 instead of adding 1, subtract -1 and similarly instead of subtracting
25838 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
25840 PR rtl-optimization/80357
25841 * haifa-sched.c (tmp_bitmap): New variable.
25842 (model_recompute): Handle duplicate use records.
25843 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
25844 (free_global_sched_pressure_data): Free it.
25846 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
25849 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
25850 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25851 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25852 instead of SYSTEM_HEADER_DIR.
25854 2017-04-18 Jeff Law <law@redhat.com>
25856 PR middle-end/80422
25857 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
25858 predecessors after walking up the insn chain.
25860 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25863 * dwarf2out.c (modified_type_die): Try harder not to emit internal
25864 sizetype type into debug info.
25866 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
25869 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
25870 unneeded test for TARGET_UPPER_REGS_SF.
25871 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
25873 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25876 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
25877 instead of gsi_after_labels.
25879 2017-04-18 Jeff Law <law@redhat.com>
25881 * regcprop.c (maybe_mode_change): Avoid creating copies of the
25885 2017-04-13 Jeff Law <law@redhat.com>
25886 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
25887 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
25889 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
25892 * config/avr/avr.c (intl.h): Include it.
25893 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
25895 2017-04-18 Martin Liska <mliska@suse.cz>
25897 PR gcov-profile/78783
25898 * gcov-tool.c (gcov_output_files): Validate that destination
25899 file is either removed by the tool or by a user.
25901 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
25902 Guy Benyei <guybe@mellanox.com>
25904 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
25905 block, and do not negate it, the stored id is already negative.
25907 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
25909 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
25911 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
25914 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
25915 masks of options that should be turned off if the VSX vector
25916 options are turned off.
25917 (OTHER_P8_VECTOR_MASKS): Likewise.
25918 (OTHER_VSX_VECTOR_MASKS): Likewise.
25919 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
25920 rs6000_disable_incompatible_switches to validate no type switches
25922 (rs6000_incompatible_switch): New function to disallow turning on
25923 other vector options if -mno-vsx, -mno-power8-vector, or
25924 -mno-power9-vector are specified.
25926 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
25928 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
25930 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
25932 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
25933 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
25934 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
25935 (ARG_POINTER_CFA_OFFSET): Likewise.
25937 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
25939 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
25940 conditions to take advantage of various optimizations.
25942 2017-04-13 Jeff Law <law@redhat.com>
25944 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
25945 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
25946 (zero_extendsidi2_dext): Likewise.
25948 2017-04-13 Jakub Jelinek <jakub@redhat.com>
25951 * fold-const.c (fold_ternary_loc): Revert
25952 use op0 instead of fold_convert_loc (loc, type, arg0) part of
25955 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
25957 PR rtl-optimization/80343
25958 * lra-remat.c (update_scratch_ops): Assign original hard reg to
25959 new scratch pseudo.
25961 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
25964 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
25965 to ubsan_encode_value.
25967 2017-04-13 Jeff Law <law@redhat.com>
25969 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
25970 appearing in DEBUG_INSNs.
25972 2017-04-13 Martin Liska <mliska@suse.cz>
25974 PR gcov-profile/80413
25975 * gcov-io.c (gcov_write_string): Copy to buffer just when
25976 allocated size is greater than zero.
25978 2017-04-13 Jakub Jelinek <jakub@redhat.com>
25981 * dwarf2out.c (decls_for_scope): Ignore declarations of
25982 current_function_decl in BLOCK_NONLOCALIZED_VARS.
25984 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
25987 * ipa-visibility.c (non_local_p): Fix typos.
25988 (localize_node): When localizing symbol in same comdat group,
25989 dissolve the group only when we know external symbols are going
25991 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
25993 2017-04-12 Jakub Jelinek <jakub@redhat.com>
25995 PR tree-optimization/79390
25996 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
25997 order does not result in usable sequence, retry with reversed operand
26003 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
26004 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
26005 op0 instead of fold_convert_loc (loc, type, arg0).
26007 2017-04-12 Jeff Law <law@redhat.com>
26009 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
26010 has a delay slot in the generated code.
26012 * config/cris/cris.md (cris_preferred_reload_class): Return
26013 GENNONACR_REGS rather than GENERAL_REGS.
26015 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26018 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
26019 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
26020 signedness of the result type.
26022 2017-04-12 Richard Biener <rguenther@suse.de>
26023 Jeff Law <law@redhat.com>
26025 PR tree-optimization/80359
26026 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
26027 trim stores to TARGET_MEM_REFs.
26029 2017-04-12 Richard Biener <rguenther@suse.de>
26031 PR tree-optimization/79390
26032 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
26033 threading case even more.
26035 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
26038 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
26039 for quad_address_p for TImode, instead of just not indexed_address.
26041 2017-04-12 Richard Biener <rguenther@suse.de>
26042 Bernd Edlinger <bernd.edlinger@hotmail.de>
26044 PR middle-end/79671
26045 * alias.c (component_uses_parent_alias_set_from): Handle
26046 TYPE_TYPELESS_STORAGE.
26047 (get_alias_set): Likewise.
26048 * tree-core.h (tree_type_common): Add typeless_storage flag.
26049 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
26050 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
26051 for types containing members with TYPE_TYPELESS_STORAGE.
26052 (place_field): Likewise.
26053 (layout_type): Likewise for ARRAY_TYPE.
26054 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
26055 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
26056 TYPE_TYPELESS_STORAGE.
26057 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
26059 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26062 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
26063 first argument to type.
26065 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26069 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
26070 CONST0_RTX (mode) rather than const0_rtx where appropriate.
26071 (rs6000_expand_binop_builtin): Likewise.
26072 (rs6000_expand_ternop_builtin): Likewise; also add missing
26073 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
26074 vshasigma built-ins.
26075 * doc/extend.texi: Document that vec_xxpermdi's third argument
26076 must be a constant.
26078 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
26080 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
26081 Use shift_const cost parameter when calculating gain of STV shifts.
26083 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
26085 PR rtl-optimization/70478
26086 * lra-constraints.c (process_alt_operands): Check memory for
26087 disfavoring memory insn operand.
26089 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26091 PR middle-end/80100
26092 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
26093 left shift in unsigned HOST_WIDE_INT type.
26095 PR rtl-optimization/80385
26096 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
26097 (not (neg X)) into (plus X -1) for complex or non-integral modes.
26100 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
26101 if they have any depend clauses.
26103 2017-04-11 Martin Liska <mliska@suse.cz>
26106 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
26107 * ipa-split.c (split_function): Create a local comdat symbol
26108 if caller is in a comdat group.
26110 2017-04-11 Martin Liska <mliska@suse.cz>
26113 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
26116 2017-04-11 Martin Sebor <msebor@redhat.com>
26118 PR middle-end/80364
26119 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
26120 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
26122 (directive::set_width, directive::set_precision, format_character):
26124 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
26127 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
26130 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
26131 conflict, set target->arch_name instead of target->cpu_name.
26133 2017-04-11 Richard Biener <rguenther@suse.de>
26135 PR tree-optimization/80374
26136 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
26137 build_zero_cst, remove fold_convertible_p check again.
26139 2017-04-11 Martin Liska <mliska@suse.cz>
26142 * ubsan.c (instrument_object_size): Do not instrument register
26145 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26148 * config/i386/i386-builtin-types.def
26149 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
26150 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
26151 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
26152 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
26153 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
26154 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
26155 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
26156 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
26157 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
26158 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
26159 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
26160 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
26161 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
26162 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
26163 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
26164 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
26165 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
26166 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
26167 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
26168 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
26169 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
26170 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
26171 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
26172 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
26173 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
26174 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
26175 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
26176 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
26177 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
26178 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
26179 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
26180 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
26181 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
26182 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
26183 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
26184 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
26185 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
26186 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
26187 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
26188 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
26189 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
26190 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
26191 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
26192 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
26193 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
26194 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
26196 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
26197 flag to second_arg_count, handle 4 argument function type _COUNT
26198 aliases, handle second_arg_count on second argument rather than last.
26200 2017-04-10 Jeff Law <law@redhat.com>
26202 PR tree-optimization/80374
26203 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
26204 record anything if we can not convert integer_zero_node to the
26207 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
26210 * config/rs6000/rs6000.c (rs6000_option_override_internal):
26211 Enhance special handling given to the TARGET_P9_MINMAX option in
26212 relation to certain other options.
26214 2017-04-10 Bin Cheng <bin.cheng@arm.com>
26216 PR tree-optimization/80153
26217 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
26218 remove POINTER_PLUS_EXPR's base part directly, rather than through
26221 2017-04-10 Richard Biener <rguenther@suse.de>
26222 Bin Cheng <bin.cheng@arm.com>
26224 PR tree-optimization/80153
26225 * tree-affine.c (aff_combination_to_tree): Get base pointer from
26226 the first element of pointer type aff_tree. Build result expr in
26228 (add_elt_to_tree): Convert to type unconditionally. Remove other
26229 fold_convert calls.
26230 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
26231 (rewrite_use_nonlinear_expr): Check invariant using iv information.
26233 2017-04-10 Richard Biener <rguenther@suse.de>
26235 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26238 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
26240 PR rtl-optimization/70478
26241 * lra-constraints.c (curr_small_class_check): New.
26242 (update_and_check_small_class_inputs): New.
26243 (process_alt_operands): Update curr_small_class_check. Disfavor
26244 alternative insn memory operands. Check available regs for small
26247 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26250 * config/mips/mips.opt (-mvirt): Update description.
26251 * doc/invoke.texi (-mvirt): Likewise.
26253 2017-04-10 Richard Biener <rguenther@suse.de>
26255 PR middle-end/80362
26256 * fold-const.c (fold_binary_loc): Look at unstripped ops when
26257 looking for NEGATE_EXPR in -A / -B to A / B folding.
26259 2017-04-10 Martin Liska <mliska@suse.cz>
26261 PR gcov-profile/80224
26262 * gcov.c (print_usage): Fix usage string.
26263 (get_gcov_intermediate_filename): Remove.
26264 (output_gcov_file): Use both for normal and intermediate format.
26265 (generate_results): Do not initialize special file for
26266 intermediate format.
26268 2017-04-10 Richard Biener <rguenther@suse.de>
26270 PR tree-optimization/80304
26271 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
26274 2017-04-10 Nathan Sidwell <nathan@acm.org>
26277 * config/rs6000/rs6000.c (rs6000_vector_type): New.
26278 (rs6000_init_builtins): Use it.
26280 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26282 * config/arm/arm.md (<mrc>): Add mode to SET source.
26283 (<mrrc>): Likewise.
26285 2017-04-10 Richard Biener <rguenther@suse.de>
26287 PR middle-end/80344
26288 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
26290 2017-04-10 Jakub Jelinek <jakub@redhat.com>
26293 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
26294 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
26295 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
26296 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
26297 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
26298 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
26299 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
26300 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
26301 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
26302 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
26303 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
26304 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
26305 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
26306 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
26307 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
26308 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
26309 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
26310 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
26311 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
26312 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
26313 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
26314 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
26315 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
26317 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
26319 PR rtl-optimization/70478
26320 * lra-constraints.c: Reverse the last patch.
26322 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
26324 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
26325 Add comment for WCHAR_T.
26327 2017-04-08 Martin Liska <mliska@suse.cz>
26330 2017-04-07 Martin Liska <mliska@suse.cz>
26333 * ipa-split.c (split_function): Add function part to a same comdat
26336 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26339 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
26341 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
26343 * rs6000/rs6000.c (vec_load_pendulum): Rename...
26344 (vec_pairing): ...to this.
26345 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
26346 (rs6000_sched_init): Adjust for name change.
26347 (struct rs6000_sched_context): Likewise.
26348 (rs6000_init_sched_context): Likewise.
26349 (rs6000_set_sched_context): Likewise.
26351 2017-04-07 Jakub Jelinek <jakub@redhat.com>
26357 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
26359 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
26360 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
26361 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
26363 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
26365 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
26367 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
26369 PR rtl-optimization/70703
26370 * ira-color.c (update_conflict_hard_regno_costs): Use
26371 int64_t instead of HOST_WIDE_INT.
26373 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
26375 PR rtl-optimization/70478
26376 * lra-constraints.c (process_alt_operands): Disfavor alternative
26377 insn memory operands.
26379 2017-04-07 Jeff Law <law@redhat.com>
26381 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
26382 CALL and NOTE_INSN_CALL_ARG_LOCATION.
26384 2017-04-07 Martin Liska <mliska@suse.cz>
26387 * config/aarch64/aarch64.c (aarch64_process_target_attr):
26388 Show error message instead of an ICE.
26390 2017-04-07 Martin Liska <mliska@suse.cz>
26393 * ipa-split.c (split_function): Add function part to a same comdat
26396 2017-04-07 Richard Biener <rguenther@suse.de>
26398 PR middle-end/80341
26399 * tree.c (get_unwidened): Also handle ! for_type case for
26401 * convert.c (do_narrow): Split out from ...
26402 (convert_to_integer_1): ... here. Do not pass final truncation
26403 type to get_unwidened for TRUNC_DIV_EXPR.
26405 2017-04-07 Richard Biener <rguenther@suse.de>
26407 * tree-affine.c (wide_int_ext_for_comb): Take type rather
26409 (aff_combination_const): Adjust.
26410 (aff_combination_scale): Likewise.
26411 (aff_combination_add_elt): Likewise.
26412 (aff_combination_add_cst): Likewise.
26413 (aff_combination_convert): Likewise.
26414 (add_elt_to_tree): Likewise. Remove unused argument.
26415 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
26417 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
26419 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
26421 * config/arm/arm.c (arm_default_short_enums): Use
26422 ARM_DEFAULT_SHORT_ENUMS.
26423 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
26425 2017-04-06 Jakub Jelinek <jakub@redhat.com>
26428 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
26429 members with redundant out-of-class redeclaration.
26431 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
26434 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
26435 * config/i386/i386.md (*zero_extendsidi2):
26436 Add (?*x,*x) and (?*v,*v) alternatives.
26438 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
26441 * config/i386/i386.c (ix86_expand_builtin)
26442 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
26443 mode from insn data. Convert operands to insn operand mode.
26444 Copy operands that don't satisfy insn predicate to a register.
26446 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
26448 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
26451 2017-04-06 Richard Biener <rguenther@suse.de>
26453 PR tree-optimization/80334
26454 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
26455 preserve alignment of accesses.
26457 2017-04-06 Richard Biener <rguenther@suse.de>
26459 PR tree-optimization/80262
26460 * tree-sra.c (build_ref_for_offset): Preserve address-space
26462 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
26463 Drop useless address-space information on MEM_REF offsets.
26465 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
26467 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
26469 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
26471 PR rtl-optimization/70703
26472 * ira-color.c (update_conflict_hard_regno_costs): Use
26473 HOST_WIDE_INT instead of long.
26475 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
26478 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
26479 not defined for x86_64 target. Add -mmmx target option when __SSE2__
26481 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
26482 for x86_64 target. Handle -m3dnowa option.
26484 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
26486 PR rtl-optimization/70703
26487 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
26488 (update_conflict_hard_regno_costs): Use long instead of unsigned
26489 arithmetic for cost calculation.
26491 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26492 Bernd Edlinger <bernd.edlinger@hotmail.de>
26495 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
26498 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
26501 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
26502 ptr_mode with Pmode throughout.
26503 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
26504 into probe_stack_range and use DImode.
26506 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
26509 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
26510 call_eh_return is true.
26512 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26514 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
26515 Initialize last_match_fntype_index.
26517 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26520 * tree-nvr.c: Include internal-fn.h.
26521 (pass_return_slot::execute): Ignore internal calls without
26524 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26525 Richard Biener <rguenther@suse.de>
26528 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
26529 captures used multiple times, except for the last use.
26530 * generic-match-head.c: Include gimplify.h.
26532 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26534 PR tree-optimization/79390
26535 * target.h (struct noce_if_info): Declare.
26536 * targhooks.h (default_noce_conversion_profitable_p): Declare.
26537 * target.def (noce_conversion_profitable_p): New target hook.
26538 * ifcvt.h (struct noce_if_info): New type, moved from ...
26539 * ifcvt.c (struct noce_if_info): ... here.
26540 (noce_conversion_profitable_p): Renamed to ...
26541 (default_noce_conversion_profitable_p): ... this. No longer
26543 (noce_try_store_flag_constants, noce_try_addcc,
26544 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
26545 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
26546 instead of noce_conversion_profitable_p.
26547 * config/i386/i386.c: Include ifcvt.h.
26548 (ix86_option_override_internal): Don't override
26549 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
26550 (ix86_noce_conversion_profitable_p): New function.
26551 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
26552 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
26553 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
26554 * doc/tm.texi: Regenerated.
26556 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26558 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
26561 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26564 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
26565 instructions for small multiply cores.
26567 2017-04-04 Jeff Law <law@redhat.com>
26569 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
26571 (mips_expand_vec_perm_const): Initialize elements in orig_perm
26572 that are not set by the loop over the elements.
26574 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26577 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
26578 int mode, convert_modes it to mode as unsigned, otherwise use
26579 lowpart_subreg to mode rather than SImode.
26580 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
26581 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
26582 Use DImode instead of SImode for the shift count operand.
26583 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
26586 2017-04-04 Richard Biener <rguenther@suse.de>
26588 PR middle-end/80281
26589 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
26590 arithmetic done for the negate or the plus. Simplify.
26591 (A - (-B) -> A + B): Likewise.
26592 * fold-const.c (split_tree): Make sure to not negate pointers.
26594 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
26596 PR rtl-optimization/60818
26597 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
26598 a compare of comparisons with the thing compared if this results
26599 in a different machine mode.
26601 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
26603 * alias.c (base_alias_check): Fix typo in comment.
26604 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
26605 * cgraphunit.c (symbol_table::compile): Likewise.
26606 * collect2.c (maybe_run_lto_and_relink): Likewise.
26607 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
26608 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
26609 * config/avr/avr.c (avr_map_op_t): Likewise.
26610 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
26611 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
26612 * config/epiphany/epiphany.md (movcc): Likewise.
26613 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
26614 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
26616 * config/mips/mips.c (mips_save_restore_reg): Likewise.
26617 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
26618 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
26619 * config/sh/sh.c (sh_rtx_costs): Likewise.
26620 * fold-const.c (fold_truth_andor): Likewise.
26621 * genautomata.c (collapse_flag): Likewise.
26622 * gengtype.h (struct type::u::s): Likewise.
26623 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
26624 * input.c (FORMAT_AMOUNT): Likewise.
26625 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
26626 (known_aggs_to_agg_replacement_list): Likewise.
26627 * ipa-inline-analysis.c: Likewise.
26628 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
26629 * ipa-polymorphic-call.c
26630 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
26631 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
26632 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
26634 * modulo-sched.c (apply_reg_moves): Likewise.
26635 * omp-expand.c (build_omp_regions_1): Likewise.
26636 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
26637 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
26638 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
26639 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
26640 * value-prof.c: Likewise.
26641 * var-tracking.c (val_reset): Likewise.
26643 2017-04-03 Richard Biener <rguenther@suse.de>
26645 PR tree-optimization/80275
26646 * fold-const.c (split_address_to_core_and_offset): Handle
26649 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
26651 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
26652 descriptors is at least equal to that of functions.
26654 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
26656 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
26658 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
26661 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
26662 (mov<IMOD4:mode>): New expander.
26663 (*mov<IMOD4:mode>_internal): New insn and split pattern.
26665 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
26667 PR rtl-optimization/79405
26668 * fwprop.c (propagations_left): New variable.
26669 (forward_propagate_into): Decrement it.
26670 (fwprop_init): Initialize it.
26671 (fw_prop): If the variable has reached zero, stop propagating.
26672 (fwprop_addr): Ditto.
26674 2017-03-31 Jakub Jelinek <jakub@redhat.com>
26677 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
26678 a FUNCTION_DECL, pass it as decl instead of origin to
26681 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
26683 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
26686 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
26689 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
26690 TARGET_VSX_SMALL_INTEGER.
26692 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26694 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
26695 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
26697 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26699 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
26700 extraction from odd-numbered MSA register.
26702 2017-03-31 Jakub Jelinek <jakub@redhat.com>
26704 PR middle-end/80173
26705 * expmed.c (store_bit_field_1): Don't attempt to create
26706 a word subreg out of hard registers wider than word if they
26707 have HARD_REGNO_NREGS of 1 for their mode.
26709 PR middle-end/80163
26710 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
26711 conversions to integer types wider than word and pointer.
26714 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
26715 (rtx_equal_for_cselib_p): Pass 0 to it.
26716 * cselib.c (cselib_hasher::equal): Likewise.
26717 (rtx_equal_for_cselib_1): Add depth argument. If depth
26718 is 128, don't look up VALUE locs and punt. Increment
26719 depth in recursive calls when walking VALUE locs.
26721 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
26723 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
26724 (make_gcov_file_name): Use the canonical path name for generating
26726 (read_line): Fix handling of files with ascii null bytes.
26728 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
26730 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
26731 to initialise a vector register instead
26732 of using a const_int.
26734 2017-03-30 Jakub Jelinek <jakub@redhat.com>
26736 PR translation/80189
26737 * gimplify.c (omp_default_clause): Use %qs instead of %s in
26738 diagnostic messages.
26740 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
26743 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
26744 (dfp_diex_<mode>): Update mode of operand 1.
26745 * doc/extend.texi (dxex, dxexq): Document change to return type.
26746 (diex, diexq): Document change to argument type.
26748 2017-03-30 Martin Jambor <mjambor@suse.cz>
26751 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
26752 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
26753 it reflects the signature changes performed at the callee side.
26754 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
26755 to cgraph_build_function_type_skip_args.
26756 (build_function_decl_skip_args): Adjust call to the above function.
26758 2017-03-30 Jakub Jelinek <jakub@redhat.com>
26761 * config/i386/sse.md
26762 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
26763 register as dest whenever it is a MEM not rtx_equal_p to the
26764 corresponding dup operand, and when forcing into reg move the
26765 reg into the memory afterwards.
26766 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
26767 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
26768 for the force_reg mode.
26769 (avx512vl_vextractf128<mode>): Use register as dest either
26770 always when a MEM, or when it is a MEM not rtx_equal_p to the
26771 corresponding dup operand, or even not when it is a CONST_VECTOR
26772 depending on the mode and lo vs. hi.
26773 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
26775 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
26776 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
26777 Likewise. Require that operands[2] is even.
26778 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
26779 Remove extraneous parens. Require that operands[2] is a multiple
26781 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
26782 operands[0] is a MEM if <mask_applied>, the predicates/constraints
26783 disallow memory then.
26785 2017-03-30 Richard Biener <rguenther@suse.de>
26787 PR tree-optimization/77498
26788 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
26789 to non-constants over backedges.
26791 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
26793 PR rtl-optimization/80233
26794 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
26795 as last_combined_insn. Do not test for BARRIER_P separately.
26797 2017-03-29 Andreas Schwab <schwab@suse.de>
26800 * calls.c (prepare_call_address): Convert funexp to Pmode before
26801 copying to temp reg.
26803 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26805 PR tree-optimization/80158
26806 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
26807 Handle possible future case of more than one alternate
26809 (replace_rhs_if_not_dup): Likewise.
26810 (replace_one_candidate): Likewise.
26812 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
26814 PR rtl-optimization/80193
26815 * ira.c (ira): Do not check allocation for LRA.
26817 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
26819 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
26820 (nvptx_output_simt_exit): Declare.
26821 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
26822 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
26823 (init_softstack_frame): Move initialization of crtl->is_leaf to...
26824 (nvptx_declare_function_name): ...here. Emit declaration of local
26825 memory space buffer for omp_simt_enter insn.
26826 (nvptx_output_unisimt_switch): New.
26827 (nvptx_output_softstack_switch): New.
26828 (nvptx_output_simt_enter): New.
26829 (nvptx_output_simt_exit): New.
26830 * config/nvptx/nvptx.h (struct machine_function): New fields
26831 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
26832 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
26833 (UNSPECV_SIMT_EXIT): Ditto.
26834 (omp_simt_enter_insn): New insn.
26835 (omp_simt_enter): New expansion.
26836 (omp_simt_exit): New insn.
26837 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
26839 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
26840 (expand_GOMP_SIMT_ENTER_ALLOC): New.
26841 (expand_GOMP_SIMT_EXIT): New.
26842 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
26843 (GOMP_SIMT_ENTER_ALLOC): Ditto.
26844 (GOMP_SIMT_EXIT): Ditto.
26845 * target-insns.def (omp_simt_enter): New insn.
26846 (omp_simt_exit): Ditto.
26847 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
26849 (lower_rec_simd_input_clauses): Implement SIMT privatization.
26850 (lower_rec_input_clauses): Likewise.
26851 (lower_lastprivate_clauses): Handle SIMT privatization.
26853 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
26854 (ompdevlow_adjust_simt_enter): New.
26855 (find_simtpriv_var_op): New.
26856 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
26857 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
26859 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
26860 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
26861 (copy_decl_for_dup_finish): Ditto.
26863 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
26865 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
26868 * config/i386/i386.c (ix86_option_override_internal): Always
26869 allow -mpreferred-stack-boundary=3 for 64-bit targets.
26871 2017-03-28 Bin Cheng <bin.cheng@arm.com>
26873 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
26875 2017-03-28 Bin Cheng <bin.cheng@arm.com>
26877 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
26878 mark new edge's irreducible flag accordign to it.
26879 (vect_do_peeling): Check loop preheader edge's irreducible flag
26880 and pass it to function slpeel_add_loop_guard.
26882 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
26884 PR tree-optimization/80218
26885 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
26886 Update block frequencies and counts.
26888 2017-03-28 Richard Biener <rguenther@suse.de>
26890 PR tree-optimization/78644
26891 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
26892 of a simplification result we may not use it at all.
26894 2017-03-28 Richard Biener <rguenther@suse.de>
26897 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
26898 without arguments, generate default definition of a SSA name.
26900 2017-03-28 Richard Biener <rguenther@suse.de>
26902 PR middle-end/80222
26903 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
26904 TYPE_REF_CAN_ALIAS_ALL references.
26905 * fold-const.c (fold_indirect_ref_1): Likewise.
26907 2017-03-28 Martin Liska <mliska@suse.cz>
26910 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
26911 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
26913 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
26914 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
26916 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
26917 (EXTRA_SPECS): Define.
26918 (SUBTARGET_EXTRA_SPECS): Likewise.
26919 (SUBTARGET_CPP_SPEC): Likewise.
26920 * config/arc/elf.h (EXTRA_SPECS): Renamed to
26921 SUBTARGET_EXTRA_SPECS.
26922 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
26924 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
26926 * config/arc/simdext.md (vst64_insn): Update pattern.
26927 (vld32wh_insn): Likewise.
26928 (vld32wl_insn): Likewise.
26929 (vld64_insn): Likewise.
26930 (vld32_insn): Likewise.
26932 2017-03-28 Marek Polacek <polacek@redhat.com>
26935 * fold-const.c (fold_comparison): Use protected_set_expr_location
26936 instead of SET_EXPR_LOCATION.
26938 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
26940 * tree.c (add_expr): Avoid name lookup warning.
26942 2017-03-27 Jeff Law <law@redhat.com>
26944 PR tree-optimization/80216
26945 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
26946 function name. Limit recursion depth.
26947 (record_temporary_equivalences): Corresponding changes.
26949 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
26951 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
26954 2017-03-27 Jakub Jelinek <jakub@redhat.com>
26957 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
26959 * cfgcleanup.c (reg_note_cfa_p): New array.
26960 (insns_have_identical_cfa_notes): New function.
26961 (old_insns_match_p): Don't cross-jump in between /f
26962 and non-/f instructions. If both i1 and i2 are frame related,
26963 verify all CFA notes, their order and content.
26965 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
26968 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
26969 HImode and SImode with zero extend to DImode to one insn.
26970 (bswap<mode>2_extenddi): Likewise.
26971 (bswapsi2_extenddi): Likewise.
26972 (bswaphi2_extendsi): Likewise.
26973 (bswaphi2): Combine bswap HImode and SImode into one insn.
26974 Separate memory insns from swapping register.
26975 (bswapsi2): Likewise.
26976 (bswap<mode>2): Likewise.
26977 (bswaphi2_internal): Delete, no longer used.
26978 (bswapsi2_internal): Likewise.
26979 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
26980 store, and gpr<-gpr swap insns.
26981 (bswap<mode>2_store): Likewise.
26982 (bswaphi2_reg): Register only splitter, combine with the splitter.
26983 (bswaphi2 splitter): Likewise.
26984 (bswapsi2_reg): Likewise.
26985 (bswapsi2 splitter): Likewise.
26986 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
26987 the insns into load, store, and register/register insns.
26988 (bswapdi2_ldbrx): Likewise.
26989 (bswapdi2_load): Likewise.
26990 (bswapdi2_store): Likewise.
26991 (bswapdi2_reg): Likewise.
26993 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
26995 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
26996 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
26998 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
27001 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
27003 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27004 special handling for target option conflicts between dform
27005 options (-mpower9-dform, -mpower9-dform-vector,
27006 -mpower9-dform-scalar) and -mno-direct-move.
27008 2017-03-27 Richard Biener <rguenther@suse.de>
27010 PR tree-optimization/80181
27011 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
27013 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27015 * config/arc/predicates.md (move_double_src_operand): Replace the
27016 call to move_double_src_operand with a call to address_operand.
27018 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27020 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
27021 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
27022 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
27024 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27026 * config/arc/predicates.md (long_immediate_loadstore_operand):
27027 Consider scaled addresses cases.
27029 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27031 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
27032 restored when in interrupt.
27033 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
27034 doesn't have delay slot.
27036 2017-03-27 Richard Biener <rguenther@suse.de>
27039 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
27040 inlined thunk clones.
27042 2017-03-27 Jakub Jelinek <jakub@redhat.com>
27045 * asan.c (instrument_derefs): Copy over last operand from
27046 original COMPONENT_REF to the new COMPONENT_REF with
27047 DECL_BIT_FIELD_REPRESENTATIVE.
27048 * ubsan.c (instrument_object_size): Likewise.
27050 2017-03-27 Richard Biener <rguenther@suse.de>
27052 PR tree-optimization/80170
27053 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
27054 sure DR/SCEV didnt fold in constants we do not see when looking
27055 at the reference base alignment.
27057 2017-03-27 Richard Biener <rguenther@suse.de>
27059 PR middle-end/80171
27060 * gimple-fold.c (fold_ctor_reference): Properly guard against
27061 NULL return value from canonicalize_constructor_val.
27063 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
27066 * config/i386/i386.c (ix86_expand_builtin)
27067 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
27068 flags reg setting and flags reg using instructions.
27069 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
27070 clobbering instructions to zero extend op2.
27072 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
27074 * doc/install.texi (Configuration) <--with-aix-soname>:
27075 Update link to AIX ld.
27077 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
27079 PR rtl-optimization/80160
27080 PR rtl-optimization/80159
27081 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
27082 reg_alternate_class into account.
27084 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
27087 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
27088 to consider in curr_insn_transform.
27090 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27092 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
27093 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
27094 and emit_mode_inner.
27096 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27098 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
27099 argument to the overloaded builtin variants. Use the new flag to
27100 deprecate certain builtin variants.
27101 * config/s390/s390-builtin-types.def: Add new builtin types.
27102 * config/s390/s390-builtins.h: Support new flags field for
27103 overloaded builtins.
27104 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
27105 (s390_macro_to_expand): Enable vector float data type.
27106 (s390_cpu_cpp_builtins_internal): Indicate support of the new
27107 builtins by incrementing the __VEC__ version number.
27108 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
27110 (s390_resolve_overloaded_builtin): Emit error messages depending
27111 on the builtin flags.
27112 * config/s390/s390.c (s390_expand_builtin): Support additional
27113 flags argument. Change error message to match the messages
27114 emitted in s390-c.c.
27115 * config/s390/s390.md: New UNSPEC_* constants.
27116 (op_type): Add new instruction types.
27117 * config/s390/vecintrin.h: Add new builtins and test data class
27119 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
27120 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
27121 (VEC_INEXACT, VEC_NOINEXACT): New constants.
27122 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
27123 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
27124 ("vec_mergel<mode>"): V_HW -> VEC_HW.
27126 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
27127 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
27128 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
27129 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
27131 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
27132 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
27133 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
27134 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
27136 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
27137 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
27138 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
27139 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
27140 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
27141 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
27142 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
27144 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
27145 ("vec_scatter_element<V_HW_4:mode>_DI")
27146 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
27147 ("vec_fpint<mode>", "vflls")
27148 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
27149 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
27150 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
27151 ("*vec_cmphe<mode>_cc"): ... these.
27153 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
27154 mode constant instead of magic value.
27156 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27158 * config/s390/s390.c (s390_expand_vec_compare): Support other
27159 vector floating point modes than just V2DF.
27160 (s390_expand_vcond): Likewise.
27161 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
27162 (s390_cannot_change_mode_class): Prevent mode changes between TF
27163 and V1TF in vector registers.
27164 * config/s390/s390.md (DF, SF): New mode attributes.
27165 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
27166 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
27167 SFmode support for VRs.
27168 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
27170 (VFT, VF_HW): New mode iterators.
27171 (vw, sdx): New mode attributes.
27172 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
27173 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
27174 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
27175 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
27176 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
27177 also the new vector floating point modes. Renaming to ...
27179 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
27180 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
27181 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
27182 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
27183 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
27184 ("vec_unordered<mode>"): ... these.
27186 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
27187 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
27188 ("*vec_extendv2df"): New insn definitions.
27190 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27192 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
27193 ("mulditi3_2", "*muldi3_sign"): New patterns.
27194 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
27195 rename the pattern definition.
27197 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27199 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
27201 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
27203 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27205 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
27206 instruction if possible.
27207 * config/s390/vector.md (vec_halfnumelts): New mode
27209 ("*vec_vllezlf<mode>"): New pattern.
27211 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27213 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
27214 ("popcountv4si2", "popcountv2di2"): Rename to ...
27215 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
27216 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
27218 ("popcount<mode>2_vxe"): New pattern.
27220 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27222 * common/config/s390/s390-common.c (processor_flags_table): Add
27224 * config.gcc: Add arch12.
27225 * config/s390/driver-native.c (s390_host_detect_local_cpu):
27226 Default to arch12 for unknown CPU model numbers.
27227 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
27228 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
27229 PROCESSOR_max sanity check.
27230 * config/s390/s390-opts.h (enum processor_type): Add
27232 * config/s390/s390.c (processor_table): Add arch12.
27233 (s390_expand_builtin): Add check for B_VXE flag.
27234 (s390_issue_rate): Add PROCESSOR_ARCH12.
27235 (s390_get_sched_attrmask): Likewise.
27236 (s390_get_unit_mask): Likewise.
27237 (s390_sched_score): Enable z13 scheduling for arch12.
27238 (s390_sched_reorder): Likewise.
27239 (s390_sched_variable_issue): Likewise.
27240 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
27242 (s390_tune_attr): Use z13 scheduling also for arch12.
27243 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
27244 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
27245 (TARGET_VXE_P): New macros.
27246 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
27247 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
27248 * config/s390/s390.opt: Add arch12 as processor_type.
27250 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27252 * config/s390/s390.md
27253 ("fixuns_truncdddi2", "fixuns_trunctddi2")
27254 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
27255 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
27257 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
27258 Rename expanders to ...
27260 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
27261 ("fixuns_truncdddi2_emu"): ... these.
27263 ("fixuns_trunc<mode>si2_emu"): New expander.
27265 ("*fixuns_truncdfdi2_z13"): Rename to ...
27266 ("*fixuns_truncdfdi2_vx"): ... this.
27268 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27270 * config/s390/2964.md: Remove the single element vector compare
27271 instructions which are no longer used.
27272 * config/s390/s390.c (s390_select_ccmode): Remove handling of
27274 (s390_canonicalize_comparison): Remove handling of DFmode
27276 (s390_expand_vec_compare_scalar): Remove function.
27277 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
27278 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
27280 ("*cmp<mode>_ccs"): Add wfcdb instruction.
27282 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27284 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
27286 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
27287 will anyway by matched by mov<mode>_64dfp.
27289 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27291 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
27292 vlef/vstef. Add missing operand to vleif.
27294 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27296 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
27297 pair for all vector types with 64 bit elements.
27298 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
27299 * config/s390/vector.md (V_HW_64): ... here.
27300 (V_128_NOSINGLE): New mode iterator.
27301 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
27302 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
27303 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
27304 ("*vec_load_pairv2di"): Change to ...
27305 ("*vec_load_pair<mode>"): ... this one.
27307 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27309 * config/s390/constraints.md: Add comments.
27310 (jKK): Reject element sizes > 8 bytes.
27311 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
27313 * config/s390/s390.md: Add the s_operand checks formerly in
27314 s390_split_ok_p to various splitters where they are still
27316 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
27317 for 128 bit vectors. Plus two splitters.
27319 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27321 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
27324 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27327 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
27328 error if the boundary argument is not constant.
27330 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27332 PR rtl-optimization/80112
27333 * loop-doloop.c (doloop_condition_get): Don't check condition
27334 if cmp isn't SET with IF_THEN_ELSE src.
27336 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27338 PR tree-optimization/80158
27339 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
27340 replacing a candidate statement, also replace it for the
27341 candidate's alternate interpretation.
27342 (replace_rhs_if_not_dup): Likewise.
27343 (replace_one_candidate): Likewise.
27345 2017-03-24 Richard Biener <rguenther@suse.de>
27347 PR tree-optimization/80167
27348 * graphite-isl-ast-to-gimple.c
27349 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
27351 (translate_isl_ast_to_gimple::get_rename): Likewise.
27353 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
27355 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
27356 handling of certain combinations of target options, including the
27357 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
27358 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
27360 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27363 * config/arm/arm.md (*load_multiple): Add reload_completed to
27364 matching condition.
27366 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27367 Richard Biener <rguenth@suse.de>
27369 PR tree-optimization/79908
27370 PR tree-optimization/80136
27371 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27372 been cast away, gimplify_and_add suffices.
27374 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
27376 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
27378 2017-03-23 Richard Biener <rguenther@suse.de>
27380 PR tree-optimization/80032
27381 * gimplify.c (gimple_push_cleanup): Forced unconditional
27382 cleanups still have to go to the conditional_cleanups
27385 2017-03-22 Jakub Jelinek <jakub@redhat.com>
27387 PR tree-optimization/80072
27388 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
27390 (next_operand_entry_id): Change type to unsigned int.
27391 (sort_by_operand_rank): Make sure to return the right return value
27392 even if unsigned fields are bigger than INT_MAX.
27393 (struct oecount): Change cnt and id type to unsigned int.
27394 (oecount_hasher::equal): Formatting fix.
27395 (oecount_cmp): Make sure to return the right return value
27396 even if unsigned fields are bigger than INT_MAX.
27397 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
27400 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
27401 TREE_READONLY on result if writing it more than once.
27404 * doc/invoke.texi (-fsanitize=thread): Document that with
27405 -fnon-call-exceptions atomics are not able to throw
27409 * tsan.c: Include tree-eh.h.
27410 (instrument_builtin_call): Call maybe_clean_eh_stmt or
27411 maybe_clean_or_replace_eh_stmt where needed.
27412 (instrument_memory_accesses): Add cfg_changed argument.
27413 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
27414 if it returned true.
27415 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
27417 PR rtl-optimization/63191
27418 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
27419 wrapper function, moved the whole old content into ...
27420 (ix86_delegitimize_address_1): ... this. New inline function.
27421 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
27422 true as last argument instead of ix86_delegitimize_address.
27424 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
27426 * config/aarch64/aarch64.c (generic_branch_cost): Copy
27427 cortexa57_branch_cost.
27429 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
27431 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
27433 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27436 * doc/md.texi (Constraints): Document wA constraint.
27437 * config/rs6000/constraints.md (wA): New.
27438 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
27439 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
27440 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
27441 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
27443 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
27446 * gimplify.c (is_oacc_declared): New function.
27447 (oacc_default_clause): Use it to set default flags for acc declared
27448 variables inside parallel regions.
27449 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
27450 declared variables.
27451 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
27452 declare attribute to any decl as necessary.
27454 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
27457 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
27458 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
27459 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
27460 (arm_arch_lpae): This.
27461 * config/arm/arm.c (arm_arch7ve): Rename into ...
27462 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
27463 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
27466 2017-03-22 Martin Liska <mliska@suse.cz>
27469 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
27470 error message instead of an ICE.
27472 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27474 * doc/extend.texi (6.11 Additional Floating Types): Revise.
27476 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
27478 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
27480 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27483 2017-03-21 Martin Sebor <msebor@redhat.com>
27485 * doc/extend.texi: Use "cannot" instead of "can't."
27486 * doc/hostconfig.texi: Same.
27487 * doc/install.texi: Same.
27488 * doc/invoke.texi: Same.
27489 * doc/loop.texi: Same.
27490 * doc/md.texi: Same.
27491 * doc/objc.texi: Same.
27492 * doc/rtl.texi: Same.
27493 * doc/tm.texi: Same.
27494 * doc/tm.texi.in: Same.
27495 * doc/trouble.texi: Same.
27497 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
27500 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
27501 (collect_checksum_attributes): Set it.
27502 (die_checksum_ordered): Use it.
27504 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27506 PR tree-optimization/79908
27507 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
27508 change: For a VA_ARG whose LHS has been cast away, use
27509 force_gimple_operand to construct the side effects.
27511 2017-03-21 David Malcolm <dmalcolm@redhat.com>
27513 PR translation/80001
27514 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
27515 more amenable to translation.
27516 (oacc_loop_auto_partitions): Likewise.
27518 2017-03-21 Marek Polacek <polacek@redhat.com>
27519 Martin Sebor <msebor@redhat.com>
27521 PR tree-optimization/80109
27522 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
27523 on INTEGRAL_TYPE_P.
27525 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27526 Segher Boessenkool <segher@kernel.crashing.org>
27529 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
27530 check reg_used_between_p between insn and one of succ or succ2
27531 depending on if succ is artificial insn not inserted into insn
27534 2017-03-21 Martin Liska <mliska@suse.cz>
27536 PR gcov-profile/80081
27537 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
27538 * doc/gcc.texi: Include gcov-dump stuff.
27539 * doc/gcov-dump.texi: New file.
27541 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
27543 PR rtl-optimization/79150
27544 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
27545 conditional jump, if the jump is the last insn of the loop.
27547 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27548 Richard Biener <rguenth@suse.de>
27550 PR tree-optimization/79908
27551 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27552 been cast away, use force_gimple_operand to construct the side
27555 2017-03-21 Martin Liska <mliska@suse.cz>
27557 PR libfortran/79956
27558 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
27561 2017-03-21 Brad Spengler <spender@grsecurity.net>
27564 * plugin.c (htab_hash_plugin): New function.
27565 (add_new_plugin): Use it and adjust.
27566 (parse_plugin_arg_opt): Adjust.
27567 (init_one_plugin): Likewise.
27569 2017-03-21 Richard Biener <rguenther@suse.de>
27571 PR tree-optimization/80032
27572 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
27573 if set force the cleanup to happen unconditionally.
27574 (gimplify_target_expr): Push inserted clobbers with force_uncond
27575 to avoid them being removed by control-dependent DCE.
27577 2017-03-21 Richard Biener <rguenther@suse.de>
27579 PR tree-optimization/80122
27580 * tree-inline.c (copy_bb): Do not expans va-arg packs or
27581 va_arg_pack_len when the inlined call stmt requires pack
27583 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
27585 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27588 * tsan.c (instrument_builtin_call): If the memory model argument
27589 is not a constant, assume it is valid.
27592 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
27595 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27597 PR rtl-optimization/79910
27598 * combine.c (can_combine_p): Do not allow combining an I0 or I1
27599 if its dest is used by an insn before I2 (other than the combined
27600 insns themselves, which are properly handled already).
27602 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27605 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
27607 * combine.c (record_used_regs): New static function.
27608 (try_combine): Handle situations where there is an additional
27609 instruction between I2 and I3 which needs to have a LOG_LINK
27613 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27615 * combine.c (try_combine): Delete redundant i1 test. Call
27616 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27618 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27621 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
27622 alternatives 13/14.
27624 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27626 PR tree-optimization/80054
27627 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
27628 the optimization if a PHI or any of its arguments is not dominated
27629 by the candidate's basis. Use gphi* rather than gimple* as
27631 (replace_profitable_candidates): Clean up a gimple* variable that
27632 should be a gphi* variable.
27634 2017-03-20 Martin Sebor <msebor@redhat.com>
27637 * doc/extend.texi (attribute constructor): Document present limitation.
27639 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
27642 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
27643 __POWER9_VECTOR__ #ifdef control, change template definition to
27644 use Power9-specific built-in function.
27645 (vec_any_eq): Likewise.
27646 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
27647 to control outcomes from this test.
27648 (vector_ae_<mode>p): For VEC_F modes, likewise.
27650 2017-03-20 Ian Lance Taylor <iant@google.com>
27652 * config/i386/i386.c (ix86_function_regparm): Save an extra
27653 register for -fsplit-stack with DECL_STATIC_CHAIN.
27655 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
27658 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
27659 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
27661 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
27663 * config/riscv/riscv.c (riscv_print_operand): Use "fence
27665 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
27668 2017-03-20 Marek Polacek <polacek@redhat.com>
27671 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
27673 2017-03-20 Richard Biener <rguenther@suse.de>
27675 PR tree-optimization/80113
27676 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
27677 allocate extra SSA name for PHI def.
27678 (add_close_phis_to_outer_loops): Likewise.
27679 (add_close_phis_to_merge_points): Likewise.
27680 (copy_loop_close_phi_args): Likewise.
27681 (copy_cond_phi_nodes): Likewise.
27683 2017-03-20 Martin Liska <mliska@suse.cz>
27685 PR middle-end/79753
27686 * tree-chkp.c (chkp_build_returned_bound): Do not build
27687 returned bounds for a LHS that's not a BOUNDED_P type.
27689 2017-03-20 Martin Liska <mliska@suse.cz>
27693 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
27694 COMPLEX_CST and VECTOR_CST.
27696 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27699 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
27700 target operand. A new splitter adds the clobber statement in case
27701 the target operand is dead anyway.
27703 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
27705 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
27706 to age-old versions of binutils and glibc.
27708 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
27710 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
27712 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
27714 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
27716 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
27718 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
27719 requirement for binutils 2.13.
27721 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27723 * combine.c (try_combine): Delete redundant i1 test. Call
27724 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27726 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
27728 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
27729 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
27731 <riscv64-*-elf>: Re-arrange section
27732 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
27733 <riscv32-*-linux>: Likewise.
27734 <riscv64-*-elf>: Likewise
27735 <riscv64-*-linux>: Likewise.
27737 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
27740 * aarch64.opt(verbose-cost-dump): Fix typo.
27742 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
27745 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
27746 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
27748 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
27750 * reload.c (find_reloads): When reloading a nonoffsettable address,
27751 use RELOAD_OTHER for it and its address reloads.
27753 PR rtl-optimization/79910
27754 * combine.c (record_used_regs): New static function.
27755 (try_combine): Handle situations where there is an additional
27756 instruction between I2 and I3 which needs to have a LOG_LINK
27759 2017-03-17 Jeff Law <law@redhat.com>
27761 PR tree-optimization/71437
27762 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
27763 conditional in the hash table first.
27764 (vrp_dom_walker::before_dom_children): Extract condition from
27765 ASSERT_EXPR. Record condition, its inverion and any implied
27766 conditions as well.
27768 2017-03-17 Marek Polacek <polacek@redhat.com>
27769 Markus Trippelsdorf <markus@trippelsdorf.de>
27771 PR tree-optimization/80079
27772 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
27775 2017-03-17 Richard Biener <rguenther@suse.de>
27777 PR middle-end/80075
27778 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
27779 Properly verify the LHS before the RHS possibly claims to be
27781 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
27784 2017-03-17 Martin Jambor <mjambor@suse.cz>
27786 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
27787 (List of -O2 options): Likewise.
27788 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
27791 2017-03-17 Tom de Vries <tom@codesourcery.com>
27793 * gcov-dump.c (print_usage): Print bug_report_url.
27795 2017-03-17 Richard Biener <rguenther@suse.de>
27797 PR middle-end/80050
27798 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
27799 (parser::peek): Likewise.
27801 2017-03-17 Richard Biener <rguenther@suse.de>
27803 PR tree-optimization/80048
27804 * sese.c (free_sese_info): Properly release rename_map and
27805 copied_bb_map elements.
27807 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
27809 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
27810 Add linked-list forward and backlinks. Insert on
27811 construction, remove on destruction.
27812 (class pass_store_merging): Add m_stores_head field.
27813 (pass_store_merging::terminate_and_process_all_chains):
27814 Iterate over m_stores_head list.
27815 (pass_store_merging::terminate_all_aliasing_chains):
27817 (pass_store_merging::execute): Check for debug stmts first.
27818 Push new chains onto the m_stores_head stack.
27820 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
27823 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
27824 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
27825 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
27827 2017-03-16 Jeff Law <law@redhat.com>
27829 PR tree-optimization/71437
27830 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
27831 member function. Implementation moved into after_dom_children
27832 member function and into the threader's thread_outgoing_edges
27834 (dom_opt_dom_walker::after_dom_children): Simplify by moving
27835 some code into new thread_outgoing_edges.
27836 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
27837 definition. Simplify marker handling (do it here). Assume we always
27838 have the available expression and the const/copies tables.
27839 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
27841 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
27842 * tree-vrp.c (equiv_stack): No longer file scoped.
27843 (vrp_dom_walker): New class.
27844 (vrp_dom_walker::before_dom_children): New member function.
27845 (vrp_dom_walker::after_dom_children): Likewise.
27846 (identify_jump_threads): Setup domwalker. Use it rather than
27847 walking edges in a random order by hand. Simplify setup/finalization.
27848 (finalize_jump_threads): Remove.
27849 (vrp_finalize): Do not call identify_jump_threads here.
27850 (execute_vrp): Do it here instead and call thread_through_all_blocks
27853 PR tree-optimization/71437
27854 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
27856 (simplify_stmt_for_jump_threading): Add basic_block argument. All
27858 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
27859 (dom_opt_dom_walker::thread_across_edge): Remove
27860 handle_dominating_asserts argument. All callers changed.
27861 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
27862 changes. Remove calls to lhs_of_dominating_assert. Other
27863 uses of handle_dominating_asserts turn into unconditional code
27864 (simplify_control_stmt_condition_1): Likewise.
27865 (simplify_control_stmt_condition): Likewise.
27866 (thread_through_normal_block, thread_across_edge): Likewise.
27867 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
27868 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
27869 object if it is not an SSA_NAME.
27870 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
27871 before calling into the VRP specific simplifiers.
27872 (identify_jump_threads): Remove handle_dominating_asserts
27875 2017-03-16 Jakub Jelinek <jakub@redhat.com>
27878 * tree-diagnostic.c (default_tree_printer): No longer static.
27879 * tree-diagnostic.h (default_tree_printer): New prototype.
27881 2017-03-16 Tamar Christina <tamar.christina@arm.com>
27883 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
27884 Change ins into fmov.
27886 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27888 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
27889 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
27890 Use h_con constraint for operand 1.
27891 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
27892 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
27894 2017-03-15 Jeff Law <law@redhat.com>
27896 PR tree-optimization/71437
27897 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
27898 (record_temporary_equivalences): Use it.
27900 PR tree-optimization/71437
27901 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
27902 tree-ssa-scopedtables.
27903 (lookup_avail_expr, build_and_record_new_cond): Likewise.
27904 (record_conditions, record_cond, vuse_eq): Likewise.
27905 (record_edge_info): Adjust to API tweak of record_conditions.
27906 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
27907 (record_temporary_equivalences, optimize_stmt): Likewise.
27908 (eliminate_redundant_computations): Likewise.
27909 (record_equivalences_from_stmt): Likewise.
27910 * tree-ssa-scopedtables.c: Include options.h and params.h.
27911 (vuse_eq): New function, moved from tree-ssa-dom.c
27912 (build_and_record_new_cond): Likewise.
27913 (record_conditions): Likewise. Accept vector of conditions rather
27914 than edge_equivalence structure for first argument.
27915 for the first argument.
27916 (avail_exprs_stack::lookup_avail_expr): New member function, moved
27917 from tree-ssa-dom.c.
27918 (avail_exprs_stack::record_cond): Likewise.
27919 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
27920 from tree-ssa-dom.c.
27921 (avail_exprs_stack): Add new member functions lookup_avail_expr
27923 (record_conditions): Declare.
27925 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
27928 * lra-constraints.c (process_alt_operands): Increase reject for
27929 reloading an input/output operand.
27931 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
27934 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
27935 insns to convert from signed/unsigned char/short to IEEE 128-bit
27937 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
27939 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
27942 * config/i386/i386.c (ix86_vector_duplicate_value): Create
27943 subreg of inner mode for values already in registers.
27945 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
27947 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
27948 iteration reg is used after the loop.
27950 2017-03-14 Martin Sebor <msebor@redhat.com>
27952 PR tree-optimization/79800
27953 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
27954 precision in negative-positive range.
27955 (format_floating): Call non-const overload with adjusted precision.
27957 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
27960 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
27963 2017-03-14 Martin Sebor <msebor@redhat.com>
27965 PR middle-end/80020
27966 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
27967 * builtins.def (aligned_alloc): Use it.
27970 * Makefile.in (GTFILES): Add calls.c.
27971 * calls.c: Include "gt-calls.h".
27973 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
27975 PR rtl-optimization/79728
27976 * regs.h (struct target_regs): New field
27977 x_contains_allocatable_regs_of_mode.
27978 (contains_allocatable_regs_of_mode): New macro.
27979 * reginfo.c (init_reg_sets_1): Initialize it, and change
27980 contains_reg_of_mode so it includes global regs as well.
27981 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
27982 rather than contains_regs_of_mode.
27984 2017-03-14 Martin Liska <mliska@suse.cz>
27986 * doc/invoke.texi: Document options that can't be combined with
27987 -fcheck-pointer-bounds.
27989 2017-03-14 Martin Liska <mliska@suse.cz>
27991 PR middle-end/79831
27992 * doc/invoke.texi (-Wchkp): Document the option.
27994 2017-03-14 Martin Liska <mliska@suse.cz>
27996 * Makefile.in: Install gcov-dump.
27998 2017-03-14 Martin Liska <mliska@suse.cz>
28000 * multiple_target.c (expand_target_clones): Bail out for
28001 an invalid attribute.
28003 2017-03-14 Richard Biener <rguenther@suse.de>
28005 * alias.c (struct alias_set_entry): Pack properly.
28006 * cfgloop.h (struct loop): Likewise.
28007 * cse.c (struct set): Likewise.
28008 * ipa-utils.c (struct searchc_env): Likewise.
28009 * loop-invariant.c (struct invariant): Likewise.
28010 * lra-remat.c (struct cand): Likewise.
28011 * recog.c (struct change_t): Likewise.
28012 * rtl.h (struct address_info): Likewise.
28013 * symbol-summary.h (function_summary): Likewise.
28014 * tree-loop-distribution.c (struct partition): Likewise.
28015 * tree-object-size.c (struct object_size_info): Likewise.
28016 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
28017 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
28018 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
28019 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
28020 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
28021 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
28022 (struct _stmt_vec_info): Likewise.
28024 2017-03-14 Martin Liska <mliska@suse.cz>
28027 * multiple_target.c (create_dispatcher_calls): Check that
28028 a target can create a function dispatcher.
28030 2017-03-14 Martin Liska <mliska@suse.cz>
28033 * multiple_target.c (expand_target_clones): Drop local.local
28034 flag for default implementation.
28036 2017-03-14 Richard Biener <rguenther@suse.de>
28038 PR tree-optimization/80030
28039 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
28041 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
28043 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
28044 gcc_fallthrough() instead of __attribute__((fallthrough));
28046 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28048 * doc/gcc.texi: Remove "up" link to (DIR).
28049 * doc/gccint.texi: Ditto.
28051 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28053 * doc/install.texi (Specific) <avr>: Remove reference to
28056 2017-03-13 Jeff Law <law@redhat.com>
28058 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
28059 attribute rather than comments.
28061 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
28062 match_scratch operand is highest.
28064 2017-03-13 Martin Liska <mliska@suse.cz>
28066 PR middle-end/78339
28067 * ipa-pure-const.c (warn_function_noreturn): If the declarations
28068 is a CHKP clone, use original declaration.
28070 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28072 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
28073 (arc_conditional_register_usage): Use a different allocation order
28074 when optimizing for size.
28075 * common/config/arc/arc-common.c (arc_option_optimization_table):
28076 Section anchors default on when optimizing for size.
28078 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28080 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
28082 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28084 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
28085 * config/arc/arc.md (cpu_facility): Add cd variant.
28086 (*movqi_insn): Add code density variant.
28087 (*movhi_insn): Likewise.
28088 (*movqi_insn): Likewise.
28089 (*addsi3_mixed): Likewise.
28090 (subsi3_insn): Likewise.
28092 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28094 * config/arc/arc.md (movsi_cond_exec): Update constraint.
28096 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28098 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
28099 expressions with MINUS and UNARY ops.
28101 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28104 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
28106 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
28107 between vec_select and vector argument.
28108 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
28109 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
28110 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
28111 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
28112 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
28113 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
28115 2017-03-13 Richard Biener <rguenther@suse.de>
28118 * params.def (vect-max-peeling-for-alignment): Fix typo.
28120 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28122 * doc/install.texi (Specific) <mips-*-*>: Remove description of
28123 issue that only occurred with binutils below 2.18.
28125 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28127 * doc/install.texi (Specific) <cris-axis-elf>: No longer
28128 refer to binutils 2.11/2.12 minimum.
28130 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28132 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
28133 ftp.kernel.org and simplify binutils requirement.
28135 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
28137 * doc/invoke.texi (Warning Options): Fix spelling of link-time
28139 (Optimize Options): Ditto. Also remove redundancy.
28141 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28143 PR translation/79848
28144 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
28146 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
28147 to G_ to avoid double translation.
28149 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28151 PR translation/79923
28152 * auto-profile.c (get_combined_location): Convert leading
28153 character of diagnostics to lower case and remove trailing period.
28154 (read_profile): Likewise for various diagnostics.
28155 * config/arm/arm.c (arm_option_override): Remove trailing period
28156 from various diagnostics.
28157 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
28158 (msp430_expand_delay_cycles): Likewise.
28160 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28163 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
28164 full command-line argument, rather than just "str".
28165 (aarch64_validate_march): Likewise.
28166 (aarch64_validate_mtune): Likewise.
28168 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
28170 PR rtl-optimization/78911
28171 * lra-assigns.c (must_not_spill_p): New function.
28172 (spill_for): Use it.
28174 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28176 PR tree-optimization/79981
28177 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
28178 ATOMIC_COMPARE_EXCHANGE ifn result.
28179 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
28180 IFN_ATOMIC_COMPARE_EXCHANGE.
28182 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28185 * opts.c (parse_sanitizer_options): Add missing question mark to
28186 "did you mean" message.
28188 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28190 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
28192 (VMULEUH_UNS): Likewise.
28193 (VMULOUB_UNS): Likewise.
28194 (VMULOUH_UNS): Likewise.
28195 * config/rs6000/rs6000.c (builtin_function_type): Remove
28196 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
28198 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28201 * read-rtl-function.c (function_reader::read_rtx_operand): Update
28202 x with result of extra_parsing_for_operand_code_0.
28203 (function_reader::extra_parsing_for_operand_code_0): Convert
28204 return type from void to rtx, returning x. When reading
28205 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
28206 larger size containing struct block_symbol.
28208 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
28210 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
28211 -mfloat128-hardware without -m64.
28213 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
28216 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
28217 entries to the case statement that marks unsigned arguments to
28218 overloaded functions.
28220 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
28222 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
28223 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
28225 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
28228 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
28229 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
28231 2017-03-10 Martin Liska <mliska@suse.cz>
28235 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
28236 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
28239 2017-03-10 Olivier Hainque <hainque@adacore.com>
28241 * tree-switch-conversion (array_value_type): Start by resetting
28242 candidate type to it's main variant.
28244 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28246 PR rtl-optimization/79909
28247 * combine.c (try_combine): Use simplify_replace_rtx on individual
28248 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
28249 of the whole CALL_INSN_FUNCTION_USAGE.
28251 PR tree-optimization/79972
28252 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
28253 get_range_info on SSA_NAMEs. Formatting fixes.
28255 2017-03-10 Richard Biener <rguenther@suse.de>
28256 Jakub Jelinek <jakub@redhat.com>
28258 PR tree-optimization/77975
28259 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
28260 edge to be constant.
28261 (get_val_for): For constant x return it. Formatting fix.
28262 (loop_niter_by_eval): Avoid pointless looping if the next iteration
28263 would use the same bases as the current one.
28265 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28267 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
28268 instead of vec_select for V1TImode.
28269 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
28271 (VSX_LE_128): Add V1TI to this mode iterator.
28272 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
28273 (*vsx_le_perm_store_<mode>): Likewise.
28274 (pre-reload splitter for VSX stores): Likewise.
28275 (post-reload splitter for VSX stores): Likewise.
28276 (*vsx_xxpermdi2_le_<mode>): Likewise.
28277 (*vsx_lxvd2x2_le_<mode>): Likewise.
28278 (*vsx_stxvd2x2_le_<mode>): Likewise.
28280 2017-03-09 Michael Eager <eager@eagercon.com>
28282 Correct failures with --enable-checking=yes,rtl.
28284 * config/microblaze/microblaze.c (microblaze_expand_shift):
28285 Replace GET_CODE test with CONST_INT_P and INTVAL test with
28286 test for const0_rtx.
28287 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
28288 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
28290 2017-03-09 Richard Biener <rguenther@suse.de>
28292 PR tree-optimization/79977
28293 * graphite-scop-detection.c (scop_detection::merge_sese):
28294 Handle the case of extra exits to blocks dominating the entry.
28296 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
28298 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28301 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
28303 PR rtl-optimization/79949
28304 * lra-constraints.c (process_alt_operands): Check memory when
28305 trying to predict a cycle. Print about the overall increase.
28307 2017-03-09 Richard Biener <rguenther@suse.de>
28309 PR middle-end/79971
28310 * gimple-expr.c (useless_type_conversion_p): Preserve
28311 TYPE_SATURATING for fixed-point types.
28313 2017-03-09 Richard Biener <rguenther@suse.de>
28316 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
28317 alignment of BLKmode params.
28319 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28322 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
28323 (VALL_NO_V2Q): Likewise.
28325 * config/aarch64/aarch64-simd.md
28326 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
28328 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
28329 VALL_NO_V2Q mode iterator.
28330 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
28332 2017-03-09 Martin Liska <mliska@suse.cz>
28334 PR tree-optimization/79631
28335 * tree-chkp-opt.c (chkp_is_constant_addr): Call
28336 tree_int_cst_sign_bit just for INTEGER constants.
28338 2017-03-09 Martin Liska <mliska@suse.cz>
28342 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
28345 2017-03-09 Marek Polacek <polacek@redhat.com>
28348 * tree.c (inchash::add_expr): Handle TREE_VEC.
28350 2017-03-09 Martin Liska <mliska@suse.cz>
28353 (chkp_narrow_size_and_offset): New function.
28354 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
28355 (void chkp_parse_bit_field_ref): New function.
28356 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
28357 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
28359 2017-03-09 Martin Liska <mliska@suse.cz>
28362 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
28363 (chkp_find_bounds_1): Remove gcc_unreachable.
28365 2017-03-09 Jakub Jelinek <jakub@redhat.com>
28368 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
28369 BUILT_IN_SYNC*, determine the access type from the size suffix and
28370 always build a MEM_REF with that type. Handle forgotten
28371 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
28374 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
28375 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
28376 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
28377 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
28378 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
28379 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
28380 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
28381 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
28382 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
28383 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
28384 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
28385 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
28386 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
28387 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
28388 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
28389 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
28390 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
28391 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
28392 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
28393 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
28394 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
28395 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
28396 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
28397 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
28398 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
28399 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
28400 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
28401 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
28402 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
28403 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
28404 definitions outside of __OPTIMIZE__ guarded section.
28407 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
28408 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
28409 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
28410 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
28413 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28415 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
28416 ("vfenez<mode>"): Add missing constraints.
28418 2017-03-08 Martin Sebor <msebor@redhat.com>
28421 * config/nds32/nds32.c (nds32_option_override):
28422 Fix misspelled diagnostic.
28424 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28427 * gimplify.c (gimplify_omp_for): Replace index var in outer
28428 taskloop statement with an artificial variable and add
28429 OMP_CLAUSE_PRIVATE clause for it.
28431 2017-03-08 Richard Biener <rguenther@suse.de>
28433 PR tree-optimization/79955
28434 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
28435 for accesses that are completely outside of the variable.
28437 2017-03-08 Andrew Haley <aph@redhat.com>
28439 PR tree-optimization/79943
28440 * tree-ssa-loop-split.c (compute_new_first_bound): When
28441 calculating the new upper bound, (END-BEG) should be added, not
28444 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28446 * config/avr/avr.md (setmemhi): Make sure match_dup
28447 operand number comes before match_scratch.
28449 2017-03-08 Richard Biener <rguenther@suse.de>
28451 PR tree-optimization/79920
28452 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
28453 with ncopies == 1 to ...
28454 (vect_transform_slp_perm_load): ... here. Properly compute
28455 all element loads by iterating VF times over the group. Do
28456 not handle ncopies (computed in a broken way) in
28457 vect_create_mask_and_perm.
28459 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28462 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
28463 is a uniform vector, use uniform_vector_p return value instead of
28464 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
28466 2017-03-07 Marek Polacek <polacek@redhat.com>
28468 PR middle-end/79809
28469 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
28470 (alloca_call_type): Likewise.
28472 2017-03-07 Martin Liska <mliska@suse.cz>
28474 * gcov.c (process_args): Put comment to correct location.
28476 2017-03-07 Martin Liska <mliska@suse.cz>
28478 PR middle-end/68270
28479 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
28480 Use array_at_struct_end_p instead of DECL_CHAIN (field).
28481 (chkp_narrow_bounds_for_field): Likewise.
28482 (chkp_parse_array_and_component_ref): Pass one more argument to
28485 2017-03-07 Richard Biener <rguenther@suse.de>
28487 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
28490 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
28492 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
28493 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
28495 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28498 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
28499 to end of description.
28500 (PARAM_MAX_STORES_TO_MERGE): Likewise.
28502 2017-03-07 Jakub Jelinek <jakub@redhat.com>
28504 PR rtl-optimization/79901
28505 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
28507 (*avx512f_<code><mode>3<mask_name>): ... this.
28508 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
28509 iterator instead of VI8_AVX2_AVX512BW.
28511 PR rtl-optimization/79901
28512 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
28513 min/max expander, expand it using expand_vec_cond_expr.
28516 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
28519 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28522 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
28523 to void * for PCH reasons.
28524 * dwarf2out.c (output_loc_operands, output_die): Cast
28525 v.val_vec.array to unsigned char *.
28527 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
28530 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
28533 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
28535 PR rtl-optimization/79571
28536 * lra-constraints.c (process_alt_operands): Calculate static
28537 reject and subtract it from overall when only addresses will be
28540 2017-03-06 Julia Koval <julia.koval@intel.com>
28543 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
28544 incoming stack boundary to 128 for 64-bit targets.
28546 2017-03-06 Richard Biener <rguenther@suse.de>
28548 PR tree-optimization/79894
28549 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
28550 to NULL after folding it.
28552 2017-03-06 Richard Biener <rguenther@suse.de>
28554 PR tree-optimization/79824
28555 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
28556 check disabling peeling for gaps.
28558 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
28560 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28561 attributes): Document gettimeofday.
28563 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28565 * config/s390/s390.c (s390_option_override_internal): Set
28566 PARAM_MIN_VECT_LOOP_BOUND
28568 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28570 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
28571 * config/s390/s390.md: Likewise.
28573 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28576 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
28577 (<avx2_avx512>_perm<mode>): Rename to ...
28578 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
28580 (<avx512>_perm<mode>_mask): Rename to ...
28581 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
28583 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
28584 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
28585 instead of VI8F_256_512.
28586 (avx512f_perm<mode>): New define_expand.
28587 (avx512f_perm<mode>_mask): Likewise.
28588 (avx512f_perm<mode>_1<mask_name>): New define_insn.
28589 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
28591 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28593 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
28594 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
28596 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
28598 2017-03-06 Martin Liska <mliska@suse.cz>
28601 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
28602 when having a SSA NAME w/o VAR_DECL assigned to it.
28604 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28606 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
28607 msa_dpsub_<su>_d): Fix MODE for vec_select.
28609 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28611 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
28613 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
28615 2017-03-06 Richard Biener <rguenther@suse.de>
28617 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
28618 * plugin.c (register_plugin_info): Likewise.
28619 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
28621 2017-03-05 Jakub Jelinek <jakub@redhat.com>
28623 * config/i386/sse.md (sse_storehps, sse_storelps,
28624 avx_<castmode><avxsizesuffix>_<castmode>,
28625 avx512f_<castmode><avxsizesuffix>_<castmode>,
28626 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
28627 in condition that at least one operand is not a MEM.
28629 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28631 PR middle-end/79805
28632 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
28633 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
28635 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
28636 gimple_call_nothrow_p flag based on whether original builtin can throw.
28637 If it can, emit following stmts on the fallthrough edge.
28638 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
28639 don't create new bb if inserting just debug stmts on the edge, try to
28640 insert them on the fallthru bb or just reset debug stmts.
28642 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
28645 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
28646 restore recog_data (including the operand rtxes inside it) around
28647 the call to get_insn_template.
28649 2017-03-03 Martin Sebor <msebor@redhat.com>
28651 PR tree-optimization/79699
28652 * context.c (context::~context): Free MPFR caches to avoid
28653 a memory leak on program exit.
28655 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28657 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
28658 Use wide_int::ulow () instead of .elt (0).
28660 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
28662 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
28663 (*pushxf): Limit oF constraint to 32bit targets and add oC
28664 constraint for 64bit targets.
28665 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
28666 (*pushdf): Change rmF constraint to rmC.
28668 2017-03-03 Martin Liska <mliska@suse.cz>
28670 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
28671 Remove unused variable.
28673 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28676 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
28677 is a memory operand, increase num_memory.
28678 (ix86_expand_args_builtin): Likewise.
28680 2017-03-03 Jan Hubicka <jh@suse.cz>
28683 * ipa-devirt.c (maybe_record_node): Properly handle
28684 __cxa_pure_virtual visibility.
28686 2017-03-03 Martin Liska <mliska@suse.cz>
28688 PR tree-optimization/79803
28689 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
28691 (pass_loop_prefetch::execute): Disabled optimization if an
28692 assumption about L1 cache size is not met.
28694 2017-03-03 Martin Liska <mliska@suse.cz>
28696 PR rtl-optimization/79574
28697 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
28698 (hash_scan_set): Likewise.
28699 (dump_hash_table): Likewise.
28700 (hoist_code): Likewise.
28702 2017-03-03 Richard Biener <rguenther@suse.de>
28704 * fixed-value.c (fixed_from_string): Restore use of elt (1)
28705 in place of uhigh ().
28706 (fixed_convert_from_real): Likewise.
28708 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
28711 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
28713 2017-03-03 Richard Biener <rguenther@suse.de>
28715 PR middle-end/79818
28716 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
28717 TYPE_OVERFLOW_UNDEFINED check.
28719 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28721 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
28723 (vector_ae_<mode>_p): Likewise.
28724 (vector_nez_<mode>_p): Likewise.
28725 (vector_ne_v2di_p): Likewise.
28726 (vector_ae_v2di_p): Likewise.
28727 (vector_ne_<mode>_p): Likewise.
28728 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
28730 (vsx_tsqrt<mode>2_fe): Likewise.
28732 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
28735 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
28737 2017-03-02 Jakub Jelinek <jakub@redhat.com>
28739 PR rtl-optimization/79780
28740 * cprop.c (one_cprop_pass): When second and further conditional trap
28741 in a single basic block is turned into an unconditional trap, turn it
28742 into a deleted note to avoid RTL verification failures.
28744 2017-03-02 Richard Biener <rguenther@suse.de>
28746 * fold-const.c (const_binop): Use ulow () instead of elt (0).
28748 2017-03-02 Richard Biener <rguenther@suse.de>
28750 PR tree-optimization/79345
28752 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
28753 param and abort the walk, returning -1 if it is hit.
28754 (walk_aliased_vdefs): Take a limit param and pass it on.
28755 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
28756 defaulting to 0 and return a signed int.
28757 * tree-ssa-uninit.c (struct check_defs_data): New struct.
28758 (check_defs): New helper.
28759 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
28760 about uninitialized memory.
28761 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
28762 bogus uninitialized warning.
28763 (fixed_convert_from_real): Likewise.
28765 2017-03-02 Bin Cheng <bin.cheng@arm.com>
28767 PR tree-optimization/66768
28768 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
28769 iv_use if base object can't be determined.
28771 2017-03-02 Jakub Jelinek <jakub@redhat.com>
28773 PR tree-optimization/79345
28774 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
28775 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
28776 (get_pattern_stats): Initialize it.
28777 * genemit.c (gen_expand): Verify match_scratch numbers come after
28778 match_operand/match_dup numbers.
28779 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
28780 match_scratch numbers.
28781 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
28783 * config/s390/s390.md (trunctdsd2): Likewise.
28785 2017-03-02 Richard Biener <rguenther@suse.de>
28787 * wide-int.h (wide_int_storage::operator=): Implement in terms
28790 2017-03-02 Richard Biener <rguenther@suse.de>
28792 PR tree-optimization/79777
28793 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
28794 the to insert expression to sth existing.
28796 2017-03-01 Martin Sebor <msebor@redhat.com>
28798 PR middle-end/79692
28799 * gimple-ssa-sprintf.c
28800 (directive::known_width_and_precision): New function.
28801 (format_integer): Use it.
28802 (get_mpfr_format_length): Consider the full range of precision
28803 when computing %g output with the # flag. Set the likely byte
28804 count to 3 rather than 1 when precision is indeterminate.
28805 (format_floating): Correct the lower bound of precision.
28807 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28809 * doc/invoke.texi: Document default code model for 64-bit Linux.
28811 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28814 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
28815 udiv rather than div since input pattern is unsigned.
28817 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
28819 * config/i386/i386.c (print_reg): Warn for values of
28820 unsupported size in integer register.
28822 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
28825 * config/rs6000/predicates.md (current_file_function_operand): Do
28826 not allow self calls to be local if the function is replaceable.
28828 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
28831 * config/rs6000/altivec.h (vec_ctz and others): Change the
28832 preprocessor macro that controls conditional compilation from
28833 _ARCH_PWR9 to __POWER9_VECTOR__.
28834 (vec_all_ne): Change parameterization of __altivec_scalar_pred
28835 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
28836 control (instead of _ARCH_PWR9 control) so that template
28837 definition uses power9-specific function.
28838 (vec_any_eq): Likewise.
28839 (vec_all_ne): Change macro definition to use a power9-specific
28840 expansion under #ifdef __POWER9_VECTOR__ control (instead of
28841 _ARCH_PWR9 control).
28842 (vec_any_eq) Likewise.
28843 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
28844 expansion for CMPNEF to remove support for xvcmpnesp instruction.
28845 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
28846 support for xvcmpnedp instruction.
28847 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
28848 macro expansion so that Power9 implementation of vec_all_ne does
28849 not use the AltiVec predicate framework.
28850 (VCMPNEH_P): Likewise.
28851 (VCMPNEW_P): Likewise.
28852 (VCMPNED_P): Likewise.
28853 (VCMPNEFP_P): Likewise.
28854 (VCMPNEDP_P): Likewise.
28855 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
28856 implementation of vec_any_eq to not use AltiVec predicate
28858 (VCMPAEH_P): Likewise.
28859 (VCMPAEW_P): Likewise.
28860 (VCMPAED_P): Likewise.
28861 (VCMPAEFP_P): Likewise.
28862 (VCMPAEDP_P): Likewise.
28863 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
28864 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
28865 not use the AltiVec predicate framework.
28866 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
28867 of vec_any_eq to not use AltiVec predicate framework.
28868 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
28869 support for predefined __POWER9_VECTOR__ macro to indicate that
28870 Power9 instruction selection is enabled.
28871 (altivec_overloaded_builtins): Remove extraneous
28872 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
28873 function argument types RS6000_BTI_bool_V16QI and
28874 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
28875 entry for overloaded function argument types RS6000_BTI_bool_V4SI
28876 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
28877 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
28878 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
28879 Power9 for implementations of vec_cmpne. Change the signature for
28880 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
28881 (representing vec_all_ne) to remove the previously described first
28882 argument of type RS6000_BTI_INTSI, as this was an artifact of
28883 reliance on the AltiVec predicate framework, which is no longer
28884 used in the implementation of these functions. Add
28885 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
28886 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
28887 since, unlike the AltiVec predicate framework implementation, we
28888 do not share function descriptors between vec_alle and vec_anyeq.
28889 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
28890 set of modes that receive special treatment even when
28891 TARGET_P9_VECTOR is true. The special treatment emits code that
28892 does not depend on Power9 instructions.
28893 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
28894 define_expand to not rely on AltiVec predicate framework.
28895 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
28897 (vector_ne_v2di_p): Change this define_expand to not rely on
28898 AltiVec predicate framework.
28899 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
28901 (vector_ne_<mode>_p): Change this define_expand to not rely on
28902 AltiVec predicate framework.
28903 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
28905 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
28906 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
28907 define_insn pattern.
28908 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
28909 define_insn pattern because the xvcmpne<VSs>. instruction is not
28911 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
28912 instruction is not supported.
28914 2017-03-01 Jakub Jelinek <jakub@redhat.com>
28916 * config/nvptx/nvptx.c: Include intl.h.
28918 2017-03-01 Martin Jambor <mjambor@suse.cz>
28921 * ipa-prop.h (ipa_bits): Removed field known.
28922 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
28923 to pointers. Adjusted their comments to warn about their sharing.
28924 (ipcp_transformation_summary): Change bits to a vector of pointers.
28925 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
28926 (ipa_get_ipa_bits_for_value): Declare.
28927 * tree-vrp.h (value_range): Mark as GTY((for_user)).
28928 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
28929 (ipa_bits_hash_table): Likewise.
28930 (ipa_vr_ggc_hash_traits): Likewise.
28931 (ipa_vr_hash_table): Likewise.
28932 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
28933 being pointers and vr_known being removed.
28934 (ipa_set_jf_unknown): Likewise.
28935 (ipa_get_ipa_bits_for_value): New function.
28936 (ipa_set_jfunc_bits): Likewise.
28937 (ipa_get_value_range): New overloaded functions.
28938 (ipa_set_jfunc_vr): Likewise.
28939 (ipa_compute_jump_functions_for_edge): Use the above functions to
28940 construct bits and vr parts of jump functions.
28941 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
28942 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
28944 (ipcp_grow_transformations_if_necessary): Also allocate
28945 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
28947 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
28948 them. Fix too long lines.
28949 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
28950 vr_known being removed.
28951 (ipa_read_jump_function): Use new setter functions to construct bits
28952 and vr parts of jump functions or set them to NULL.
28953 (write_ipcp_transformation_info): Adjust for bits being pointers.
28954 (read_ipcp_transformation_info): Likewise.
28955 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
28957 Include gt-ipa-prop.h.
28958 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
28960 (ipcp_store_bits_results): Likewise.
28961 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
28962 Do not write to existing jump functions but use a temporary instead.
28964 2017-03-01 Jakub Jelinek <jakub@redhat.com>
28967 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
28968 attempt to use its first operand as BIT_FIELD_REF base.
28970 2017-03-01 Richard Biener <rguenther@suse.de>
28972 PR middle-end/79721
28973 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
28974 interpolating formula in wrapping arithmetic.
28975 (chrec_apply): Convert chrec_evaluate return value to wanted type.
28977 2017-03-01 Jakub Jelinek <jakub@redhat.com>
28979 PR tree-optimization/79734
28980 * tree-vect-generic.c (expand_vector_condition): Optimize
28981 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
28982 Handle VEC_COND_EXPR where comparison has different inner width from
28983 type's inner width.
28985 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
28987 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
28988 markup, and similar issues. Remove @opindex entries for things
28989 that aren't options. Add missing -mmpy-option entries.
28991 2017-02-28 Jakub Jelinek <jakub@redhat.com>
28993 PR tree-optimization/79737
28994 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
28995 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
28996 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
28997 instead of byte_size. Formatting fix.
28998 (shift_bytes_in_array_right): Formatting fix.
29000 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
29003 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
29004 condition on optimize for the leaf function test.
29006 2017-02-28 Martin Liska <mliska@suse.cz>
29009 * read-rtl-function.c (function_reader::handle_unknown_directive):
29010 Bail out when one uses -flto.
29012 2017-02-28 Martin Liska <mliska@suse.cz>
29014 * common.opt: Replace space with tabular for options of <number>
29016 * config/i386/i386.opt: Show <number> value for
29017 -mlarge-data-threshold.
29018 * opts.c (print_filtered_help): Do not display number in hexadecimal
29021 2017-02-28 Martin Liska <mliska@suse.cz>
29023 * common.opt: Fix --help=option -Q for options which are of
29026 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
29028 * config/i386/i386.c (print_reg): Error out for values
29029 of 8-bit size in invalid integer register.
29031 2017-02-28 Martin Sebor <msebor@redhat.com>
29033 PR tree-optimization/79691
29034 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
29036 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29039 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
29040 gcc_unreachable with output_operand_lossage.
29042 2017-02-28 Richard Biener <rguenther@suse.de>
29044 PR tree-optimization/79740
29045 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
29047 (visit_nary_op): Insert the nary into the hashtable if we
29048 pattern-matched sth.
29049 * tree-ssa-pre.c (eliminate_insert): Robustify.
29051 2017-02-28 Richard Biener <rguenther@suse.de>
29053 PR middle-end/79731
29054 * fold-const.c (decode_field_reference): Reject out-of-bound
29057 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29059 * config/i386/i386.c: Include intl.h.
29060 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
29061 instead of just cond ? "..." : "...".
29062 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
29063 * coverage.c (read_counts_file): Likewise.
29064 * omp-offload.c: Include intl.h.
29065 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
29066 of just cond ? "..." : "...".
29067 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
29068 of just cond ? "..." : "...".
29070 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
29073 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
29075 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
29076 'tune for' CPU name.
29077 * config/arm/arm-cpu-data.h: Regenerated.
29079 2017-02-28 Richard Biener <rguenther@suse.de>
29081 PR tree-optimization/79732
29082 * tree-inline.c (expand_call_inline): Do not shadow var.
29084 2017-02-28 Richard Biener <rguenther@suse.de>
29086 PR tree-optimization/79723
29087 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
29088 address-space properly.
29090 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
29092 * doc/optinfo.texi (Optimization groups): Fix option used for
29094 * doc/invoke.texi (-fopt-info): Document "omp".
29095 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
29096 (OPTGROUP_ALL): Add OPTGROUP_OMP.
29097 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
29098 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
29099 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
29101 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
29103 * dumpfile.c (optgroup_options): Instead of "openmp", associate
29104 OPTGROUP_OMP with "omp".
29106 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
29109 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
29110 for arithmetic shift of unsigned V2DI.
29112 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
29114 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
29115 arc/linux.h headers.
29116 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
29117 (LINK_SPEC): Likewise.
29118 (ARC_TLS_EXTRA_START_SPEC): Likewise.
29119 (EXTRA_SPECS): Likewise.
29120 (STARTFILE_SPEC): Likewise.
29121 (ENDFILE_SPEC): Likewise.
29122 (LIB_SPEC): Likewise.
29123 (TARGET_SDATA_DEFAULT): Likewise.
29124 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
29125 (MULTILIB_DEFAULTS): Likewise.
29126 (DWARF2_UNWIND_INFO): Likewise.
29127 * config/arc/big.h: New file.
29128 * config/arc/elf.h: Likewise.
29129 * config/arc/linux.h: Likewise.
29130 * config/arc/t-uClibc: Remove.
29132 2017-02-27 Bin Cheng <bin.cheng@arm.com>
29134 PR tree-optimization/77536
29135 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
29136 (tree_transform_and_unroll_loop): Use above function to compute the
29137 estimated niter of unrolled loop and use it when scaling profile.
29138 Also use count info rather than frequency if it's non-zero.
29139 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
29140 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
29141 (vect_transform_loop): Call above function.
29143 2017-02-27 Richard Biener <rguenther@suse.de>
29145 PR tree-optimization/45397
29146 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
29147 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
29148 (visit_nary_op): Add pattern matching for CSEing sign-changed
29149 or truncated operations with wider ones.
29151 2017-02-27 Richard Biener <rguenther@suse.de>
29153 PR tree-optimization/79690
29154 * tree-vect-stmts.c (vectorizable_store): Use vector type
29155 built from the DR with address-space.
29157 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
29159 * doc/invoke.texi (Optimize Options): Refine the description
29160 of asan-use-after-return.
29162 2017-02-25 Alan Modra <amodra@gmail.com>
29164 PR rtl-optimization/79584
29165 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
29166 base, not ad->base_term, the reg within base. Remove assertion
29167 that ad->base == ad->base_term. Replace gen_int_mode using
29168 bogus mode with const0_rtx.
29170 2017-02-25 Jakub Jelinek <jakub@redhat.com>
29172 PR middle-end/79396
29173 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
29174 FMA_EXPR like tcc_binary or tcc_unary.
29176 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
29179 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
29181 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
29182 (output_loc_operands): Handle DW_OP_call_ref and
29183 DW_OP_GNU_variable_value.
29184 (struct variable_value_struct): New type.
29185 (struct variable_value_hasher): Likewise.
29186 (variable_value_hash): New variable.
29187 (string_types): Remove.
29188 (copy_loc_descr): New function.
29189 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
29190 (prepend_loc_descr_to_each): New function.
29191 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
29192 instead of add_loc_descr_to_each if the first argument is single
29193 location list and the second has multiple.
29194 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
29195 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
29196 when looking for variable value which doesn't have other location info.
29197 (loc_list_from_tree): Formatting fix.
29198 (gen_array_type_die): Simplify DW_AT_string_length handling.
29199 (adjust_string_types): Remove.
29200 (gen_subprogram_die): Don't call adjust_string_types nor test/set
29201 string_types. Call resolve_variable_values.
29202 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
29203 (resolve_addr_in_expr): Likewise. Add A argument.
29204 (copy_deref_exprloc): Remove deref argument. Adjust for the
29205 original expression being DW_OP_GNU_variable_value with optionally
29206 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
29207 optionally after it.
29208 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
29209 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
29210 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
29211 (variable_value_hasher::hash, variable_value_hasher::equal): New
29213 (resolve_variable_value_in_expr, resolve_variable_value,
29214 resolve_variable_values, note_variable_value_in_expr,
29215 note_variable_value): New functions.
29216 (dwarf2out_early_finish): Call note_variable_value on all toplevel
29219 2017-02-24 Jakub Jelinek <jakub@redhat.com>
29222 * opts.h (handle_generated_option): Add GENERATED_P argument.
29223 * opts-common.c (handle_option): Adjust function comment.
29224 (handle_generated_option): Add GENERATED_P argument, pass it to
29226 (control_warning_option): Pass false to handle_generated_option
29228 * opts.c (maybe_default_option): Pass true to handle_generated_option
29230 * optc-gen.awk: Likewise.
29232 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29234 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
29235 a REG, look at the REG it is a SUBREG of.
29236 (splitter for cmpeqsi_t): Ditto.
29238 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29240 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
29241 the special USEs with the pattern of the insn, not the insn itself.
29243 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
29246 * doc/invoke.texi: Document -mload-store-pairs.
29248 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29249 Sandra Loosemore <sandra@codesourcery.com>
29251 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
29252 argument isn't a CONST_INT.
29253 (nios2_alternate_compare_const): Assert op is a CONST_INT.
29254 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
29255 (nios2_validate_compare): Bypass alternate compare logic if *op2
29256 is not a CONST_INT.
29257 (ldstwm_operation_p): Return false if first_base is not a REG or
29258 if first_offset is not a CONST_INT.
29260 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29262 * config/cris/cris.md: Use correct operand in a define_peephole2.
29264 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29266 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
29268 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29270 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
29271 this_insn if it is an INSN or JUMP_INSN.
29272 (force_offsettable): Look at base, not at addr.
29273 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
29274 on things that aren't necessarily CONST_INTs.
29276 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
29278 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
29279 -mfpmath=sse is the default also for x86-32 targets with SSE2
29280 instruction set when @option{-ffast-math} is enabled
29282 2017-02-24 Jeff Law <law@redhat.com>
29284 PR rtl-optimizatoin/79286
29285 * ira.c (update_equiv_regs): Drop may_trap_p exception to
29288 2017-02-24 Richard Biener <rguenther@suse.de>
29290 PR tree-optimization/79389
29291 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
29294 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
29296 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
29297 function comment to reflect reality.
29298 (loop_exits_before_overflow): Fix typo in function description.
29300 2017-02-24 Richard Biener <rguenther@suse.de>
29302 PR tree-optimization/79389
29303 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
29304 properly that a threading opportunity exists. Detect conditional
29305 copy/constant propagation opportunities.
29307 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
29309 * config/visium/visium.md (type): Add trap.
29310 (b): New mode attribute.
29311 (*btst): Rename into...
29312 (*btst<mode>): ...this and adjust.
29313 (*cbranchsi4_btst_insn): Rename into...
29314 (*cbranch<mode>4_btst_insn): ...this and adjust.
29315 (trap): New define_insn.
29317 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29319 PR tree-optimization/79389
29320 * ifcvt.c (struct noce_if_info): Add rev_cond field.
29321 (noce_reversed_cond_code): New function.
29322 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
29323 reversed_comparison_code. Formatting fix.
29324 (noce_try_store_flag): Test rev_cond != NULL in addition to
29325 reversed_comparison_code.
29326 (noce_try_store_flag_constants): Likewise.
29327 (noce_try_store_flag_mask): Likewise.
29328 (noce_try_addcc): Use rev_cond if non-NULL instead of
29329 reversed_comparison_code.
29330 (noce_try_cmove_arith): Likewise. Formatting fixes.
29331 (noce_try_minmax, noce_try_abs): Clear rev_cond.
29332 (noce_find_if_block): Initialize rev_cond.
29333 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
29334 instead of false as last argument never attempt to reverse it
29337 2017-02-23 Bin Cheng <bin.cheng@arm.com>
29339 PR tree-optimization/79663
29340 * tree-predcom.c (combine_chains): Process refs in reverse order
29341 only for ZERO length chains, and add explaining comment.
29343 2017-02-23 Jeff Law <law@redhat.com>
29345 PR tree-optimization/79578
29346 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
29347 in call to operand_equal_p.
29349 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
29352 * config/i386/cpuid.h: Fix another undefined behavior.
29354 2017-02-23 Richard Biener <rguenther@suse.de>
29356 PR tree-optimization/79683
29357 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
29358 vector types for data-refs.
29360 2017-02-23 Martin Liska <mliska@suse.cz>
29362 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
29364 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29366 PR middle-end/79665
29367 * internal-fn.c (get_range_pos_neg): Moved to ...
29368 * tree.c (get_range_pos_neg): ... here. No longer static.
29369 * tree.h (get_range_pos_neg): New prototype.
29370 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
29371 are known to be in between 0 and signed maximum inclusive, try to
29372 expand both unsigned and signed divmod and use the cheaper one from
29375 2017-02-22 Jeff Law <law@redhat.com>
29377 PR tree-optimization/79578
29378 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
29379 to compare base operands.
29381 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
29384 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
29385 gpc_reg_operand instead of fpr_reg_operand.
29387 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
29389 * config/mips/mips.c (mips_return_in_memory): Force FP
29390 vector types to be returned in memory for o32 ABI.
29392 2017-02-22 Jakub Jelinek <jakub@redhat.com>
29394 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
29395 instead of DW_TAG_member for static data member declarations and don't
29396 set no_linkage_name for static inline data members.
29397 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
29400 2017-02-22 Martin Liska <mliska@suse.cz>
29402 * doc/invoke.texi: Replace inequality signs with square brackets
29405 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
29408 * lra-constraints.c (simplify_operand_subreg): Handle
29409 WORD_REGISTER_OPERATIONS targets.
29411 2017-02-22 Jakub Jelinek <jakub@redhat.com>
29414 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
29415 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
29416 elimination by swapping fld*.
29418 2017-02-22 Richard Biener <rguenther@suse.de>
29420 PR tree-optimization/79673
29421 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
29422 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
29423 irrelevant address-space qualifiers and avoiding a
29424 ADDR_SPACE_CONVERT_EXPR from fold_convert.
29426 2017-02-22 Richard Biener <rguenther@suse.de>
29428 PR tree-optimization/79666
29429 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
29430 to not symbolically negate if that may introduce undefined
29433 2017-02-22 Martin Liska <mliska@suse.cz>
29436 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
29437 * data-streamer-out.c (streamer_write_gcov_count_stream):
29439 * value-prof.c (stream_out_histogram_value): Make assert more
29440 precise based on type of counter.
29442 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
29445 * config/i386/i386.md (standard_x87sse_constant_load splitter):
29446 Use nonimmediate_operand instead of memory_operand for operand 1.
29447 (float-extend standard_x87sse_constant_load splitter): Ditto.
29449 2017-02-21 Jeff Law <law@redhat.com>
29451 PR tree-optimization/79621
29452 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
29453 blocks with edges to themselves.
29455 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29458 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
29459 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
29460 Use gimple_call_builtin_p.
29463 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
29464 on temporarily removed DEBUG_INSNs.
29466 PR tree-optimization/79649
29467 * tree-loop-distribution.c (classify_partition): Give up on
29468 non-generic address space loads/stores.
29470 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
29472 * doc/loop.texi (Loop manipulation): Remove nonexistent
29473 tree_ssa_loop_version from the documentation.
29474 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
29476 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29479 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
29480 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
29481 * config/rs6000/rs6000.c: Include except.h.
29482 (rs6000_expand_split_stack_prologue): Call
29483 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
29485 2017-02-21 Martin Jambor <mjambor@suse.cz>
29488 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
29489 have been analyzed.
29491 2017-02-21 Martin Jambor <mjambor@suse.cz>
29493 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
29494 for backward compatibility only.
29495 * doc/invoke.texi (Option Summary): Remove all references to
29496 -fipa-cp-alignment.
29498 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
29502 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29504 * lra-constraints.c (curr_insn_transform): Handle
29505 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29507 2017-02-21 Martin Liska <mliska@suse.cz>
29509 * config/i386/i386.opt: Replace -masm-dialect with -masm.
29511 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
29513 PR translation/79638
29514 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
29516 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
29519 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
29520 (arm_function_ok_for_sibcall): Return false for an indirect call by
29521 descriptor if all the argument registers are used.
29522 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
29523 alignment of the function.
29525 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29527 PR tree-optimization/61441
29528 * simplify-rtx.c (simplify_const_unary_operation): For
29529 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
29530 the sNaN unmodified.
29532 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
29534 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29535 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29536 instead of SYSTEM_HEADER_DIR.
29538 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
29539 Martin Liška <mliska@suse.cz>
29541 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
29542 Fix typos and grammar, use active voice, and clarify.
29544 2017-02-20 Marek Polacek <polacek@redhat.com>
29546 PR middle-end/79537
29547 * gimplify.c (gimplify_expr): Handle unused *&&L;.
29550 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
29552 2017-02-20 Jakub Jelinek <jakub@redhat.com>
29555 * config/i386/i386.c (ix86_expand_builtin): Handle
29556 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
29557 ix86_builtins_isa[fcode].isa as a requirement of those
29558 flags and any other flag in the bitmask.
29559 (ix86_init_mmx_sse_builtins): Use 0 instead of
29560 ~OPTION_MASK_ISA_64BIT as mask.
29561 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
29562 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
29563 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
29564 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
29566 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29569 * lra-constraints.c (split_reg): Check requested split mode
29570 is supported by the register.
29572 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29574 * lra-constraints.c (simplify_operand_subreg): Remove early
29577 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29580 * lra-constraints.c (curr_insn_transform): Tighten condition
29581 for converting SUBREG reloads from OP_OUT to OP_INOUT.
29583 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29586 * lra-constraints.c (curr_insn_transform): Handle
29587 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29589 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
29592 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
29594 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
29596 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
29599 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
29602 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29604 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
29607 2017-02-18 Jakub Jelinek <jakub@redhat.com>
29609 * final.c (last_columnnum, override_columnnum): New variables.
29610 (final_start_function): Set last_columnnum, pass it to begin_prologue
29611 hook and pass 0 to dwarf2out_begin_prologue.
29612 (final_scan_insn): Update override_columnnum. Pass last_columnnum
29613 to source_line debug hook.
29614 (notice_source_line): Compute last_columnnum and for debug_column_info
29615 return true on column changes.
29616 * debug.h (struct gcc_debug_hooks): Add column argument to
29617 source_line and begin_prologue hooks.
29618 (debug_nothing_int_charstar_int_bool): Remove prototype.
29619 (debug_nothing_int_int_charstar,
29620 debug_nothing_int_int_charstar_int_bool): New prototypes.
29621 (dwarf2out_begin_prologue): Add column argument.
29622 * debug.c (do_nothing_debug_hooks): Adjust source_line and
29623 begin_prologue hooks.
29624 (debug_nothing_int_charstar_int_bool): Remove.
29625 (debug_nothing_int_int_charstar,
29626 debug_nothing_int_int_charstar_int_bool): New functions.
29627 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
29628 through to dwarf2out_source_line.
29629 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
29630 (dwarf2out_source_line): Add column argument, emit it if requested.
29631 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
29633 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29634 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29635 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
29636 through to dwarf2out_begin_prologue.
29637 (vmsdbgout_source_line): Add column argument, pass it through to
29638 dwarf2out_source_line.
29639 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
29640 dbxout_source_line caller.
29641 (dbxout_source_line): Add column argument.
29643 * common.opt (gno-column-info, gcolumn-info): New options.
29644 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
29645 (check_die): Also test for multiple DW_AT_decl_column attributes.
29646 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
29647 DW_AT_decl_column if requested.
29648 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
29650 (gen_variable_die): Likewise.
29651 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
29652 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
29655 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
29656 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
29657 (ix86_handle_option): Handle OPT_m3dnowa.
29658 * doc/invoke.texi (-m3dnowa): Document.
29659 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
29660 -m3dnowa instead of -m3dnow -march=athlon.
29663 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
29664 instead of gcc_assert for K, r and R code checks. Formatting fixes.
29666 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29669 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
29670 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
29671 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
29672 generator for vsx_xxpermdi_<mode>_be.
29673 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
29674 force big-endian semantics.
29675 (vsx_xxpermdi_<mode>_be): New define_expand with same
29676 implementation as previous version of vsx_xxpermdi_<mode>.
29678 2017-02-17 Jakub Jelinek <jakub@redhat.com>
29680 PR tree-optimization/79327
29681 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
29682 variable, its initialization and use.
29684 2017-02-17 Julia Koval <julia.koval@intel.com>
29686 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
29687 (OPTION_MASK_ISA_PKU_UNSET): New.
29688 (ix86_handle_option): Handle -mrdpid.
29689 * config/i386/cpuid.h (bit_RDPID): New.
29690 * config/i386/driver-i386.c (host_detect_local_cpu):
29691 Detect RDPID feature.
29692 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
29693 * config/i386/i386-c.c (ix86_target_macros_internal):
29695 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
29696 (ix86_valid_target_attribute_inner_p): Add "rdpid".
29697 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
29698 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
29699 * config/i386/i386.md (define_insn "rdpid"): New.
29700 * config/i386/i386.opt Add -mrdpid.
29701 * config/i386/immintrin.h (_rdpid_u32): New.
29703 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
29705 PR rtl-optimization/79541
29706 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
29707 instead of transforming it into USE.
29709 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
29711 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
29712 If HONOR_SNANS (SFmode) force the input to a register.
29713 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
29714 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
29715 an frsp or similar insn.
29717 2017-02-17 Martin Liska <mliska@suse.cz>
29719 PR rtl-optimization/79577
29720 * params.def (selsched-max-sched-times): Increase minimum to 1.
29722 2017-02-17 Martin Liska <mliska@suse.cz>
29724 PR rtl-optimization/79574
29725 * gcse.c (want_to_gcse_p): Prevent integer overflow.
29727 2017-02-17 Martin Liska <mliska@suse.cz>
29729 PR tree-optimization/79529
29730 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
29731 ssa_defined_default_def_p to handle cases which are implicitly
29733 * tree-ssa.c (ssa_defined_default_def_p): New function.
29734 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
29735 which are implicitly defined.
29736 * tree-ssa.h (ssa_defined_default_def_p): Declare.
29738 2017-02-17 Richard Biener <rguenther@suse.de>
29740 PR middle-end/79576
29741 * params.def (max-ssa-name-query-depth): Limit to 10.
29743 2017-02-17 Richard Biener <rguenther@suse.de>
29745 PR tree-optimization/79552
29746 * tree-ssa-structalias.c (visit_loadstore): Properly verify
29749 2017-02-17 Richard Biener <rguenther@suse.de>
29752 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
29754 2017-02-17 Marek Polacek <polacek@redhat.com>
29756 PR middle-end/79536
29757 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
29758 (fold_negate_expr): New wrapper.
29760 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
29762 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
29763 Correct terminology and de-emphasize pre-standard behavior.
29765 2017-02-16 Alan Modra <amodra@gmail.com>
29767 PR rtl-optimization/79286
29768 * ira.c (def_dominates_uses): New function.
29769 (update_equiv_regs): Don't create an equivalence for insns that
29770 may trap where the register def does not dominate the use.
29772 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
29774 PR rtl-optimization/78127
29775 * lra.c (lra): Call lra_eliminate before finish the loop after
29778 2017-02-16 Richard Biener <rguenther@suse.de>
29780 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
29782 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
29783 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
29784 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
29785 (isl_val_int_from_wi): New function.
29786 (extract_affine_gmp): Rename to ...
29787 (extract_affine_wi): ... this, take a widest_int.
29788 (extract_affine_int): Just wrap extract_affine_wi.
29789 (add_param_constraints): Use isl_val_int_from_wi.
29790 (add_loop_constraints): Likewise, and extract_affine_wi.
29792 2017-02-15 Jeff Law <law@redhat.com>
29794 PR middle-end/79521
29795 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
29796 ira_init_register_move_cost_if_necessary.
29798 2017-02-15 Martin Sebor <msebor@redhat.com>
29800 PR middle-end/32003
29801 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
29802 removed in a prior commit.
29804 2017-02-15 Bin Cheng <bin.cheng@arm.com>
29806 PR tree-optimization/79347
29807 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
29808 counters during peeling.
29810 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
29812 * Makefile.in (site.exp): Remove "set ISLVER".
29814 2017-02-15 Jakub Jelinek <jakub@redhat.com>
29817 * real.c (real_from_integer): Call real_convert even for decimal.
29819 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
29822 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
29824 2017-02-14 Andrew Pinski <apinski@cavium.com>
29826 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
29827 cores and change the partno/implementer to be correct.
29828 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
29829 the 'B" as the implementer.
29830 * config/aarch64/aarch64-tune.md: Regenerate.
29832 2017-02-14 Carl Love <cel@us.ibm.com>
29834 * config/rs6000/rs6000.c: Add case statement entry to make the
29835 xvcvuxdsp built-in argument unsigned.
29836 * config/rs6000/vsx.md: Fix the source and return operand types so they
29837 match the instruction definitions from the ISA document. Fix typo
29838 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
29841 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
29844 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
29845 member early_clobber_alts.
29846 * lra-lives.c (reg_early_clobber_p): New.
29847 (process_bb_lives): Use it.
29848 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
29849 (debug_operand_data): Initialize early_clobber_alts.
29850 (setup_operand_alternative): Set up early_clobber_alts.
29851 (collect_non_operand_hard_regs): Ditto. Pass early clobber
29852 alternatives to new_insn_reg.
29853 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
29855 (lra_update_insn_regno_info): Pass the new arg.
29857 2017-02-14 Jakub Jelinek <jakub@redhat.com>
29859 PR middle-end/79505
29860 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
29861 (new_oacc_loop_raw): Don't clear already cleared fields.
29864 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
29865 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
29866 _mm512_prefetch_i64gather_ps): New inline functions and macros.
29868 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
29871 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
29873 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
29876 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
29877 the extra instruction to the right place to store 128-bit constant
29880 2017-02-14 Martin Sebor <msebor@redhat.com>
29882 PR middle-end/79448
29883 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
29884 warning for strings of unknown length.
29886 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
29888 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
29890 2017-02-14 Jeff Law <law@redhat.com>
29893 * ira-costs.c (scan_one_insn): Initialize register move costs
29894 for pseudos seen in USE/CLOBBER insns.
29896 PR tree-optimization/79095
29897 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
29898 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
29899 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
29900 if the operands are known to be not equal, then the resulting range
29902 (intersect_ranges): If the new range is ~[0,0] and the old range is
29903 wide, then prefer ~[0,0].
29904 * tree-vrp.c (overflow_comparison_p_1): New function.
29905 (overflow_comparison_p): New function.
29906 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
29907 if NAME is used in an overflow test.
29908 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
29909 overflow check that can be expressed as an equality test, then adjust
29910 ops to be that equality test.
29912 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29914 * config/s390/s390-builtin-types.def: Remove flags argument.
29915 * config/s390/s390.c (s390_init_builtins): Likewise.
29917 2017-02-14 Martin Liska <mliska@suse.cz>
29919 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
29920 vector. Fix trailing white spaces.
29922 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
29924 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
29927 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29929 PR rtl-optimization/68664
29930 * config/arm/arm.c (arm_sched_can_speculate_insn):
29931 New function. Declare prototype.
29932 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
29934 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29936 PR rtl-optimization/68664
29937 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
29939 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
29941 2017-02-14 Amit Pawar <amit.pawar@amd.com>
29943 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
29944 max skip bytes for function, loop and jump.
29946 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29948 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
29949 ABS_EXPR for gimple dump.
29951 2017-02-14 Jakub Jelinek <jakub@redhat.com>
29954 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
29956 PR tree-optimization/79408
29957 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
29958 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
29959 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
29960 also if rhs1 is INTEGER_CST.
29962 2017-02-14 Richard Biener <rguenther@suse.de>
29964 PR middle-end/79432
29965 * tree-into-ssa.c (insert_phi_nodes): When the function can
29966 have abnormal edges rewrite SSA names with broken use-def
29967 dominance out of SSA and register them for PHI insertion.
29969 2017-02-13 Martin Sebor <msebor@redhat.com>
29971 PR middle-end/79496
29972 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
29973 clearing info.nowrite flag when snprintf size argument is a range.
29975 2017-02-13 Jakub Jelinek <jakub@redhat.com>
29977 * cprop.c (cprop_jump): Add missing space in string literal.
29978 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
29979 (get_constraint_for_component_ref): Likewise.
29980 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
29981 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
29982 * lra-constraints.c (process_alt_operands): Likewise.
29983 * ipa-inline.c (inline_small_functions): Likewise.
29984 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
29985 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
29986 * trans-mem.c (diagnose_tm_1_op): Likewise.
29987 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
29988 (grid_parallel_clauses_gridifiable): Likewise.
29990 * config/nvptx/mkoffload.c (process): Add space in between
29993 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
29994 "MOD4_SSE_REGS" and "ALL_REGS".
29996 * spellcheck.c (test_data): Add , in between "foo" and "food".
29998 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30001 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
30002 boundary crossing check and subsequent code generation agree.
30004 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30006 * config/aarch64/aarch64.c (has_memory_op): Delete.
30007 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
30010 2017-02-13 Jakub Jelinek <jakub@redhat.com>
30012 PR rtl-optimization/79388
30013 PR rtl-optimization/79450
30014 * combine.c (distribute_notes): When removing TEM_INSN for which
30015 corresponding dest has last value recorded, invalidate that last
30018 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30020 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
30021 of explicit '@'. Add missing assembly comment marker on branch costs
30024 2017-02-13 Nathan Sidwell <nathan@acm.org>
30026 * gengtype-lex.l (<in_struct>): Add '/'.
30028 2017-02-13 Martin Liska <mliska@suse.cz>
30031 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
30033 2017-02-13 Richard Biener <rguenther@suse.de>
30035 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
30037 * configure: Re-generate.
30038 * config.in: Likewise.
30039 * graphite-dependences.c: Simplify as if
30040 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
30041 * graphite-isl-ast-to-gimple.c: Likewise.
30042 * graphite-optimize-isl.c: Likewise.
30043 * graphite-poly.c: Likewise.
30044 * graphite-sese-to-poly.c: Likewise.
30045 * graphite.h: Likewise.
30046 * toplev.c: Include isl/version.h and use isl_version () for
30047 printing the ISL version.
30048 * doc/install.texi: Update ISL requirement.
30050 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30052 * doc/standards.texi (Standards): Update reference to
30055 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30057 * doc/extend.texi (Named Address Spaces): sourceware.org now
30059 * doc/install.texi (Binaries): Ditto.
30062 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30064 * doc/cpp.texi: Replace "stringify"/"stringification" with C
30065 standard terminology "stringize"/"stringizing" throughout.
30066 * doc/cppinternals.texi: Likewise.
30068 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30070 * doc/extend.texi: Fix some spelling mistakes and typos.
30071 * doc/invoke.texi: Likewise.
30073 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30076 * params.def (inline-min-speedup) Change from 10 to 8.
30078 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30080 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
30083 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30086 * ipa-inline-analysis.c (get_minimal_bb): New function.
30087 (record_modified): Use it.
30088 (remap_edge_change_prob): Handle also ancestor functions.
30090 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
30092 * doc/contrib.texi (Contributors): Remove broken link into
30093 the Mauve CVS repository.
30095 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30097 PR middle-end/79454
30098 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
30099 result computation whenever lhs doesn't have vector mode, not
30100 just when it has BLKmode.
30102 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30104 * doc/makefile.texi (profiledbootstrap): Refer to the
30105 installation instructions only in textual form.
30107 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30110 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
30112 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30114 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
30115 (Specific): Update mingw-w64 reference.
30117 (Specific): Remove broken link to Renesas RX processor.
30119 2017-02-10 Richard Biener <rguenther@suse.de>
30121 * toplev.c (process_options): Do not mention obsolete graphite
30122 options when printing sorry message about missing graphite support.
30123 Mention -floop-nest-optimize.
30125 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
30127 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
30128 (vtst_p16): Likewise.
30129 (vtstq_p8): Likewise.
30130 (vtstq_p16): Likewise.
30132 (vtstq_p64): Likewise.
30133 * config/arm/arm_neon.h (vgetq_lane_p64): New.
30134 (vset_lane_p64): New.
30135 (vsetq_lane_p64): New.
30137 2017-02-10 Jakub Jelinek <jakub@redhat.com>
30139 PR tree-optimization/79411
30140 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
30141 stmt operands are SSA_NAMEs used in abnormal phis.
30142 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
30145 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30148 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
30151 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30153 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
30155 2017-02-09 Jakub Jelinek <jakub@redhat.com>
30157 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
30161 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
30162 not arbitrary TREE_CONSTANT.
30165 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
30166 "omp declare target link" attribute unless is_global_var.
30167 * omp-offload.c (find_link_var_op): Likewise.
30169 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
30170 Chung-Lin Tang <cltang@codesourcery.com>
30172 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
30174 (gimplify_adjust_omp_clauses): Don't delete TILE.
30175 (gimplify_omp_for): Deal with TILE.
30176 * internal-fn.c (expand_GOACC_TILE): New function.
30177 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
30179 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
30180 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
30182 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
30183 avoid DIV for outermost collapse var.
30184 (expand_oacc_for): Insert tile element loop as needed. Adjust.
30185 Remove out of date comments, fix whitespace.
30186 * omp-general.c (omp_extract_for_data): Deal with tiling.
30187 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
30188 adjust OLF_DIM_BASE value.
30189 (struct omp_for_data): Add tiling field.
30190 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
30191 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
30192 for auto loops. Remove default auto determining, moved to
30193 oacc_loop_fixed_partitions.
30194 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
30195 stmts, add e_mask field.
30196 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
30197 (oacc_thread_numbers): Use oacc_dim_call.
30198 (oacc_xform_tile): New.
30199 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
30200 (finish_oacc_loop): Adjust for ifns vector.
30201 (oacc_loop_discover_walk): Append loop abstraction sites to list,
30202 add case for GOACC_TILE fns.
30203 (oacc_loop_xform_loop): Delete.
30204 (oacc_loop_process): Iterate over call list directly, and add
30205 handling for GOACC_TILE fns.
30206 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
30208 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
30209 vector partitioning to outer loops. Assign 2 partitions to loops
30210 when available. Add TILE handling.
30211 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
30212 (execite_oacc_device_lower): Process GOACC_TILE fns,
30213 ignore unknown specs.
30214 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
30215 * tree.c (omp_clause_num_ops): Adjust TILE ops.
30216 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
30218 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
30220 * configure.ac (ACX_BUGURL): Update.
30221 * configure: Regenerate.
30223 2017-02-09 Richard Biener <rguenther@suse.de>
30225 PR tree-optimization/69823
30226 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
30227 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
30229 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30231 * config/arc/arc-c.def: Add __NPS400__ definition.
30232 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
30233 (TARGET_NPS400): Define.
30235 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30237 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
30239 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
30240 pointer, arch_info.
30241 (arc_cpu_types): Fill the arch_info field with a pointer into the
30242 arc_arch_types table.
30243 (arc_selected_cpu): Declare.
30244 * config/arc/arc.c (arc_selected_cpu): Make global.
30245 (arc_selected_arch): Delete.
30246 (arc_base_cpu): Delete.
30247 (arc_override_options): Remove references to deleted variables,
30248 update access to arch information.
30249 (ARC_OPT): Update access to arch information.
30250 (ARC_OPTX): Likewise.
30251 * config/arc/arc.h (arc_base_cpu): Remove declaration.
30252 (TARGET_ARC600): Update access to arch information.
30253 (TARGET_ARC601): Likewise.
30254 (TARGET_ARC700): Likewise.
30255 (TARGET_EM): Likewise.
30256 (TARGET_HS): Likewise.
30257 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
30260 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
30263 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
30264 condition/operands for integer GE/LE/GEU/LEU operations.
30266 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
30268 PR translation/79397
30269 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
30272 2017-02-08 Martin Jambor <mjambor@suse.cz>
30275 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
30276 whether allocation happened.
30277 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
30278 nothing was allocated.
30280 2017-02-08 Jakub Jelinek <jakub@redhat.com>
30282 PR tree-optimization/79408
30283 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
30284 constant, but SSA_NAME with a known integer range, use the minimum
30285 of that range instead of op1 to determine if modulo can be replaced
30286 with its first operand.
30288 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30290 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
30292 2017-02-08 Richard Biener <rguenther@suse.de>
30294 PR tree-optimization/71824
30295 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
30296 Check all loops contained in the merged region.
30298 2017-02-07 Andrew Pinski <apinski@cavium.com>
30300 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
30302 2017-02-07 Andrew Pinski <apinski@cavium.com>
30304 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
30305 (thunderxt88): Likewise.
30306 (thunderxt81): Disable LSE and change v8.1 to v8.
30307 (thunderxt83): Likewise.
30309 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30310 Richard Biener <rguenther@suse.de>
30312 PR middle-end/79399
30313 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
30314 type from int to size_t.
30315 * ira-costs.c (struct_costs_size): Change type from int to size_t.
30317 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30319 PR rtl-optimization/79386
30320 * cprop.c (bypass_conditional_jumps): Initialize
30321 bypass_last_basic_block already before splitting bbs after
30322 unconditional traps...
30323 (bypass_conditional_jumps): ... rather than here.
30326 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
30327 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
30328 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
30329 fix -masm=intel patterns.
30331 2017-02-07 Richard Biener <rguenther@suse.de>
30333 PR tree-optimization/79256
30334 PR middle-end/79278
30335 * builtins.c (get_object_alignment_2): Use min_align_of_type
30336 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
30337 and ADJUST_FIELD_ALIGN.
30339 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
30341 * doc/tm.texi: Regenerate.
30342 * stor-layout.c (layout_decl): Adjust.
30343 (update_alignment_for_field): Likewise.
30344 (place_field): Likewise.
30345 (min_align_of_type): Likewise.
30346 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
30347 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
30348 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
30349 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
30350 * config/frv/frv.c (frv_adjust_field_align): Likewise.
30351 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
30352 * config/i386/i386.c (x86_field_alignment): Likewise.
30353 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
30354 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
30355 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
30356 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
30357 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
30358 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
30362 2017-01-30 Richard Biener <rguenther@suse.de>
30364 PR tree-optimization/79256
30365 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
30366 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
30369 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
30371 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
30372 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
30373 builtins to SImode and emit a zero-extend, if necessary.
30375 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30377 * docs/invoke.texi (RISC-V Options): Alphabetize.
30379 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30381 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
30384 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30386 * config/riscv/riscv.c: New file.
30387 * common/config/riscv/riscv-common.c: Likewise.
30388 * config.gcc: Likewise.
30389 * config/riscv/constraints.md: Likewise.
30390 * config/riscv/elf.h: Likewise.
30391 * config/riscv/generic.md: Likewise.
30392 * config/riscv/linux.h: Likewise.
30393 * config/riscv/multilib-generator: Likewise.
30394 * config/riscv/peephole.md: Likewise.
30395 * config/riscv/pic.md: Likewise.
30396 * config/riscv/predicates.md: Likewise.
30397 * config/riscv/riscv-builtins.c: Likewise.
30398 * config/riscv/riscv-c.c: Likewise.
30399 * config/riscv/riscv-ftypes.def: Likewise.
30400 * config/riscv/riscv-modes.def: Likewise.
30401 * config/riscv/riscv-opts.h: Likewise.
30402 * config/riscv/riscv-protos.h: Likewise.
30403 * config/riscv/riscv.h: Likewise.
30404 * config/riscv/riscv.md: Likewise.
30405 * config/riscv/riscv.opt: Likewise.
30406 * config/riscv/sync.md: Likewise.
30407 * config/riscv/t-elf-multilib: Likewise.
30408 * config/riscv/t-linux: Likewise.
30409 * config/riscv/t-linux-multilib: Likewise.
30410 * config/riscv/t-riscv: Likewise.
30411 * configure.ac: Likewise.
30412 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
30413 Waterman as RISC-V maintainers.
30414 * doc/install.texi: Add RISC-V entries.
30415 * doc/invoke.texi: Add RISC-V options section.
30416 * doc/md.texi: Add RISC-V constraints section.
30417 * configure: Regenerated.
30419 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
30422 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
30423 false values to be constant vectors with all 0 or all 1 bits set.
30424 (vcondu<mode><mode>): Likewise.
30425 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
30427 (fpmask_comparison_operator): Update comment.
30428 (vecint_comparison_operator): New predicate.
30429 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
30430 vector conditionals when the true and false values are constant
30431 vectors with all 0 bits or all 1 bits set.
30433 2017-02-06 Martin Sebor <msebor@redhat.com>
30435 PR tree-optimization/79376
30436 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
30438 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
30440 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
30441 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
30442 to simplify split condition.
30444 2017-02-06 Jakub Jelinek <jakub@redhat.com>
30446 * omp-expand.c (oxpand_omp_atomic_fetch_op,
30447 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
30450 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
30452 PR rtl-optimization/68664
30453 * target.def (can_speculate_insn): New hook.
30454 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
30455 * doc/tm.texi: Regenerate.
30456 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
30457 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
30458 (rs6000_sched_can_speculate_insn): New function.
30460 2017-02-06 Jakub Jelinek <jakub@redhat.com>
30462 PR tree-optimization/79284
30463 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
30464 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
30465 vectorizable_mask_load_store, vectorizable_operation,
30466 vect_is_simple_cond, get_same_sized_vectype): Use it instead
30467 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
30468 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
30469 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
30470 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
30471 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
30472 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
30473 is_gimple_assign (stmt). Replace another such test with
30474 is_gimple_assign (stmt).
30476 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
30479 * config/avr/avr.c (rtl-iter.h): Include it.
30480 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
30481 (avr_legitimate_combined_insn): ...and implementation.
30483 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
30485 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
30486 * config/s390/s390.c (s390_const_operand_ok)
30487 (s390_canonicalize_comparison, s390_extract_part)
30488 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
30489 (s390_contiguous_bitmask_p, s390_rtx_costs)
30490 (legitimize_pic_address): Likewise.
30491 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
30492 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
30493 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
30494 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
30495 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
30497 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
30499 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
30500 REGNO($0) == REGNO($1).
30502 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30504 * config/s390/linux.h(SIZE_TYPE): Add comment.
30506 2017-02-06 Julian Brown <julian@codesourcery.com>
30507 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30508 Virendra Pathak <virendra.pathak@broadcom.com>
30510 * config/aarch64/aarch64-cores.def: Change the scheduler
30512 * config/aarch64/aarch64.md: Include thunderx2t99.md.
30513 * config/aarch64/thunderx2t99.md: New file.
30515 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30517 * doc/standards.texi (Go Language): Update link to language
30520 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30522 * tree-eh.c (lower_resx): Sanitize profile.
30523 (cleanup_empty_eh_move_lp): Likewise.
30525 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30528 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
30530 * cfgloopmanip.h (loop_version): Update prototype.
30531 * modulo-sched.c (sms_schedule): Update call of loop_version.
30532 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
30533 * tree-parloops.c (gen_parallel_loop): Likewise.
30534 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
30535 * tree-ssa-loop-split.c (split_loop): Likewise.
30536 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
30537 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
30539 2017-02-05 Martin Liska <mliska@suse.cz>
30542 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
30544 (print_operand_address): Initialize a struct to zero.
30546 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30548 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
30549 garbage collector only in textual form.
30551 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30553 * doc/extend.texi (x86 specific memory model extensions for
30554 transactional memory): Simplify a phrase.
30556 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
30559 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
30560 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
30561 (atomic_storedi_1): Likewise.
30563 2017-02-04 Jakub Jelinek <jakub@redhat.com>
30565 PR tree-optimization/79338
30566 * tree-parloops.c (gather_scalar_reductions): Don't call
30567 vect_analyze_loop_form for loop->inner before destroying loop's
30570 2017-02-03 Martin Sebor <msebor@redhat.com>
30572 PR tree-optimization/79327
30573 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
30574 when precision has resulted in leading zeros.
30575 (format_integer): Adjust the likely counter to assume an unknown
30576 argument that may be zero is non-zero.
30578 2017-02-03 Jason Merrill <jason@redhat.com>
30581 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
30582 avoid copying non-taken branch.
30584 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30586 PR tree-optimization/79340
30587 * tree-vect-loop.c (vectorizable_reduction): Release
30588 vec_defs elements after safe_splicing them into other vectors.
30591 PR tree-optimization/79327
30592 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
30593 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
30595 (format_integer): Use wide_int_to_tree instead of build_int_cst
30596 + to_?hwi. If argmin is NULL, just set argmin and argmax to
30597 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
30598 of shortest and longest sequence.
30600 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
30602 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
30603 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
30605 2017-02-03 Walter Lee <walt@tilera.com>
30608 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
30609 after initial stackframe link reg save.
30610 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
30612 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30615 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
30616 wu for stxssp alternative.
30618 2017-02-03 Martin Sebor <msebor@redhat.com>
30620 PR tree-optimization/79352
30621 * gimple-fold.c (get_range_strlen): Add argument.
30622 (get_range_strlen): Change return type to bool.
30623 (get_maxval_strlen): Pass in a dummy argument.
30624 * gimple-fold.h (get_range_strlen): Change return type to bool.
30625 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
30626 * tree.h (array_at_struct_end_p): Add argument.
30627 * tree.c (array_at_struct_end_p): Handle it.
30629 2017-02-03 Martin Liska <mliska@suse.cz>
30632 * multiple_target.c (create_dispatcher_calls): Redirect edge
30633 from a caller of a dispatcher.
30634 (expand_target_clones): Make the clones local.
30635 (ipa_target_clone): Do both target clones and resolvers.
30636 (ipa_dispatcher_calls): Remove the pass.
30637 (pass_dispatcher_calls::gate): Likewise.
30638 (make_pass_dispatcher_calls): Likewise.
30639 * passes.def (pass_target_clone): Put as very first IPA early
30642 2017-02-03 Martin Liska <mliska@suse.cz>
30644 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
30645 in case of a function with ifunc attribute.
30647 2017-02-03 Martin Liska <mliska@suse.cz>
30649 * cgraph.c (cgraph_node::dump): Dump function version info.
30650 * symtab.c (symtab_node::dump_base): Add missing new line.
30652 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
30654 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
30655 (ifcombine_ifandif): Use it.
30657 2017-02-03 Martin Liska <mliska@suse.cz>
30659 * doc/invoke.texi: Document default value for
30660 use-after-scope-direct-emission-threshold.
30662 2017-02-03 Martin Liska <mliska@suse.cz>
30664 PR tree-optimization/79339
30665 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
30666 (format_floating): Likewise.
30668 2017-02-03 Martin Liska <mliska@suse.cz>
30671 * ipa-prop.c (ipa_node_params_t::insert): Remove current
30673 (ipa_node_params_t::remove): Likewise.
30674 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
30675 initialization from removed ipa_node_params_t::insert.
30676 (ipa_node_params::~ipa_node_params): Move from removed
30677 ipa_node_params_t::release.
30678 * symbol-summary.h (symbol_summary::m_released): New member.
30679 Do not release a summary twice. Do not allow to call finalizer
30680 for types of a summary that live in GGC memory.
30682 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30684 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
30687 2017-02-02 Martin Sebor <msebor@redhat.com>
30689 PR middle-end/79275
30690 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
30691 (format_string): Tighten up the range of output for non-constant
30692 strings and correct the expected range for wide non-constant strings.
30694 2017-02-02 Martin Sebor <msebor@redhat.com>
30696 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
30698 PR middle-end/32003
30699 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
30701 (-fdump-tree-@var): Add to index and document how to come up
30702 with pass-specific option and dump file names.
30703 (-fdump-passes): Clarify where to look for output.
30705 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
30707 PR middle-end/77445
30708 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
30709 statistics of the analyzed path; allow threading for speed when
30710 any of BBs along the path are optimized for speed.
30712 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
30714 PR middle-end/78468
30715 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
30716 settings of the virtual registers.
30719 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
30721 * explow.c (get_dynamic_stack_size): Take known alignment of stack
30722 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
30725 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30727 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
30728 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
30730 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30732 * config/s390/s390.md: Add missing comments with the expanded
30734 * config/s390/vector.md: Likewise.
30735 * config/s390/vx-builtins.md: Likewise.
30737 2017-02-02 Jakub Jelinek <jakub@redhat.com>
30740 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
30741 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
30742 conditions on a single line.
30744 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30746 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
30747 __S390_VX__ to __VX__.
30749 2017-02-01 Andrew Pinski <apinski@cavium.com>
30751 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
30752 stmt_info to record_stmt_cost.
30753 (vect_get_known_peeling_cost): Pass stmt_info if known to
30755 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
30756 cpu_vector_cost field into
30757 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
30758 field into vec_int_stmt_cost and vec_fp_stmt_cost.
30759 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
30760 splitting of scalar_stmt_cost and vec_stmt_cost.
30761 (thunderx_vector_cost): Likewise.
30762 (cortexa57_vector_cost): LIkewise.
30763 (exynosm1_vector_cost): Likewise.
30764 (xgene1_vector_cost): Likewise.
30765 (thunderx2t99_vector_cost): Improve after the splitting of the two
30767 (aarch64_builtin_vectorization_cost): Update for the splitting of
30768 scalar_stmt_cost and vec_stmt_cost.
30770 2017-02-01 Torvald Riegel <triegel@redhat.com>
30771 Richard Henderson <rth@redhat.com>
30773 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
30774 conditional on existance of a fast atomic load.
30775 * optabs-query.c (can_atomic_load_p): New function.
30776 * optabs-query.h (can_atomic_load_p): Declare it.
30777 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
30778 no fast atomic load is available for the particular size of access.
30779 (expand_atomic_compare_and_swap): Likewise.
30780 (expand_atomic_load): Likewise.
30781 (expand_atomic_store): Likewise.
30782 (expand_atomic_fetch_op): Likewise.
30783 * testsuite/lib/target-supports.exp
30784 (check_effective_target_sync_int_128): Remove x86 because it provides
30785 no fast atomic load.
30786 (check_effective_target_sync_int_128_runtime): Likewise.
30788 2017-02-01 Richard Biener <rguenther@suse.de>
30790 * graphite.c: Include tree-vectorizer.h for find_loop_location.
30791 (graphite_transform_loops): Provide opt-info for optimized nests.
30792 * tree-parloop.c (parallelize_loops): Provide opt-info for
30793 parallelized loops.
30795 2017-02-01 Richard Biener <rguenther@suse.de>
30797 PR middle-end/79315
30798 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
30799 was not set before.
30801 2017-02-01 Richard Biener <rguenther@suse.de>
30803 PR tree-optimization/71824
30804 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
30805 Verify the loops are valid in the merged SESE region.
30806 (scop_detection::can_represent_loop_1): Check analyzing the
30807 evolution of the number of iterations in the region succeeds.
30809 2017-01-31 Ian Lance Taylor <iant@golang.org>
30811 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
30812 REG_ARGS_SIZE note to 32-bit push insns and call insn.
30814 2017-01-31 David Malcolm <dmalcolm@redhat.com>
30816 PR preprocessor/79210
30817 * input.c (get_substring_ranges_for_loc): Replace line_width
30818 assertion with error-handling.
30820 2017-01-31 Richard Biener <rguenther@suse.de>
30822 PR tree-optimization/77318
30823 * graphite-sese-to-poly.c (extract_affine): Fix assert.
30824 (create_pw_aff_from_tree): Take loop parameter.
30825 (add_condition_to_pbb): Pass loop of the condition to
30826 create_pw_aff_from_tree.
30828 2017-01-31 Jakub Jelinek <jakub@redhat.com>
30830 * config/s390/s390.c (s390_asan_shadow_offset): New function.
30831 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
30833 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
30837 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
30839 (convert_int_to_float128): Likewise.
30840 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
30841 (convert_int_to_float128): Likewise.
30842 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
30843 (UNSPEC_IEEE128_CONVERT): Likewise.
30844 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
30845 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
30846 Use local variables for IBM extended format.
30847 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
30848 (fix_trunc<mode>si2_fprs): Likewise.
30849 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
30850 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
30851 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
30852 to know that we can now have integers of all sizes in vector
30854 (fix<uns>_<mode>di2_hw): Likewise.
30855 (float<uns>_<mode>si2_hw): Likewise.
30856 (fix_<mode>si2_hw): Likewise.
30857 (fixuns_<mode>si2_hw): Likewise.
30858 (float<uns>_<mode>di2_hw): Likewise.
30859 (float_<mode>di2_hw): Likewise.
30860 (float_<mode>si2_hw): Likewise.
30861 (floatuns_<mode>di2_hw): Likewise.
30862 (floatuns_<mode>si2_hw): Likewise.
30863 (xscvqp<su>wz_<mode>): Delete, no longer used.
30864 (xscvqp<su>dz_<mode>): Likewise.
30865 (xscv<su>dqp_<mode>): Likewise.
30866 (ieee128_mfvsrd_64bit): Likewise.
30867 (ieee128_mfvsrd_32bit): Likewise.
30868 (ieee128_mfvsrwz): Likewise.
30869 (ieee128_mtvsrw): Likewise.
30870 (ieee128_mtvsrd_64bit): Likewise.
30871 (ieee128_mtvsrd_32bit): Likewise.
30873 2017-01-31 Martin Liska <mliska@suse.cz>
30876 * ipa-prop.c (ipa_free_all_node_params): Call release method
30877 instead of ~sumbol_summary to not to trigger double times
30880 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
30882 PR tree-optimization/71691
30883 * bitmap.h (class auto_bitmap): New.
30884 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
30885 is_maybe_undefined instead of ssa_undefined_value_p.
30887 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30889 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
30890 __S390_ARCH_LEVEL__ to __ARCH__.
30892 2017-01-31 Jakub Jelinek <jakub@redhat.com>
30894 PR tree-optimization/79267
30895 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
30896 if should_remove_lhs_p is true.
30898 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
30901 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
30902 (add_alignment_attribute): New.
30903 (base_type_die): Add alignment attribute.
30904 (subrange_type_die): Likewise.
30905 (modified_type_die): Likewise.
30906 (gen_array_type_die): Likewise.
30907 (gen_descr_array_type_die: Likewise.
30908 (gen_enumeration_type_die): Likewise.
30909 (gen_subprogram_die): Likewise.
30910 (gen_variable_die): Likewise.
30911 (gen_field_die): Likewise.
30912 (gen_ptr_to_mbr_type_die): Likewise.
30913 (gen_struct_or_union_type_die): Likewise.
30914 (gen_subroutine_type_die): Likewise.
30915 (gen_typedef_die): Likewise.
30916 (base_type_cmp): Compare alignment attribute.
30918 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30921 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
30922 (setb_unsigned) New pattern for setb with CCUNS.
30923 * config/rs6000/rs6000.c (expand_block_compare): Use a different
30924 subfc./subfe sequence to avoid overflow problems. Generate a
30925 shorter sequence with cmpld/setb for power9.
30926 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
30927 for generating subfc. instruction.
30928 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
30929 now uses this instruction.
30931 2017-01-30 Ian Lance Taylor <iant@google.com>
30934 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
30935 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
30937 2017-01-30 Martin Sebor <msebor@redhat.com>
30939 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
30940 Move constant to the right of a relational operator.
30941 (get_mpfr_format_length, format_character, format_string): Ditto.
30942 (should_warn_p, maybe_warn): Same.
30944 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
30946 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
30949 * asan.c (get_translation_unit_decl): Remove function.
30950 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
30952 2017-01-30 Martin Liska <mliska@suse.cz>
30954 PR gcov-profile/79259
30955 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
30956 -fprofile-generate.
30958 2017-01-30 Martin Liska <mliska@suse.cz>
30961 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
30962 Initialize variables with NULL value.
30964 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
30967 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
30969 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
30971 2017-01-30 Richard Biener <rguenther@suse.de>
30973 PR tree-optimization/79276
30974 * tree-vrp.c (process_assert_insertions): Properly adjust common
30975 when removing a duplicate.
30977 2017-01-30 Richard Biener <rguenther@suse.de>
30979 PR tree-optimization/79256
30980 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
30981 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
30983 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
30985 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
30988 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
30989 ("*r<noxa>sbg_<mode>_sll_bitmask")
30990 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
30991 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
30992 Use contiguous_bitmask_nowrap_operand.
30994 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30997 * config/rs6000/altivec.h (vec_xl): Revise #define.
30998 (vec_xst): Likewise.
31000 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
31002 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
31004 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
31006 PR rtl-optimization/79194
31007 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
31008 traps before call to bypass_conditional_jumps.
31010 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31012 PR tree-optimization/71374
31013 * lra-constraints.c (check_conflict_input_operands): New.
31014 (match_reload): Use it.
31016 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31019 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
31020 account to calculate conflict_set.
31022 2017-01-27 Bin Cheng <bin.cheng@arm.com>
31024 PR rtl-optimization/78559
31025 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
31026 other_insn in combine.
31028 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31030 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
31031 uint16_type_node for BT_UINT16.
31033 2017-01-27 David Malcolm <dmalcolm@redhat.com>
31035 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
31036 "RTL Tests" to menu.
31037 (GIMPLE Tests): New node.
31038 (RTL Tests): New node.
31040 2017-01-27 Richard Biener <rguenther@suse.de>
31042 PR tree-optimization/79245
31043 * tree-loop-distribution.c (distribute_loop): Apply cost
31044 modeling also to detected patterns.
31046 2017-01-27 Richard Biener <rguenther@suse.de>
31048 PR tree-optimization/71433
31049 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
31050 (compare_assert_loc): New function.
31051 (process_assert_insertions): Sort and optimize assert locations
31052 to remove duplicates and push down identical assertions on
31053 edges to their destination block.
31055 2017-01-27 Richard Biener <rguenther@suse.de>
31057 PR tree-optimization/79244
31058 * tree-vrp.c (remove_range_assertions): Forcefully propagate
31059 out SSA names even if abnormal.
31061 2017-01-27 Jakub Jelinek <jakub@redhat.com>
31063 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
31064 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
31065 instead of MPFR_RNDN.
31067 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
31070 * arm.c (arm_option_override): Don't call build_target_option_node
31071 until after doing all option overrides.
31072 (arm_valid_target_attribute_tree): Likewise.
31074 2017-01-27 Martin Liska <mliska@suse.cz>
31076 * doc/invoke.texi (-fprofile-arcs): Document profiling support
31077 for {cd}tors and C++ {cd}tors.
31079 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
31081 * config/s390/s390.md ("*setmem_long_and")
31082 ("*setmem_long_and_31z"): Use zero_extend instead of and.
31084 2017-01-26 Martin Sebor <msebor@redhat.com>
31086 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
31089 2017-01-26 Martin Sebor <msebor@redhat.com>
31091 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
31092 HAVE_DFmode before using XFmode or DFmode.
31093 (parse_directive): Avoid using the z length modifier to avoid
31094 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
31096 PR middle-end/78703
31097 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
31098 to accept adjustment as an array.
31099 (get_int_range): New function.
31100 (struct directive): Make width and prec arrays.
31101 (directive::set_width, directive::set_precision): Call get_int_range.
31102 (format_integer, format_floating): Handle width and precision ranges.
31103 (format_string, parse_directive): Same.
31105 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31108 * dwarf2out.c (generate_skeleton_bottom_up): For children with
31109 comdat_type_p set, just clone them, but keep the children in the
31113 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
31114 which have direct callers with -fvar-tracking-assignments enabled
31116 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
31117 inside of type units.
31119 2017-01-26 Martin Sebor <msebor@redhat.com>
31121 PR middle-end/78703
31122 * gimple-ssa-sprintf.c (struct result_range): Add likely and
31124 (struct format_result): Replace number_chars, number_chars_min,
31125 and number_chars_max with a single member of struct result_range.
31127 (format_result::operator+=): Adjust.
31128 (struct fmtresult): Remove bounded. Handle likely and unlikely
31130 (fmtresult::adjust_for_width_or_precision): New function.
31131 (fmtresult:type_max_digits): New function.
31132 (bytes_remaining): Handle likely and unlikely counters.
31133 (min_bytes_remaining): Remove.
31134 (format_percent): Simplify.
31135 (format_integer, format_floating): Set likely and unlikely counters.
31136 (get_string_length, format_character, format_string): Same.
31137 (format_plain, should_warn_p): New function.
31138 (maybe_warn): Call should_warn_p. Update diagnostic messages
31139 and handle those for all directives, including plain strings.
31140 (format_directive): Handle likely and unlikely counters.
31141 Remove unnecessary quoting from diagnostics. Add an informational
31143 (add_bytes): Remove.
31144 (pass_sprintf_length::compute_format_length): Simplify.
31145 (try_substitute_return_value): Handle likely and unlikely counters.
31147 2017-01-26 Carl Love <cel@us.ibm.com>
31149 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
31150 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
31152 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
31155 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
31156 endianess for subregs into account.
31157 * lra-constraints.c (lra_constraints): Do risky transformations
31158 always on the first iteration.
31159 * lra-lives.c (check_pseudos_live_through_calls): Add arg
31160 last_call_used_reg_set.
31161 (process_bb_lives): Define and use last_call_used_reg_set.
31162 * lra.c (lra): Always continue after lra_constraints on the first
31165 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
31167 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
31169 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
31171 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31173 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
31174 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
31175 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
31176 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
31177 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
31178 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
31179 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
31180 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
31181 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
31183 2017-01-26 Marek Polacek <polacek@redhat.com>
31186 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
31187 for the third operand.
31189 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31191 PR middle-end/79236
31192 * omp-low.c (struct omp_context): Add simt_stmt field.
31193 (scan_omp_for): Return omp_context *.
31194 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
31195 context to the _simt_ SIMD stmt.
31196 (lower_omp_for): For combined SIMD with sibling _simt_
31197 SIMD, make sure to use the same decls in _looptemp_
31198 clauses as in the sibling.
31200 2017-01-26 David Sherwood <david.sherwood@arm.com>
31202 PR middle-end/79212
31203 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
31206 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31209 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
31210 emit fld b; fld a; if possible.
31212 * brig-builtins.def: Update copyright years.
31213 * config/arm/arm_acle_builtins.def: Update copyright years.
31215 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
31218 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
31219 constraint instead of o for the stxsd instruction.
31221 2017-01-25 Carl Love <cel@us.ibm.com>
31223 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
31224 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
31226 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
31228 * doc/invoke.texi (C++ Dialect Options): Fix typo.
31230 2017-01-25 Richard Biener <rguenther@suse.de>
31232 PR tree-optimization/69264
31233 * target.def (vector_alignment_reachable): Improve documentation.
31234 * doc/tm.texi: Regenerate.
31235 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
31237 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
31238 earlier changes with respect to TYPE_USER_ALIGN.
31239 (vector_alignment_reachable_p): Likewise. Improve dumping.
31241 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31244 * config/arm/arm.md (xordi3): Force constant operand into a register
31247 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31249 * doc/invoke.texi (-fstore-merging): Correct default optimization
31250 levels at which it is enabled.
31251 (-O): Move -fstore-merging from list to...
31254 2017-01-25 Richard Biener <rguenther@suse.de>
31257 * omp-expand.c: Include debug.h.
31258 (expand_omp_taskreg): Make sure to generate early debug before
31259 outlining anything from a function.
31260 (expand_omp_target): Likewise.
31261 (grid_expand_target_grid_body): Likewise.
31263 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
31266 * asan.c (get_translation_unit_decl): New function.
31267 (asan_add_global): Extract modules file name from globals
31268 TRANSLATION_UNIT_DECL name.
31270 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
31273 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
31274 for long calls with APCS frame and VFP.
31276 2017-01-24 David Malcolm <dmalcolm@redhat.com>
31278 * cfg.c (original_copy_tables_initialized_p): New function.
31279 * cfg.h (original_copy_tables_initialized_p): New decl.
31280 * cfgrtl.c (relink_block_chain): Guard the call to
31281 free_original_copy_tables with a call to
31282 original_copy_tables_initialized_p.
31283 * cgraph.h (symtab_node::native_rtl_p): New decl.
31284 * cgraphunit.c (symtab_node::native_rtl_p): New function.
31285 (symtab_node::needed_p): Don't assert for early assembly output
31286 for __RTL functions.
31287 (cgraph_node::finalize_function): Set "force_output" for __RTL
31289 (cgraph_node::analyze): Bail out early for __RTL functions.
31290 (analyze_functions): Update assertion to support __RTL functions.
31291 (cgraph_node::expand): Bail out early for __RTL functions.
31292 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
31294 * function.h (struct function): Update comment for field
31296 * gimple-expr.c: Include "tree-pass.h".
31297 (gimple_has_body_p): Return false for __RTL functions.
31298 * Makefile.in (OBJS): Add run-rtl-passes.o.
31299 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
31301 (gcc::pass_manager::get_clean_slate): New accessor.
31302 * passes.c: Include "insn-addr.h".
31303 (should_skip_pass_p): Add logging. Update logic for running
31304 "expand" to be compatible with both __GIMPLE and __RTL. Guard
31305 property-provider override so it is only done for gimple passes.
31307 (skip_pass): New function.
31308 (execute_one_pass): Call skip_pass when skipping passes.
31309 * read-md.c (md_reader::read_char): Support filtering
31310 the input to a subset of line numbers.
31311 (md_reader::md_reader): Initialize fields
31312 m_first_line and m_last_line.
31313 (md_reader::read_file_fragment): New function.
31314 * read-md.h (md_reader::read_file_fragment): New decl.
31315 (md_reader::m_first_line): New field.
31316 (md_reader::m_last_line): New field.
31317 * read-rtl-function.c (function_reader::create_function): Only
31318 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
31319 curr_properties. Set DECL_INITIAL to a dummy block.
31320 (read_rtl_function_body_from_file_range): New function.
31321 * read-rtl-function.h (read_rtl_function_body_from_file_range):
31323 * run-rtl-passes.c: New file.
31324 * run-rtl-passes.h: New file.
31326 2017-01-24 Jeff Law <law@redhat.com>
31328 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
31331 2017-01-24 Bin Cheng <bin.cheng@arm.com>
31333 PR tree-optimization/79159
31334 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
31335 (record_nonwrapping_iv): Improve boundary using above function if no
31336 value range information.
31338 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
31339 Martin Jambor <mjambor@suse.cz>
31341 * brig-builtins.def: New file.
31342 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
31343 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
31344 (DEF_HSAIL_SAT_BUILTIN): Likewise.
31345 (DEF_HSAIL_INTR_BUILTIN): Likewise.
31346 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
31347 * builtin-types.def (BT_INT8): New.
31348 (BT_INT16): Likewise.
31349 (BT_UINT8): Likewise.
31350 (BT_UINT16): Likewise.
31351 (BT_FN_ULONG): Likewise.
31352 (BT_FN_UINT_INT): Likewise.
31353 (BT_FN_UINT_ULONG): Likewise.
31354 (BT_FN_UINT_LONG): Likewise.
31355 (BT_FN_UINT_PTR): Likewise.
31356 (BT_FN_ULONG_PTR): Likewise.
31357 (BT_FN_INT8_FLOAT): Likewise.
31358 (BT_FN_INT16_FLOAT): Likewise.
31359 (BT_FN_UINT32_FLOAT): Likewise.
31360 (BT_FN_UINT16_FLOAT): Likewise.
31361 (BT_FN_UINT8_FLOAT): Likewise.
31362 (BT_FN_UINT64_FLOAT): Likewise.
31363 (BT_FN_UINT16_UINT32): Likewise.
31364 (BT_FN_UINT32_UINT16): Likewise.
31365 (BT_FN_UINT16_UINT16_UINT16): Likewise.
31366 (BT_FN_INT_PTR_INT): Likewise.
31367 (BT_FN_UINT_PTR_UINT): Likewise.
31368 (BT_FN_LONG_PTR_LONG): Likewise.
31369 (BT_FN_ULONG_PTR_ULONG): Likewise.
31370 (BT_FN_VOID_UINT64_UINT64): Likewise.
31371 (BT_FN_UINT8_UINT8_UINT8): Likewise.
31372 (BT_FN_INT8_INT8_INT8): Likewise.
31373 (BT_FN_INT16_INT16_INT16): Likewise.
31374 (BT_FN_INT_INT_INT): Likewise.
31375 (BT_FN_UINT_FLOAT_UINT): Likewise.
31376 (BT_FN_FLOAT_UINT_UINT): Likewise.
31377 (BT_FN_ULONG_UINT_UINT): Likewise.
31378 (BT_FN_ULONG_UINT_PTR): Likewise.
31379 (BT_FN_ULONG_ULONG_ULONG): Likewise.
31380 (BT_FN_UINT_UINT_UINT): Likewise.
31381 (BT_FN_VOID_UINT_PTR): Likewise.
31382 (BT_FN_UINT_UINT_PTR: Likewise.
31383 (BT_FN_UINT32_UINT64_PTR): Likewise.
31384 (BT_FN_INT_INT_UINT_UINT): Likewise.
31385 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
31386 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
31387 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
31388 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
31389 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
31390 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
31391 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
31392 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
31393 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
31394 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
31395 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
31396 * doc/frontends.texi: List BRIG FE.
31397 * doc/install.texi (Testing): Add BRIG tesring requirements.
31398 * doc/invoke.texi (Overall Options): Mention BRIG.
31399 * doc/standards.texi (Standards): Doucment BRIG HSA version.
31401 2017-01-24 Richard Biener <rguenther@suse.de>
31403 PR translation/79208
31404 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
31406 2017-01-24 Martin Jambor <mjambor@suse.cz>
31409 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
31410 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
31411 and known_contexts.
31413 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
31415 PR middle-end/79123
31416 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
31417 casts from signed to unsigned really don't have a range.
31419 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
31421 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
31422 GMP_RNDx for compatiblity.
31424 2017-01-24 Martin Liska <mliska@suse.cz>
31427 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
31428 that would prevent us to call alloca with -1 as argument.
31430 2017-01-24 Jakub Jelinek <jakub@redhat.com>
31432 * dwarf2out.c (output_compilation_unit_header, output_file_names):
31433 Avoid -Wformat-security warning.
31435 2017-01-23 Andrew Pinski <apinski@cavium.com>
31437 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
31440 2017-01-23 Martin Sebor <msebor@redhat.com>
31442 PR middle-end/78703
31443 * gimple-ssa-sprintf.c (warn_level): New global.
31444 (format_integer): Use it here and throughout the rest of the file.
31445 Use the same switch to compute sign as base.
31446 (maybe_warn): New function.
31447 (format_directive): Factor out warnings into maybe_warn.
31448 Add debugging output. Use warn_level.
31449 (add_bytes): Use warn_level.
31450 (pass_sprintf_length::compute_format_length): Add debugging output.
31451 (try_substitute_return_value): Same.
31452 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
31454 PR middle-end/78703
31455 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
31456 (struct fmtresult, format_integer, format_floating): Adjust.
31457 (fmtresult::fmtresult): Set max correctly in two argument ctor.
31458 (get_string_length, format_string,format_directive): Same.
31459 (pass_sprintf_length::compute_format_length): Same.
31460 (try_substitute_return_value): Simplify slightly.
31462 PR middle-end/78703
31463 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
31464 (fmtresult::operator+=): Outlined.
31465 (struct fmtresult): Add ctors.
31466 (struct conversion_spec): Rename...
31467 (struct directive): ...to this. Add and remove data members.
31468 (directive::set_width, directive::set_precision): New functions.
31469 (format_percent): Use fmtresult ctor.
31470 (get_width_and_precision): Remove.
31471 (format_integer): Make naming changes. Avoid computing width and
31473 (format_floating): Same. Adjust indentation.
31474 (format_character, format_none): New functions.
31475 (format_string): Moved character handling to format_character.
31476 (format_directive): Remove arguments, change return type.
31477 (parse_directive): New function.
31478 (pass_sprintf_length::compute_format_length): Move directive
31479 parsing to parse_directive.
31481 2017-01-23 Jakub Jelinek <jakub@redhat.com>
31483 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
31484 (assign_assembler_name_if_needed): ... this.
31485 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
31486 (assign_assembler_name_if_needed): ... this.
31487 (free_lang_data_in_cgraph): Adjust callers.
31488 * cgraphunit.c (cgraph_node::analyze): Likewise.
31489 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
31492 2017-01-23 Richard Biener <rguenther@suse.de>
31494 PR tree-optimization/79088
31495 PR tree-optimization/79188
31496 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
31497 resetting loop bounds after last path deletion. Reset loop
31498 bounds of the target loop, make code match the comments.
31499 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
31500 Make sure loops need no fixups.
31502 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
31504 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
31505 exponent support with double type for first argument.
31506 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
31507 type returned by __builtin_vec_extract_sig,
31508 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
31509 functions from "vector int" to "vector unsigned int" or from
31510 "vector long long int" to "vector unsigned long long int".
31511 Changed type returned by __builtin_vec_extract_exp,
31512 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
31513 functions from "vector int" to "vector unsigned int" or from
31514 "vector long long int" to "vector unsigned long long int".
31515 Changed return type of __builtin_vec_test_data_class,
31516 __builtin_vec_test_data_class_sp, and
31517 __builtin_vec_test_data_class_dp from "vector int" to
31518 "vector bool int" or from "vector long long int" to "vector bool
31519 long long int" and changed second argument type from "unsigned
31520 int" to "int". Added new overloaded function forms "vector float
31521 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
31522 "vector float __builtin_vec_insert_exp_sp (vector float, vector
31523 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
31524 double, vector unsigned long long int)" and "vector double
31525 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
31526 long int)". Changed return type of
31527 __builtin_scalar_test_data_class and
31528 __builtin_scalar_test_data_class_sp and
31529 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
31530 int" and changed second argument from "unsigned int" to "int".
31531 Changed type returned by __builtin_scalar_test_neg,
31532 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
31533 from "int" to "bool int". Added new overloaded function form
31534 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
31535 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
31536 exponent double-precision with floating point first argument.
31537 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
31538 documentation of scalar_test_data_class, scalar_test_neg,
31539 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
31540 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
31541 vec_test_data_class built-in functions to reflect refinements in
31542 their type signatures.
31544 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
31546 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
31548 (aarch64_elf_asm_destructor): Likewise.
31550 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
31552 PR rtl-optimization/78634
31553 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
31554 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
31555 * ifcvt.c (noce_try_cmove): Add missing cost check.
31557 PR rtl-optimization/71724
31558 * combine.c (if_then_else_cond): Look for situations where it is
31559 beneficial to undo the work of one of the recursive calls.
31561 2017-01-23 Bin Cheng <bin.cheng@arm.com>
31563 PR tree-optimization/70754
31564 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
31565 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
31566 combined stmt before it if not NULL.
31567 (combine_chains): Process refs reversely and compute dominance point
31570 2017-01-23 Martin Liska <mliska@suse.cz>
31572 PR tree-optimization/79196
31573 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
31574 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
31576 (strlen_optimize_stmt): Call the renamed function.
31578 2017-01-23 Michael Matz <matz@suse.de>
31580 PR tree-optimization/78384
31581 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
31583 2017-01-23 Richard Biener <rguenther@suse.de>
31585 PR tree-optimization/79186
31586 * tree-vrp.c (register_new_assert_for): Make sure we've seen
31587 both incoming edges before moving an assert.
31589 2017-01-23 Martin Jambor <mjambor@suse.cz>
31591 * ipa-prop.c (load_from_param_1): Removed.
31592 (load_from_unmodified_param): Bits from load_from_param_1 put back
31594 (load_from_param): Removed.
31595 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
31596 with stmt. Reverted back to use of load_from_unmodified_param.
31598 2017-01-23 Martin Jambor <mjambor@suse.cz>
31601 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
31602 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
31603 field a pointer to garbage collected vector, mark lattices and
31604 ipcp_orig_node with GTY((skip)).
31605 (ipa_get_param_count): Adjust to descriptors being a pointer.
31606 (ipa_get_param): Likewise.
31607 (ipa_get_type): Likewise.
31608 (ipa_get_param_move_cost): Likewise.
31609 (ipa_set_param_used): Likewise.
31610 (ipa_get_controlled_uses): Likewise.
31611 (ipa_set_controlled_uses): Likewise.
31612 (ipa_is_param_used): Likewise.
31613 (ipa_node_params_t): Move into garbage collector. New methods insert
31615 (ipa_node_params_sum): Annotate wth GTY(()).
31616 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
31618 (ipa_load_from_parm_agg): Adjust declaration.
31619 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
31620 * ipa-profile.c (ipa_profile): Likewise.
31621 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
31622 (ipa_populate_param_decls): Make descriptors parameter garbage
31624 (ipa_dump_param): Adjust to descriptors being a pointer.
31625 (ipa_alloc_node_params): Likewise.
31626 (ipa_initialize_node_params): Likewise.
31627 (load_from_param_1): Make descriptors parameter garbage collected.
31628 (load_from_unmodified_param): Likewise.
31629 (load_from_param): Likewise.
31630 (ipa_load_from_parm_agg): Likewise.
31631 (ipa_node_params::~ipa_node_params): Removed.
31632 (ipa_free_all_node_params): Remove call to delete operator.
31633 (ipa_node_params_t::insert): New.
31634 (ipa_node_params_t::remove): Likewise.
31635 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
31636 copy known_csts and known_contexts vectors.
31637 (ipa_read_node_info): Adjust to descriptors being a pointer.
31638 (ipcp_modif_dom_walker): Make m_descriptors field garbage
31640 (ipcp_transform_function): Make descriptors variable garbage
31643 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
31645 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
31646 * config/i386/avx512dqintrin.h: Ditto.
31647 * config/i386/avx512fintrin.h: Ditto.
31648 * config/i386/i386.c: Handle new builtins.
31649 * config/i386/i386-builtin.def: Add new builtins.
31650 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
31651 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
31653 2017-01-23 Jakub Jelinek <jakub@redhat.com>
31654 Martin Liska <mliska@suse.cz>
31656 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
31657 * asan.c (asan_expand_poison_ifn): Support stores and use
31658 appropriate ASAN report function.
31659 * internal-fn.c (expand_ASAN_POISON_USE): New function.
31660 * internal-fn.def (ASAN_POISON_USE): Declare.
31661 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
31662 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
31663 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
31664 ASAN_POISON calls w/o LHS.
31665 * tree-ssa.c (execute_update_addresses_taken): Create clobber
31666 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
31667 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
31668 * gimplify.c (asan_poison_variables): Add attribute
31669 use_after_scope_memory to variables that really needs to live
31671 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
31672 having the attribute.
31674 2017-01-23 Martin Liska <mliska@suse.cz>
31676 * asan.c (create_asan_shadow_var): New function.
31677 (asan_expand_poison_ifn): Likewise.
31678 * asan.h (asan_expand_poison_ifn): New declaration.
31679 * internal-fn.c (expand_ASAN_POISON): Likewise.
31680 * internal-fn.def (ASAN_POISON): New builtin.
31681 * sanopt.c (pass_sanopt::execute): Expand
31682 asan_expand_poison_ifn.
31683 * tree-inline.c (copy_decl_for_dup_finish): Make function
31685 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
31686 * tree-ssa.c (is_asan_mark_p): New function.
31687 (execute_update_addresses_taken): Rewrite local variables
31688 (identified just by use-after-scope as addressable) into SSA.
31690 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
31692 * doc/install.texi (Specific): opensource.apple.com uses https
31693 now. Remove trailing slash.
31695 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
31697 * README.Portability: Remove note on an Irix compatibility issue.
31699 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
31701 * gcov.c (INCLUDE_ALGORITHM): Define.
31702 (INCLUDE_VECTOR): Define.
31703 No longer include <vector> and <algorithm> directly.
31705 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
31707 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
31709 * doc/invoke.texi (Code Gen Options): Ditto.
31711 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
31714 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
31716 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
31718 rtl-optimization/79125
31719 * cprop.c (local_cprop_pass): Handle cases where we make an
31720 unconditional trap.
31722 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
31726 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
31727 read from, for big endian.
31729 2017-01-20 Jiong Wang <jiong.wang@arm.com>
31731 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
31732 register pauth builtins for LP64 only.
31734 2017-01-20 Marek Polacek <polacek@redhat.com>
31737 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
31740 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
31742 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
31744 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
31745 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
31746 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
31748 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31751 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
31752 in big-endian mode when they are not a single duplicated value.
31754 2017-01-20 Richard Biener <rguenther@suse.de>
31756 * BASE-VER: Bump to 7.0.1.
31758 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
31760 * omp-low.c (omplow_simd_context): New struct. Use it...
31761 (lower_rec_simd_input_clauses): ...here and...
31762 (lower_rec_input_clauses): ...here to hold common data. Adjust all
31763 references to idx, lane, max_vf, is_simt.
31765 2017-01-20 Graham Markall <graham.markall@embecosm.com>
31767 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
31770 2017-01-20 Martin Jambor <mjambor@suse.cz>
31772 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
31773 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
31775 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
31776 (GTFILES): Rename hsa.c to hsa-common.c.
31777 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
31778 * hsa-dump.c: Likewise.
31779 * hsa-gen.c: Likewise.
31780 * hsa-regalloc.c: Likewise.
31781 * ipa-hsa.c: Likewise.
31782 * omp-expand.c: Likewise.
31783 * omp-low.c: Likewise.
31784 * toplev.c: Likewise.
31786 2017-01-20 Marek Polacek <polacek@redhat.com>
31789 * doc/invoke.texi: Document -Wduplicated-branches.
31790 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
31791 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
31792 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
31793 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
31794 return 0 only when not OEP_LEXICOGRAPHIC.
31795 (fold_build_cleanup_point_expr): Use the expression
31796 location when building CLEANUP_POINT_EXPR.
31797 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
31798 * tree.c (add_expr): Handle error_mark_node.
31800 2017-01-20 Martin Liska <mliska@suse.cz>
31803 * tree-profile.c (init_ic_make_global_vars): Do not call
31805 (gimple_init_gcov_profiler): Likewise.
31807 2017-01-20 Martin Liska <mliska@suse.cz>
31810 * cgraph.h (maybe_create_reference): Remove argument and
31812 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
31814 * ipa-cp.c (create_specialized_node): Likewise.
31815 * symtab.c (symtab_node::maybe_create_reference): Handle
31816 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
31818 2017-01-20 Martin Liska <mliska@suse.cz>
31820 * read-rtl-function.c (function_reader::create_function): Use
31821 build_decl instread of build_decl_stat.
31823 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
31825 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
31826 * config/i386/avx512dqintrin.h: Ditto.
31827 * config/i386/avx512fintrin.h: Ditto.
31828 * config/i386/i386-builtin-types.def: Add new types.
31829 * config/i386/i386.c: Handle new types.
31830 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
31831 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
31832 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
31833 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
31834 (__builtin_ia32_kshiftridi): New.
31835 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
31837 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
31841 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
31842 define to rs6000_init_stack_protect_guard.
31843 (rs6000_init_stack_protect_guard): New function.
31845 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
31846 Yunqiang Su <yunqiang.su@imgtec.com>
31848 * config.gcc (supported_defaults): Add madd4.
31849 (with_madd4): Add validation.
31850 (all_defaults): Add madd4.
31851 * config/mips/mips.opt (mmadd4): New option.
31852 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
31854 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
31856 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
31857 (ISA_HAS_FUSED_MADD4): Likewise.
31858 * doc/invoke.texi (-mmadd4): Document the new option.
31859 * doc/install.texi (--with-madd4): Document the new option.
31861 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31863 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
31864 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
31865 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
31866 (aarch64_init_pauth_hint_builtins): New.
31867 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
31868 (aarch64_expand_builtin): Expand new builtins.
31870 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31872 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
31873 * combine-stack-adj.c (no_unhandled_cfa): Handle
31874 REG_CFA_TOGGLE_RA_MANGLE.
31875 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
31876 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
31877 info for return address signing.
31878 (aarch64_expand_epilogue): Likewise.
31880 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31882 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
31883 * config/aarch64/aarch64-protos.h
31884 (aarch64_return_address_signing_enabled): New declaration.
31885 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
31887 (aarch64_expand_prologue): Sign return address before it's pushed onto
31889 (aarch64_expand_epilogue): Authenticate return address fetched from
31891 (aarch64_override_options): Sanity check for ILP32 and ISA level.
31892 (aarch64_attributes): New function attributes for "sign-return-address".
31893 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
31894 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
31895 ("*do_return"): Generate combined instructions according to key index.
31896 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
31897 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
31899 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
31900 * config/aarch64/aarch64.opt (msign-return-address=): New.
31901 * doc/extend.texi (AArch64 Function Attributes): Documents
31902 "sign-return-address=".
31903 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
31905 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
31907 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
31908 overall option summary.
31910 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31912 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
31913 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
31914 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
31915 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
31917 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
31919 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
31920 -mpower9-minmax by default for -mcpu=power9.
31921 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
31922 128-bit floating point.
31924 2017-01-20 Alan Modra <amodra@gmail.com>
31926 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
31927 optimizing for size.
31929 2017-01-20 Alan Modra <amodra@gmail.com>
31932 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
31933 for strcmp and strncmp from corresponding builtin decl.
31935 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
31937 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
31938 instead of i386/rtems-64.h.
31939 * config/i386/rtems-64.h: Remove.
31941 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
31945 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
31947 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
31949 2017-01-19 Tamar Christina <tamar.christina@arm.com>
31951 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
31952 Change int to HOST_WIDE_INT.
31953 * config/aarch64/aarch64-protos.h
31954 (aarch64_simd_gen_const_vector_dup): Likewise.
31955 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
31957 2017-01-19 David Malcolm <dmalcolm@redhat.com>
31959 * langhooks-def.h (lhd_type_for_size): New decl.
31960 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
31961 * langhooks.c (lhd_type_for_size): New function, taken from
31964 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
31966 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
31967 define_bypass for CR latency.
31968 (power9-cracked-alu): Update bypass latency and remove power9-branch.
31969 (power9-alu2): Add define_bypass for CR latency.
31971 (power9-mul): Update insn latency.
31972 (power9-mul-compare): Update insn latency, bypass latency and remove
31975 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31977 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
31979 * config/aarch64/aarch64.md
31980 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
31981 aarch64_nopcrelative_literal_loads.
31982 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
31984 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
31986 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
31987 TARGET_LOONGSON_3A.
31988 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
31990 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
31993 * config.gcc (supported_defaults): Add lxc1-sxc1.
31994 (with_lxc1_sxc1): Add validation.
31995 (all_defaults): Add lxc1-sxc1.
31996 * config/mips/mips.opt (mlxc1-sxc1): New option.
31997 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
31999 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
32000 __mips_no_lxc1_sxc1.
32001 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
32002 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
32003 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
32005 2017-01-19 Richard Biener <rguenther@suse.de>
32007 PR tree-optimization/72488
32008 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
32009 sure to restore SSA info.
32010 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
32012 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
32014 PR rtl-optimization/79121
32015 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
32016 of the inner type when shifting an extended value.
32018 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
32021 * symtab.c (symtab_node::equal_address_to): Fix comparing of
32022 interposable aliases.
32024 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
32027 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
32028 Use the evmergelohi instruction.
32029 (mov_si<mode>_e500_subreg4_2_le): Likewise.
32030 (mov_sitf_e500_subreg8_2_be): Likewise.
32031 (mov_sitf_e500_subreg12_2_le): Likewise.
32032 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
32033 (mov_si<mode>_e500_subreg4_2_be): Likewise.
32034 (mov_sitf_e500_subreg8_2_le): Likewise.
32035 (mov_sitf_e500_subreg12_2_be): Likewise.
32037 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32039 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
32040 attribute from vecsimple to vecperm.
32041 (altivec_vbpermq2): Likewise.
32043 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32046 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
32048 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32049 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
32050 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
32051 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
32052 case where N arg is SIZE_MAX.
32053 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
32054 (cmpstrsi): Add pattern.
32056 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
32058 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
32059 __builtin_vec_revb builtins.
32060 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
32061 built-in functions to support generation of the ISA 3.0 XXBR<x>
32062 vector byte reverse instructions.
32063 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
32064 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
32065 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
32066 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
32067 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
32068 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
32069 (P9V_BUILTIN_VEC_REVB): Likewise.
32070 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
32071 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
32072 (p9_xxbrq_v16qi): Likewise.
32073 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
32074 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
32075 (p9_xxbrh_v8hi): Likewise.
32076 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
32077 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
32078 vec_revb built-in functions.
32080 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
32082 PR rtl-optimization/78952
32083 * config/i386/i386.md (any_extract): New code iterator.
32084 (*insvqi_2): Use any_extract for source operand.
32085 (*insvqi_3): Use any_shiftrt for source operand.
32087 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
32089 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
32091 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
32093 2017-01-18 Matthias Klose <doko@ubuntu.com>
32095 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
32097 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32099 * config/rs6000/altivec.h (vec_bperm): Change #define.
32100 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
32101 (altivec_vbpermq2): New define_insn.
32102 (altivec_vbpermd): Likewise.
32103 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
32104 function interface.
32105 (VBPERMD): Likewise.
32106 (VBPERM): New polymorphic function interface.
32107 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
32108 Add entries for P9V_BUILTIN_VEC_VBPERM.
32109 * doc/extend.texi: Add interfaces for vec_bperm.
32111 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32113 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
32114 first letter of error messages.
32115 (s390_resolve_overloaded_builtin): Likewise.
32116 * config/s390/s390.c (s390_expand_builtin): Likewise.
32117 (s390_invalid_arg_for_unprototyped_fn): Likewise.
32118 (s390_valid_target_attribute_inner_p): Likewise.
32119 * config/s390/s390.md ("tabort"): Likewise.
32121 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
32123 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
32124 (ISA_AVOID_DIV_HILO): New macro.
32125 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
32126 (ISA_HAS_DDIV): Likewise.
32128 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32130 * doc/invoke.texi (fabi-version): Correct number of occurrences.
32132 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32134 * doc/invoke.texi (fabi-version): Spelling fix.
32136 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32139 * doc/invoke.texi (fabi-version): Mention mangling fix for
32142 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32145 * doc/invoke.texi (fabi-version): Document discriminator mangling.
32147 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
32150 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
32151 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
32153 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
32155 (stack_protect_test): Ditto.
32156 * config/rs6000/rs6000.opt (mstack-protector-guard=,
32157 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
32159 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
32160 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
32161 -mstack-protector-guard-offset=.
32162 (RS/6000 and PowerPC Options): Ditto.
32164 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32166 * config/i386/i386.h (MASK_CLASS_P): New define.
32167 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
32168 there are no registers from different register sets also when
32169 mask registers are used. Update function comment.
32170 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
32171 to (*k/*r) and (*k/*km) alternatives.
32173 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
32175 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
32176 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
32177 (EH_RETURN_HANDLER_RTX): New define.
32178 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
32179 Force frame pointer in EH return functions.
32180 (aarch64_expand_epilogue): Add barrier for eh_return.
32181 (aarch64_final_eh_return_addr): Remove.
32182 (aarch64_eh_return_handler_rtx): New function.
32183 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
32185 (aarch64_eh_return_handler_rtx): New prototype.
32187 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32189 * config/rs6000/altivec.h (vec_rlmi): New #define.
32190 (vec_vrlnm): Likewise.
32191 (vec_rlnm): Likewise.
32192 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
32193 (UNSPEC_VRLNM): Likewise.
32194 (VIlong): New mode iterator.
32195 (altivec_vrl<VI_char>mi): New define_insn.
32196 (altivec_vrl<VI_char>nm): Likewise.
32197 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
32199 (VRLDNM): Likewise.
32200 (RLNM): New polymorphic function entry.
32201 (VRLWMI): New monomorphic function entry.
32202 (VRLDMI): Likewise.
32203 (RLMI): New polymorphic function entry.
32204 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
32205 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
32206 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
32209 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32212 * dwarf2out.c (field_byte_offset): Restore the
32213 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
32214 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
32215 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
32218 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32221 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
32223 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32226 * dwarf2out.c (add_data_member_location_attribute): For constant
32227 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
32228 instead of DW_AT_data_member_location, DW_AT_bit_offset and
32229 DW_AT_byte_size attributes.
32231 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32233 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
32234 after forcing to constant memory when the code model is medium.
32236 2017-01-17 Julia Koval <julia.koval@intel.com>
32239 * config/i386/avx512fintrin.h
32240 (_mm512_i32gather_ps): Change __addr type to void const*.
32241 (_mm512_mask_i32gather_ps): Ditto.
32242 (_mm512_i32gather_pd): Ditto.
32243 (_mm512_mask_i32gather_pd): Ditto.
32244 (_mm512_i64gather_ps): Ditto.
32245 (_mm512_mask_i64gather_ps): Ditto.
32246 (_mm512_i64gather_pd): Ditto.
32247 (_mm512_mask_i64gather_pd): Ditto.
32248 (_mm512_i32gather_epi32): Ditto.
32249 (_mm512_mask_i32gather_epi32): Ditto.
32250 (_mm512_i32gather_epi64): Ditto.
32251 (_mm512_mask_i32gather_epi64): Ditto.
32252 (_mm512_i64gather_epi32): Ditto.
32253 (_mm512_mask_i64gather_epi32): Ditto.
32254 (_mm512_i64gather_epi64): Ditto.
32255 (_mm512_mask_i64gather_epi64): Ditto.
32256 (_mm512_i32scatter_ps): Change __addr type to void*.
32257 (_mm512_mask_i32scatter_ps): Ditto.
32258 (_mm512_i32scatter_pd): Ditto.
32259 (_mm512_mask_i32scatter_pd): Ditto.
32260 (_mm512_i64scatter_ps): Ditto.
32261 (_mm512_mask_i64scatter_ps): Ditto.
32262 (_mm512_i64scatter_pd): Ditto.
32263 (_mm512_mask_i64scatter_pd): Ditto.
32264 (_mm512_i32scatter_epi32): Ditto.
32265 (_mm512_mask_i32scatter_epi32): Ditto.
32266 (_mm512_i32scatter_epi64): Ditto.
32267 (_mm512_mask_i32scatter_epi64): Ditto.
32268 (_mm512_i64scatter_epi32): Ditto.
32269 (_mm512_mask_i64scatter_epi32): Ditto.
32270 (_mm512_i64scatter_epi64): Ditto.
32271 (_mm512_mask_i64scatter_epi64): Ditto.
32272 * config/i386/avx512pfintrin.h
32273 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
32274 (_mm512_mask_prefetch_i32gather_ps): Ditto.
32275 (_mm512_mask_prefetch_i64gather_pd): Ditto.
32276 (_mm512_mask_prefetch_i64gather_ps): Ditto.
32277 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
32278 (_mm512_prefetch_i32scatter_ps): Ditto.
32279 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
32280 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
32281 (_mm512_prefetch_i64scatter_pd): Ditto.
32282 (_mm512_prefetch_i64scatter_ps): Ditto.
32283 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
32284 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
32285 * config/i386/avx512vlintrin.h
32286 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
32287 (_mm_mmask_i32gather_ps): Ditto.
32288 (_mm256_mmask_i32gather_pd): Ditto.
32289 (_mm_mmask_i32gather_pd): Ditto.
32290 (_mm256_mmask_i64gather_ps): Ditto.
32291 (_mm_mmask_i64gather_ps): Ditto.
32292 (_mm256_mmask_i64gather_pd): Ditto.
32293 (_mm_mmask_i64gather_pd): Ditto.
32294 (_mm256_mmask_i32gather_epi32): Ditto.
32295 (_mm_mmask_i32gather_epi32): Ditto.
32296 (_mm256_mmask_i32gather_epi64): Ditto.
32297 (_mm_mmask_i32gather_epi64): Ditto.
32298 (_mm256_mmask_i64gather_epi32): Ditto.
32299 (_mm_mmask_i64gather_epi32): Ditto.
32300 (_mm256_mmask_i64gather_epi64): Ditto.
32301 (_mm_mmask_i64gather_epi64): Ditto.
32302 (_mm256_i32scatter_ps): Change __addr type to void*.
32303 (_mm256_mask_i32scatter_ps): Ditto.
32304 (_mm_i32scatter_ps): Ditto.
32305 (_mm_mask_i32scatter_ps): Ditto.
32306 (_mm256_i32scatter_pd): Ditto.
32307 (_mm256_mask_i32scatter_pd): Ditto.
32308 (_mm_i32scatter_pd): Ditto.
32309 (_mm_mask_i32scatter_pd): Ditto.
32310 (_mm256_i64scatter_ps): Ditto.
32311 (_mm256_mask_i64scatter_ps): Ditto.
32312 (_mm_i64scatter_ps): Ditto.
32313 (_mm_mask_i64scatter_ps): Ditto.
32314 (_mm256_i64scatter_pd): Ditto.
32315 (_mm256_mask_i64scatter_pd): Ditto.
32316 (_mm_i64scatter_pd): Ditto.
32317 (_mm_mask_i64scatter_pd): Ditto.
32318 (_mm256_i32scatter_epi32): Ditto.
32319 (_mm256_mask_i32scatter_epi32): Ditto.
32320 (_mm_i32scatter_epi32): Ditto.
32321 (_mm_mask_i32scatter_epi32): Ditto.
32322 (_mm256_i32scatter_epi64): Ditto.
32323 (_mm256_mask_i32scatter_epi64): Ditto.
32324 (_mm_i32scatter_epi64): Ditto.
32325 (_mm_mask_i32scatter_epi64): Ditto.
32326 (_mm256_i64scatter_epi32): Ditto.
32327 (_mm256_mask_i64scatter_epi32): Ditto.
32328 (_mm_i64scatter_epi32): Ditto.
32329 (_mm_mask_i64scatter_epi32): Ditto.
32330 (_mm256_i64scatter_epi64): Ditto.
32331 (_mm256_mask_i64scatter_epi64): Ditto.
32332 (_mm_i64scatter_epi64): Ditto.
32333 (_mm_mask_i64scatter_epi64): Ditto.
32334 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
32335 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
32336 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
32337 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
32338 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
32339 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
32340 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
32341 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
32342 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
32343 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
32344 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
32345 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
32346 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
32347 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
32348 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
32349 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
32350 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
32351 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
32352 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
32353 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
32354 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
32355 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
32356 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
32357 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
32358 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
32359 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
32360 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
32361 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
32362 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
32363 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
32364 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
32365 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
32366 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
32367 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
32368 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
32369 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
32370 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
32371 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
32372 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
32373 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
32374 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
32375 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
32376 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
32377 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
32378 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
32379 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
32380 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
32381 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
32382 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
32383 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
32384 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
32385 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
32386 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
32387 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
32388 definitions accordingly.
32390 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
32391 Kuan-Lin Chen <kuanlinchentw@gmail.com>
32394 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
32397 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
32400 * ira-conflicts.c (ira_build_conflicts): Update total conflict
32401 hard regs for inner regno.
32403 2017-01-17 Martin Liska <mliska@suse.cz>
32406 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
32407 assumption and add comment.
32409 2017-01-17 Nathan Sidwell <nathan@acm.org>
32411 * ipa-visibility.c (localize_node): New function, broken out of ...
32412 (function_and_variable_visibility): ... here. Call it.
32414 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
32416 PR middle-end/77445
32417 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
32418 correctly set frequency of oudgoing edge.
32419 (duplicate_thread_path): Fix profile updating.
32421 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32424 * configure.ac: Add GCC_BASE_VER.
32425 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
32426 version from BASE-VER file.
32427 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
32428 (gcc.o): Depend on $(BASEVER).
32429 * common.opt (dumpfullversion): New option.
32430 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
32431 * doc/invoke.texi: Document -dumpfullversion.
32432 * doc/install.texi: Document --with-gcc-major-version-only.
32433 * configure: Regenerated.
32435 2017-01-17 Richard Biener <rguenther@suse.de>
32437 PR tree-optimization/71433
32438 * tree-vrp.c (register_new_assert_for): Merge same asserts
32439 on all incoming edges.
32440 (process_assert_insertions_for): Handle insertions at the
32443 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
32445 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
32446 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
32448 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
32451 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
32454 2017-01-17 Alan Modra <amodra@gmail.com>
32457 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
32458 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
32459 symbolic stack limit when pic.
32461 2017-01-16 Martin Sebor <msebor@redhat.com>
32463 PR tree-optimization/78608
32464 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
32466 2017-01-16 Jeff Law <law@redhat.com>
32469 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
32470 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
32471 for several include directories that may be relative to sysroot.
32472 * config/i386/x-mingw32 (gplus_includedir): Define.
32473 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
32474 (native_system_includedir): Likewise.
32475 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
32476 override if TARGET_SYSTEM_ROOT is defined.
32477 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
32479 PR tree-optimization/79090
32480 PR tree-optimization/33562
32481 PR tree-optimization/61912
32482 PR tree-optimization/77485
32483 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
32484 and computed trims into the dump file.
32486 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32488 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
32490 2017-01-16 Jakub Jelinek <jakub@redhat.com>
32493 * gimplify.c (gimplify_init_constructor): If want_value and
32494 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
32498 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
32499 sequence. Formatting fixes.
32500 (doloop_optimize): Formatting fixes.
32503 * gcc.c (debug_level_greater_than_spec_func): New function.
32504 (static_spec_functions): Add debug-level-gt spec function.
32505 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
32507 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32508 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32509 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
32510 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
32511 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
32512 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
32514 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
32516 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
32517 QImode fixups to general and mask registers only.
32519 2017-01-16 Carl Love <cel@us.ibm.com>
32521 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
32522 for built-in functions
32523 vector signed char vec_nabs (vector signed char)
32524 vector signed short vec_nabs (vector signed short)
32525 vector signed int vec_nabs (vector signed int)
32526 vector signed long long vec_nabs (vector signed long long)
32527 vector float vec_nabs (vector float)
32528 vector double vec_nabs (vector double)
32529 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
32531 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
32532 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
32533 * doc/extend.texi: Update the documentation file for the new built-in
32536 2017-01-16 Martin Sebor <msebor@redhat.com>
32538 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
32541 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32543 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
32544 UNSPEC_VSX__XXSPLTD to require special splat handling.
32546 2017-01-16 David Malcolm <dmalcolm@redhat.com>
32549 * system.h: Poison strndup.
32551 2017-01-16 Alan Modra <amodra@gmail.com>
32554 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
32557 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
32559 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
32561 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
32563 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
32564 call recog here. Assert that INSN_CODE (insn) is non-negative.
32566 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32569 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
32571 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
32572 in the currently scheduled RTL fragment.
32574 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32576 PR rtl-optimization/78751
32577 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
32580 2017-01-14 Jeff Law <law@redhat.com>
32582 PR tree-optimization/79090
32583 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
32584 variable length stores.
32585 (compute_trims): Delete dead assignment to *trim_tail.
32586 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
32589 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
32591 PR rtl-optimization/78626
32592 PR rtl-optimization/78727
32593 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
32594 of a block, and split such blocks after everything else is finished.
32596 2017-01-14 Alan Modra <amodra@gmail.com>
32599 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
32600 target legitimate_combined_insn.
32601 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
32602 (rs6000_legitimate_combined_insn): New function.
32603 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
32605 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
32606 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
32607 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
32609 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
32611 * doc/frontends.texi (G++ and GCC): Remove references to Java.
32613 2017-01-13 Jeff Law <law@redhat.com>
32615 PR tree-optimization/33562
32616 PR tree-optimization/61912
32617 PR tree-optimization/77485
32618 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
32620 (delete_dead_assignment): Likewise.
32621 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
32622 statement to delete_dead_call and delete_dead_assignment.
32624 2017-01-13 David Malcolm <dmalcolm@redhat.com>
32627 * substring-locations.c (format_warning_va): Strengthen case 1 so
32628 that both endpoints of the substring must be within the format
32629 range for just the substring to be printed.
32631 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
32633 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
32634 * config/i386/i386.c (ix86_target_string): Add missing options
32635 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
32636 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
32637 flags_other and ix86_target_other to flags2_other. Display unknown
32639 (ix86_valid_target_attribute_inner_p): Add missing options and
32640 reorder options by implied ISAs, as in ix86_target_string.
32642 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
32644 * hash-table.h (hash_table::too_empty_p): New function.
32645 (hash_table::expand): Use it.
32646 (hash_table::traverse): Likewise.
32647 (hash_table::empty_slot): Use sizeof (value_type) instead of
32648 sizeof (PTR) to convert bytes to elements. Shrink the table
32649 if the current size is excessive for the current number of
32652 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
32654 * ira-costs.c (record_reg_classes): Break from the inner loop
32655 early once alt_fail is known to be true. Update outer loop
32656 handling accordingly.
32658 2017-01-13 Jeff Law <law@redhat.com>
32660 * tree-ssa-dse.c (decrement_count): New function.
32661 (increment_start_addr, maybe_trim_memstar_call): Likewise.
32662 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
32663 when we know the partially dead statement is a mem* function.
32665 PR tree-optimization/61912
32666 PR tree-optimization/77485
32667 * tree-ssa-dse.c: Include expr.h.
32668 (maybe_trim_constructor_store): New function.
32669 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
32671 PR tree-optimization/33562
32672 PR tree-optimization/61912
32673 PR tree-optimization/77485
32674 * doc/invoke.texi: Document new dse-max-object-size param.
32675 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
32676 * tree-ssa-dse.c: Include params.h.
32677 (dse_store_status): New enum.
32678 (initialize_ao_ref_for_dse): New, partially extracted from
32680 (valid_ao_ref_for_dse, normalize_ref): New.
32681 (setup_live_bytes_from_ref, compute_trims): Likewise.
32682 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
32683 (maybe_trim_partially_dead_store): Likewise.
32684 (maybe_trim_complex_store): Likewise.
32685 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
32686 Track what bytes live from the original store. Return tri-state
32687 for dead, partially dead or live.
32688 (dse_dom_walker): Add constructor, destructor and new private members.
32689 (delete_dead_call, delete_dead_assignment): New extracted from
32691 (dse_optimize_stmt): Make a member of dse_dom_walker.
32692 Use initialize_ao_ref_for_dse.
32694 PR tree-optimization/33562
32695 PR tree-optimization/61912
32696 PR tree-optimization/77485
32697 * sbitmap.h (bitmap_count_bits): Prototype.
32698 (bitmap_clear_range, bitmap_set_range): Likewise.
32699 * sbitmap.c (bitmap_clear_range): New function.
32700 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
32702 2017-01-13 Martin Liska <mliska@suse.cz>
32705 * function.c (set_cfun): Add new argument force.
32706 * function.h (set_cfun): Likewise.
32707 * ipa-inline-transform.c (inline_call): Use the function when
32708 strict alising from is dropped for function we inline to.
32710 2017-01-13 Richard Biener <rguenther@suse.de>
32712 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
32713 for dumping GIMPLE INTEGER_CSTs.
32715 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
32717 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
32718 to 201112L since C++17.
32720 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
32723 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
32724 if -fsanitize=kernel-address is present.
32726 2017-01-13 Richard Biener <rguenther@suse.de>
32728 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
32729 as _Literal ( type ) number in case usual suffixes do not
32730 preserve all information.
32732 2017-01-13 Richard Biener <rguenther@suse.de>
32734 PR tree-optimization/77283
32735 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
32736 and ssa-iterators.h.
32737 (is_feasible_trace): Implement a cost model based on joiner
32740 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
32743 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
32744 char or short to __float128/_Float128 directly.
32746 2017-01-12 Martin Sebor <msebor@redhat.com>
32748 to -Wformat-overflow.
32749 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
32750 (min_bytes_remaining): Same.
32751 (get_string_length): Same.
32752 (format_string): Same.
32753 (format_directive): Same.
32755 (pass_sprintf_length::handle_gimple_call): Same.
32757 2017-01-12 Jakub Jelinek <jakub@redhat.com>
32759 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
32760 info.nowrite calls with no lhs that can't throw. Return bool
32761 whether gsi_remove has been called or not.
32762 (pass_sprintf_length::handle_gimple_call): Return bool whether
32763 try_substitute_return_value called gsi_remove. Formatting fix.
32764 (pass_sprintf_length::execute): Don't use gsi_remove if
32765 handle_gimple_call returned true.
32768 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
32769 be removed due to side-effects, don't remove following barrier nor
32770 turn the successor edge into fallthru edge.
32772 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32775 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
32776 element-reversing loads and stores as not swappable.
32778 2017-01-12 Nathan Sidwell <nathan@acm.org>
32779 Nicolai Stange <nicstange@gmail.com>
32781 * combine.c (try_combine): Don't ignore result of overlap checking
32782 loop. Combine overlap & asm check into single loop.
32784 2017-01-12 Richard Biener <rguenther@suse.de>
32786 * tree-pretty-print.c (dump_generic_node): Provide -gimple
32787 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
32789 2017-01-12 Richard Biener <rguenther@suse.de>
32791 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
32792 and TS_TARGET_OPTION directly derive from TS_BASE.
32793 * tree-core.h (tree_optimization_option): Derive from tree_base.
32794 (tree_target_option): Likewise.
32796 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
32798 * config/i386/i386.c (memory_address_length): Increase len
32799 only when rip_relative_addr_p returns false.
32801 2017-01-11 Julia Koval <julia.koval@intel.com>
32803 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
32804 (OPTION_MASK_ISA_SGX_SET): New.
32805 (ix86_handle_option): Handle OPT_msgx.
32806 * config.gcc: Added sgxintrin.h.
32807 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
32808 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
32809 * config/i386/i386.c (ix86_target_string): Add -msgx.
32811 (ix86_option_override_internal): Handle new options.
32812 (ix86_valid_target_attribute_inner_p): Add sgx.
32813 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
32814 * config/i386/i386.opt: Add msgx.
32815 * config/i386/sgxintrin.h: New file.
32816 * config/i386/x86intrin.h: Add sgxintrin.h.
32818 2017-01-11 Jakub Jelinek <jakub@redhat.com>
32821 * fold-const.c (maybe_nonzero_address): Return 1 for function
32823 (tree_single_nonzero_warnv_p): Don't handle function local objects
32827 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
32830 2017-01-11 David Malcolm <dmalcolm@redhat.com>
32833 * opts.c: Include "spellcheck.h"
32834 (struct string_fragment): New struct.
32835 (struct edit_distance_traits<const string_fragment &>): New
32837 (get_closest_sanitizer_option): New function.
32838 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
32840 2017-01-11 Jakub Jelinek <jakub@redhat.com>
32842 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
32844 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
32845 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
32846 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
32847 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
32848 for initial die_offset if dwarf_split_debug_info.
32849 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
32850 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
32852 (output_skeleton_debug_sections): Formatting fix. Use
32853 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
32854 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
32856 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
32858 * config/arm/cortex-a53.md: Add bypasses for
32859 cortex_a53_r2f_cvt.
32860 (cortex_a53_r2f): Only use for transfers.
32861 (cortex_a53_f2r): Likewise.
32862 (cortex_a53_r2f_cvt): Add reservation for conversions.
32863 (cortex_a53_f2r_cvt): Likewise.
32865 2017-01-11 Tamar Christina <tamar.christina@arm.com>
32867 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
32868 to all inlined functions, change static to extern.
32870 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
32873 * config/arm/arm.c (legitimize_pic_address): Handle reference to
32875 (arm_assemble_integer): Likewise.
32877 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
32879 * config.gcc: Use new awk script to check CPU, FPU and architecture
32880 parameters for --with-... options.
32881 * config/arm/parsecpu.awk: New file
32882 * config/arm/arm-cpus.in: New file.
32883 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
32885 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
32887 * config/arm/t-arm: Update dependency rules.
32888 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
32889 of processing .def files.
32890 * config/arm/genopt.sh: Deleted.
32891 * config/arm/gentune.sh: Deleted.
32892 * config/arm/arm-cores.def: Deleted.
32893 * config/arm/arm-arches.def: Deleted.
32894 * config/arm/arm-fpus.def: Deleted.
32895 * config/arm/arm-tune.md: Regenerated.
32896 * config/arm/arm-tables.opt: Regenerated.
32897 * config/arm/arm-cpu.h: New generated file.
32898 * config/arm/arm-cpu-data.h: New generated file.
32899 * config/arm/arm-cpu-cdata.h: New generated file.
32901 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
32904 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
32906 (input_varpool_node): Unpack dynamically_initialized bit.
32908 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
32910 PR rtl-optimization/79032
32911 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
32912 the alignment of the adjusted memory reference against that of MODE,
32913 instead of the alignment of the original memory reference.
32915 2017-01-11 Martin Jambor <mjambor@suse.cz>
32917 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
32919 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
32920 decorated functions.
32922 2017-01-11 Richard Biener <rguenther@suse.de>
32924 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
32925 set range/nonnull info for PHI results. Do not set it on
32926 stmts marked for removal.
32928 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
32930 * expr.c (store_field): In the bitfield case, fetch the return value
32931 from the registers before applying a single big-endian adjustment.
32932 Always do a final load for a BLKmode value not larger than a word.
32934 2017-01-10 David Malcolm <dmalcolm@redhat.com>
32937 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
32938 that we correctly handle column numbers greater than
32939 LINE_MAP_MAX_COLUMN_NUMBER.
32941 2017-01-10 Martin Sebor <msebor@redhat.com>
32943 PR middle-end/78245
32944 * gimple-ssa-sprintf.c (get_destination_size): Call
32945 {init,fini}object_sizes.
32946 * tree-object-size.c (addr_object_size): Adjust.
32947 (pass_through_call): Adjust.
32948 (pass_object_sizes::execute): Adjust.
32949 * tree-object-size.h (fini_object_sizes): Declare.
32951 2017-01-10 Martin Sebor <msebor@redhat.com>
32953 PR tree-optimization/78775
32954 * builtins.c (get_size_range): Move...
32955 * calls.c: ...to here.
32956 (alloc_max_size): Accept zero argument.
32957 (operand_signed_p): Remove.
32958 (maybe_warn_alloc_args_overflow): Call get_size_range.
32959 * calls.h (get_size_range): Declare.
32961 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
32963 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
32964 from TI's devices.csv file as of September 2016.
32965 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
32967 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
32969 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
32970 * doc/invoke.texi: Likewise.
32971 * doc/md.texi: Likewise.
32972 * doc/objc.texi: Likewise.
32974 2017-01-10 Joshua Conner <joshconner@google.com>
32976 * config/arm/fuchsia-elf.h: New file.
32977 * config/fuchsia.h: New file.
32978 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
32979 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
32981 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
32983 2016-01-10 Richard Biener <rguenther@suse.de>
32985 PR tree-optimization/79034
32986 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
32987 Propagate out degenerate PHIs in the joiner.
32989 2017-01-10 Martin Liska <mliska@suse.cz>
32991 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
32992 (sort_congruence_classes_by_decl_uid): Likewise.
32993 (sort_congruence_class_groups_by_decl_uid): Likewise.
32994 (sem_item_optimizer::merge_classes): Sort class, groups in these
32995 classes and members in the groups by DECL_UID of declarations.
32996 This would make merge operations stable.
32998 2017-01-10 Martin Liska <mliska@suse.cz>
33000 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
33001 usage of m_classes_vec.
33002 (sem_item_optimizer::~sem_item_optimizer): Likewise.
33003 (sem_item_optimizer::get_group_by_hash): Likewise.
33004 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
33005 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
33006 (sem_item_optimizer::verify_classes): Likewise.
33007 (sem_item_optimizer::process_cong_reduction): Likewise.
33008 (sem_item_optimizer::dump_cong_classes): Likewise.
33009 (sem_item_optimizer::merge_classes): Likewise.
33010 * ipa-icf.h (congruence_class_hash): Rename from
33011 congruence_class_group_hash. Remove declaration of m_classes_vec.
33013 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
33015 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
33016 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
33017 * config.gcc: Add avx512vpopcntdqintrin.h.
33018 * config/i386/avx512vpopcntdqintrin.h: New.
33019 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
33020 * config/i386/i386-builtin-types.def: Add new types.
33021 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
33022 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
33023 __builtin_ia32_vpopcountq_v8di_mask): New.
33024 * config/i386/i386-c.c (ix86_target_macros_internal): Define
33025 __AVX512VPOPCNTDQ__.
33026 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
33027 (PTA_AVX512VPOPCNTDQ): Define.
33028 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
33029 TARGET_AVX512VPOPCNTDQ_P): Define.
33030 * config/i386/i386.opt: Add mavx512vpopcntdq.
33031 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
33032 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
33034 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33036 PR middle-end/77484
33037 * predict.def (PRED_CALL): Set to 67.
33039 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
33041 * expr.c (store_field): In the bitfield case, if the value comes from
33042 a function call and is of an aggregate type returned in registers, do
33043 not modify the field mode; extract the value in all cases if the mode
33044 is BLKmode and the size is not larger than a word.
33046 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
33049 * config/i386/cpuid.h: Fix undefined behavior.
33051 2017-01-04 Jeff Law <law@redhat.com>
33053 PR tree-optimization/79007
33054 PR tree-optimization/67955
33055 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
33056 conservative for pt.null when flag_non_call_exceptions is on.
33058 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33060 PR translation/79019
33061 PR translation/79020
33062 * params.def (PARAM_INLINE_MIN_SPEEDUP,
33063 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
33064 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
33066 * config/avr/avr.opt (maccumulate-args): Likewise.
33067 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
33068 * common.opt (freport-bug): Likewise.
33069 * cif-code.def (CIF_FINAL_ERROR): Likewise.
33070 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
33071 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
33072 translatable string.
33073 * config/i386/i386.c (function_value_32): Likewise.
33074 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
33075 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
33077 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
33078 * common/config/msp430/msp430-common.c (msp430_handle_option):
33080 * symtab.c (symtab_node::verify_base): Likewise.
33081 * opts.c (set_debug_level): Likewise.
33082 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
33083 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
33084 missing whitespace to translatable strings.
33085 * config/avr/avr.md (bswapsi2): Fix typo in comment.
33086 * config/sh/superh.h: Likewise.
33087 * config/i386/xopintrin.h: Likewise.
33088 * config/i386/znver1.md: Likewise.
33089 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
33090 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
33091 * double-int.h (struct double_int): Likewise.
33092 * double-int.c (div_and_round_double): Likewise.
33093 * wide-int.cc: Likewise.
33094 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
33095 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
33096 * cfgcleanup.c (crossjumps_occured): Renamed to ...
33097 (crossjumps_occurred): ... this.
33098 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
33101 PR tree-optimization/78899
33102 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
33103 returning bool return struct loop *, NULL for failure and the new
33105 (versionable_outer_loop_p): Don't version outer loop if it has
33106 dont_vectorized bit set.
33107 (tree_if_conversion): When versioning outer loop, ensure
33108 tree_if_conversion is performed also on the inner loop of the
33109 non-vectorizable outer loop copy.
33110 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
33111 LOOP_VECTORIZED in inner loop of the scalar outer loop and
33112 prevent vectorization of it.
33113 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
33114 the outer loop vectorization of the non-scalar version is attempted
33115 before vectorization of the inner loop in scalar version. If
33116 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
33117 vectorization of its inner loop.
33118 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
33119 has 2 inner loops, rename also on edges from bb whose single pred
33120 is outer_loop->header. Fix typo in function comment.
33122 2017-01-09 Martin Sebor <msebor@redhat.com>
33125 * asan.c (asan_emit_stack_protection): Increase local buffer size
33126 to avoid snprintf truncation warning.
33128 2017-01-09 Andrew Pinski <apinski@cavium.com>
33130 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
33131 to reference thunderx2t99 for the tuning structure
33132 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
33134 (thunderx2t99_extra_costs): This.
33135 * config/aarch64/aarch64-tune.md: Regenerate.
33136 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
33137 (vulcan_addrcost_table): This.
33138 (vulcan_regmove_cost): Rename to ...
33139 (thunderx2t99_regmove_cost): This.
33140 (vulcan_vector_cost): Rename to ...
33141 (thunderx2t99_vector_cost): this.
33142 (vulcan_branch_cost): Rename to ...
33143 (thunderx2t99_branch_cost): This.
33144 (vulcan_tunings): Rename to ...
33145 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
33146 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
33148 2017-01-09 Martin Jambor <mjambor@suse.cz>
33152 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
33153 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
33154 (propagate_vr_accross_jump_function): Use the above function for all
33155 value range computations for pass-through jump functions and type
33156 converasion from explicit value range values.
33157 (ipcp_propagate_stage): Do not attempt to deduce types of formal
33158 parameters from TYPE_ARG_TYPES.
33159 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
33160 (ipa_write_node_info): Stream type of the actual argument.
33161 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
33163 2017-01-09 Martin Liska <mliska@suse.cz>
33166 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
33167 (lookup_compiler): Do not show error message with have_E.
33169 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33171 PR tree-optimization/78938
33172 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
33173 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
33174 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
33175 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
33178 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33180 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
33183 2017-01-09 Richard Biener <rguenther@suse.de>
33185 PR tree-optimization/78997
33186 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
33187 name condition properly.
33189 2017-01-09 Richard Biener <rguenther@suse.de>
33192 * dwarf2out.c (is_cxx): New overload with context.
33193 (is_naming_typedef_decl): Use it.
33195 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33197 * invoke.texi (Option Summary): Correct spacing in option lists
33198 and add line breaks to fix over-long lines.
33200 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33202 PR middle-end/17660
33204 * extend.texi (Common Variable Attributes): Add xref to GCC
33205 Internals manual to explain mode attribute keywords.
33207 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33210 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
33211 and Preprocessor Options.
33212 (Options for Linking): Document -pthread here....
33213 (RS/6000 and PowerPC Options): ...not here.
33214 (Solaris 2 Options): ...or here.
33215 * doc/cppopts.texi: Document -pthread.
33217 2017-01-08 Martin Sebor <msebor@redhat.com>
33219 PR middle-end/77708
33220 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
33221 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
33222 New member functions.
33223 (format_directive): Used them.
33225 (pass_sprintf_length::handle_gimple_call): Same.
33226 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
33227 to avoid truncation for any argument.
33228 (extract_affine_mul): Same.
33229 * tree.c (get_file_function_name): Same.
33231 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33233 PR middle-end/77484
33234 * predict.def (PRED_INDIR_CALL): Set to 86.
33236 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33238 PR preprocessor/54124
33239 * doc/cppopts.texi: Reformat -d subtable to list the full name
33240 of the options. Add cross-reference to the docs for the general
33241 compiler -d options.
33242 * doc/invoke.texi (Developer Options): Add cross-reference to the
33243 preprocessor-specific -d option documentation.
33245 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33247 PR preprocessor/13498
33248 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
33249 redudant material, and reflect new command-line options.
33250 (System Headers): Likewise.
33252 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33254 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
33255 -isystem, and -idirafter. Copy-edit.
33256 * doc/cppopts.texi: Copy-edit. Remove contradiction about
33257 default for -ftrack-macro-expansion. Delete obsolete and
33258 badly-formatted implementation details about -fdebug-cpp output.
33259 * doc/cppwarnopts.texi: Copy-edit.
33261 2017-01-07 David Malcolm <dmalcolm@redhat.com>
33264 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
33265 that the transition from a max line width >= 1<<10 to narrower
33266 lines works correctly.
33268 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
33270 * doc/options.texi (PerFunction): New.
33271 * opt-functions.awk (switch_flags): Map both Optimization and
33272 PerFunction to CL_OPTIMIZATION.
33273 * opth-gen.awk: Test for PerFunction flag along with
33275 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
33276 it only when the latter is present. Skip those that don't in
33277 the hash function generator.
33278 * common.opt (fvar-tracking): Mark as PerFunction instead of
33280 (fvar-tracking-assignments): Likewise.
33281 (fvar-tracking-assignments-toggle): Likewise.
33282 (fvar-tracking-uninit): Likewise.
33284 2017-01-07 Jakub Jelinek <jakub@redhat.com>
33286 PR translation/79018
33287 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
33290 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
33293 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
33294 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
33295 TARGET_LONG_JUMP_TABLE_OFFSETS.
33296 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
33297 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
33298 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
33299 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
33300 * config/m68k/m68k.md (tablejump expander): Likewise.
33301 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
33302 TARGET_LONG_JUMP_TABLE_OFFSETS.
33303 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
33304 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
33306 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
33307 David Holsgrove <david.holsgrove@xilinx.com>
33309 * common/config/microblaze/microblaze-common.c
33310 (TARGET_EXCEPT_UNWIND_INFO): Remove.
33311 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
33313 * config/microblaze/microblaze.c (microblaze_must_save_register)
33314 (microblaze_expand_epilogue, microblaze_return_addr): Handle
33316 (microblaze_eh_return): New function.
33317 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
33318 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
33319 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
33320 * config/microblaze/microblaze.md (eh_return): New pattern.
33322 2017-01-06 Jakub Jelinek <jakub@redhat.com>
33324 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
33325 GCC_DIAGNOSTIC_STRINGIFY): Define.
33327 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
33329 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33331 * config/arm/arm.md (<mcrr>): New.
33333 * config/arm/arm.c (arm_arch5te): New.
33334 (arm_option_override): Set arm_arch5te.
33335 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
33337 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
33338 (arm_mcrr_qualifiers): ... this. New.
33339 (MRRC_QUALIFIERS): Define to...
33340 (arm_mrrc_qualifiers): ... this. New.
33341 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
33343 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
33344 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
33345 (MRRCI, mrrc, MRRC): New.
33346 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
33347 VUNSPEC_MRRC2): New.
33349 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33351 * config/arm/arm.md (<mcr>): New.
33353 * config/arm/arm.c (arm_coproc_builtin_available): Add
33354 support for mcr, mrc, mcr2 and mrc2.
33355 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
33356 (arm_mcr_qualifiers): ... this. New.
33357 (MRC_QUALIFIERS): Define to ...
33358 (arm_mrc_qualifiers): ... this. New.
33359 (MCR_QUALIFIERS): Define to ...
33360 (arm_mcr_qualifiers): ... this. New.
33361 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
33363 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
33364 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
33365 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
33366 VUNSPEC_MRC2): New.
33368 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33370 * config/arm/arm.md (*ldc): New.
33374 * config/arm/arm.c (arm_coproc_builtin_available): Add
33375 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
33376 (arm_coproc_ldc_stc_legitimate_address): New.
33377 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
33378 'qualifier_const_pointer'.
33379 (LDC_QUALIFIERS): Define to...
33380 (arm_ldc_qualifiers): ... this. New.
33381 (STC_QUALIFIERS): Define to...
33382 (arm_stc_qualifiers): ... this. New.
33383 * config/arm/arm-protos.h
33384 (arm_coproc_ldc_stc_legitimate_address): New.
33385 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
33386 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
33387 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
33388 stc2, stcl, stc2l): New.
33389 * config/arm/constraints.md (Uz): New.
33390 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
33391 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
33392 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
33393 VUNSPEC_STC2L): New.
33395 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33397 * config/arm/arm.md (<cdp>): New.
33398 * config/arm/arm.c (neon_const_bounds): Rename this ...
33399 (arm_const_bounds): ... this.
33400 (arm_coproc_builtin_available): New.
33401 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
33402 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
33403 (CDP_QUALIFIERS): Define to...
33404 (arm_cdp_qualifiers): ... this. New.
33406 (arm_expand_builtin_args): Add case for 6 arguments.
33407 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
33408 (arm_const_bounds): ... this.
33409 (arm_coproc_builtin_available): New.
33410 * config/arm/arm_acle.h (__arm_cdp): New.
33412 * config/arm/arm_acle_builtins.def (cdp): New.
33414 * config/arm/iterators.md (CDPI,CDP,cdp): New.
33415 * config/arm/neon.md: Rename all 'neon_const_bounds' to
33416 'arm_const_bounds'.
33417 * config/arm/types.md (coproc): New.
33418 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
33419 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
33420 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
33421 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
33423 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33425 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
33426 (UBINOP_QUALIFIERS): New.
33428 (acle_builtin_data): New. Change comment.
33429 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
33430 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
33431 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
33432 arm_acle_builtins.def.
33433 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
33434 (arm_init_acle_builtins): New.
33435 (CRC32_BUILTIN): Remove.
33436 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
33437 crc32cb, crc32ch and crc32cw.
33438 (arm_init_crc32_builtins): Remove.
33439 (arm_init_builtins): Use arm_init_acle_builtins rather
33440 than arm_init_crc32_builtins.
33441 (arm_expand_acle_builtin): New.
33442 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
33443 * config/arm/arm_acle_builtins.def: New.
33445 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33447 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
33448 (arm_builtin_datum): ... this.
33449 (arm_init_neon_builtin): Rename to ...
33450 (arm_init_builtin): ... this. Add a new parameters PREFIX
33451 and USE_SIG_IN_NAME.
33452 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
33453 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
33454 'arm_builtin_datum'.
33455 (arm_init_vfp_builtins): Likewise.
33456 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
33457 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
33458 (arm_expand_neon_args): Rename to ...
33459 (arm_expand_builtin_args): ... this. Rename builtin_arg
33460 enum values and differentiate between ARG_BUILTIN_MEMORY
33461 and ARG_BUILTIN_NEON_MEMORY.
33462 (arm_expand_neon_builtin_1): Rename to ...
33463 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
33464 values, arm_expand_builtin_args and add bool parameter NEON.
33465 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
33466 (arm_expand_vfp_builtin): Likewise.
33467 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
33469 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33471 PR middle-end/77484
33472 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
33473 * predict.c (tree_estimate_probability_bb): Reverse direction of
33474 polymorphic call predictor.
33476 2017-01-06 David Malcolm <dmalcolm@redhat.com>
33478 * passes.c (execute_one_pass): Split out pass-skipping logic into...
33479 (determine_pass_name_match): ...this new function and...
33480 (should_skip_pass_p): ...this new function.
33482 2017-01-06 Nathan Sidwell <nathan@acm.org>
33484 * ipa-visibility.c (function_and_variable_visibility): Reformat
33485 comments and long lines. Remove extrneous if.
33486 * symtab.c (symtab_node::make_decl_local): Fix code format.
33487 (symtab_node::set_section_for_node): Fix comment typo.
33489 2017-01-06 Martin Liska <mliska@suse.cz>
33492 * lra-constraints.c: Rename invariant to lra_invariant.
33493 * predict.c (set_even_probabilities): Initialize e to NULL.
33495 2017-01-05 Martin Sebor <msebor@redhat.com>
33497 PR tree-optimization/78910
33498 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
33499 (format_integer): Correct off-by-one error in the handling
33500 of precision with negative numbers in signed conversions..
33502 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
33504 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
33506 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33508 PR tree-optimization/71016
33509 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
33510 factor_out_conditional_conversion. Formatting fix.
33511 (factor_out_conditional_conversion): Add cond_stmt argument.
33512 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
33513 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
33516 2017-01-05 David Malcolm <dmalcolm@redhat.com>
33518 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
33519 read-rtl-function.o, and selftest-rtl.o.
33520 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
33521 (selftest::aarch64_test_loading_full_dump): New function.
33522 (selftest::aarch64_run_selftests): New function.
33523 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
33524 selftest::aarch64_run_selftests.
33525 * config/i386/i386.c
33526 (selftest::ix86_test_loading_dump_fragment_1): New function.
33527 (selftest::ix86_test_loading_call_insn): New function.
33528 (selftest::ix86_test_loading_full_dump): New function.
33529 (selftest::ix86_test_loading_unspec): New function.
33530 (selftest::ix86_run_selftests): Call the new functions.
33531 * emit-rtl.c (maybe_set_max_label_num): New function.
33532 * emit-rtl.h (maybe_set_max_label_num): New decl.
33533 * function.c (instantiate_decls): Guard call to
33534 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
33535 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
33537 * gensupport.c (gen_reader::gen_reader): Pass "false"
33538 for new "compact" param of rtx_reader.
33539 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
33540 rather than an empty string for NULL strings.
33541 * read-md.c: Potentially include config.h rather than bconfig.h.
33542 Wrap include of errors.h with #ifdef GENERATOR_FILE.
33543 (have_error): New global, copied from errors.c.
33544 (md_reader::read_name): Rename to...
33545 (md_reader::read_name_1): ...this, adding "out_loc" param,
33546 and converting "missing name or number" to returning false, rather
33548 (md_reader::read_name): Reimplement in terms of read_name_1.
33549 (md_reader::read_name_or_nil): New function.
33550 (md_reader::read_string): Handle "(nil)" by returning NULL.
33551 (md_reader::md_reader): Add new param "compact".
33552 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
33553 (md_reader::read_file): New method.
33554 * read-md.h (md_reader::md_reader): Add new param "compact".
33555 (md_reader::read_file): New method.
33556 (md_reader::is_compact): New accessor.
33557 (md_reader::read_name): Convert return type from void to file_location.
33558 (md_reader::read_name_or_nil): New decl.
33559 (md_reader::read_name_1): New decl.
33560 (md_reader::m_compact): New field.
33561 (noop_reader::noop_reader): Pass "false" for new "compact" param
33563 (rtx_reader::rtx_reader): Add new "compact" param.
33564 (rtx_reader::read_rtx_operand): Make virtual and convert return
33565 type from void to rtx.
33566 (rtx_reader::read_until): New decl.
33567 (rtx_reader::handle_any_trailing_information): New virtual function.
33568 (rtx_reader::postprocess): New virtual function.
33569 (rtx_reader::finalize_string): New virtual function.
33570 (rtx_reader::m_in_call_function_usage): New field.
33571 (rtx_reader::m_reuse_rtx_by_id): New field.
33572 * read-rtl-function.c: New file.
33573 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
33574 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
33575 (selftest::verify_three_block_rtl_cfg): New decl.
33576 * read-rtl-function.h: New file.
33577 * read-rtl.c: Potentially include config.h rather than bconfig.h.
33578 For host, include function.h, memmodel.h, and emit-rtl.h.
33579 (one_time_initialization): New function.
33580 (struct compact_insn_name): New struct.
33581 (compact_insn_names): New array.
33582 (find_code): Handle insn codes in compact dumps.
33583 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
33584 (bind_subst_iter_and_attr): Likewise.
33585 (add_condition_to_string): Likewise.
33586 (add_condition_to_rtx): Likewise.
33587 (apply_attribute_uses): Likewise.
33588 (add_current_iterators): Likewise.
33589 (apply_iterators): Likewise.
33590 (initialize_iterators): Guard usage of apply_subst_iterator with
33591 #ifdef GENERATOR_FILE.
33592 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
33593 (md_reader::read_mapping): Likewise.
33594 (add_define_attr_for_define_subst): Likewise.
33595 (add_define_subst_attr): Likewise.
33596 (read_subst_mapping): Likewise.
33597 (check_code_iterator): Likewise.
33598 (rtx_reader::read_rtx): Likewise. Move one-time initialization
33600 (one_time_initialization): New function.
33601 (rtx_reader::read_until): New method.
33602 (read_flags): New function.
33603 (parse_reg_note_name): New function.
33604 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
33605 Handle reuse_rtx ids.
33606 Wrap iterator lookup within #ifdef GENERATOR_FILE.
33607 Add parsing support for RTL dumps, mirroring the special-cases in
33608 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
33609 values, and calling handle_any_trailing_information.
33610 (rtx_reader::read_rtx_operand): Convert return type from void
33611 to rtx, returning return_rtx. Handle case 'e'. Call
33612 finalize_string on XSTR and XTMPL fields.
33613 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
33614 "(nil)" values were omitted. Call the postprocess vfunc on the
33616 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
33617 class ctor. Initialize m_in_call_function_usage. Call
33618 one_time_initialization.
33619 * rtl-tests.c (selftest::test_uncond_jump): Call
33620 set_new_first_and_last_insn.
33621 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
33622 * selftest-rtl.c: New file.
33623 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
33624 (selftest::get_insn_by_uid): New decl.
33625 * selftest-run-tests.c (selftest::run_tests): Call
33626 read_rtl_function_c_tests.
33627 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
33628 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
33631 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
33633 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
33634 operands in a special way. Assert that pos+len <= mode precision.
33636 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33638 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
33639 3 argument Alias with unlimited for the negative form.
33640 (fno-vect-cost-model): Removed.
33642 2017-01-05 Martin Liska <mliska@suse.cz>
33644 * hsa-gen.c (gen_hsa_divmod): New function.
33645 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
33647 2017-01-05 Martin Liska <mliska@suse.cz>
33650 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
33653 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33655 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
33656 small constant length operands.
33658 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33660 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
33661 between loop iterations.
33663 2017-01-05 Martin Liska <mliska@suse.cz>
33666 * gimplify.c (gimplify_decl_expr): Compare to
33667 asan_poisoned_variables instread of checking flags.
33668 (gimplify_target_expr): Likewise.
33669 (gimplify_expr): Likewise.
33670 (gimplify_function_tree): Conditionally initialize
33671 asan_poisoned_variables.
33673 2017-01-04 Jeff Law <law@redhat.com>
33675 PR tree-optimizatin/78812
33676 * rtl.h (contains_mem_rtx_p): Prototype.
33677 * ifcvt.c (containts_mem_rtx_p): Move from here to...
33678 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
33679 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
33680 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
33681 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
33683 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33685 * input.c (assert_char_at_range): Default-initialize actual_range.
33687 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33689 * df-scan.c (df_ref_create_structure): Make regno unsigned,
33690 to match the caller.
33692 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33694 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
33695 insns after final jump in test to emit dummy move.
33697 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33699 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
33700 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
33702 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33704 * multiple_target.c (create_dispatcher_calls): Init e_next.
33705 * tree-ssa-loop-split.c (split_loop): Init border.
33706 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
33709 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
33714 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
33715 (altivec_register_operand): Do not return true if the operand
33716 contains a SUBREG mixing SImode and SFmode.
33717 (vsx_register_operand): Likewise.
33718 (vsx_reg_sfsubreg_ok): New predicate.
33719 (vfloat_operand): Do not return true if the operand contains a
33720 SUBREG mixing SImode and SFmode.
33721 (vint_operand): Likewise.
33722 (vlogical_operand): Likewise.
33723 (gpc_reg_operand): Likewise.
33724 (int_reg_operand): Likewise.
33725 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
33726 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
33727 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
33729 (rs6000_emit_move_si_sf_subreg): New helper function.
33730 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
33731 fixup SUBREGs involving SImode and SFmode.
33732 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
33733 numbers for the new peephole2 optimization.
33734 (peephole2 for SFmode unions): New peephole2 to optimize cases in
33735 the GLIBC math library that do AND/IOR/XOR operations on single
33736 precision floating point.
33737 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
33738 target macros to say whether we need to avoid SUBREGs mixing
33740 (TARGET_ALLOW_SF_SUBREG): Likewise.
33741 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
33742 (UNSPEC_SI_FROM_SF): Likewise.
33743 (iorxor): Change spacing.
33744 (and_ior_xor): New iterator for AND, IOR, and XOR.
33745 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
33746 (movdi_from_sf_zero_ext): Likewise.
33747 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
33748 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
33749 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
33750 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
33751 (fms<mode>4): Likewise.
33752 (fnma<mode>4): Likewise.
33753 (fnms<mode>4): Likewise.
33754 (nfma<mode>4): Likewise.
33755 (nfms<mode>4): Likewise.
33757 2017-01-04 Marek Polacek <polacek@redhat.com>
33760 * doc/invoke.texi: Document -Wpointer-compare.
33762 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33764 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
33767 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
33768 descriptions for -gdwarf-5 and emit them as uleb128 instead of
33771 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
33774 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
33775 documentation of the powerpc_popcntb_ok attribute.
33776 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
33777 code to issue warning messages if a requested CPU configuration is
33778 not supported by the binary (assembler and loader) toolchain.
33779 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
33780 made to define a built-in function that has been disabled.
33781 (paired_init_builtins): Add assertion to prevent ICE if attempt is
33782 made to define a built-in function that has been disabled.
33783 (altivec_init_builtins): Add comment explaining why definition
33784 of the DST built-in functions is not preceded by an assertion
33785 check. Add assertions to prevent ICE if attempts are made to
33786 define an altivec predicate or an abs* built-in function that has
33788 (htm_init_builtins): Add comment explaining why definition of the
33789 htm built-in functions is not preceded by an assertion check.
33791 2017-01-04 Jeff Law <law@redhat.com>
33793 PR tree-optimizatin/67955
33794 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
33795 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
33796 the points-to solution does not include pt_null. Use DECL_PT_UID
33799 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
33801 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
33802 Use gen_int_mode instead of gen_lopwart for const_int operands.
33804 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33806 PR tree-optimization/71563
33807 * match.pd: Simplify X << Y into X if Y is known to be 0 or
33808 out of range value - has low bits known to be zero.
33810 2017-01-04 Alan Modra <amodra@gmail.com>
33812 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
33813 * configure: Regenerate.
33814 * config.in: Regenerate.
33816 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33819 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
33820 a substring of the message, but strcmp with the whole message. Ifdef
33821 ENABLE_NLS, translate the message first using dgettext.
33823 2017-01-03 Jeff Law <law@redhat.com>
33825 PR tree-optimizatin/78856
33826 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
33827 (mark_threaded_blocks): Remove code to truncate thread paths that
33828 cross multiple loop headers. Instead invalidate the cached loop
33829 iteration information and handle case of a thread path walking
33830 into an irreducible region.
33832 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
33835 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
33836 assertions. Add support for doing the signbit if the IEEE 128-bit
33837 floating point value is in a GPR.
33838 * config/rs6000/rs6000.md (Fsignbit): Delete.
33839 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
33840 Update the length attribute if the value is in a GPR.
33841 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
33842 the sign or zero extension instruction, since the value is always 0/1.
33843 (signbit<mode>2_dm2): Delete using <Fsignbit>.
33846 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
33847 extracting SImode to a GPR register so that we can generate a
33848 store, limit the vector to be in a traditional Altivec register
33849 for the vextuwrx instruction.
33851 2017-01-03 Ian Lance Taylor <iant@google.com>
33853 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
33855 2017-01-03 Martin Sebor <msebor@redhat.com>
33857 PR tree-optimization/78696
33858 * gimple-ssa-sprintf.c (format_floating): Correct handling of
33859 precision. Use MPFR for %f for greater fidelity. Correct handling
33861 (pass_sprintf_length::compute_format_length): Set width and precision
33862 specified by asrerisk to void_node for vararg functions.
33863 (try_substitute_return_value): Adjust dump output.
33865 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
33867 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
33869 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
33871 * doc/invoke.texi (SPARC options): Document -mlra as the default.
33872 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
33873 -mlra/-mno-lra was passed to the compiler.
33875 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
33877 PR rtl-optimization/65618
33878 * emit-rtl.c (try_split): Move initialization of "before" and
33879 "after" to just before the call to emit_insn_after_setloc.
33881 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
33883 * doc/md.texi (Standard Names): Remove reference to Java frontend.
33885 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
33887 * dwarf2out.c (gen_enumeration_type_die): When
33888 -gno-strict-dwarf, add a DW_AT_encoding attribute.
33890 2017-01-03 Jakub Jelinek <jakub@redhat.com>
33892 PR tree-optimization/78965
33893 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
33894 Change first argument from const call_info & to call_info &. For %n
33895 set info.nowrite to false.
33897 PR middle-end/78901
33898 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
33899 possibly throwing calls.
33901 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
33902 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
33903 and fns handling, rather than in a separate case SSA_NAME.
33905 2017-01-02 Jeff Law <law@redhat.com>
33907 * config/darwin-driver.c (darwin_driver_init): Const-correctness
33908 fixes for first_period and second_period variables.
33910 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
33913 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
33914 (*insvqi_1): New insn pattern.
33915 (*insvqi_1_mem_rex64): Ditto.
33916 (*insvqi_2): Ditto.
33917 (*insvqi_3): Rename from *insvqi.
33919 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
33921 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
33923 * doc/cfg.texi (Edges): Remove reference to Java.
33924 (Maintaining the CFG): Ditto.
33926 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33928 PR middle-end/77674
33929 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
33930 transparent aliases.
33932 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33934 PR middle-end/77484
33935 * predict.def (PRED_CALL): Update hitrate.
33936 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
33937 * predict.c (tree_estimate_probability_bb): Split CALL predictor
33938 into direct/indirect/polymorphic variants.
33940 2017-01-01 Jakub Jelinek <jakub@redhat.com>
33942 Update copyright years.
33944 * gcc.c (process_command): Update copyright notice dates.
33945 * gcov-dump.c (print_version): Ditto.
33946 * gcov.c (print_version): Ditto.
33947 * gcov-tool.c (print_version): Ditto.
33948 * gengtype.c (create_file): Ditto.
33949 * doc/cpp.texi: Bump @copying's copyright year.
33950 * doc/cppinternals.texi: Ditto.
33951 * doc/gcc.texi: Ditto.
33952 * doc/gccint.texi: Ditto.
33953 * doc/gcov.texi: Ditto.
33954 * doc/install.texi: Ditto.
33955 * doc/invoke.texi: Ditto.
33957 Copyright (C) 2017 Free Software Foundation, Inc.
33959 Copying and distribution of this file, with or without modification,
33960 are permitted in any medium without royalty provided the copyright
33961 notice and this notice are preserved.