1 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
3 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
5 2017-11-07 Jakub Jelinek <jakub@redhat.com>
8 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
9 (*cmp<mode>_ccz_1): New insn with $k alternative.
12 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
13 RTX_COMM_COMPARE as commutative as well.
14 (ix86_binary_operator_ok): Formatting fix.
15 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
16 *<code><mode>3<mask_name><round_saeonly_name>,
17 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
18 *<s>mul<mode>3_highpart<mask_name>,
19 *vec_widen_umult_even_v16si<mask_name>,
20 *vec_widen_umult_even_v8si<mask_name>,
21 *vec_widen_umult_even_v4si<mask_name>,
22 *vec_widen_smult_even_v16si<mask_name>,
23 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
24 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
25 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
26 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
27 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
28 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
29 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
30 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
31 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
32 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
33 ix86_binary_operator_ok. Formatting fixes.
34 (*<plusminus_insn><mode>3<mask_name><round_name>,
35 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
38 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
40 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
41 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
42 eq and ne if TARGET_ISEL.
43 (cmp): New code_iterator.
44 (UNS, UNSU_, UNSIK): New code_attrs.
45 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
46 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
48 ("eq<mode>3"): ... this.
49 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
51 ("ne<mode>3"): ... this.
53 2017-11-07 Julia Koval <julia.koval@intel.com>
56 * common/config/i386/i386-common.c
57 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
58 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
59 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
60 * config/i386/i386.opt: Ditto.
61 * config/i386/i386.c (ix86_target_string): Ditto.
62 (ix86_option_override_internal): Ditto.
63 (ix86_init_mpx_builtins): Move MPX to args2.
64 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
65 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
66 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
67 __builtin_ia32_vgf2p8affineinvqb_v32qi,
68 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
69 __builtin_ia32_vgf2p8affineinvqb_v16qi,
70 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
72 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
75 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
76 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
77 (zero-extendsidi peephole2): Remove peephole.
79 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
82 * stor-layout.c: Include attribs.h.
83 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
84 explicit lookup of "aligned" attribute.
86 2017-11-07 Andrew Waterman <andrew@sifive.com>
88 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
89 (riscv_expand_block_move): Likewise.
90 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
92 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
93 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
94 * config/riscv/riscv.c (riscv_block_move_straight): New function.
95 (riscv_adjust_block_mem): Likewise.
96 (riscv_block_move_loop): Likewise.
97 (riscv_expand_block_move): Likewise.
98 gcc/config/riscv/riscv.md (movmemsi): New pattern.
100 2017-11-07 Michael Clark <michaeljclark@mac.com>
102 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
103 (MUSL_DYNAMIC_LINKER): Likewise.
105 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
107 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
108 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
111 2017-11-07 Richard Biener <rguenther@suse.de>
113 * match.pd: Fix build.
115 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
116 Jackson Woodruff <jackson.woodruff@arm.com>
118 PR tree-optimization/71026
119 * match.pd: Canonicalize negate in division.
121 2017-11-07 Sudakshina Das <sudi.das@arm.com>
124 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
126 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
128 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
129 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
131 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
133 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
134 non-scalar integral types.
135 * match.pd (negate_expr_p): Handle MINUS_EXPR.
136 (-(A-B), -(~A)): New transformations.
138 2017-11-07 Tom de Vries <tom@codesourcery.com>
140 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
141 semicolon after "do {} while (0)".
142 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
143 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
144 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
145 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
146 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
148 2017-11-07 Tom de Vries <tom@codesourcery.com>
150 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
151 after "do {} while (0)".
152 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
153 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
154 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
155 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
156 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
158 2017-11-07 Tom de Vries <tom@codesourcery.com>
161 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
163 (arm_rtx_costs_internal): Add missing semicolon after
164 HANDLE_NARROW_SHIFT_ARITH call.
166 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
168 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
169 disable isel if it was not set explicitly.
171 2017-11-06 James Bowman <james.bowman@ftdichip.com>
173 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
175 (add_type_attribute) likewise.
177 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
179 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
180 of struct ix86_frame.
181 (ix86_initial_elimination_offset): Likewise.
182 (ix86_expand_split_stack_prologue): Likewise.
184 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
186 * tree-vrp.h (enum value_range_type): Update stale comment.
188 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
189 Alan Hayward <alan.hayward@arm.com>
190 David Sherwood <david.sherwood@arm.com>
192 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
193 (aarch64_expand_vec_perm_const): Take the number of units too.
194 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
195 (aarch64_expand_vec_perm_const): Likewise.
196 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
197 (vec_perm<mode>): Update accordingly.
199 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
200 Alan Hayward <alan.hayward@arm.com>
201 David Sherwood <david.sherwood@arm.com>
203 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
204 Take the number of units too.
205 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
206 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
207 but check for a vector mode before rather than after the call.
208 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
209 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
210 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
211 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
212 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
213 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
214 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
215 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
216 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
217 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
218 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
219 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
220 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
221 (aarch64_sqdmull2_n<mode>): Update accordingly.
223 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
224 Alan Hayward <alan.hayward@arm.com>
225 David Sherwood <david.sherwood@arm.com>
227 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
228 the number of units too.
229 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
230 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
231 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
232 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
233 (vec_store_lanesxi<mode>): Update accordingly.
235 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
236 Alan Hayward <alan.hayward@arm.com>
237 David Sherwood <david.sherwood@arm.com>
239 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
240 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
241 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
242 of units rather than the mode.
243 * config/aarch64/iterators.md (nunits): New mode attribute.
244 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
245 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
246 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
247 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
248 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
249 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
250 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
251 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
252 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
253 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
254 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
255 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
256 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
257 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
258 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
259 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
260 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
261 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
262 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
263 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
264 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
265 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
266 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
267 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
268 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
269 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
270 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
271 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
272 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
273 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
274 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
275 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
276 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
277 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
278 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
279 (aarch64_simd_vec_setv2di): Likewise.
281 2017-11-06 Carl Love <cel@us.ibm.com>
283 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
285 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
286 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
288 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
290 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
291 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
292 Add power 8 macro expansions.
293 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
294 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
295 power 8 instructions. (VSX_XXBR): Add iterator.
297 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
299 * config/arm/arm.md (predicable_short_it): Change default to "no",
300 improve documentation, remove uses that are identical to the default.
301 (enabled_for_depr_it): Rename to enabled_for_short_it.
302 * gcc/config/arm/arm-fixed.md (predicable_short_it): Remove default uses.
303 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
304 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
305 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
306 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
308 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
311 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
312 float128 helper macros, which are no longer used after deleting
313 the old 'q' built-in functions, and moving the round to odd
314 built-in functions to being special built-in functions.
315 (BU_FLOAT128_2): Likewise.
316 (BU_FLOAT128_1_HW): Likewise.
317 (BU_FLOAT128_2_HW): Likewise.
318 (BU_FLOAT128_3_HW): Likewise.
319 (FABSQ): Delete old 'q' built-in functions.
320 (COPYSIGNQ): Likewise.
321 (SQRTF128_ODD): Move round to odd built-in functions to be
322 special built-in functions, so that we can handle
323 -mabi=ieeelongdouble.
324 (TRUNCF128_ODD): Likewise.
325 (ADDF128_ODD): Likewise.
326 (SUBF128_ODD): Likewise.
327 (MULF128_ODD): Likewise.
328 (DIVF128_ODD): Likewise.
329 (FMAF128_ODD): Likewise.
330 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
331 built-in names to 'f128'.
332 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
333 old 'q' built-in functions, as the machine independent code for
334 'f128' built-in functions handles this.
335 (rs6000_expand_builtin): Add expansion for float128 round to odd
336 functions, keying off on -mabi=ieeelongdouble of whether to use
337 the KFmode or TFmode variant.
338 (rs6000_init_builtins): Initialize the _Float128 round to odd
340 * doc/extend.texi (PowerPC Built-in Functions): Document the old
341 _Float128 'q' built-in functions are now mapped into the new
342 'f128' built-in functions.
344 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
346 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
347 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
349 2017-11-06 David Malcolm <dmalcolm@redhat.com>
352 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
353 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
354 * toplev.c: Include "ipa-fnsummary.h".
355 (toplev::finalize): Call ipa_fnsummary_c_finalize.
357 2017-11-06 Jakub Jelinek <jakub@redhat.com>
359 PR tree-optimization/82838
360 * gimple-ssa-store-merging.c
361 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
362 on a separate gimple_seq which is then appended to seq.
364 2017-11-06 Jeff Law <law@redhat.com>
367 * config/i386/i386.c (PROBE_INTERVAL): Remove.
368 (get_probe_interval): New functions.
369 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
370 (ix86_adjust_stack_and_probe): Likewise.
371 (output_adjust_stack_and_probe): Likewise.
372 (ix86_emit_probe_stack_range): Likewise.
373 (ix86_expand_prologue): Likewise.
375 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
377 PR tree-optimization/82816
378 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
379 if the modes of the two types are the same.
380 (convert_plusminus_to_widen): Likewise.
382 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
384 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
386 (usadv16qi): New define_expand.
387 (usadv8hi): New define_expand.
389 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
392 * ipa-inline-transform.c (update_noncloned_frequencies): Always
394 (inline_transform): Likewise.
395 * predict.c (counts_to_freqs): Remove useless conditional.
396 * profile-count.h (profile_count::apply_scale): Move sanity check.
397 * tree-inline.c (copy_bb): Always scale.
398 (copy_cfg_body): Likewise.
400 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
403 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
405 (*cmp_ite0): Add enabled_for_depr_it attribute.
406 (*cmp_ite1): Likewise.
408 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
410 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
413 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
415 * tree-vrp.c (vrp_int_const_binop): Return true on success and
416 return the value by pointer.
417 (extract_range_from_multiplicative_op_1): Update accordingly.
418 Return as soon as an operation fails.
420 2017-11-05 Tom de Vries <tom@codesourcery.com>
423 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
424 (DEF_SANITIZER_BUILTIN): ... here.
425 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
426 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
428 2017-11-05 Tom de Vries <tom@codesourcery.com>
431 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
433 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
434 ASM_OUTPUT_BEFORE_CASE_LABEL call.
435 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
437 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
438 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
439 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
441 2017-11-05 Tom de Vries <tom@codesourcery.com>
444 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
447 2017-11-04 Michael Clark <michaeljclark@mac.com>
449 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
450 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
452 (*addsi3_extended): Likewise.
453 (*addsi3_extended2): Likewise.
454 (<optab>si3): Likewise.
455 (<optab>di3): Likewise.
456 (<optab><mode>3): Likewise.
457 (<*optabe>si3_internal): Likewise.
458 (zero_extendqi<SUPERQI:mode>2): Likewise.
459 (*add<mode>hi3): Likewise.
460 (*xor<mode>hi3): Likewise.
461 (<optab>di3): Likewise.
462 (*<optab>si3_extend): Likewise.
463 (*sge<u>_<X:mode><GPR:mode>): Likewise.
464 (*slt<u>_<X:mode><GPR:mode>): Likewise.
465 (*sle<u>_<X:mode><GPR:mode>): Likewise.
467 2017-11-04 Andrew Waterman <andrew@sifive.com>
469 * config/riscv/riscv.c (riscv_option_override): Conditionally set
470 TARGET_STRICT_ALIGN based upon -mtune argument.
472 2017-11-04 Andrew Waterman <andrew@sifive.com>
474 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
476 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
478 config/i386/i386.c (choose_basereg): Use optional scratch
479 register and add assertion.
480 (x86_emit_outlined_ms2sysv_save): Use scratch register when
481 needed, and don't allocate stack.
482 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
483 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
484 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
486 2017-11-03 Jeff Law <law@redhat.com>
488 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
489 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
490 to probe at the start of a noreturn function.
492 2017-11-03 Jakub Jelinek <jakub@redhat.com>
494 PR tree-optimization/78821
495 * gimple-ssa-store-merging.c: Update the file comment.
496 (MAX_STORE_ALIAS_CHECKS): Define.
497 (struct store_operand_info): New type.
498 (store_operand_info::store_operand_info): New constructor.
499 (struct store_immediate_info): Add rhs_code and ops data members.
500 (store_immediate_info::store_immediate_info): Add rhscode, op0r
501 and op1r arguments to the ctor, initialize corresponding data members.
502 (struct merged_store_group): Add load_align_base and load_align
504 (merged_store_group::merged_store_group): Initialize them.
505 (merged_store_group::do_merge): Update them.
506 (merged_store_group::apply_stores): Pick the constant for
507 encode_tree_to_bitpos from one of the two operands, or skip
508 encode_tree_to_bitpos if neither operand is a constant.
509 (class pass_store_merging): Add process_store method decl. Remove
510 bool argument from terminate_all_aliasing_chains method decl.
511 (pass_store_merging::terminate_all_aliasing_chains): Remove
512 var_offset_p argument and corresponding handling.
513 (stmts_may_clobber_ref_p): New function.
514 (compatible_load_p): New function.
515 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
516 if there is overlap and rhs_code is not INTEGER_CST. For
517 non-overlapping stores terminate group if rhs is not mergeable.
518 (get_alias_type_for_stmts): Change first argument from
519 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
520 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
521 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
523 (get_location_for_stmts): Change first argument from
524 auto_vec<gimple *> & to vec<gimple *> &.
525 (struct split_store): Remove orig_stmts data member, add orig_stores.
526 (split_store::split_store): Create orig_stores rather than orig_stmts.
527 (find_constituent_stmts): Renamed to ...
528 (find_constituent_stores): ... this. Change second argument from
529 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
530 to info structures rather than the statements.
531 (split_group): Rename ALLOW_UNALIGNED argument to
532 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
533 it. Adjust find_constituent_stores caller.
534 (imm_store_chain_info::output_merged_store): Handle rhs_code other
535 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
536 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
537 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
538 (mem_valid_for_store_merging): New function.
539 (handled_load): New function.
540 (pass_store_merging::process_store): New method.
541 (pass_store_merging::execute): Use process_store method. Adjust
542 terminate_all_aliasing_chains caller.
544 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
546 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
547 Return true for more constants, symbols and label references.
548 (aarch64_valid_floating_const): Remove unused function.
550 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
553 * config/aarch64/aarch64.c (aarch64_layout_frame):
554 Undo forcing of LR at bottom of frame.
556 2017-11-03 Jeff Law <law@redhat.com>
559 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
560 for int_registers_saved.
562 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
563 extracted from tree-ssa-dom.c.
564 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
565 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
566 (record_equivalences_from_incoming_edge): Add additional argument
567 to single_pred_edge_ignoring_loop_edges call.
568 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
569 (uncprop_dom_walker::before_dom_children): Add additional argument
570 to single_pred_edge_ignoring_loop_edges call.
571 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
572 single_pred_edge_ignoring_loop_edges rather than open coding.
573 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
575 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
577 * match.pd (-(-A)): Rewrite.
579 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
581 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
582 (rs6000_emit_int_cmove): New declaration.
583 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
584 (rs6000_emit_sISEL): Delete.
585 (rs6000_emit_int_cmove): Make non-static.
586 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
587 instead of rs6000_emit_sISEL.
589 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
591 * asan.c (create_cond_insert_point): Maintain profile.
592 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
594 * basic-block.h (struct basic_block_def): Remove frequency.
595 (EDGE_FREQUENCY): Use to_frequency
596 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
598 (find_traces): Update to use to_frequency.
599 (find_traces_1_round): Likewise; use only IPA counts.
600 (bb_to_key): Likewise.
601 (connect_traces): Use IPA counts only.
602 (copy_bb_p): Update to use to_frequency.
603 (fix_up_crossing_landing_pad): Likewise.
604 (sanitize_hot_paths): Likewise.
605 * bt-load.c (basic_block_freq): Likewise.
606 * cfg.c (init_flow): Set count_max to uninitialized.
607 (check_bb_profile): Remove frequencies; check counts.
608 (dump_bb_info): Do not dump frequencies.
609 (update_bb_profile_for_threading): Update counts only.
610 (scale_bbs_frequencies_int): Likewise.
611 (MAX_SAFE_MULTIPLIER): Remove.
612 (scale_bbs_frequencies_gcov_type): Update counts only.
613 (scale_bbs_frequencies_profile_count): Update counts only.
614 (scale_bbs_frequencies): Update counts only.
615 * cfg.h (struct control_flow_graph): Add count-max.
616 (update_bb_profile_for_threading): Update prototype.
617 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
618 (find_many_sub_basic_blocks): Likewise.
619 * cfgcleanup.c (try_forward_edges): Likewise.
620 (try_crossjump_to_edge): Likewise.
621 * cfgexpand.c (expand_gimple_cond): Likewise.
622 (expand_gimple_tailcall): Likewise.
623 (construct_init_block): Likewise.
624 (construct_exit_block): Likewise.
625 * cfghooks.c (verify_flow_info): Check consistency of counts.
626 (dump_bb_for_graph): Do not dump frequencies.
627 (split_block_1): Do not update frequencies.
628 (split_edge): Do not update frequencies.
629 (make_forwarder_block): Do not update frequencies.
630 (duplicate_block): Do not update frequencies.
631 (account_profile_record): Do not update frequencies.
632 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
633 for global heuristics.
634 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
635 (expected_loop_iterations_unbounded): Use counts only.
636 * cfgloopmanip.c (scale_loop_profile): Simplify.
637 (create_empty_loop_on_edge): Simplify
639 (duplicate_loop_to_header_edge): Simplify
640 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
641 (update_br_prob_note): Take care of removing note when profile
643 (relink_block_chain): Do not dump frequency.
644 (rtl_account_profile_record): Use to_frequency.
645 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
646 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
647 (cgraph_update_edges_for_call_stmt_node): Likewise.
648 (cgraph_edge::verify_count_and_frequency): Update.
649 (cgraph_node::verify_node): Temporarily disable frequency verification.
650 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
652 (cgraph_edge::rebuild_edges): Convert to ipa counts.
653 * cgraphunit.c (init_lowered_empty_function): Do not initialize
655 (cgraph_node::expand_thunk): Update profile.
656 * except.c (dw2_build_landing_pads): Do not update frequency.
657 * final.c (compute_alignments): Use to_frequency.
658 (dump_basic_block_info): Do not dump frequency.
659 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
660 (dump_gimple_bb_header): Do not dump frequency.
661 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
663 * gimple-streamer-in.c (input_bb): Do not stream frequency.
664 * gimple-streamer-out.c (output_bb): Do not stream frequency.
665 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
666 (init_before_recovery): Do not update frequency.
667 (sched_create_recovery_edges): Do not update frequency.
668 * hsa-gen.c (convert_switch_statements): Do not update frequency.
669 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
670 (ipa_cp_c_finalize): Set max_count to uninitialized.
671 * ipa-fnsummary.c (get_minimal_bb): Use counts.
672 (param_change_prob): Use counts.
673 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
675 * ipa-split.c (consider_split): Use to_frequency.
676 (split_function): Use to_frequency.
677 * ira-build.c (loop_compare_func): Likewise.
678 (mark_loops_for_removal): Likewise.
679 (mark_all_loops_for_removal): Likewise.
680 * loop-doloop.c (doloop_modify): Do not update frequency.
681 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
683 * lto-streamer-in.c (input_function): Update count_max.
684 * omp-expand.c (expand_omp_taskreg): Update count_max.
685 * omp-simd-clone.c (simd_clone_adjust): Update profile.
686 * predict.c (maybe_hot_frequency_p): Use to_frequency.
687 (maybe_hot_count_p): Use ipa counts only.
688 (maybe_hot_bb_p): Simplify.
689 (maybe_hot_edge_p): Simplify.
690 (probably_never_executed): Do not take frequency argument.
691 (probably_never_executed_bb_p): Do not pass frequency.
692 (probably_never_executed_edge_p): Likewise.
693 (combine_predictions_for_bb): Check that profile is nonzero.
694 (propagate_freq): Do not set frequency.
695 (drop_profile): Simplify.
696 (counts_to_freqs): Simplify.
697 (expensive_function_p): Use to_frequency.
698 (propagate_unlikely_bbs_forward): Simplify.
699 (determine_unlikely_bbs): Simplify.
700 (estimate_bb_frequencies): Add hack to silence graphite issues.
701 (compute_function_frequency): Use ipa counts.
702 (pass_profile::execute): Update.
703 (rebuild_frequencies): Use counts only.
704 (force_edge_cold): Use counts only.
705 * profile-count.c (profile_count::dump): Dump new count types.
706 (profile_count::differs_from_p): Check compatiblity.
707 (profile_count::to_frequency): New function.
708 (profile_count::to_cgraph_frequency): New function.
709 * profile-count.h (struct function): Declare.
710 (enum profile_quality): Add profile_guessed_local and
711 profile_guessed_global0.
712 (class profile_proability): Decrease number of bits to 29;
713 update from_reg_br_prob_note and to_reg_br_prob_note.
714 (class profile_count: Update comment; decrease number of bits
715 to 61. Check compatibility.
716 (profile_count::compatible_p): New private member function.
717 (profile_count::ipa_p): New member function.
718 (profile_count::operator<): Handle global zero correctly.
719 (profile_count::operator>): Handle global zero correctly.
720 (profile_count::operator<=): Handle global zero correctly.
721 (profile_count::operator>=): Handle global zero correctly.
722 (profile_count::nonzero_p): New member function.
723 (profile_count::force_nonzero): New member function.
724 (profile_count::max): New member function.
725 (profile_count::apply_scale): Handle IPA scalling.
726 (profile_count::guessed_local): New member function.
727 (profile_count::global0): New member function.
728 (profile_count::ipa): New member function.
729 (profile_count::to_frequency): Declare.
730 (profile_count::to_cgraph_frequency): Declare.
731 * profile.c (OVERLAP_BASE): Delete.
732 (compute_frequency_overlap): Delete.
733 (compute_branch_probabilities): Do not use compute_frequency_overlap.
734 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
735 * sched-ebb.c (rank): Use counts only.
736 * shrink-wrap.c (handle_simple_exit): Use counts only.
737 (try_shrink_wrapping): Use counts only.
738 (place_prologue_for_one_component): Use counts only.
739 * tracer.c (find_best_predecessor): Use to_frequency.
740 (find_trace): Use to_frequency.
741 (tail_duplicate): Use to_frequency.
742 * trans-mem.c (expand_transaction): Do not update frequency.
743 * tree-call-cdce.c: Do not update frequency.
744 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
745 (gimple_merge_blocks): Likewise.
746 (gimple_split_edge): Likewise.
747 (gimple_duplicate_sese_region): Likewise.
748 (gimple_duplicate_sese_tail): Likewise.
749 (move_sese_region_to_fn): Likewise.
750 (gimple_account_profile_record): Likewise.
751 (insert_cond_bb): Likewise.
752 * tree-complex.c (expand_complex_div_wide): Likewise.
753 * tree-eh.c (lower_resx): Update profile.
754 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
756 (initialize_cfun): Do not initialize frequencies
757 (freqs_to_counts): Delete.
758 (copy_cfg_body): Ignore count parameter.
760 (expand_call_inline): Update count_max.
761 (optimize_inline_calls): Update count_max.
762 (tree_function_versioning): Update count_max.
763 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
764 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
766 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
767 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
768 (try_peel_loop): Likewise.
769 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
771 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
772 (tree_transform_and_unroll_loop): Do not use frequencies
773 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
774 Use reliable prediction only.
775 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
776 * tree-ssa-sink.c (select_best_block): Use to_frequency.
777 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
779 * tree-ssa-threadupdate.c (create_block_for_threading): Do
781 (any_remaining_duplicated_blocks): Likewise.
782 (update_profile): Likewise.
783 (estimated_freqs_path): Delete.
784 (freqs_to_counts_path): Delete.
785 (clear_counts_path): Delete.
786 (ssa_fix_duplicate_block_edges): Likewise.
787 (duplicate_thread_path): Likewise.
788 * tree-switch-conversion.c (gen_inbound_check): Use counts.
789 * tree-tailcall.c (decrease_profile): Do not update frequency.
790 (eliminate_tail_call): Likewise.
791 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
792 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
793 (optimize_mask_stores): Likewise.
794 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
795 * ubsan.c (ubsan_expand_null_ifn): Update profile.
796 (ubsan_expand_ptr_ifn): Update profile.
797 * value-prof.c (gimple_ic): Simplify.
798 * value-prof.h (gimple_ic): Update prototype.
799 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
800 * ipa-inline.c (compute_uninlined_call_time): Be sure that
802 (want_inline_self_recursive_call_p): Likewise.
803 (resolve_noninline_speculation): Only cummulate defined counts.
804 (inline_small_functions): Use nonzero_p.
805 (ipa_inline): Do not access freed node.
807 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
809 * config/aarch64/aarch64.c (aarch64_override_options_internal):
810 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
812 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
814 * config/riscv/riscv.c (riscv_legitimize_move): Handle
815 non-legitimate address.
817 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
819 * config/rs6000/rs6000.md (*lt0_disi): Delete.
820 (*lt0_<mode>di, *lt0_<mode>si): New.
822 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
824 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
827 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
828 Jim Wilson <jim.wilson@linaro.org>
830 * config/aarch64/aarch64-cores.def (saphira): New CPU.
831 * config/aarch64/aarch64-tune.md: Regenerated.
832 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
833 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
835 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
837 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
838 (arc_expand_prologue): Restore blink for millicode.
839 * config/arc/linux.h (LINK_EH_SPEC): Defined.
841 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
844 * config/i386/i386.c (ix86_vector_duplicate_value): Use
845 gen_vec_duplicate after forcing the scalar into a register.
847 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
849 * combine (try_combine): Print the insns input to try_combine to the
852 2017-11-02 Steve Ellcey <sellcey@cavium.com>
855 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
856 Remove second argument from aarch64_process_target_attr call.
857 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
859 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
861 (aarch64_handle_attr_arch): Remove second argument.
862 (aarch64_handle_attr_cpu): Ditto.
863 (aarch64_handle_attr_tune): Ditto.
864 (aarch64_handle_attr_isa_flags): Ditto.
865 (aarch64_process_one_target_attr): Ditto.
866 (aarch64_process_target_attr): Ditto.
867 (aarch64_option_valid_attribute_p): Remove second argument.
868 on aarch64_process_target_attr call.
870 2017-11-02 David Malcolm <dmalcolm@redhat.com>
872 * diagnostic.c: Include "selftest-diagnostic.h".
873 (selftest::assert_location_text): New function.
874 (selftest::test_diagnostic_get_location_text): New function.
875 (selftest::diagnostic_c_tests): Call it.
877 2017-11-02 David Malcolm <dmalcolm@redhat.com>
879 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
880 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
881 (class selftest::test_diagnostic_context): Move to...
882 * selftest-diagnostic.c: New file.
883 * selftest-diagnostic.h: New file.
885 2017-11-02 James Bowman <james.bowman@ftdichip.com>
887 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
888 offset range for FT32B.
889 * config/ft32/ft32.h: option "mcompress" enables relaxation.
890 * config/ft32/ft32.md: Add TARGET_NOPM.
891 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
892 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
894 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
896 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
898 2017-11-02 Jeff Law <law@redhat.com>
900 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
901 virtual keyword on FINAL OVERRIDE members.
903 * tree-ssa-propagate.h (ssa_propagation_engine): Group
904 virtuals together. Add virtual destructor.
905 (substitute_and_fold_engine): Similarly.
907 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
909 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
911 2017-11-02 Richard Biener <rguenther@suse.de>
913 PR tree-optimization/82795
914 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
916 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
918 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
920 (gcc_GAS_CHECK_FEATURE): Remove.
921 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
922 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
923 * configure: Regenerate.
925 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
927 * config/arc/arc.c (hwloop_optimize): Account for empty
930 2017-11-02 Richard Biener <rguenther@suse.de>
933 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
934 Truncate ARRAY_REF index and element size.
936 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
938 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
940 2017-11-01 Jeff Law <law@redhat.com>
942 * tree-ssa-ccp.c (ccp_folder): New class derived from
943 substitute_and_fold_engine.
944 (ccp_folder::get_value): New member function.
945 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
946 (ccp_fold_stmt): Remove prototype.
947 (ccp_finalize): Call substitute_and_fold from the ccp_class.
948 * tree-ssa-copy.c (copy_folder): New class derived from
949 substitute_and_fold_engine.
950 (copy_folder::get_value): Renamed from get_value.
951 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
952 * tree-vrp.c (vrp_folder): New class derived from
953 substitute_and_fold_engine.
954 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
955 (vrp_folder::get_value): New member function.
956 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
957 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
958 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
959 provide a class interface to folder/substitute routines.
960 (ssa_prop_fold_stmt_fn): Remove typedef.
961 (ssa_prop_get_value_fn): Likewise.
962 (subsitute_and_fold): Remove prototype.
963 (replace_uses_in): Likewise.
964 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
965 Renamed from replace_uses_in. Call the virtual member function
966 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
967 (substitute_and_fold_dom_walker): Remove initialization of
968 data member entries for calbacks. Add substitute_and_fold_engine
969 member and initialize it.
970 (substitute_and_fold_dom_walker::before_dom_children0: Use the
971 member functions for get_value, replace_phi_args_in c
972 replace_uses_in, and fold_stmt calls.
973 (substitute_and_fold_engine::substitute_and_fold): Renamed from
974 substitute_and_fold. Remove assert. Update ctor call.
976 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
977 (ssa_prop_visit_phi_fn): Likewise.
978 (class ssa_propagation_engine): New class to provide an interface
980 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
982 (ssa_prop_visit_phi): Likewise.
983 (ssa_propagation_engine::simulate_stmt): Moved into class.
984 Call visit_phi/visit_stmt from the class rather than via
985 file scoped static variables.
986 (ssa_propagation_engine::simulate_block): Moved into class.
987 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
988 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
989 set file scoped statics for the visit_stmt/visit_phi callbacks.
990 * tree-complex.c (complex_propagate): New class derived from
991 ssa_propagation_engine.
992 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
993 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
994 (tree_lower_complex): Call ssa_propagate via the complex_propagate
996 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
997 ssa_propagation_engine.
998 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
999 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
1000 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
1001 * tree-ssa-copy.c (copy_prop): New class derived from
1002 ssa_propagation_engine.
1003 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
1004 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
1005 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
1006 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
1007 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
1008 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
1009 (execute_vrp): Call ssa_propagate from the vrp_prop class.
1011 2017-11-01 Jakub Jelinek <jakub@redhat.com>
1013 PR rtl-optimization/82778
1014 PR rtl-optimization/82597
1015 * compare-elim.c (struct comparison): Add in_a_setter field.
1016 (find_comparison_dom_walker::before_dom_children): Remove killed
1017 bitmap and df_simulate_find_defs call, instead walk the defs.
1018 Compute last_setter and initialize in_a_setter. Merge definitions
1019 with first initialization for a few variables.
1020 (try_validate_parallel): Use insn_invalid_p instead of
1021 recog_memoized. Return insn rather than just the pattern.
1022 (try_merge_compare): Fix up comment. Don't uselessly test if
1023 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
1025 (execute_compare_elim_after_reload): Remove df_chain_add_problem
1028 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1029 Alan Hayward <alan.hayward@arm.com>
1030 David Sherwood <david.sherwood@arm.com>
1032 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
1033 aarch64_hard_regno_nregs to get the number of registers
1036 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1037 Alan Hayward <alan.hayward@arm.com>
1038 David Sherwood <david.sherwood@arm.com>
1040 * config/aarch64/constraints.md (Upl): Rename to...
1042 * config/aarch64/aarch64.md
1043 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
1046 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1047 Alan Hayward <alan.hayward@arm.com>
1048 David Sherwood <david.sherwood@arm.com>
1050 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
1051 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
1054 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1055 Alan Hayward <alan.hayward@arm.com>
1056 David Sherwood <david.sherwood@arm.com>
1058 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
1059 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
1060 (aarch64_evpc_dup): Generate rtl direcly, rather than using
1062 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
1063 of a single element.
1064 * config/aarch64/iterators.md: Add a comment above the permute
1065 unspecs to say that they are generated directly by
1066 aarch64_expand_vec_perm_const.
1067 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
1069 2017-11-01 Nathan Sidwell <nathan@acm.org>
1071 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
1073 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1075 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
1076 and medany code models, and describe what they do.
1078 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1080 Revert accidental duplicate:
1082 * combine.c (can_change_dest_mode): Reject changes in
1083 REGMODE_NATURAL_SIZE.
1085 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
1087 PR rtl-optimization/64682
1088 PR rtl-optimization/69567
1089 PR rtl-optimization/69737
1090 PR rtl-optimization/82683
1091 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
1092 register mentioned in the note, drop the note, unless it came from I3,
1093 in which case it should go to I3 again.
1095 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1097 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
1098 and return false if not.
1099 (clear_bytes_written_by, live_bytes_read): Update accordingly.
1101 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1103 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
1104 range is known to be empty.
1106 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1107 Alan Hayward <alan.hayward@arm.com>
1108 David Sherwood <david.sherwood@arm.com>
1110 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
1111 and CONST_VECTOR_NUNITS instead of computing the number of units from
1112 the byte sizes of the vector and element.
1113 (simplify_binary_operation_1): Likewise.
1114 (simplify_const_binary_operation): Likewise.
1115 (simplify_ternary_operation): Likewise.
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 * var-tracking.c (INT_MEM_OFFSET): Replace with...
1122 (int_mem_offset): ...this new function.
1123 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
1124 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
1125 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
1128 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1129 Alan Hayward <alan.hayward@arm.com>
1130 David Sherwood <david.sherwood@arm.com>
1132 * lower-subreg.c (interesting_mode_p): New function.
1133 (compute_costs, find_decomposable_subregs, decompose_register)
1134 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
1135 (resolve_clobber, dump_choices): Use it.
1137 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1138 Alan Hayward <alan.hayward@arm.com>
1139 David Sherwood <david.sherwood@arm.com>
1141 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
1143 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1144 Alan Hayward <alan.hayward@arm.com>
1145 David Sherwood <david.sherwood@arm.com>
1147 * alias.c (find_base_value, find_base_term): Only process integer
1148 truncations. Check the precision rather than the size.
1150 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1151 Alan Hayward <alan.hayward@arm.com>
1152 David Sherwood <david.sherwood@arm.com>
1154 * machmode.h (is_narrower_int_mode): New function
1155 * optabs.c (expand_float, expand_fix): Use it.
1156 * dwarf2out.c (rotate_loc_descriptor): Likewise.
1158 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1159 Alan Hayward <alan.hayward@arm.com>
1160 David Sherwood <david.sherwood@arm.com>
1162 * rtl.h (narrower_subreg_mode): New function.
1163 * ira-color.c (update_costs_from_allocno): Use it.
1165 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1166 Alan Hayward <alan.hayward@arm.com>
1167 David Sherwood <david.sherwood@arm.com>
1169 * optabs-query.h (convert_optab_p): New function, split out from...
1170 (convert_optab_handler): ...here.
1171 (widening_optab_handler): Delete.
1172 (find_widening_optab_handler): Remove permit_non_widening parameter.
1173 (find_widening_optab_handler_and_mode): Likewise. Provide an
1174 override that operates on mode class wrappers.
1175 * optabs-query.c (widening_optab_handler): Delete.
1176 (find_widening_optab_handler_and_mode): Remove permit_non_widening
1177 parameter. Assert that the two modes are the same class and that
1178 the "from" mode is narrower than the "to" mode. Use
1179 convert_optab_handler instead of widening_optab_handler.
1180 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
1181 instead of widening_optab_handler.
1182 * expr.c (expand_expr_real_2): Update calls to
1183 find_widening_optab_handler.
1184 * optabs.c (expand_widen_pattern_expr): Likewise.
1185 (expand_binop_directly): Take the insn_code as a parameter.
1186 (expand_binop): Only call find_widening_optab_handler for
1187 conversion optabs; use optab_handler otherwise. Update calls
1188 to find_widening_optab_handler and expand_binop_directly.
1189 Use convert_optab_handler instead of widening_optab_handler.
1190 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
1191 find_widening_optab_handler and use scalar_mode rather than
1193 (convert_plusminus_to_widen): Likewise.
1195 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1196 Alan Hayward <alan.hayward@arm.com>
1197 David Sherwood <david.sherwood@arm.com>
1199 * machmode.h (fixed_size_mode): New class.
1200 * rtl.h (get_pool_mode): Return fixed_size_mode.
1201 * gengtype.c (main): Add fixed_size_mode.
1202 * target.def (get_raw_result_mode): Return a fixed_size_mode.
1203 (get_raw_arg_mode): Likewise.
1204 * doc/tm.texi: Regenerate.
1205 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
1206 * targhooks.c (default_get_reg_raw_mode): Likewise.
1207 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
1208 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
1209 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
1210 (msp430_get_raw_result_mode): Likewise.
1211 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
1212 * dbxout.c (dbxout_parms): Require fixed-size modes.
1213 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
1214 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
1215 * omp-low.c (lower_oacc_reductions): Likewise.
1216 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
1217 (simplify_subreg): Update accordingly.
1218 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
1219 (force_const_mem): Update accordingly. Return NULL_RTX for modes
1220 that aren't fixed-size.
1221 (get_pool_mode): Return a fixed_size_mode.
1222 (output_constant_pool_2): Take a fixed_size_mode.
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 * doc/rtl.texi (vec_series): Document.
1229 (const): Say that the operand can be a vec_series.
1230 * rtl.def (VEC_SERIES): New rtx code.
1231 * rtl.h (const_vec_series_p_1): Declare.
1232 (const_vec_series_p): New function.
1233 * emit-rtl.h (gen_const_vec_series): Declare.
1234 (gen_vec_series): Likewise.
1235 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
1236 (gen_vec_series): Likewise.
1237 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
1238 * simplify-rtx.c (simplify_unary_operation): Handle negations
1240 (simplify_binary_operation_series): New function.
1241 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
1242 (test_vector_ops_series): New function.
1243 (test_vector_ops): Call it.
1244 * config/powerpcspe/altivec.md (altivec_lvsl): Use
1245 gen_const_vec_series.
1246 (altivec_lvsr): Likewise.
1247 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
1249 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1250 Alan Hayward <alan.hayward@arm.com>
1251 David Sherwood <david.sherwood@arm.com>
1253 * doc/rtl.texi (const): Update description of address constants.
1254 Say that vector constants are allowed too.
1255 * common.md (E, F): Use CONSTANT_P instead of checking for
1257 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
1258 checking for CONST_VECTOR.
1259 * expmed.c (make_tree): Use build_vector_from_val for a CONST
1261 * expr.c (expand_expr_real_2): Check for vector modes instead
1262 of checking for CONST_VECTOR.
1263 * rtl.h (const_vec_p): New function.
1264 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
1265 (unwrap_const_vec_duplicate): Handle them here too.
1267 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1268 David Malcolm <dmalcolm@redhat.com>
1269 Alan Hayward <alan.hayward@arm.com>
1270 David Sherwood <david.sherwood@arm.com>
1272 * rtl.h (vec_duplicate_p): New function.
1273 * selftest-rtl.c (assert_rtx_eq_at): New function.
1274 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
1275 (assert_rtx_eq_at): Declare.
1276 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
1277 * selftest-run-tests.c (selftest::run_tests): Call it.
1278 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
1279 (simplify_unary_operation_1): Recursively handle vector duplicates.
1280 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
1282 (simplify_subreg): Handle subregs of vector duplicates.
1283 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
1284 (selftest::simplify_rtx_c_tests): New functions.
1286 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1287 Alan Hayward <alan.hayward@arm.com>
1288 David Sherwood <david.sherwood@arm.com>
1290 * emit-rtl.h (gen_const_vec_duplicate): Declare.
1291 (gen_vec_duplicate): Likewise.
1292 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
1294 (gen_const_vector): ...here.
1295 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
1296 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
1297 whose elements are all equal.
1298 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
1299 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
1300 (simplify_relational_operation): Likewise.
1301 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
1303 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
1304 (aarch64_expand_vector_init): Likewise.
1305 * config/arm/arm.c (neon_vdup_constant): Likewise.
1306 (neon_expand_vector_init): Likewise.
1307 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
1308 (arm_block_set_unaligned_vect): Likewise.
1309 (arm_block_set_aligned_vect): Likewise.
1310 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
1311 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
1312 (expand_vec_perm_even_odd_pack): Likewise.
1313 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
1314 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
1315 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
1316 gen_const_vec_duplicate.
1317 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
1318 * config/mips/mips.c (mips_gen_const_int_vector): Use
1319 gen_const_vec_duplicate.
1320 (mips_expand_vector_init): Use CONST0_RTX.
1321 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
1322 (define_split): Use gen_const_vec_duplicate.
1323 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
1324 (define_split): Use gen_const_vec_duplicate.
1325 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
1326 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
1327 * config/spu/spu.c (spu_const): Likewise.
1329 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1330 Alan Hayward <alan.hayward@arm.com>
1331 David Sherwood <david.sherwood@arm.com>
1333 * combine.c (can_change_dest_mode): Reject changes in
1334 REGMODE_NATURAL_SIZE.
1336 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
1338 * configure.ac (--enable-libssp): New.
1339 (gcc_cv_libc_provides_ssp): Check for explicit setting before
1340 trying to determine target-specific default. Adjust indentation.
1341 * configure: Regenerated.
1342 * doc/install.texi (Configuration): Expand --disable-libssp
1345 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
1347 config/i386/i386.c (ix86_expand_epilogue): Correct stack
1350 2017-10-31 Martin Jambor <mjambor@suse.cz>
1353 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
1355 2017-10-31 David Malcolm <dmalcolm@redhat.com>
1357 * auto-profile.c (autofdo_source_profile::read): Use
1358 UNKNOWN_LOCATION rather than 0.
1359 * diagnostic-core.h (warning_at_rich_loc): Rename to...
1360 (warning_at): ...this overload.
1361 (warning_at_rich_loc_n): Rename to...
1362 (warning_n): ...this overload.
1363 (error_at_rich_loc): Rename to...
1364 (error_at): ...this overload.
1365 (pedwarn_at_rich_loc): Rename to...
1366 (pedwarn): ...this overload.
1367 (permerror_at_rich_loc): Rename to...
1368 (permerror): ...this overload.
1369 (inform_at_rich_loc): Rename to...
1370 (inform): ...this overload.
1371 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
1372 (diagnostic_n_impl_richloc): Rename to...
1373 (diagnostic_n_impl): ...this rich_location *-based decl.
1374 (inform_at_rich_loc): Rename to...
1375 (inform): ...this, and add an assertion.
1376 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
1377 (warning_at_rich_loc): Rename to...
1378 (warning_at): ...this, and add an assertion.
1379 (warning_at_rich_loc_n): Rename to...
1380 (warning_n): ...this, and add an assertion.
1381 (warning_n): Update location_t-based implementation for removal of
1382 location_t-based diagnostic_n_impl.
1383 (pedwarn_at_rich_loc): Rename to...
1384 (pedwarn): ...this, and add an assertion.
1385 (permerror_at_rich_loc): Rename to...
1386 (permerror): ...this, and add an assertion.
1387 (error_n): Update for removal of location_t-based diagnostic_n_impl.
1388 (error_at_rich_loc): Rename to...
1389 (error_at): ...this, and add an assertion.
1390 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
1391 (driver::do_spec_on_infiles): Likewise.
1392 * substring-locations.c (format_warning_va): Update for renaming
1393 of inform_at_rich_loc.
1395 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
1397 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
1398 _Float<N>X built-in functions so that the variant without the
1399 "__builtin_" prefix is only enabled for the GNU C and Objective C
1400 languages when they are in non-strict ANSI/ISO mode.
1401 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
1402 * target.def (floatn_builtin_p): Add a target hook to control
1403 whether _Float<N> and _Float<N>X built-in functions without the
1404 "__builtin_" prefix are enabled, and return true for C and
1405 Objective C in the default hook. Include langhooks.h in
1407 * targhooks.h (default_floatn_builtin_p): Likewise.
1408 * targhooks.c (default_floatn_builtin_p): Likewise.
1409 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
1410 floatn_builtin_p target hook.
1411 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
1413 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
1414 Eric Botcazou <ebotcazou@adacore.com>
1416 PR rtl-optimization/81803
1417 * lra-constraints.c (curr_insn_transform): Also reload the whole
1418 register for a strict subreg no wider than a word if this is for
1419 a WORD_REGISTER_OPERATIONS target.
1421 2017-10-31 Jason Merrill <jason@redhat.com>
1423 * gdbinit.in: Skip over inlines from timevar.h.
1425 2017-10-31 Martin Liska <mliska@suse.cz>
1427 * doc/gcov.texi: Document new option.
1428 * gcov.c (print_usage): Likewise print it.
1429 (process_args): Support the argument.
1430 (format_count): New function.
1431 (format_gcov): Use the function.
1433 2017-10-31 Martin Liska <mliska@suse.cz>
1435 * gcov.c (struct name_map): do not use typedef.
1436 Define operator== and operator<.
1437 (name_search): Remove.
1438 (name_sort): Remove.
1439 (main): Do not allocate names.
1440 (process_file): Add vertical space.
1441 (generate_results): Use std::find.
1442 (release_structures): Do not release memory.
1443 (find_source): Use std::find.
1445 2017-10-31 Martin Liska <mliska@suse.cz>
1447 * gcov.c (struct line_info): Remove it's typedef.
1448 (line_info::line_info): Add proper ctor.
1449 (line_info::has_block): Do not use a typedef.
1450 (struct source_info): Do not use typedef.
1451 (circuit): Likewise.
1452 (get_cycles_count): Likewise.
1453 (output_intermediate_file): Iterate via vector iterator.
1454 (add_line_counts): Use std::vector methods.
1455 (accumulate_line_counts): Likewise.
1456 (output_lines): Likewise.
1458 2017-10-31 Martin Liska <mliska@suse.cz>
1460 * gcov.c (struct source_info): Remove typedef.
1461 (source_info::source_info): Add proper ctor.
1462 (accumulate_line_counts): Use struct, not it's typedef.
1463 (output_gcov_file): Likewise.
1464 (output_lines): Likewise.
1465 (main): Do not allocate an array.
1466 (output_intermediate_file): Use size of vector container.
1467 (process_file): Resize the vector.
1468 (generate_results): Do not preallocate, use newly added vector
1470 (release_structures): Do not release sources.
1471 (find_source): Use vector methods.
1472 (add_line_counts): Do not use typedef.
1474 2017-10-31 Martin Liska <mliska@suse.cz>
1476 * doc/gcov.texi: Document that.
1477 * gcov.c (add_line_counts): Mark lines with a non-executed
1479 (output_line_beginning): Handle such lines.
1480 (output_lines): Pass new argument.
1481 (output_intermediate_file): Print it in intermediate format.
1483 2017-10-31 Martin Liska <mliska@suse.cz>
1485 * color-macros.h: New file.
1486 * diagnostic-color.c: Factor out color related to macros to
1488 * doc/gcov.texi: Document -k option.
1489 * gcov.c (INCLUDE_STRING): Include string.h.
1490 (print_usage): Add -k option.
1491 (process_args): Parse it.
1492 (pad_count_string): New function.
1493 (output_line_beginning): Likewise.
1494 (DEFAULT_LINE_START): New macro.
1495 (output_lines): Support color output.
1497 2017-10-31 Martin Liska <mliska@suse.cz>
1499 PR gcov-profile/82633
1500 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
1501 their interaction with GCOV infrastructure.
1502 * configure.ac: Add -fkeep-{inline,static}-functions to
1504 * configure: Regenerate.
1506 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
1509 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
1511 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
1514 * config/rs6000/rs6000.md (allocate_stack): Force update interval
1515 into a register if it does not fit into an immediate offset field.
1517 2017-10-31 Olivier Hainque <hainque@adacore.com>
1519 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
1521 2017-10-31 Julia Koval <julia.koval@intel.com>
1523 * config.gcc: Add gfniintrin.h.
1524 * config/i386/gfniintrin.h: New.
1525 * config/i386/i386-builtin-types.def
1526 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1527 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1528 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1529 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1530 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1531 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
1532 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
1533 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
1534 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
1535 V64QI_FTYPE_V64QI_V64QI_INT): New types.
1536 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1537 * config/i386/immintrin.h: Include gfniintrin.h.
1538 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
1540 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
1542 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
1544 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1546 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
1547 (arm_ashldi3_1bit): Remove pattern.
1548 (ashrdi3): Remove shift by 1 expansion.
1549 (arm_ashrdi3_1bit): Remove pattern.
1550 (lshrdi3): Remove shift by 1 expansion.
1551 (arm_lshrdi3_1bit): Remove pattern.
1552 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
1553 cost of ashldi3 by 1.
1554 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
1555 (<shift>di3_neon): Likewise.
1557 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
1559 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
1560 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
1561 and (*aarch64_simd_mov<VQ:mode>).
1562 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
1563 pattern alternative.
1564 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
1565 attributes to match pattern alternative.
1567 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
1569 * config.gcc (powerpc*-*-*): Add emmintrin.h.
1570 * config/rs6000/emmintrin.h: New file.
1571 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
1573 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1575 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
1576 * (movdi_vfp_cortexa8): Remove pattern.
1578 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1580 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
1582 (Specific, alpha*-dec-osf5.1): Remove.
1583 (Specific, mips-sgi-irix5): Remove.
1584 (Specific, mips-sgi-irix6): Remove.
1586 2017-10-30 Jakub Jelinek <jakub@redhat.com>
1589 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
1590 arguments to clear_bit_region_be.
1592 2017-10-30 Jim Wilson <wilson@tuliptree.org>
1594 * gimplify.c: Include memmodel.h.
1596 2017-10-30 Martin Jambor <mjambor@suse.cz>
1598 * omp-grid.c (grid_attempt_target_gridification): Also insert a
1599 condition whether loop should be executed at all.
1601 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
1603 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1604 gimple folding of vec_madd() intrinsics.
1605 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
1606 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
1607 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
1609 2017-10-30 Richard Biener <rguenther@suse.de>
1611 PR tree-optimization/82762
1613 2017-10-23 Richard Biener <rguenther@suse.de>
1615 PR tree-optimization/82129
1617 2017-08-01 Richard Biener <rguenther@suse.de>
1619 PR tree-optimization/81181
1620 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
1621 (compute_antic): ... end of iteration here.
1623 2017-10-30 Joseph Myers <joseph@codesourcery.com>
1625 * doc/invoke.texi (C Dialect Options): Document -std=c17,
1626 -std=iso9899:2017 and -std=gnu17.
1627 * doc/standards.texi (C Language): Document C17 support.
1628 * doc/cpp.texi (Overview): Mention -std=c17.
1629 (Standard Predefined Macros): Document C11 and C17 values of
1630 __STDC_VERSION__. Do not refer to C99 support as incomplete.
1631 * doc/extend.texi (Inline): Do not list individual options for
1632 standards newer than C99.
1633 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
1635 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
1638 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
1640 * asan.c (asan_finish_file): Align asan globals array by shadow
1643 2017-10-30 Jakub Jelinek <jakub@redhat.com>
1646 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
1647 (struct store_immediate_info): Add bitregion_start and bitregion_end
1649 (store_immediate_info::store_immediate_info): Add brs and bre
1650 arguments and initialize bitregion_{start,end} from those.
1651 (struct merged_store_group): Add bitregion_start, bitregion_end,
1652 align_base and mask fields. Drop unnecessary struct keyword from
1653 struct store_immediate_info. Add do_merge method.
1654 (clear_bit_region_be): Use memset instead of loop storing zeros.
1655 (merged_store_group::do_merge): New method.
1656 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
1657 stores as long as the surrounding bitregions have no gaps.
1658 (merged_store_group::merge_overlapping): Use do_merge.
1659 (merged_store_group::apply_stores): Test that bitregion_{start,end}
1660 is byte aligned, rather than requiring that start and width are
1661 byte aligned. Drop unnecessary struct keyword from
1662 struct store_immediate_info. Allocate and populate also mask array.
1663 Make start of the arrays relative to bitregion_start rather than
1664 start and size them according to bitregion_{end,start} difference.
1665 (struct imm_store_chain_info): Drop unnecessary struct keyword from
1666 struct store_immediate_info.
1667 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
1668 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
1669 struct keyword from struct store_immediate_info.
1670 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
1671 between stores as long as the surrounding bitregions have no gaps.
1673 (struct split_store): Add orig non-static data member.
1674 (split_store::split_store): Initialize orig to false.
1675 (find_constituent_stmts): Return store_immediate_info *, non-NULL
1676 if there is exactly a single original stmt. Change stmts argument
1677 to pointer from reference, if NULL, don't push anything to it. Add
1678 first argument, use it to optimize skipping over orig stmts that
1679 are known to be before bitpos already. Simplify.
1680 (split_group): Return unsigned int count how many stores are or
1681 would be needed rather than a bool. Add allow_unaligned argument.
1682 Change split_stores argument from reference to pointer, if NULL,
1683 only do a dry run computing how many stores would be produced.
1684 Rewritten algorithm to use both alignment and misalign if
1685 !allow_unaligned and handle bitfield stores with gaps.
1686 (imm_store_chain_info::output_merged_store): Set start_byte_pos
1687 from bitregion_start instead of start. Compute allow_unaligned
1688 here, if true, do 2 split_group dry runs to compute which one
1689 produces fewer stores and prefer aligned if equal. Punt if
1690 new count is bigger or equal than original before emitting any
1691 statements, rather than during that. Remove no longer needed
1692 new_ssa_names tracking. Replace num_stmts with
1693 split_stores.length (). Use 32-bit stack allocated entries
1694 in split_stores auto_vec. Try to reuse original store lhs/rhs1
1695 if possible. Handle bitfields with gaps.
1696 (pass_store_merging::execute): Ignore bitsize == 0 stores.
1697 Compute bitregion_{start,end} for the stores and construct
1698 store_immediate_info with that. Formatting fixes.
1700 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
1703 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
1704 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
1706 2017-10-29 Jim Wilson <wilson@tuliptree.org>
1708 * gimplify.c: Include tm_p.h.
1710 * common.opt (gcoff): Re-add as ignored option.
1711 (gcoff1, gcoff2, gcoff3): Likewise.
1713 * Makefile.in (OBJS): Delete sdbout.o.
1714 (GTFILES): Delete $(srcdir)/sdbout.c.
1715 * debug.h: Delete sdb_debug_hooks.
1716 * final.c: Delete sdbout.h include.
1717 (final_scan_insn): Delete SDB_DEBUG check.
1718 (rest_of_clean_state): Likewise.
1719 * output.h: Delete sdb_begin_function_line.
1722 * toplev.c: Delete sdbout.h include.
1723 (process_options): Delete SDB_DEBUG check.
1724 * tree-core.h (tree_type_common): Delete pointer field of
1726 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
1727 TYPE_SYMTAB_POINTER.
1728 * tree.h (TYPE_SYMTAB_POINTER): Delete.
1729 (TYPE_SYMTAB_IS_POINTER): Delete.
1730 (TYPE_SYMTAB_IS_DIE): Renumber.
1731 * xcoffout.c: Refer to former sdbout.c file.
1732 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
1734 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
1735 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
1736 (Debugging Options): Delete -gcoff.
1737 (-gstabs): Delete SDB reference.
1739 (-gcoff@var{level}): Delete.
1740 * doc/passes.texi (Debugging information output): Delete SDB and
1741 sdbout.c references.
1742 * doc/tm.texi: Regenerate.
1743 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
1744 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
1746 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
1747 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
1748 (SDB_DEBUGGING_INFO): Delete.
1749 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
1750 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
1751 * target.def (output_source_filename): Delete COFF reference.
1753 * common.opt (gcoff): Delete.
1754 (gxcoff+): Update Negative chain.
1755 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
1757 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
1758 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
1759 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
1761 (expand_function_start): Change sdb reference to past tense.
1762 (expand_function_end): Change sdb reference to past tense.
1763 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
1764 * opts.c (debug_type_names): Delete coff entry.
1765 (common_handle_option): Delete OPT_gcoff case.
1766 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
1768 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
1769 * config/cris/cris.h: Delete SDB reference in comment.
1770 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
1771 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
1772 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
1773 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
1775 (ix86_expand_prologue): Likewise.
1776 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
1777 * config/ia64/ia64.h: Likewise.
1778 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
1779 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
1781 * config/mmix/mmix.h: Likewise.
1782 * config/nds32/nds32.c: Likewise.
1783 * config/stormy/storym16.h: Likewise.
1784 * config/visium/visium.h: Likewise.
1785 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
1787 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
1789 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
1790 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
1791 FRAME_POINTER_REGNUM point at high end of local var area.
1793 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
1795 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
1796 Move comment around. Do not reset best_edge for a copiable
1797 destination if the copy would cause a partition change.
1798 (better_edge_p): Remove redundant check.
1800 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
1802 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
1804 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
1806 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
1807 for math functions that have _Float<N> and _Float<N>X variants.
1808 (mathfn_built_in_2): Add support for math functions that have
1809 _Float<N> and _Float<N>X variants.
1810 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
1811 (expand_builtin_mathfn_ternary): Add support for fma with
1812 _Float<N> and _Float<N>X variants.
1813 (expand_builtin): Likewise.
1814 (fold_builtin_3): Likewise.
1815 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
1816 create math function _Float<N> and _Float<N>X variants as external
1818 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
1819 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
1820 the __builtin_ prefix and if not strict ansi, without the prefix.
1821 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
1822 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
1823 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
1824 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
1825 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
1826 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
1827 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
1828 function signatures for fma _Float<N> and _Float<N>X variants.
1829 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
1830 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
1831 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
1832 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
1833 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
1834 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
1835 * gencfn-macros.c (print_case_cfn): Add support for math functions
1836 that have _Float<N> and _Float<N>X variants.
1837 (print_define_operator_list): Likewise.
1838 (fltfn_suffixes): Likewise.
1840 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
1841 for math functions that have _Float<N> and _Float<N>X variants.
1842 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
1843 and _Float<N>X variants.
1844 (COPYSIGN): Likewise.
1847 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
1848 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
1850 (integer_valued_read_call_p): Likewise.
1851 * fold-const-call.c (fold_const_call_ss): Likewise.
1852 (fold_const_call_sss): Add support for copysign, fmin, and fmax
1853 _Float<N> and _Float<N>X variants.
1854 (fold_const_call_ssss): Add support for fma _Float<N> and
1855 _Float<N>X variants.
1856 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
1857 support for copysign and fma _Float<N> and _Float<N>X variants.
1858 (backprop::process_builtin_call_use): Likewise.
1859 * tree-call-cdce.c (can_test_argument_range); Add support for
1860 sqrt _Float<N> and _Float<N>X variants.
1861 (edom_only_function): Likewise.
1862 (get_no_error_domain): Likewise.
1863 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
1864 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
1865 copysign _Float<N> and _Float<N>X variants.
1866 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
1867 handled by machine independent code.
1868 (FMAF128): Likewise.
1869 * doc/cpp.texi (Common Predefined Macros): Document defining
1870 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
1871 fma _Float<N> and _Float<N>X variants.
1873 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
1876 * config/i386/i386-modes.def (CCFPU): Remove definition.
1877 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
1878 (ix86_cc_modes_compatible): Ditto.
1879 (ix86_expand_carry_flag_compare): Ditto.
1880 (ix86_expand_int_movcc): Ditto.
1881 (ix86_expand_int_addcc): Ditto.
1882 (ix86_reverse_condition): Ditto.
1883 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
1884 Return true/false for unordered/ordered fp comparisons.
1885 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
1886 (ix86_prepare_fp_compare_args): Update for rename.
1887 (ix86_expand_fp_compare): Update for rename. Generate unordered
1888 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
1889 (ix86_expand_sse_compare_and_jump): Ditto.
1890 * config/i386/predicates.md (fcmov_comparison_operator):
1891 Remove CCFPU mode handling.
1892 (ix86_comparison_operator): Ditto.
1893 (ix86_carry_flag_operator): Ditto.
1894 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
1895 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
1896 (*cmpu<mode>_cc_i387): Ditto.
1897 (FPCMP): Remove mode iterator.
1898 (unord): Remove mode attribute.
1899 (unord_subst): New define_subst transformation
1900 (unord): New define_subst attribute.
1902 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
1903 (*cmpi<unord>xf_i387): Ditto.
1904 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
1905 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
1906 using unord_subst transformation.
1907 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
1908 (round_saeonly): Also handle CCFP mode.
1909 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
1910 Remove UNSPEC_SAHF unspec handling.
1912 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
1914 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
1916 2017-10-27 Jeff Law <law@redhat.com>
1918 * gimple-ssa-sprintf.c: Include domwalk.h.
1919 (class sprintf_dom_walker): New class, derived from dom_walker.
1920 (sprintf_dom_walker::before_dom_children): New function.
1921 (struct call_info): Moved into sprintf_dom_walker class
1922 (compute_formath_length, handle_gimple_call): Likewise.
1923 (sprintf_length::execute): Call the dominator walker rather
1924 than walking the statements.
1926 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
1927 gimple statement locations.
1928 (check_array_bounds): Corresponding changes. Get the statement's
1929 location directly from wi->stmt.
1931 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
1934 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
1936 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
1938 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
1939 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
1941 2017-10-27 Jakub Jelinek <jakub@redhat.com>
1944 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
1945 * config/i386/i386.c (maybe_get_pool_constant): Removed.
1946 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
1947 maybe_get_pool_constant.
1948 * config/i386/predicates.md (zero_extended_scalar_load_operand):
1951 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1953 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
1954 2.26 caveat. Update gas and gld versions.
1955 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
1958 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1960 * cgraph.h (set_malloc_flag): Declare.
1961 * cgraph.c (set_malloc_flag_1): New function.
1962 (set_malloc_flag): Likewise.
1963 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
1964 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
1966 (read_ipa_call_summary): Add support for reading is_return_callee.
1967 (write_ipa_call_summary): Stream is_return_callee.
1968 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
1969 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
1970 ipa-prop.h, ipa-fnsummary.h.
1971 (pure_const_names): Change to static.
1972 (malloc_state_e): Define.
1973 (malloc_state_names): Define.
1974 (funct_state_d): Add field malloc_state.
1975 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
1976 (check_retval_uses): New function.
1977 (malloc_candidate_p): Likewise.
1978 (analyze_function): Add support for malloc attribute.
1979 (pure_const_write_summary): Stream malloc_state.
1980 (pure_const_read_summary): Add support for reading malloc_state.
1981 (dump_malloc_lattice): New function.
1982 (propagate_malloc): New function.
1983 (warn_function_malloc): New function.
1984 (ipa_pure_const::execute): Call propagate_malloc and
1985 ipa_free_fn_summary.
1986 (pass_local_pure_const::execute): Add support for malloc attribute.
1987 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
1988 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
1990 2017-10-27 Martin Liska <mliska@suse.cz>
1992 PR gcov-profile/82457
1993 * doc/invoke.texi: Document that one needs a non-strict ISO mode
1994 for fork-like functions to be properly instrumented.
1996 2017-10-27 Richard Biener <rguenther@suse.de>
1999 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
2000 info when we redirected EH.
2002 2017-10-26 Michael Collison <michael.collison@arm.com>
2004 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
2006 (<optab>_trunchf<GPI:mode>2: New pattern.
2007 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
2008 * config/aarch64/iterators.md (wv): New mode attribute.
2009 (vf, VF): New mode attributes.
2010 (vgp, VGP): New mode attributes.
2011 (s): Update attribute with SImode and DImode prefixes.
2013 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2015 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
2016 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
2017 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
2018 (nios2_option_overide): Initialize it. Don't allow R0-relative
2019 addressing with PIC.
2020 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
2021 (nios2_symbolic_constant_p): Likewise.
2022 (nios2_legitimate_address_p): Likewise.
2023 (nios2_r0rel_section_name_p): New.
2024 (nios2_symbol_ref_in_r0rel_data_p): New.
2025 (nios2_emit_move_sequence): Handle r0rel_constant_p.
2026 (r0rel_constant_p): New.
2027 (nios2_print_operand_address): Handle r0rel_constant_p.
2028 (nios2_cdx_narrow_form_p): Likewise.
2029 * config/nios2/nios2.opt (mr0rel-sec=): New option.
2030 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
2031 (Nios II Options): Document -mr0rel-sec.
2033 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2035 * config/nios2/nios2.c: Include xregex.h.
2036 (nios2_gprel_sec_regex): New.
2037 (nios2_option_overide): Initialize it. Don't allow GP-relative
2038 addressing with PIC.
2039 (nios2_small_section_name_p): Check for regex match.
2040 * config/nios2/nios2.opt (mgprel-sec=): New option.
2041 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
2042 (Nios II Options): Document -mgprel-sec.
2044 2017-10-26 Jim Wilson <wilson@tuliptree.org>
2046 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
2048 2017-10-26 Tom de Vries <tom@codesourcery.com>
2050 PR tree-optimization/82707
2051 * gimple.c (gimple_copy): Fix unsharing of
2052 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
2054 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
2056 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
2057 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
2058 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
2059 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
2060 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
2061 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
2062 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
2063 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
2064 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
2065 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
2066 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
2067 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
2068 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
2069 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
2070 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
2071 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
2072 _mm512_mask_cmpunord_ps_mask): New intrinsics.
2074 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2076 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
2078 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
2079 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
2080 warning to rs6000.c. Remove the Undocumented flag, since it has
2082 (-mabi=ibmlongdouble): Likewise.
2083 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
2084 already set, set the default format for long double.
2085 (rs6000_debug_reg_global): Print whether long double is IBM or
2087 (rs6000_option_override_internal): Rework setting long double
2088 format. Only warn if the user is changing the long double default
2089 and they did not use -Wno-psabi.
2090 * doc/invoke.texi (PowerPC options): Update the documentation for
2091 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
2093 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2094 Alan Hayward <alan.hayward@arm.com>
2095 David Sherwood <david.sherwood@arm.com>
2097 * rtl.h (wider_subreg_mode): New function.
2098 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
2099 rather than an unsigned int *.
2100 * ira-color.c (regno_max_ref_width): Replace with...
2101 (regno_max_ref_mode): ...this new variable.
2102 (coalesced_pseudo_reg_slot_compare): Update accordingly.
2103 Use wider_subreg_mode.
2104 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
2105 rather than an unsigned int *.
2106 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
2107 (process_alt_operands): Likewise.
2108 (invariant_p): Likewise.
2109 * lra-spills.c (assign_mem_slot): Likewise.
2110 (add_pseudo_to_slot): Likewise.
2111 * lra.c (collect_non_operand_hard_regs): Likewise.
2112 (add_regs_to_insn_regno_info): Likewise.
2113 * reload1.c (regno_max_ref_width): Replace with...
2114 (regno_max_ref_mode): ...this new variable.
2115 (reload): Update accordingly. Update call to
2116 ira_sort_regnos_for_alter_reg.
2117 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
2118 (init_eliminable_invariants): Update to use regno_max_ref_mode.
2119 (scan_paradoxical_subregs): Likewise.
2121 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2123 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
2124 (aarch64_frame): Add emit_frame_chain boolean.
2125 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2126 Move eh_return case to aarch64_layout_frame.
2127 (aarch64_layout_frame): Initialize emit_frame_chain.
2128 (aarch64_expand_prologue): Use emit_frame_chain.
2130 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2132 * config/aarch64/aarch64.c (aarch64_layout_frame):
2133 Ensure LR is always stored at the bottom of the callee-saves.
2134 Remove rarely used frame layout which saves callee-saves at top of
2135 frame, so the store of LR can be used as a valid probe in all cases.
2137 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2139 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
2140 Improve unaligned TImode/TFmode base/offset split.
2142 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2143 Alan Hayward <alan.hayward@arm.com>
2144 David Sherwood <david.sherwood@arm.com>
2146 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
2147 * combine.c (find_single_use_1): Likewise.
2148 (expand_field_assignment): Likewise.
2149 (move_deaths): Likewise.
2150 * lra-constraints.c (simplify_operand_subreg): Likewise.
2151 (curr_insn_transform): Likewise.
2152 * lra.c (collect_non_operand_hard_regs): Likewise.
2153 (add_regs_to_insn_regno_info): Likewise.
2154 * rtlanal.c (reg_referenced_p): Likewise.
2155 (covers_regno_no_parallel_p): Likewise.
2157 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2159 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
2160 Don't print any bits outside the precision of the value.
2161 * wide-int.cc (test_printing): Add some new tests.
2163 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2165 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
2166 supports -xbrace_comment option.
2167 * configure: Regenerate.
2168 * config.in: Regenerate.
2169 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
2170 (ASM_CPU_SPEC): Use it.
2172 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2174 * target.def (static_rtx_alignment): New hook.
2175 * targhooks.h (default_static_rtx_alignment): Declare.
2176 * targhooks.c (default_static_rtx_alignment): New function.
2177 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
2178 * doc/tm.texi: Regenerate.
2179 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
2180 instead of targetm.constant_alignment. Remove call to
2182 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2183 (cris_preferred_mininum_alignment): New function, split out from...
2184 (cris_constant_alignment): ...here.
2185 (cris_static_rtx_alignment): New function.
2186 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
2188 (ix86_constant_alignment): ...here.
2189 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2190 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2191 (mmix_static_rtx_alignment): New function.
2192 * config/spu/spu.c (spu_static_rtx_alignment): New function.
2193 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2195 2017-10-26 Tamar Christina <tamar.christina@arm.com>
2198 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
2199 Add flag_trapping_math and flag_fp_int_builtin_inexact.
2201 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
2203 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
2204 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
2205 mark as a sign-extending load.
2206 (local_pic_load_u): Define.
2208 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
2211 * fold-const.c (operand_equal_for_comparison_p): Also return true
2212 if ARG0 is a simple variant of ARG1 with narrower precision.
2213 (fold_ternary_loc): Always pass unstripped operands to the predicate.
2215 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
2217 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
2219 * i386.h (processor_costs): Add gather_static, gather_per_elt,
2220 scatter_static, scatter_per_elt.
2221 * x86-tune-costs.h: Add new cost entries.
2223 2017-10-25 Richard Biener <rguenther@suse.de>
2225 * tree-ssa-sccvn.h (vn_eliminate): Declare.
2226 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
2227 class pass_fre): Move to ...
2228 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
2229 class pass_fre): ... here and adjust for statistics.
2231 2017-10-25 Jakub Jelinek <jakub@redhat.com>
2234 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
2235 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
2236 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
2237 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
2240 2017-10-25 Richard Biener <rguenther@suse.de>
2242 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
2243 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
2244 eliminate_push_avail, eliminate_insert): Move inside...
2245 (class eliminate_dom_walker): ... this class in preparation
2247 (fini_eliminate): Remove by merging with ...
2248 (eliminate): ... this function. Adjust for class changes.
2249 (pass_pre::execute): Remove fini_eliminate call.
2250 (pass_fre::execute): Likewise.
2252 2017-10-24 Jakub Jelinek <jakub@redhat.com>
2255 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
2256 (VPERMI2, VPERMI2I): New mode iterators.
2257 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
2258 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
2260 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
2261 mode iterator. Remove 3 old define_insn patterns.
2262 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
2263 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
2264 VPERMI2 mode iterator, remove the other two expanders.
2265 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
2266 to use VPERMI2 mode iterator, add another alternative for vpermi2*
2267 instructions, remove the other two patterns.
2268 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
2269 mode iterator, remove the other two patterns.
2270 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
2271 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
2272 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
2273 and adjust argument order accordingly.
2274 (ix86_expand_vec_perm): Adjust caller.
2275 (expand_vec_perm_1): Likewise.
2276 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
2277 (expand_vec_perm_vpermt2_vpshub2): ... this.
2278 (ix86_expand_vec_perm_const_1): Adjust caller.
2279 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
2282 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
2283 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
2284 (vec_shl_<mode>): Remove unused expander.
2285 (avx512bw_<shift_insn><mode>3): New define_insn.
2286 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
2287 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
2289 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
2292 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
2295 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
2297 PR rtl-optimization/82396
2298 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
2299 (autopref_multipass_init): Simplify initialization.
2300 (autopref_rank_data): Simplify sort order.
2301 * gcc/sched-int.h (autopref_multipass_data_): Remove
2302 multi_mem_insn_p, min_offset and max_offset.
2304 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
2307 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2308 Check special value of flag_omit_frame_pointer.
2309 (aarch64_can_eliminate): Likewise.
2310 (aarch64_override_options_after_change_1): Simplify handling of
2311 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
2313 2017-10-24 Richard Biener <rguenther@suse.de>
2315 PR tree-optimization/82697
2316 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
2317 zero for conditional load and unconditional store.
2319 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
2321 * doc/install.texi: Document bootstrap-cet.
2323 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
2326 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
2327 ENDBR instruction at function entrance if function is only
2330 2017-10-24 Jakub Jelinek <jakub@redhat.com>
2333 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
2334 patterns to better describe from which operation the CF is computed.
2335 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
2336 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
2337 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
2338 is 0, use _0 suffixed expanders instead of emitting a comparison
2341 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
2343 * config/i386/i386.md(*movsf_internal, *movdf_internal):
2344 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
2346 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
2349 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
2350 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
2351 * loop-iv.c (iv_get_reaching_def): Likewise.
2352 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
2353 variable is promoted and the partition contains undefined values.
2355 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2357 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
2359 (nios2_address_cost): Define.
2360 (nios2_legitimize_address): Recognize (exp + constant) directly.
2361 (TARGET_ADDRESS_COST): Define.
2363 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2365 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
2366 (nios2_symbolic_memory_operand_p): Declare.
2367 (nios2_split_large_constant): Declare.
2368 (nios2_split_symbolic_memory_operand): Declare.
2369 * config/nios2/nios2.c: Adjust includes.
2370 (nios2_symbolic_constant_allowed): New.
2371 (nios2_symbolic_constant_p): New.
2372 (nios2_plus_symbolic_constant_p): New.
2373 (nios2_valid_addr_expr_p): Recognize addresses involving
2375 (nios2_legitimate_address_p): Likewise, also LO_SUM.
2376 (nios2_symbolic_memory_operand_p): New.
2377 (nios2_large_constant_p): New.
2378 (nios2_split_large_constant): New.
2379 (nios2_split_plus_large_constant): New.
2380 (nios2_split_symbolic_memory_operand): New.
2381 (nios2_legitimize_address): Code refactoring. Handle addresses
2382 involving symbolic constants.
2383 (nios2_emit_move_sequence): Likewise.
2384 (nios2_print_operand): Improve error output.
2385 (nios2_print_operand_address): Handle LO_SUM.
2386 (nios2_cdx_narrow_form_p): Likewise.
2387 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
2388 operands involving symbolic constants.
2389 (movhi_internal, movsi_internal): Likewise.
2390 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
2391 (extendhisi2, extendqi<mode>2): Likewise.
2393 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2395 * tree-pass.h (PROP_rtl_split_insns): Define.
2396 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
2398 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2400 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
2402 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2405 * target.def (const_not_ok_for_debug_p): Default to
2406 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
2407 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
2408 * targhooks.c (default_const_not_ok_for_debug_p): New function.
2409 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
2410 which targetm.const_not_ok_for_debug_p returned true.
2411 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
2413 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
2415 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
2416 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
2417 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
2419 (ix86_const_not_ok_for_debug_p): New function.
2420 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
2421 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
2423 2017-10-23 David Malcolm <dmalcolm@redhat.com>
2426 * system.h: Conditionally include "unique-ptr.h" if
2427 INCLUDE_UNIQUE_PTR is defined.
2428 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
2429 of defining INCLUDE_UNIQUE_PTR before including "system.h".
2431 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
2433 * config/rl78/rl78.md: New define_expand "subdi3".
2435 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
2438 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
2439 DF_REF_INSN if DF_REF_INSN_INFO is false.
2441 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
2443 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
2444 xmm_move instead of sse_move.
2445 (sse_store_index): New function.
2446 (ix86_register_move_cost): Be more sensible about mismatch stall;
2447 model AVX moves correctly; make difference between sse->integer and
2449 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
2450 moves; make difference between SSE and AVX.
2451 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
2452 and zmm_move. Increase size of sse load and store tables;
2453 add unaligned load and store tables; add ssemmx_to_integer.
2454 * x86-tune-costs.h: Update all entries according to real
2455 move latencies from Agner Fog's manual and chip documentation.
2457 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2460 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
2461 * config/i386/constraints.md (Wf): New constraint.
2462 * config/i386/i386.md (UNSPEC_SBB): New unspec.
2463 (cmp<dwi>_doubleword): Removed.
2464 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
2465 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
2466 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
2467 expand with cmp<dwi>_doubleword. For LTU and GEU use
2468 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
2470 * common.opt (gcolumn-info): Enable by default.
2471 * doc/invoke.texi (gcolumn-info): Document new default.
2473 2017-10-23 Richard Biener <rguenther@suse.de>
2475 PR tree-optimization/82672
2476 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
2477 Fold the stmt if we propagated into it.
2479 2017-10-23 Richard Biener <rguenther@suse.de>
2481 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
2482 (bitmap_remove_expr_from_set): ... this. All callers call this
2483 for non-constant values.
2484 (bitmap_set_subtract): Rename to...
2485 (bitmap_set_subtract_expressions): ... this. Adjust and
2487 (bitmap_set_contains_value): Remove superfluous check.
2488 (bitmap_set_replace_value): Inline into single caller ...
2489 (bitmap_value_replace_in_set): ... here and simplify.
2490 (dependent_clean): Merge into ...
2491 (clean): ... this using an overload. Adjust.
2492 (prune_clobbered_mems): Adjust.
2493 (compute_antic_aux): Likewise.
2494 (compute_partial_antic_aux): Likewise.
2496 2017-10-23 Richard Biener <rguenther@suse.de>
2498 PR tree-optimization/82129
2500 2017-08-01 Richard Biener <rguenther@suse.de>
2502 PR tree-optimization/81181
2503 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
2504 (compute_antic): ... end of iteration here.
2506 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2508 * target.def (starting_frame_offset): New hook.
2509 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
2510 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
2511 * doc/tm.texi.in: Regenerate.
2512 * hooks.h (hook_hwi_void_0): Declare.
2513 * hooks.c (hook_hwi_void_0): New function.
2514 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
2515 STARTING_FRAME_OFFSET.
2516 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
2517 * reload1.c (reload): Likewise.
2518 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
2519 instead of STARTING_FRAME_OFFSET.
2520 * function.c (try_fit_stack_local): Likewise.
2521 (assign_stack_local_1): Likewise
2522 (instantiate_virtual_regs): Likewise.
2523 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
2524 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
2525 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
2526 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
2527 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
2528 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
2529 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
2530 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
2531 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
2532 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
2533 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
2534 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
2535 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
2536 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
2537 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
2538 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
2539 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
2540 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
2541 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
2542 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
2543 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
2544 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
2545 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
2546 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
2547 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
2548 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
2549 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
2550 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
2551 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
2552 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
2553 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
2554 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
2555 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
2556 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
2557 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
2558 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
2559 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
2560 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
2561 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
2562 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
2563 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
2564 * config/avr/avr.c (avr_starting_frame_offset): Make static and
2565 return a HOST_WIDE_INT.
2566 (avr_builtin_setjmp_frame_value): Use it instead of
2567 STARTING_FRAME_OFFSET.
2568 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2569 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
2570 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
2572 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2573 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
2574 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
2575 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2576 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
2577 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
2578 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2579 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
2580 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
2581 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2582 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
2583 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
2585 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2586 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
2587 * config/mips/mips.c (mips_compute_frame_info): Refer to
2588 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
2589 (mips_starting_frame_offset): New function.
2590 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2591 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
2592 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
2593 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
2594 and return a HOST_WIDE_INT.
2595 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2596 (mmix_initial_elimination_offset): Refer to
2597 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
2598 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
2599 * config/pa/pa.c (pa_starting_frame_offset): New function.
2600 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
2601 (pa_expand_prologue): Likewise.
2602 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2603 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
2604 !FRAME_GROWS_DOWNWARD handling to...
2605 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2606 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
2607 !FRAME_GROWS_DOWNWARD handling to...
2608 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2609 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
2610 !FRAME_GROWS_DOWNWARD handling to...
2611 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2612 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
2614 (rs6000_starting_frame_offset): New function.
2615 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
2616 !FRAME_GROWS_DOWNWARD handling to...
2617 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2618 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
2619 !FRAME_GROWS_DOWNWARD handling to...
2620 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2621 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
2622 !FRAME_GROWS_DOWNWARD handling to...
2623 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2624 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
2625 (rs6000_starting_frame_offset): New function.
2626 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
2627 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
2628 * config/vax/vax.c (vax_starting_frame_offset): New function.
2629 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
2630 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2631 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
2632 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
2633 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2634 * system.h (STARTING_FRAME_OFFSET): Poison.
2636 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2638 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
2639 SCALAR_TYPE_MODE instead of TYPE_MODE.
2641 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2642 Alan Hayward <alan.hayward@arm.com>
2643 David Sherwood <david.sherwood@arm.com>
2645 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
2647 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2648 Alan Hayward <alan.hayward@arm.com>
2649 David Sherwood <david.sherwood@arm.com>
2651 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
2653 2017-10-23 Richard Biener <rguenther@suse.de>
2655 PR tree-optimization/82129
2656 * tree-ssa-pre.c (bitmap_set_and): Remove.
2657 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
2658 canonicalizing expressions in the set to those with lowest
2659 ID rather than taking that from the first edge.
2661 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2663 * combine.c (rtx_equal_for_field_assignment_p): Use
2664 byte_lowpart_offset.
2666 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2667 Alan Hayward <alan.hayward@arm.com>
2668 David Sherwood <david.sherwood@arm.com>
2670 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
2671 to a SUBREG_PROMOTED_VAR.
2673 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2674 Alan Hayward <alan.hayward@arm.com>
2675 David Sherwood <david.sherwood@arm.com>
2677 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
2678 (expand_debug_source_expr): Likewise.
2679 * combine.c (combine_simplify_rtx): Likewise.
2680 * cse.c (fold_rtx): Likewise.
2681 * optabs.c (expand_float): Likewise.
2682 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2683 (simplify_binary_operation_1): Likewise.
2685 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2686 Alan Hayward <alan.hayward@arm.com>
2687 David Sherwood <david.sherwood@arm.com>
2689 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
2690 (record_promoted_value): Likewise.
2691 * expr.c (expand_expr_real_2): Likewise.
2692 * ree.c (update_reg_equal_equiv_notes): Likewise.
2693 (combine_set_extension): Likewise.
2694 * rtlanal.c (low_bitmask_len): Likewise.
2695 * simplify-rtx.c (neg_const_int): Likewise.
2696 (simplify_binary_operation_1): Likewise.
2698 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2699 Alan Hayward <alan.hayward@arm.com>
2700 David Sherwood <david.sherwood@arm.com>
2702 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
2703 * regcprop.c (maybe_mode_change): Likewise.
2704 * reload1.c (alter_reg): Likewise.
2706 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2708 * inchash.h (inchash::hash::add_wide_int): New function.
2709 * lto-streamer-out.c (hash_tree): Use it.
2711 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2713 * inchash.h (inchash::hash::add_wide_int): Rename to...
2714 (inchash::hash::add_hwi): ...this.
2715 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
2716 (polymorphic_call_target_hasher::hash): Likewise.
2717 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
2718 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
2719 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
2720 * lto-streamer-out.c (hash_tree): Likewise.
2721 * optc-save-gen.awk: Likewise.
2722 * tree.c (add_expr): Likewise.
2724 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
2727 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
2728 for ordered inequality comparisons even with TARGET_IEEE_FP.
2730 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
2733 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
2734 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
2735 Expand with cmp<dwi>_doubleword.
2737 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2739 * extend.texi: Add x86 specific to 'nocf_check' attribute.
2740 List CET intrinsics.
2741 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
2742 specific to -fcf-protection option.
2744 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2746 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
2747 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
2748 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
2749 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
2750 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
2751 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
2752 (extra_objs): Add cet.o for Linux/x86 targets.
2753 (tmake_file): Add i386/t-cet for Linux/x86 targets.
2754 * config/i386/cet.c: New file.
2755 * config/i386/cetintrin.h: Likewise.
2756 * config/i386/t-cet: Likewise.
2757 * config/i386/cpuid.h (bit_SHSTK): New.
2758 (bit_IBT): Likewise.
2759 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
2760 pass IBT and SHSTK bits.
2761 * config/i386/i386-builtin-types.def
2762 (VOID_FTYPE_UNSIGNED_PVOID): New.
2763 (VOID_FTYPE_UINT64_PVOID): Likewise.
2764 * config/i386/i386-builtin.def: Add CET intrinsics.
2765 * config/i386/i386-c.c (ix86_target_macros_internal): Add
2766 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
2767 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
2768 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
2770 * config/i386/i386.c (rest_of_insert_endbranch): New.
2771 (pass_data_insert_endbranch): Likewise.
2772 (pass_insert_endbranch): Likewise.
2773 (make_pass_insert_endbranch): Likewise.
2774 (ix86_notrack_prefixed_insn_p): Likewise.
2775 (ix86_target_string): Add -mibt, -mshstk flags.
2776 (ix86_option_override_internal): Add flag_cf_protection
2778 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
2779 (ix86_print_operand): Add 'notrack' prefix output.
2780 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
2781 (ix86_expand_builtin): Expand CET intrinsics.
2782 (x86_output_mi_thunk): Add 'endbranch' instruction.
2783 * config/i386/i386.h (TARGET_IBT): New.
2784 (TARGET_IBT_P): Likewise.
2785 (TARGET_SHSTK): Likewise.
2786 (TARGET_SHSTK_P): Likewise.
2787 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
2788 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
2789 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
2790 (builtin_setjmp_setup): New pattern.
2791 (builtin_longjmp): Likewise.
2792 (rdssp<mode>): Likewise.
2793 (incssp<mode>): Likewise.
2794 (saveprevssp): Likewise.
2795 (rstorssp): Likewise.
2796 (wrss<mode>): Likewise.
2797 (wruss<mode>): Likewise.
2798 (setssbsy): Likewise.
2799 (clrssbsy): Likewise.
2800 (nop_endbr): Likewise.
2801 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
2803 * config/i386/immintrin.h: Include <cetintrin.h>.
2804 * config/i386/linux-common.h
2805 (file_end_indicate_exec_stack_and_cet): New prototype.
2806 (TARGET_ASM_FILE_END): New.
2808 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
2810 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
2811 latencies instead of having separate table; make difference between
2812 integer and float costs.
2813 * i386.h (processor_costs): Remove scalar_stmt_cost,
2814 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
2815 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
2817 * x86-tune-costs.h: Remove entries which has been removed in
2818 procesor_costs from all tables; make cond_taken_branch_cost
2819 and cond_not_taken_branch_cost COST_N_INSNS based.
2821 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
2823 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
2825 2017-10-20 Jakub Jelinek <jakub@redhat.com>
2827 * config/i386/i386.md (isa): Remove fma_avx512f.
2828 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
2829 <avx512>_fmadd_<mode>_mask3<round_name>,
2830 <avx512>_fmsub_<mode>_mask<round_name>,
2831 <avx512>_fmsub_<mode>_mask3<round_name>,
2832 <avx512>_fnmadd_<mode>_mask<round_name>,
2833 <avx512>_fnmadd_<mode>_mask3<round_name>,
2834 <avx512>_fnmsub_<mode>_mask<round_name>,
2835 <avx512>_fnmsub_<mode>_mask3<round_name>,
2836 <avx512>_fmaddsub_<mode>_mask<round_name>,
2837 <avx512>_fmaddsub_<mode>_mask3<round_name>,
2838 <avx512>_fmsubadd_<mode>_mask<round_name>,
2839 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
2840 (*vec_widen_umult_even_v16si<mask_name>,
2841 *vec_widen_smult_even_v16si<mask_name>): Likewise.
2842 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
2844 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2846 * extend.texi: Add 'nocf_check' documentation.
2847 * gimple.texi: Add second parameter to
2848 gimple_build_call_from_tree.
2849 * invoke.texi: Add -fcf-protection documentation.
2850 * rtl.texi: Add REG_CALL_NOTRACK documenation.
2852 2017-10-20 Richard Biener <rguenther@suse.de>
2854 PR tree-optimization/82473
2855 * tree-vect-loop.c (vectorizable_reduction): Properly get at
2856 the largest input type.
2858 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2860 * c-attribs.c (handle_nocf_check_attribute): New function.
2861 (c_common_attribute_table): Add 'nocf_check' handling.
2862 * gimple-parser.c: Add second argument NULL to
2863 gimple_build_call_from_tree.
2864 * attrib.c (comp_type_attributes): Check nocf_check attribute.
2865 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
2867 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
2868 * common.opt: Add fcf-protection flag.
2869 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
2870 * flag-types.h: Add enum cf_protection_level.
2871 * gimple.c (gimple_build_call_from_tree): Add second parameter.
2872 Add 'nocf_check' attribute propagation to gimple call.
2873 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
2874 (gimple_build_call_from_tree): Update prototype.
2875 (gimple_call_nocf_check_p): New function.
2876 (gimple_call_set_nocf_check): Likewise.
2877 * gimplify.c: Add second argument to gimple_build_call_from_tree.
2878 * ipa-icf.c: Add nocf_check attribute in statement hash.
2879 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
2880 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
2881 * toplev.c (process_options): Add flag_cf_protection handling.
2883 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
2885 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
2887 2017-10-20 Richard Biener <rguenther@suse.de>
2889 PR tree-optimization/82603
2890 * tree-if-conv.c (predicate_mem_writes): Make sure to only
2891 remove false predicated stores.
2893 2017-10-20 Richard Biener <rguenther@suse.de>
2895 * graphite-isl-ast-to-gimple.c
2896 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
2897 Remove return value and simplify, dump copied stmt after lhs
2899 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
2900 Reduce dump verbosity.
2901 (gsi_insert_earliest): Likewise.
2902 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
2903 * graphite.c (print_global_statistics): Adjust dumping.
2904 (print_graphite_scop_statistics): Likewise.
2905 (print_graphite_statistics): Do not dump loops here.
2906 (graphite_transform_loops): But here.
2908 2017-10-20 Nicolas Roche <roche@adacore.com>
2910 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
2911 * configure: Regenerate.
2913 2017-10-20 Jakub Jelinek <jakub@redhat.com>
2916 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
2917 functions when optimizing replace GIMPLE_RETURN stmts with
2918 calls to __builtin_unreachable ().
2921 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
2922 for -fsanitize=thread link of executables.
2923 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
2924 link of executables.
2927 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
2929 (<shift_insn><mode>3<mask_name>): Change the last of the 3
2930 define_insns for logical vector shifts to use VI248_AVX512BW
2931 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
2932 condition, useless isa and prefix attributes. Change the first
2933 2 of these define_insns to ...
2934 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
2935 define_insn for avx512vl.
2936 (<shift_insn><mode>3): ... and this, new define_insn without
2937 masking for non-avx512vl.
2940 * config/i386/sse.md (*andnot<mode>3,
2941 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
2942 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
2943 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
2944 not applied use empty suffix even for TARGET_AVX512VL.
2945 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
2946 is applied, supply evex,evex or evex,evex,evex instead of just
2949 2017-10-20 Julia Koval <julia.koval@intel.com>
2951 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
2952 (OPTION_MASK_ISA_GFNI_UNSET): New.
2953 (ix86_handle_option): Handle OPT_mgfni.
2954 * config/i386/cpuid.h (bit_GFNI): New.
2955 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
2956 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
2957 * config/i386/i386.c (ix86_target_string): Add -mgfni.
2958 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
2959 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
2960 * config/i386/i386.opt: Add mgfni.
2962 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
2964 * config/msp430/msp430.c (msp430_option_override): Disable
2965 -fdelete-null-pointer-checks.
2966 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
2968 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
2970 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
2971 of x87 and SSE instructions.
2973 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
2975 * asan.c (create_cond_insert_point): Do not update edge count.
2976 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
2977 (afdo_propagate_circuit): Likewise.
2978 (afdo_calculate_branch_prob): Likewise.
2979 (afdo_annotate_cfg): Likewise.
2980 * basic-block.h (struct edge_def): Remove count.
2981 (edge_def::count): New accessor.
2982 * bb-reorder.c (rotate_loop): Update.
2983 (find_traces_1_round): Update.
2984 (connect_traces): Update.
2985 (sanitize_hot_paths): Update.
2986 * cfg.c (unchecked_make_edge): Update.
2987 (make_single_succ_edge): Update.
2988 (check_bb_profile): Update.
2989 (dump_edge_info): Update.
2990 (update_bb_profile_for_threading): Update.
2991 (scale_bbs_frequencies_int): Update.
2992 (scale_bbs_frequencies_gcov_type): Update.
2993 (scale_bbs_frequencies_profile_count): Update.
2994 (scale_bbs_frequencies): Update.
2995 * cfganal.c (connect_infinite_loops_to_exit): Update.
2996 * cfgbuild.c (compute_outgoing_frequencies): Update.
2997 (find_many_sub_basic_blocks): Update.
2998 * cfgcleanup.c (try_forward_edges): Update.
2999 (try_crossjump_to_edge): Update
3000 * cfgexpand.c (expand_gimple_cond): Update
3001 (expand_gimple_tailcall): Update
3002 (construct_exit_block): Update
3003 * cfghooks.c (verify_flow_info): Update
3004 (redirect_edge_succ_nodup): Update
3005 (split_edge): Update
3006 (make_forwarder_block): Update
3007 (duplicate_block): Update
3008 (account_profile_record): Update
3009 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
3010 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
3011 * cfgloopmanip.c (scale_loop_profile): Update.
3013 (lv_adjust_loop_entry_edge): Update.
3014 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
3015 (force_nonfallthru_and_redirect): Update.
3016 (purge_dead_edges): Update.
3017 (rtl_flow_call_edges_add): Update.
3018 * cgraphunit.c (init_lowered_empty_function): Update.
3019 (cgraph_node::expand_thunk): Update.
3020 * gimple-pretty-print.c (dump_probability): Update.
3021 (dump_edge_probability): Update.
3022 * gimple-ssa-isolate-paths.c (isolate_path): Update.
3023 * haifa-sched.c (sched_create_recovery_edges): Update.
3024 * hsa-gen.c (convert_switch_statements): Update.
3025 * ifcvt.c (dead_or_predicable): Update.
3026 * ipa-inline-transform.c (inline_transform): Update.
3027 * ipa-split.c (split_function): Update.
3028 * ipa-utils.c (ipa_merge_profiles): Update.
3029 * loop-doloop.c (add_test): Update.
3030 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
3031 * lto-streamer-in.c (input_cfg): Update.
3032 (input_function): Update.
3033 * lto-streamer-out.c (output_cfg): Update.
3034 * modulo-sched.c (sms_schedule): Update.
3035 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
3036 * predict.c (maybe_hot_edge_p): Update.
3037 (unlikely_executed_edge_p): Update.
3038 (probably_never_executed_edge_p): Update.
3039 (dump_prediction): Update.
3040 (drop_profile): Update.
3041 (propagate_unlikely_bbs_forward): Update.
3042 (determine_unlikely_bbs): Update.
3043 (force_edge_cold): Update.
3044 * profile.c (compute_branch_probabilities): Update.
3045 * reg-stack.c (better_edge): Update.
3046 * shrink-wrap.c (handle_simple_exit): Update.
3047 * tracer.c (better_p): Update.
3048 * trans-mem.c (expand_transaction): Update.
3049 (split_bb_make_tm_edge): Update.
3050 * tree-call-cdce.c: Update.
3051 * tree-cfg.c (gimple_find_sub_bbs): Update.
3052 (gimple_split_edge): Update.
3053 (gimple_duplicate_sese_region): Update.
3054 (gimple_duplicate_sese_tail): Update.
3055 (gimple_flow_call_edges_add): Update.
3056 (insert_cond_bb): Update.
3057 (execute_fixup_cfg): Update.
3058 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
3059 * tree-complex.c (expand_complex_div_wide): Update.
3060 * tree-eh.c (lower_resx): Update.
3061 (unsplit_eh): Update.
3062 (cleanup_empty_eh_move_lp): Update.
3063 * tree-inline.c (copy_edges_for_bb): Update.
3064 (freqs_to_counts): Update.
3065 (copy_cfg_body): Update.
3066 * tree-ssa-dce.c (remove_dead_stmt): Update.
3067 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
3068 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
3069 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
3070 (unloop_loops): Update.
3071 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3072 * tree-ssa-loop-split.c (connect_loops): Update.
3073 (split_loop): Update.
3074 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
3075 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
3076 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
3077 * tree-ssa-reassoc.c (branch_fixup): Update.
3078 * tree-ssa-tail-merge.c (replace_block_by): Update.
3079 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
3080 (compute_path_counts): Update.
3081 (update_profile): Update.
3082 (recompute_probabilities): Update.
3083 (update_joiner_offpath_counts): Update.
3084 (estimated_freqs_path): Update.
3085 (freqs_to_counts_path): Update.
3086 (clear_counts_path): Update.
3087 (ssa_fix_duplicate_block_edges): Update.
3088 (duplicate_thread_path): Update.
3089 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
3090 (case_bit_test_cmp): Update.
3091 (collect_switch_conv_info): Update.
3092 (gen_inbound_check): Update.
3093 (do_jump_if_equal): Update.
3094 (emit_cmp_and_jump_insns): Update.
3095 * tree-tailcall.c (decrease_profile): Update.
3096 (eliminate_tail_call): Update.
3097 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
3098 (vect_do_peeling): Update.
3099 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3100 * ubsan.c (ubsan_expand_null_ifn): Update.
3101 (ubsan_expand_ptr_ifn): Update.
3102 * value-prof.c (gimple_divmod_fixed_value): Update.
3103 (gimple_mod_pow2): Update.
3104 (gimple_mod_subtract): Update.
3105 (gimple_ic): Update.
3106 (gimple_stringop_fixed_value): Update.
3108 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3111 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
3113 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
3115 PR rtl-optimization/82395
3116 * ira-color.c (allocno_priority_compare_func): Fix comparison step
3117 based on non_spilled_static_chain_regno_p.
3119 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3121 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
3122 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
3123 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
3125 2017-10-19 Martin Sebor <msebor@redhat.com>
3127 PR tree-optimization/82596
3128 * tree.c (array_at_struct_end_p): Handle STRING_CST.
3130 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3132 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
3133 (get_mem_refs_of_builtin_call): Likewise.
3134 * builtins.c (expand_builtin_apply): Adjust call to
3135 allocate_dynamic_stack_space.
3136 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
3137 the third argument to allocate_dynamic_stack_space, otherwise -1.
3138 (expand_builtin): Deal with all alloca variants.
3139 (is_inexpensive_builtin): Likewise.
3140 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
3141 * calls.c (special_function_p): Deal with all alloca variants.
3142 (initialize_argument_information): Adjust call to
3143 allocate_dynamic_stack_space.
3144 (expand_call): Likewise.
3145 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
3146 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
3147 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
3148 use it for the stack usage computation.
3149 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
3150 * function.c (gimplify_parameters): Call build_alloca_call_expr.
3151 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
3152 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
3153 (in_loop_p): Remove first argument and useless check.
3154 (pass_walloca::execute): Remove useless test and adjust call to above.
3155 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
3156 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
3157 (gimplify_call_expr): Deal with all alloca variants.
3158 * hsa-gen.c (gen_hsa_alloca): Likewise.
3159 (gen_hsa_insns_for_call): Likewise.
3160 * ipa-pure-const.c (special_builtin_state): Likewise.
3161 * tree-chkp.c (chkp_build_returned_bound): Likewise.
3162 * tree-object-size.c (alloc_object_size): Likewise.
3163 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
3164 (call_may_clobber_ref_p_1): Likewise.
3165 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
3166 (ccp_fold_stmt): Likewise.
3167 (optimize_stack_restore): Likewise.
3168 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
3169 (mark_all_reaching_defs_necessary_1): Likewise.
3170 (propagate_necessity): Likewise.
3171 (eliminate_unnecessary_stmts): Likewise.
3172 * tree.c (build_common_builtin_nodes): Build
3173 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
3174 (build_alloca_call_expr): New function.
3175 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
3176 (CASE_BUILT_IN_ALLOCA): Likewise.
3177 (build_alloca_call_expr): Declare.
3178 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
3180 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3183 * dwarf2out.c (new_die_raw): New static inline function.
3184 (new_die): Use it to create the DIE.
3185 (add_AT_external_die_ref): Likewise.
3186 (clone_die): Likewise.
3187 (clone_as_declaration): Likewise.
3188 (dwarf2out_vms_debug_main_pointer): Likewise.
3189 (base_type_die): Likewise. Remove early return for corner cases.
3190 Do not call add_pubtype on the DIE here.
3191 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
3192 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
3193 typedefs for base types with DW_AT_endianity. Make sure a DIE with
3194 native order exists for base types, attach the DIE manually and call
3195 add_pubtype on it. Do not equate a reverse order DIE to the type.
3197 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
3199 * config/arm/arm.c (align_ok_ldrd_strd): New function.
3200 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
3202 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
3204 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3206 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
3207 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
3208 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
3209 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
3210 * opts.c (sanitizer_opts): Add builtin.
3211 * ubsan.c (instrument_builtin): New function.
3212 (pass_ubsan::execute): Call it.
3213 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
3214 * doc/invoke.texi: Document -fsanitize=builtin.
3216 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
3217 builtins, store max (log2 (align), 0) into uchar field instead of
3218 align into uptr field.
3219 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
3220 store uchar 0 field instead of uptr 0 field.
3221 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
3222 instead of passing one address of struct with 2 locations pass
3223 two addresses of structs with 1 location each.
3224 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
3225 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
3226 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
3227 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
3228 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
3229 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
3230 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
3231 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
3233 2017-10-19 Martin Liska <mliska@suse.cz>
3236 * file-find.c (remove_prefix): Remove.
3237 * file-find.h (remove_prefix): Likewise.
3238 * gcc-ar.c: Remove smartness of lookup.
3240 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
3242 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
3243 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
3244 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
3246 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3249 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
3250 (setcc + and to xor + setcc): New peephole2.
3252 2017-10-19 Tom de Vries <tom@codesourcery.com>
3254 * doc/sourcebuild.texi (Test Directives, Variants of
3255 dg-require-support): Add dg-require-stack-size.
3257 2017-10-19 Martin Liska <mliska@suse.cz>
3260 * gimplify.c (gimplify_decl_expr): Do not instrument variables
3261 that have a large alignment.
3262 (gimplify_target_expr): Likewise.
3264 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
3266 PR rtl-optimization/82602
3267 * ira.c (rtx_moveable_p): Return false for volatile asm.
3269 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
3272 * config/i386/i386-modes.def (CCGZ): New CC mode.
3273 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
3274 * config/i386/predicates.md (ix86_comparison_operator):
3276 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
3277 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
3278 with double-word subtraction.
3279 (put_condition_code): Handle CCGZmode.
3281 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
3283 * wide-int.cc (debug (const wide_int &)): New.
3284 (debug (const wide_int *)): New.
3285 (debug (const widest_int &)): New.
3286 (debug (const widest_int *)): New.
3288 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
3291 * lra-constraints.c (curr_insn_transform): Use non-input operand
3292 instead of output one for matched reload.
3294 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3296 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
3297 (tree-ssa-loop-ivopts.h): New header file.
3298 (struct builtin_info): New fields.
3299 (classify_builtin_1): Compute and record base and offset parts for
3300 memset builtin partition by calling strip_offset.
3301 (offset_cmp, fuse_memset_builtins): New functions.
3302 (finalize_partitions): Fuse adjacent memset partitions by calling
3304 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
3305 Expose the interface.
3306 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
3308 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3310 PR tree-optimization/82574
3311 * tree-loop-distribution.c (find_single_drs): New parameter. Check
3312 that data reference must be executed exactly once per iteration
3313 against the outermost loop in nest.
3314 (classify_partition): Update call to above function.
3316 2017-10-18 Richard Biener <rguenther@suse.de>
3318 PR tree-optimization/82591
3319 * graphite.c (graphite_transform_loops): Move code gen message
3321 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3322 Here. Handle scop_to_isl_ast failing.
3323 (scop_to_isl_ast): Limit the number of ISL operations.
3325 2017-10-18 Richard Biener <rguenther@suse.de>
3327 * graphite-isl-ast-to-gimple.c
3328 (translate_isl_ast_to_gimple::set_rename): Simplify.
3329 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
3330 (graphite_copy_stmts_from_block): ... here.
3331 (copy_bb_and_scalar_dependences): Simplify.
3332 (add_parameters_to_ivs_params): Canonicalize.
3333 (generate_entry_out_of_ssa_copies): Simplify.
3334 * graphite-sese-to-poly.c (extract_affine_name): Simplify
3335 by passing in ISL dimension.
3336 (parameter_index_in_region_1): Rename to ...
3337 (parameter_index_in_region): ... this.
3338 (extract_affine): Adjust assert, pass down parameter index.
3339 (add_param_constraints): Use range-info when available.
3340 (build_scop_context): Adjust.
3341 * sese.c (new_sese_info): Adjust.
3342 (free_sese_info): Likewise.
3343 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
3344 Remove unused typedefs.
3345 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
3347 2017-10-18 Martin Liska <mliska@suse.cz>
3349 * combine.c (simplify_compare_const): Add gcc_fallthrough.
3351 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
3353 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
3354 (s390_sched_init): Do not reset s390_sched_state if we entered the
3355 current basic block via a fallthru edge and all others are unlikely.
3357 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
3359 * config/s390/s390.c (NUM_SIDES): New variable.
3360 (LONGRUNNING_THRESHOLD): New variable.
3361 (LATENCY_FACTOR): New variable.
3362 (s390_sched_score): Decrease score for long-running instructions on
3364 (s390_sched_variable_issue): Perform bookkeeping for long-running
3367 2017-10-18 Richard Biener <rguenther@suse.de>
3369 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
3370 Simplify with removal of the parameter rename map.
3371 (set_rename): Likewise.
3372 (should_copy_to_new_region): Likewise.
3373 (graphite_copy_stmts_from_block): Likewise.
3374 (copy_bb_and_scalar_dependences): Remove initialization of
3375 unused copied_bb_map.
3377 (copy_internal_parameters): Likewise.
3378 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
3379 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
3380 Use INTEGRAL_TYPE_P.
3381 (parameter_index_in_region_1): Rename to ...
3382 (assign_parameter_index_in_region): ... this. Assert we have
3383 a parameter we handle.
3384 (scan_tree_for_params): Adjust.
3385 * sese.h (parameter_rename_map_t): Remove.
3386 (struct sese_info_t): Remove unused parameter_rename_map and
3387 copied_bb_map members.
3388 * sese.c (new_sese_info): Adjust.
3389 (free_sese_info): Likewise.
3391 2017-10-18 Martin Liska <mliska@suse.cz>
3394 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
3395 on an abnormal edge.
3397 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
3399 * doc/invoke.texi (ffunction-sections and fdata-sections):
3402 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3404 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
3405 the use statement can throw internally.
3407 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3409 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
3410 any RTX present on the RHS of a SET.
3411 * compare-elim.c (try_eliminate_compare): Restore comment.
3413 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3415 * langhooks.h (struct lang_hooks): Document that tree_size langhook
3416 may be also called on tcc_type nodes.
3417 * langhooks.c (lhd_tree_size): Likewise.
3419 2017-10-17 David Malcolm <dmalcolm@redhat.com>
3421 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
3422 format_warning_at_substring.
3423 (maybe_warn): Convert source_range * param to a location_t. Pass
3424 UNKNOWN_LOCATION rather than NULL to fmtwarn.
3425 (format_directive): Remove code to extract source_ranges and
3426 source_range * in favor of just a location_t.
3427 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
3429 * substring-locations.c (format_warning_va): Convert
3430 source_range * param to a location_t.
3431 (format_warning_at_substring): Likewise.
3432 * substring-locations.h (format_warning_va): Likewise.
3433 (format_warning_at_substring): Likewise.
3435 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
3437 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
3439 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
3440 and scatter/gather ops.
3442 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
3443 vec_gather_load and vec_scatter_store.
3444 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
3445 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
3446 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
3447 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
3448 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
3449 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
3451 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
3453 * reg-stack.c (compare_for_stack_reg): Add bool argument.
3454 Detect FTST instruction and handle its register pops. Only pop
3455 second operand if can_pop_second_op is true.
3456 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
3457 set can_pop_second_op to false in the compare_for_stack_reg call.
3459 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
3460 output_fp_compare for stack register operands.
3461 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
3462 instructions here. Do not emit stack register pops here. Assert
3463 that FCOMPP pops next to top stack register. Rewrite function.
3465 2017-10-17 Nathan Sidwell <nathan@acm.org>
3468 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
3469 use DECL_ASSEMBLER_NAME_RAW.
3472 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
3475 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
3476 Wilco Dijkstra <wilco.dijkstra@arm.com>
3478 * builtins.c (expand_builtin_update_setjmp_buf): Add a
3479 converstion to Pmode from the buf_addr.
3481 2017-10-17 Richard Biener <rguenther@suse.de>
3483 * graphite-dependences.c (scop_get_reads_and_writes): Change
3484 output parameters to references.
3486 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
3488 PR 71026/tree-optimization
3489 * fold-const.c (distribute_real_division): Removed.
3490 (fold_binary_loc): Remove calls to distribute_real_divison.
3492 2017-10-17 Richard Biener <rguenther@suse.de>
3494 * graphite-scop-detection.c
3495 (scop_detection::stmt_has_simple_data_refs_p): Always use
3496 the full nest as region.
3497 (try_generate_gimple_bb): Likewise.
3498 * sese.c (scalar_evolution_in_region): Simplify now that
3499 SCEV can handle instantiation in regions.
3500 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
3501 in the non-loop part of a function if requested.
3503 2017-10-17 Richard Biener <rguenther@suse.de>
3505 PR tree-optimization/82563
3506 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
3508 (graphite_regenerate_ast_isl): Call it.
3509 * graphite-scop-detection.c (build_scops): Remove entry edge split.
3511 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3513 PR tree-optimization/82549
3514 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
3515 Formatting fixes. Instead of calling make_bit_field_ref with negative
3518 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
3520 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
3521 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
3522 _mm_maskz_reduce_ss): New.
3523 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
3524 __builtin_ia32_reducess_mask): Ditto..
3525 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
3526 * config/i386/sse.md (reduces<mode>): Renamed to ...
3527 (reduces<mode><mask_scalar_name>): ... this.
3528 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
3530 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
3531 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
3533 2017-10-16 David Malcolm <dmalcolm@redhat.com>
3535 * Makefile.in (OBJS): Add unique-ptr-tests.o.
3536 * selftest-run-tests.c (selftest::run_tests): Call
3537 selftest::unique_ptr_tests_cc_tests.
3538 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
3539 * unique-ptr-tests.cc: New file.
3541 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
3544 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
3546 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
3547 (make_hard_regno_born, make_hard_regno_dead): Update
3548 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
3550 2017-10-16 Jeff Law <law@redhat.com>
3552 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
3554 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
3556 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
3558 2017-10-16 Olivier Hainque <hainque@adacore.com>
3560 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
3561 with_cpu if we were configured for an e500v2 target cpu name.
3563 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
3565 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
3566 * doc/invoke.texi: Document +nodsp as a valid extension for
3569 2017-10-16 Martin Liska <mliska@suse.cz>
3571 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
3572 (test_set_range): Likewise.
3573 (test_range_functions): Rename to ...
3574 (test_bit_in_range): ... this.
3575 (sbitmap_c_tests): Add new test.
3577 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3579 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
3581 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
3582 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
3584 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3586 * config/aarch64/aarch64-builtins.c
3587 (aarch64_types_quadopu_lane_qualifiers): New.
3588 (TYPES_QUADOPU_LANE): New.
3589 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
3590 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
3591 (aarch64_<sur>dot_laneq<vsi2qi>): New.
3592 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
3593 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
3594 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
3595 (Vdottype, DOTPROD): New.
3596 (sur): Add SDOT and UDOT.
3598 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3600 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
3601 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
3602 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
3604 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
3605 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
3606 Enable TARGET_DOTPROD.
3607 (cortex-a75.cortex-a55): Likewise.
3608 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
3610 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3612 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
3613 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
3615 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
3617 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
3618 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
3619 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
3620 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
3621 * config/arm/types.md (neon_dot, neon_dot_q): New.
3622 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
3624 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3626 * config/arm/arm.h (TARGET_DOTPROD): New.
3627 * config/arm/arm.c (arm_arch_dotprod): New.
3628 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
3629 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
3630 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
3631 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
3632 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
3633 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
3634 * doc/invoke.texi (armv8.2-a): Document dotprod
3636 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
3638 * i386.c (ix86_vec_cost): New function.
3639 (ix86_rtx_costs): Handle vector operations better.
3640 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
3641 * x86-tune-costs.h: Add new costs to all tables.
3643 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
3645 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
3647 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
3648 divsd, sqrtss and sqrtsd
3649 * x86-tune-costs.h: Add new entries to all costs.
3650 (znver1_cost): Fix to match real instruction latencies.
3652 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3653 Michael Collison <michael.collison@arm.com>
3655 * compare-elim.c: Include emit-rtl.h.
3656 (can_merge_compare_into_arith): New function.
3657 (try_validate_parallel): Likewise.
3658 (try_merge_compare): Likewise.
3659 (try_eliminate_compare): Call the above when no previous clobber
3661 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
3664 2017-10-14 Jakub Jelinek <jakub@redhat.com>
3668 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
3669 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
3673 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
3674 to be any operand_equal_p operands. For & (B - 1) require
3675 B to be power of 2. Recognize
3676 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
3678 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3681 * optabs.c (expand_memory_blockage): Fix call of
3682 targetm.have_memory_blockage.
3684 2017-10-14 Jakub Jelinek <jakub@redhat.com>
3687 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
3688 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
3689 objects to extra_objs instead of overwriting it.
3691 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3693 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
3694 Use any_fp_register_operand as operand[3] predicate. Simplify
3695 equality test for operands[2] and operands[4] memory location.
3696 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
3697 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
3698 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
3699 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
3700 any_fp_register_operand as operand[1] predicate. Simplify
3701 equality test for operands[0] and operands[3] memory location.
3702 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
3703 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
3704 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
3706 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3708 * target-insns.def: Add memory_blockage.
3709 * optabs.c (expand_memory_blockage): New function.
3710 (expand_asm_memory_barrier): Rename ...
3711 (expand_asm_memory_blockage): ... to this.
3712 (expand_mem_thread_fence): Call expand_memory_blockage
3713 instead of expand_asm_memory_barrier.
3714 (expand_mem_singnal_fence): Ditto.
3715 (expand_atomic_load): Ditto.
3716 (expand_atomic_store): Ditto.
3717 * doc/md.texi (Standard Pattern Names For Generation):
3718 Document memory_blockage instruction pattern.
3720 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
3722 * config/rl78/rl78.c (rl78_emit_libcall): New function.
3723 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
3724 * config/rl78/rl78.md: New define_expand "adddi3".
3726 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
3728 * cfghooks.c (verify_flow_info): Disable check that all probabilities
3731 2017-10-13 Jeff Law <law@redhat.com>
3733 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
3735 2017-10-13 Jakub Jelinek <jakub@redhat.com>
3738 * internal-fn.c (expand_mul_overflow): If both operands have
3739 the same highpart of -1 or 0 and the topmost bit of lowpart
3740 is different, overflow is if res <= 0 rather than res < 0.
3742 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
3744 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
3745 TARGET_P9_VECTOR code for unaligned_load case.
3747 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
3749 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
3751 2017-10-13 Nathan Sidwell <nathan@acm.org>
3753 * tree-core.h (tree_contains_struct): Make bool.
3754 * tree.c (tree_contains_struct): Likewise.
3755 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
3756 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
3757 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
3758 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
3759 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
3761 2017-10-13 Richard Biener <rguenther@suse.de>
3763 * graphite-isl-ast-to-gimple.c
3764 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
3765 parameters and dominance check.
3766 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
3767 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
3768 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
3769 Do not update SSA form here or do intermediate IL verification.
3770 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
3771 (graphite_initialize): Remove check on the number of loops in
3772 the function and inline into graphite_transform_loops.
3773 (graphite_finalize): Inline into graphite_transform_loops.
3774 (graphite_transform_loops): Perform SSA update and IL verification
3776 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
3778 2017-10-13 Richard Biener <rguenther@suse.de>
3780 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
3781 graphite_expression_type_precision): Avoid global constructor
3783 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
3784 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
3785 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
3786 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
3787 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
3788 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
3790 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
3793 * config/i386/i386.h (ix86_red_zone_size): New.
3794 * config/i386/i386.md (push peephole2s): Replace
3795 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
3797 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3798 Alan Hayward <alan.hayward@arm.com>
3799 David Sherwood <david.sherwood@arm.com>
3801 * combine.c (can_change_dest_mode): Reject changes in
3802 REGMODE_NATURAL_SIZE.
3804 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3805 Alan Hayward <alan.hayward@arm.com>
3806 David Sherwood <david.sherwood@arm.com>
3808 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
3809 (expand_debug_source_expr): Likewise.
3810 * combine.c (combine_simplify_rtx): Likewise.
3811 * cse.c (fold_rtx): Likewise.
3812 * fwprop.c (canonicalize_address): Likewise.
3813 * targhooks.c (default_shift_truncation_mask): Likewise.
3815 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3816 Alan Hayward <alan.hayward@arm.com>
3817 David Sherwood <david.sherwood@arm.com>
3819 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
3820 (widened_mode): Likewise.
3821 (expand_unop): Likewise.
3822 * ree.c (transform_ifelse): Likewise.
3823 (merge_def_and_ext): Likewise.
3824 (combine_reaching_defs): Likewise.
3825 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3827 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3828 Alan Hayward <alan.hayward@arm.com>
3829 David Sherwood <david.sherwood@arm.com>
3831 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
3832 * combine.c (gen_lowpart_for_combine): Likewise.
3833 * dwarf2out.c (rtl_for_decl_location): Likewise.
3834 * final.c (alter_subreg): Likewise.
3835 * rtlhooks.c (gen_lowpart_general): Likewise.
3836 (gen_lowpart_if_possible): Likewise.
3838 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3839 Alan Hayward <alan.hayward@arm.com>
3840 David Sherwood <david.sherwood@arm.com>
3842 * calls.c (expand_call): Use subreg_lowpart_offset.
3843 * cse.c (cse_insn): Likewise.
3844 * regcprop.c (copy_value): Likewise.
3845 (copyprop_hardreg_forward_1): Likewise.
3847 2017-10-13 Jakub Jelinek <jakub@redhat.com>
3850 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
3851 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
3852 =Q constraints to +Q and into insn condition add check
3853 that operands[0] and operands[1] are equal.
3854 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
3855 =Q constraints to +Q and into insn condition add check
3856 that operands[0] is equal to either operands[1] or operands[2].
3859 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
3860 instead of handling MINUS_EXPR twice (once for each argument),
3861 canonicalize operand order and handle just once, use rtype where
3862 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
3865 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
3866 any values of __C while still being pattern recognizable as a simple
3869 2017-10-13 Richard Biener <rguenther@suse.de>
3871 PR tree-optimization/82451
3873 2017-10-02 Richard Biener <rguenther@suse.de>
3875 PR tree-optimization/82355
3876 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
3877 a mapping for the enclosing loop but avoid generating one for
3879 (copy_bb_and_scalar_dependences): Remove premature codegen
3880 error on PHIs in blocks duplicated into multiple places.
3881 * graphite-scop-detection.c
3882 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
3883 in the region use it as loop and nest to analyze the DR in.
3884 (try_generate_gimple_bb): Likewise.
3885 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
3886 (add_loop_constraints): For blocks in a loop not in the region
3887 create a dimension with a single iteration.
3888 * sese.h (gbb_loop_at_index): Remove assert.
3890 * cfgloop.c (loop_preheader_edge): For the loop tree root
3891 return the single successor of the entry block.
3892 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3893 Reset the SCEV hashtable and niters.
3894 * graphite-scop-detection.c
3895 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
3896 assert that we only have POLYNOMIAL_CHREC that vary in loops
3897 contained in the region.
3898 (scop_detection::graphite_can_represent_expr): Adjust.
3899 (scop_detection::stmt_has_simple_data_refs_p): For loops
3900 not in the region set loop to NULL. The nest is now the
3901 entry edge to the region.
3902 (try_generate_gimple_bb): Likewise.
3903 * sese.c (scalar_evolution_in_region): Adjust for
3904 instantiate_scev change.
3905 * tree-data-ref.h (graphite_find_data_references_in_stmt):
3906 Make nest parameter the edge into the region.
3907 (create_data_ref): Likewise.
3908 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
3909 entry edge into a region and adjust instantiate_scev calls.
3910 (create_data_ref): Likewise.
3911 (graphite_find_data_references_in_stmt): Likewise.
3912 (find_data_references_in_stmt): Pass the loop preheader edge
3913 from the nest argument.
3914 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
3915 parameter the edge into the region.
3916 (instantiate_parameters): Use the loop preheader edge as entry.
3917 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
3919 (get_instantiated_value_entry): Make instantiate_below parameter
3920 the edge into the region.
3921 (instantiate_scev_name): Likewise. Adjust dominance checks,
3922 when we cannot use loop-based instantiation instantiate by
3923 walking use-def chains.
3924 (instantiate_scev_poly): Adjust.
3925 (instantiate_scev_binary): Likewise.
3926 (instantiate_scev_convert): Likewise.
3927 (instantiate_scev_not): Likewise.
3928 (instantiate_array_ref): Remove.
3929 (instantiate_scev_3): Likewise.
3930 (instantiate_scev_2): Likewise.
3931 (instantiate_scev_1): Likewise.
3932 (instantiate_scev_r): Do not blindly handle N-operand trees.
3933 Do not instantiate array-refs. Handle all constants and invariants.
3934 (instantiate_scev): Make instantiate_below parameter
3935 the edge into the region.
3936 (resolve_mixers): Use the loop preheader edge for the region
3937 parameter to instantiate_scev_r.
3938 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
3940 2017-10-13 Richard Biener <rguenther@suse.de>
3942 PR tree-optimization/82525
3943 * graphite-isl-ast-to-gimple.c
3944 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
3946 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
3947 Fail code generation when we cannot represent the isl integer.
3948 (binary_op_to_tree): Elide modulo operations that are no-ops
3949 in the type we code generate. Remove now superfluous code
3952 2017-10-13 Richard Biener <rguenther@suse.de>
3954 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
3955 (scop_detection::harmful_loop_in_region): Remove premature
3956 IV type restriction.
3957 (scop_detection::graphite_can_represent_scev): We can handle
3958 pointer IVs just fine.
3960 2017-10-13 Alan Modra <amodra@gmail.com>
3962 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
3963 "Clobbers and Scratch Registers". Add paragraph on
3964 alternative to clobbers for scratch registers and OpenBLAS
3967 2017-10-13 Alan Modra <amodra@gmail.com>
3969 * doc/extend.texi (Clobbers): Correct vax example. Delete old
3970 example of a memory input for a string of known length. Move
3971 commentary out of table. Add a number of new examples
3972 covering array memory inputs.
3974 2017-10-12 Martin Liska <mliska@suse.cz>
3976 PR tree-optimization/82493
3977 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
3978 (test_range_functions): New function.
3979 (sbitmap_c_tests): Likewise.
3980 * selftest-run-tests.c (selftest::run_tests): Run new tests.
3981 * selftest.h (sbitmap_c_tests): New function.
3983 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
3985 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
3987 * config/rs6000/amo.h: Fix spacing issue.
3989 2017-10-12 Jakub Jelinek <jakub@redhat.com>
3992 * config/i386/i386.md (*ashl<mode>3_mask_1,
3993 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
3994 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
3997 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
3999 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
4000 (profile_probability): Set max_probability
4001 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
4003 * profile-count.c (profile_probability::differs_from_p): Do not
4004 rely on max_probaiblity == 10000
4006 2017-10-12 Jeff Law <law@redhat.com>
4008 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
4011 2017-10-12 Martin Sebor <msebor@redhat.com>
4015 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
4016 (handle_alias_pairs): Call it.
4017 * common.opt (-Wattribute-alias): New option.
4018 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
4019 * doc/invoke.texi (-Wattribute-alias): Document.
4021 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
4024 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4026 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4028 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4029 (make_hard_regno_born, make_hard_regno_dead): Update
4030 bb_killed_pseudos and bb_gen_pseudos.
4032 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4034 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
4036 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
4038 * config/alpha/alpha.c (alpha_split_conditional_move):
4039 Use std::swap instead of manually swapping.
4040 (alpha_stdarg_optimize_hook): Ditto.
4041 (alpha_canonicalize_comparison): Ditto.
4043 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4045 * tree-loop-distribution.c (struct builtin_info): New struct.
4046 (struct partition): Refactor fields into struct builtin_info.
4047 (partition_free): Free struct builtin_info.
4048 (build_size_arg_loc, build_addr_arg_loc): Delete.
4049 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
4050 information from struct builtin_info.
4051 (find_single_drs): New function refactored from classify_partition.
4052 Also moved builtin validity checks to this function.
4053 (compute_access_range, alloc_builtin): New functions.
4054 (classify_builtin_st, classify_builtin_ldst): New functions.
4055 (classify_partition): Refactor code into functions find_single_drs,
4056 classify_builtin_st and classify_builtin_ldst.
4057 (distribute_loop): Don't do runtime alias check when distributing
4059 (find_seed_stmts_for_distribution): New function.
4060 (pass_loop_distribution::execute): Refactor code finding seed
4061 stmts into above function. Support distribution for the innermost
4062 two-level loop nest. Adjust dump information.
4064 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4066 * tree-loop-distribution.c: Adjust the general comment.
4067 (NUM_PARTITION_THRESHOLD): New macro.
4068 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
4069 (classify_partition): Skip builtin pattern of loop nest's inner loop.
4070 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
4071 in call to build_partition_graph.
4072 (finalize_partitions): New parameter. Make loop distribution more
4073 conservative by fusing more partitions.
4074 (distribute_loop): Don't do runtime alias check in case of loop nest
4076 (find_seed_stmts_for_distribution): New function.
4077 (prepare_perfect_loop_nest): New function.
4078 (pass_loop_distribution::execute): Refactor code finding seed stmts
4079 and loop nest into above functions. Support loop nest distribution.
4080 Adjust dump information accordingly.
4082 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4084 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
4085 and set PTYPE_SEQUENTIAL for merged partition.
4087 2017-10-12 Richard Biener <rguenther@suse.de>
4089 PR tree-optimization/69728
4091 2017-09-19 Richard Biener <rguenther@suse.de>
4093 PR tree-optimization/69728
4094 * graphite-sese-to-poly.c (schedule_error): New global.
4095 (add_loop_schedule): Handle empty domain by failing the
4097 (build_original_schedule): Handle schedule_error.
4099 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
4100 domain by returning an unchanged schedule.
4102 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4104 * genrecog.c (validate_pattern): For VEC_SELECT verify that
4105 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
4107 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
4109 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
4112 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4115 * expr.c (store_field): Don't optimize away bitsize == 0 store
4116 from CALL_EXPR with addressable return type.
4118 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4120 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
4121 * config/rs6000/rs6000.md (sel): Delete mode attribute.
4122 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
4123 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
4124 TARGET_ISEL instead of TARGET_ISEL<sel>.
4126 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
4128 * config/rs6000/rs6000.c
4129 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
4131 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4133 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
4135 (reg_or_cint_operand): Fix comment.
4136 (reg_or_zero_operand): New predicate.
4137 * config/rs6000/rs6000-protos.h (output_isel): Delete.
4138 * config/rs6000/rs6000.c (output_isel): Delete.
4139 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
4140 instead of reg_or_cint_operand. Output instruction directly (not via
4142 (isel_unsigned_<mode>): Ditto.
4143 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
4144 gpc_reg_operand. Add an instruction alternative for this. Output
4145 instruction directly.
4146 (*isel_reversed_unsigned_<mode>): Ditto.
4148 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4150 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
4151 (TARGET_CANONICALIZE_COMPARISON): Define.
4153 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
4156 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
4157 Check whether the dest is REG before adding REG_EQUIV note.
4159 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4162 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4164 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4165 (make_hard_regno_born, make_hard_regno_dead): Update
4166 bb_killed_pseudos and bb_gen_pseudos.
4168 2017-10-11 Nathan Sidwell <nathan@acm.org>
4170 * incpath.h (enum incpath_kind): Name enum, prefix values.
4171 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
4172 * incpath.c (heads, tails): Use INC_MAX.
4173 (add_env_var_paths, add_standard_paths): Use incpath_kind.
4174 (merge_include_chains, split_quote_chain,
4175 register_include_chains): Update incpath_kind names.
4176 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
4177 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
4179 (add_framework_path, darwin_register_objc_includes): Likewise.
4180 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
4182 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4184 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
4185 Do not use float_operator operator predicate.
4186 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
4187 * config/i386/predicates.md (float_operator): Remove predicate.
4189 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4191 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
4192 (*jccxf_i387): Ditto.
4193 (*jcc<mode>_i387): Ditto.
4194 (*jccu<mode>_i387): Ditto.
4195 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
4196 (*jcc_*_i387 splitters): Remove.
4197 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
4198 * config/i386/i386.c (ix86_split_fp_branch): Remove.
4199 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
4202 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4204 * profile-count.h (slow_safe_scale_64bit): New function.
4205 (safe_scale_64bit): New inline.
4206 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
4207 * profile-count.c: Include wide-int.h
4208 (slow_safe_scale_64bit): New.
4210 2017-10-11 Nathan Sidwell <nathan@acm.org>
4212 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
4213 HAS_DECL_ASSEMBLER_NAME_P.
4214 * gimple-expr.c (gimple_decl_printable_name: Check
4215 HAS_DECL_ASSEMBLER_NAME_P too.
4216 * ipa-utils.h (type_in_anonymous_namespace_p): Check
4217 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
4218 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
4219 * passes.c (rest_of_decl_compilation): Check
4220 HAS_DECL_ASSEMBLER_NAME_P too.
4221 * recog.c (verify_changes): Likewise.
4222 * tree-pretty-print.c (dump_decl_name): Likewise.
4223 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
4225 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
4226 (SET_DECL_ASSEMBLER_NAME): Use it.
4227 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
4228 (COPY_DECL_ASSEMBLER_NAME): Likewise.
4229 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
4231 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4233 * config.gcc (i386, x86_64): Add extra objects.
4234 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
4235 (ix86_min_insn_size): Declare.
4236 (ix86_issue_rate): Declare.
4237 (ix86_adjust_cost): Declare.
4238 (ia32_multipass_dfa_lookahead): Declare.
4239 (ix86_macro_fusion_p): Declare.
4240 (ix86_macro_fusion_pair_p): Declare.
4241 (ix86_bd_has_dispatch): Declare.
4242 (ix86_bd_do_dispatch): Declare.
4243 (ix86_core2i7_init_hooks): Declare.
4244 (ix86_atom_sched_reorder): Declare.
4245 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
4246 (COSTS_N_BYTES): Move to x86-tune-costs.h.
4247 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
4248 (rip_relative_addr_p): Rename to ...
4249 (ix86_rip_relative_addr_p): ... this one; export.
4250 (memory_address_length): Update.
4251 (ix86_issue_rate): Move to x86-tune-sched.c.
4252 (ix86_flags_dependent): Move to x86-tune-sched.c.
4253 (ix86_agi_dependent): Move to x86-tune-sched.c.
4254 (exact_dependency_1): Move to x86-tune-sched.c.
4255 (exact_store_load_dependency): Move to x86-tune-sched.c.
4256 (ix86_adjust_cost): Move to x86-tune-sched.c.
4257 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
4258 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
4259 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
4260 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
4261 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
4262 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
4263 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
4264 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
4265 (min_insn_size): Rename to ...
4266 (ix86_min_insn_size): ... this one; export.
4267 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
4268 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
4269 (core2i7_first_cycle_multipass_backtrack): Move to
4270 x86-tune-sched-core.c.
4271 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
4272 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
4273 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
4274 (ix86_avoid_jump_mispredicts): Update.
4275 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
4276 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
4277 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
4278 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
4279 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
4280 (MAX_INSN): Move to ix86-tune-sched-bd.c.
4281 (MAX_IMM): Move to ix86-tune-sched-bd.c.
4282 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
4283 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
4284 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
4285 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
4286 (MAX_STORE): Move to ix86-tune-sched-bd.c.
4287 (BIG): Move to ix86-tune-sched-bd.c.
4288 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
4289 (enum insn_path): Move to ix86-tune-sched-bd.c.
4290 (get_mem_group): Move to ix86-tune-sched-bd.c.
4291 (is_cmp): Move to ix86-tune-sched-bd.c.
4292 (dispatch_violation): Move to ix86-tune-sched-bd.c.
4293 (is_branch): Move to ix86-tune-sched-bd.c.
4294 (is_prefetch): Move to ix86-tune-sched-bd.c.
4295 (init_window): Move to ix86-tune-sched-bd.c.
4296 (allocate_window): Move to ix86-tune-sched-bd.c.
4297 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
4298 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
4299 (process_end_window): Move to ix86-tune-sched-bd.c.
4300 (allocate_next_window): Move to ix86-tune-sched-bd.c.
4301 (find_constant): Move to ix86-tune-sched-bd.c.
4302 (get_num_immediates): Move to ix86-tune-sched-bd.c.
4303 (has_immediate): Move to ix86-tune-sched-bd.c.
4304 (get_insn_path): Move to ix86-tune-sched-bd.c.
4305 (get_insn_group): Move to ix86-tune-sched-bd.c.
4306 (count_num_restricted): Move to ix86-tune-sched-bd.c.
4307 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
4308 (add_insn_window): Move to ix86-tune-sched-bd.c.
4309 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
4310 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
4311 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
4312 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
4313 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
4314 (do_dispatch): Move to ix86-tune-sched-bd.c.
4315 (has_dispatch): Move to ix86-tune-sched-bd.c.
4316 * config/i386/t-i386: Add new object files.
4317 * config/i386/x86-tune-costs.h: New file.
4318 * config/i386/x86-tune-sched-atom.c: New file.
4319 * config/i386/x86-tune-sched-bd.c: New file.
4320 * config/i386/x86-tune-sched-core.c: New file.
4321 * config/i386/x86-tune-sched.c: New file.
4323 2017-10-11 Liu Hao <lh_mouse@126.com>
4325 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
4326 the generic version below instead.
4327 (should_colorize): Recognize Windows consoles as terminals
4329 * pretty-print.c [__MINGW32__] (write_all): New function.
4330 [__MINGW32__] (find_esc_head): Likewise.
4331 [__MINGW32__] (find_esc_terminator): Likewise.
4332 [__MINGW32__] (eat_esc_sequence): Likewise.
4333 [__MINGW32__] (mingw_ansi_fputs): New function that handles
4335 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
4338 2017-10-11 Richard Biener <rguenther@suse.de>
4340 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
4341 Properly call analyze_scalar_evolution with the loop of the stmt.
4343 2017-10-11 Richard Biener <rguenther@suse.de>
4345 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
4346 * tree-core.h (tree_base): Add chrec_var union member.
4347 * tree.h (CHREC_VAR): Remove.
4348 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
4349 * tree-chrec.h (build_polynomial_chrec): Adjust.
4350 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
4351 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
4353 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
4355 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
4356 * match.pd: ... here.
4357 ((T) X == (T) Y): Relax condition.
4359 2017-10-11 Bin Cheng <bin.cheng@arm.com>
4361 PR tree-optimization/82472
4362 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
4364 (break_alias_scc_partitions): Update postorder number.
4366 2017-10-11 Martin Liska <mliska@suse.cz>
4369 * opts.c (parse_no_sanitize_attribute): Do not use error_value
4371 * opts.h (parse_no_sanitize_attribute): Remove last argument.
4373 2017-10-11 Martin Liska <mliska@suse.cz>
4375 * print-rtl.c (print_insn): Move declaration of idbuf
4376 to same scope as name.
4378 2017-10-11 Martin Liska <mliska@suse.cz>
4383 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
4386 2017-10-11 Martin Liska <mliska@suse.cz>
4389 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
4392 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
4394 * config.gcc (default_gnu_indirect_function): Default to yes for
4395 arm*-*-linux* with glibc.
4397 2017-10-11 Richard Biener <rguenther@suse.de>
4399 * tree-scalar-evolution.c (get_scalar_evolution): Handle
4400 default-defs and types we do not want to analyze.
4401 (interpret_loop_phi): Replace unreachable code with an assert.
4402 (compute_scalar_evolution_in_loop): Remove and inline ...
4403 (analyze_scalar_evolution_1): ... here, replacing condition with
4404 what makes the intent clearer. Remove handling of cases
4405 get_scalar_evolution now handles.
4407 2017-10-10 Jim Wilson <wilson@tuliptree.org>
4409 PR rtl-optimization/81434
4410 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
4411 comment for main loop. In sched_group_found if, also add checks for
4412 pass and min_cost_group.
4414 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
4416 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
4417 (rs6000_insn_cost): New function.
4418 * config/rs6000/rs6000.md (cost): New attribute.
4420 2017-10-10 Jakub Jelinek <jakub@redhat.com>
4421 H.J. Lu <hongjiu.lu@intel.com>
4425 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
4426 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
4427 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
4428 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
4429 to OPTION_MASK_ISA_AVX512VL - builtins that have both
4430 OPTION_MASK_ISA_MMX and some other bit set require both
4431 mmx and the ISAs without the mmx bit.
4432 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
4433 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
4434 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
4435 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
4436 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
4437 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
4438 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
4439 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
4440 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
4441 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
4442 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
4443 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
4444 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
4445 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
4446 Add OPTION_MASK_ISA_MMX.
4448 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
4450 * config.gcc (armv7*-*-freebsd*): New target.
4451 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
4453 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
4455 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
4456 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
4459 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
4461 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
4462 a template parameter.
4463 (WIDE_INT_REF_FOR): Update accordingly.
4464 * tree.h (wi::int_traits <const_tree>): Delete.
4465 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
4466 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
4467 (wi::tree_to_wide_ref): New typedef.
4468 (wi::to_wide): New function.
4469 * calls.c (get_size_range): Use wi::to_wide when operating on
4471 * cgraph.c (cgraph_node::create_thunk): Likewise.
4472 * config/i386/i386.c (ix86_data_alignment): Likewise.
4473 (ix86_local_alignment): Likewise.
4474 * dbxout.c (stabstr_O): Likewise.
4475 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
4476 * expr.c (const_vector_from_tree): Likewise.
4477 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
4478 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
4479 (fold_negate_expr_1, int_const_binop_1, const_binop)
4480 (fold_convert_const_int_from_real, optimize_bit_field_compare)
4481 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
4482 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
4483 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
4484 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
4485 (fold_not_const, round_up_loc): Likewise.
4486 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
4487 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
4488 (alloca_call_type): Likewise.
4489 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
4490 * godump.c (go_output_typedef): Likewise.
4491 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
4492 * internal-fn.c (get_min_precision): Likewise.
4493 * ipa-cp.c (ipcp_store_vr_results): Likewise.
4494 * ipa-polymorphic-call.c
4495 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
4496 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
4497 (ipa_modify_call_arguments): Likewise.
4498 * match.pd: Likewise.
4499 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
4500 * print-tree.c (print_node_brief, print_node): Likewise.
4501 * stmt.c (expand_case): Likewise.
4502 * stor-layout.c (layout_type): Likewise.
4503 * tree-affine.c (tree_to_aff_combination): Likewise.
4504 * tree-cfg.c (group_case_labels_stmt): Likewise.
4505 * tree-data-ref.c (dr_analyze_indices): Likewise.
4506 (prune_runtime_alias_test_list): Likewise.
4507 * tree-dump.c (dequeue_and_dump): Likewise.
4508 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
4509 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
4510 * tree-pretty-print.c (dump_generic_node): Likewise.
4511 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
4512 (simple_iv_with_niters): Likewise.
4513 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
4514 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
4515 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
4516 * tree-ssa-loop-niter.c (split_to_var_and_offset)
4517 (refine_value_range_using_guard, number_of_iterations_ne_max)
4518 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
4519 (get_cst_init_from_scev, record_nonwrapping_iv)
4520 (scev_var_range_cant_overflow): Likewise.
4521 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
4522 * tree-ssa-pre.c (compute_avail): Likewise.
4523 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
4524 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
4525 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
4526 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
4527 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
4528 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
4529 (dump_case_nodes, try_switch_expansion): Likewise.
4530 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
4531 (vect_do_peeling): Likewise.
4532 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
4533 * tree-vect-stmts.c (vectorizable_load): Likewise.
4534 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
4535 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
4536 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
4537 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
4538 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
4539 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
4540 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
4541 (evrp_dom_walker::before_dom_children): Likewise.
4542 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
4543 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
4544 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
4545 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
4546 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
4547 (get_range_pos_neg): Likewise.
4548 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
4549 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
4550 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
4551 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
4552 * config/avr/avr.c (avr_fold_builtin): Likewise.
4553 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
4554 * config/msp430/msp430.c (msp430_attr): Likewise.
4555 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
4556 * config/powerpcspe/powerpcspe-c.c
4557 (altivec_resolve_overloaded_builtin): Likewise.
4558 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
4559 (rs6000_expand_ternop_builtin): Likewise.
4560 * config/rs6000/rs6000-c.c
4561 (altivec_resolve_overloaded_builtin): Likewise.
4562 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
4563 (rs6000_expand_ternop_builtin): Likewise.
4564 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
4566 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4568 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
4569 when copying loop nest with only one inner loop.
4571 2017-10-10 Richard Biener <rguenther@suse.de>
4573 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
4574 blocks if SCEV is active.
4575 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
4577 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
4578 (scev_initialize): Assert we are not yet initialized.
4580 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4582 * tree-loop-distribution.c (generate_loops_for_partition): Remove
4583 inner loop's exit stmt by making it always exit the loop, otherwise
4584 we would generate an infinite empty loop.
4586 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4588 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
4589 renaming variables in new preheader if it's deleted.
4591 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4593 * tree-loop-distribution.c (struct partition): Remove unused field
4594 loops of the structure.
4595 (partition_alloc, partition_free): Ditto.
4596 (build_rdg_partition_for_vertex): Ditto.
4598 2017-10-09 Jeff Law <law@redhat.com>
4600 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
4601 return type to match prototype and documentation.
4603 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4605 * config/rs6000/rs6000.c (processor_costs): Move to ...
4606 * config/rs6000/rs6000.h: ... here.
4607 (rs6000_cost): Declare.
4609 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
4611 * except.c (setjmp_fn): New global variable.
4612 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
4613 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
4614 if DONT_USE_BUILTIN_SETJMP is defined.
4616 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4618 * target.def (insn_cost): New hook.
4619 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
4620 * doc/tm.texi: Regenerate.
4621 * rtlanal.c (insn_cost): Use the new hook.
4623 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4625 * combine.c (combine_validate_cost): Compute the new insn_cost,
4626 not just pattern_cost.
4627 (try_combine): Adjust comment.
4629 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4631 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
4633 * combine.c (uid_insn_cost): Adjust comment.
4634 (combine_validate_cost): Adjust comment. Use pattern_cost instead
4636 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
4637 * dse.c (find_shift_sequence): Ditto.
4638 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
4639 (bb_valid_for_noce_process_p): Use pattern_cost.
4640 * rtl.h (insn_rtx_cost): Delete.
4641 (pattern_cost): New prototype.
4642 (insn_cost): New prototype.
4643 * rtlanal.c (insn_rtx_cost): Rename to...
4644 (pattern_cost): ... this.
4647 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
4649 * config/i386/i386.md (*jcc_2): Remove insn pattern.
4650 (*jcc<mode>_0_r_i387): Ditto.
4651 (*jccxf_r_i387): Ditto.
4652 (*jcc<mode>_r_i387): Ditto.
4653 (*jccu<mode>_r_i387): Ditto.
4654 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
4655 (*jcc): Rename from *jcc_1.
4657 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4659 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
4660 deferred rescans after the lvx/stvx recombination pre-pass.
4662 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
4664 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
4665 memory operation instruction support.
4666 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
4667 (rs6000-ibm-aix[789]*): Likewise.
4668 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
4669 Document new functions.
4671 2017-10-09 Richard Biener <rguenther@suse.de>
4673 PR tree-optimization/82397
4674 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
4675 equality only for semantically equal trees.
4677 2017-10-09 Richard Biener <rguenther@suse.de>
4679 PR tree-optimization/82449
4680 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
4681 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
4682 allow constant addresses.
4683 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
4686 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4688 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
4691 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4694 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
4697 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4700 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
4702 2017-10-09 Jakub Jelinek <jakub@redhat.com>
4705 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
4706 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
4708 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
4710 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
4711 (WI_BINARY_PREDICATE_RESULT): Likewise.
4712 (wi::binary_traits::operator_result): New type.
4713 (wi::binary_traits::predicate_result): Likewise.
4714 (generic_wide_int::operator~, unary generic_wide_int::operator-)
4715 (generic_wide_int::operator==, generic_wide_int::operator!=)
4716 (generic_wide_int::operator&, generic_wide_int::and_not)
4717 (generic_wide_int::operator|, generic_wide_int::or_not)
4718 (generic_wide_int::operator^, generic_wide_int::operator+
4719 (binary generic_wide_int::operator-, generic_wide_int::operator*):
4721 (operator~, unary operator-, operator==, operator!=, operator&)
4722 (operator|, operator^, operator+, binary operator-, operator*): New
4724 * expr.c (get_inner_reference): Use wi::bit_and_not.
4725 * fold-const.c (fold_binary_loc): Likewise.
4726 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
4727 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
4728 (bit_value_binop): Likewise.
4729 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
4730 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
4731 (extract_range_from_binary_expr_1): Likewise.
4732 (masked_increment): Likewise.
4733 (simplify_bit_ops_using_ranges): Likewise.
4735 2017-10-09 Martin Jambor <mjambor@suse.cz>
4738 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
4739 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
4740 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
4742 (hsa_fixup_mov_insn_type): New function.
4743 (hsa_op_with_type::get_in_type): Use it.
4744 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
4745 immediates in an assert.
4746 (hsa_op_with_type::extend_int_to_32bit): New method.
4747 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
4748 types. Convert to dest type if necessary.
4749 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
4750 (reg_for_gimple_ssa): Pass false as min32int to
4751 hsa_type_for_scalar_tree_type.
4752 (gen_hsa_addr): Fixup type when creating addresable temporary.
4753 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
4754 (gen_hsa_unary_operation): Extend operands and convert to dest type if
4755 necessary. Call hsa_fixup_mov_insn_type.
4756 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
4757 extend operands and convert to dest type if necessary.
4758 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
4759 to dest type if necessary.
4760 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
4761 if conversion nt necessary and size matches.
4762 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
4763 to dest type if necessary.
4764 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
4765 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
4767 (gen_hsa_clrsb): Likewise.
4768 (gen_hsa_ffs): Likewise.
4769 (gen_hsa_divmod): Extend operands and convert to dest type if
4771 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
4773 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
4775 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
4776 Remove empty default arguments. Use a brace block as output
4778 (conditional return): Ditto.
4780 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
4781 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
4783 (group_ending_nop): Ditto.
4784 (doloop_end): Ditto.
4785 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
4786 (splitters for those): Ditto.
4788 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
4790 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
4791 a conditional jump (and the compare for it) so that pc_rtx is the
4793 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
4794 for the deleted and renamed ctr<mode>_internal[234] patterns.
4795 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
4796 Delete second conditional return pattern.
4797 (ctr<mode>_internal2): Delete this second bdnz pattern.
4798 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
4799 (ctr<mode>_internal4): Delete this second bdz pattern.
4801 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
4803 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
4804 (always_initialized_rtx_for_ssa_name_p): New predicate.
4805 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
4806 (finish_out_of_ssa): Free new field of SA.
4807 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
4808 * tree-ssa-coalesce.c: Include tree-ssa.h.
4809 (get_parm_default_def_partitions): Remove extern keyword.
4810 (get_undefined_value_partitions): New function.
4811 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
4812 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
4814 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
4816 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
4818 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
4820 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
4822 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
4823 for targets that preffer 128bit.
4825 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
4827 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
4829 2017-10-08 Olivier Hainque <hainque@adacore.com>
4831 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
4832 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
4834 (thumb_set_return_address): Likewise.
4836 2017-10-08 Olivier Hainque <hainque@adacore.com>
4838 * common/config/arm/arm-common.c (arm_except_unwind_info):
4839 Handle DWARF2_UNWIND_INFO.
4841 2017-10-07 Michael Collison <michael.collison@arm.com>
4843 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
4846 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
4848 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
4849 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
4850 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
4851 defined, force the creation of a new block for a dispatch label.
4853 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
4855 * invoke.texi (Wsuggest-attribute=cold): Document.
4856 * common.opt (Wsuggest-attribute=cold): New
4857 * ipa-pure-const.c (warn_function_cold): New function.
4858 * predict.c (compute_function_frequency): Use it.
4859 * predict.h (warn_function_cold): Declare.
4861 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
4863 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
4866 2017-10-06 Martin Liska <mliska@suse.cz>
4868 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
4869 keyword for member functions.
4870 (struct sanopt_tree_couple): New struct.
4871 (struct sanopt_tree_couple_hash): New function.
4872 (struct sanopt_ctx): Add new hash_map.
4873 (has_dominating_ubsan_ptr_check): New function.
4874 (record_ubsan_ptr_check_stmt): Likewise.
4875 (maybe_optimize_ubsan_ptr_ifn): Likewise.
4876 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
4877 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
4879 2017-10-06 Sudakshina Das <sudi.das@arm.com>
4882 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
4883 aarch64_simd_valid_immediate on CONST_VECTORs.
4884 (aarch64_reg_or_bic_imm): Likewise.
4886 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
4888 PR rtl-optimization/82396
4889 * haifa-sched.c (ready_sort_real): Disable qsort checking.
4891 2017-10-06 Sebastian Pop <sebpop@gmail.com>
4893 * graphite-dependences.c (scop_get_reads): Move code to...
4894 (scop_get_must_writes): Move code to...
4895 (scop_get_may_writes): Move code to...
4896 (scop_get_reads_and_writes): ... here.
4897 (scop_get_dependences): Call scop_get_reads_and_writes.
4899 2017-10-06 Jakub Jelinek <jakub@redhat.com>
4901 PR tree-optimization/82434
4902 * fold-const.h (can_native_encode_type_p,
4903 can_native_encode_string_p): Remove.
4904 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
4905 don't encode anything, just return what would be otherwise returned.
4906 (native_encode_fixed, native_encode_complex, native_encode_vector):
4908 (native_encode_string): Likewise. Inline by hand
4909 can_native_encode_string_p.
4910 (can_native_encode_type_p): Remove.
4911 (can_native_encode_string_p): Remove.
4912 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
4913 STRING_CSTs using can_native_encode_string_p, test all
4914 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
4915 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
4916 argument from native_encode_expr.
4917 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
4918 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
4921 2017-10-06 Richard Biener <rguenther@suse.de>
4923 PR tree-optimization/82397
4924 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
4925 operand_equal_p but rely on data_ref_compare_tree for detecting
4927 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
4928 to match up with dr_group_sort_cmp.
4930 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4933 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
4935 * config/s390/s390-builtin-types.def: Regenerate.
4937 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4940 * config/s390/s390-builtin-types.def: Regenerate.
4941 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
4942 Change flag from B_VXE to B_VX.
4943 (s390_vec_min_dbl): Remove B_VXE flag.
4945 2017-10-06 Richard Biener <rguenther@suse.de>
4947 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
4948 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
4949 translate_isl_ast_to_gimple::is_valid_rename,
4950 translate_isl_ast_to_gimple::get_rename,
4951 translate_isl_ast_to_gimple::get_def_bb_for_const,
4952 translate_isl_ast_to_gimple::get_new_name,
4953 translate_isl_ast_to_gimple::collect_all_ssa_names,
4954 translate_isl_ast_to_gimple::copy_loop_phi_args,
4955 translate_isl_ast_to_gimple::collect_all_ssa_names,
4956 translate_isl_ast_to_gimple::copy_loop_phi_args,
4957 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
4958 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
4959 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
4960 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
4961 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
4962 translate_isl_ast_to_gimple::copy_cond_phi_args,
4963 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
4964 translate_isl_ast_to_gimple::edge_for_new_close_phis,
4965 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
4966 translate_isl_ast_to_gimple::rename_uses,
4967 translate_isl_ast_to_gimple::rename_all_uses): Remove.
4968 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
4969 (set_rename_for_each_def): Likewise.
4970 (graphite_copy_stmts_from_block): Handle debug stmt resetting
4971 here. Handle rewriting SCEV analyzable uses here.
4972 (copy_bb_and_scalar_dependences): Generate code for PHI
4974 (graphite_regenerate_ast_isl): Adjust.
4975 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
4976 (add_write, add_read): New functions.
4977 (build_cross_bb_scalars_def): Use it and simplify.
4978 (build_cross_bb_scalars_use): Likewise.
4979 (graphite_find_cross_bb_scalar_vars): Inline into...
4980 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
4981 simulating out-of-SSA. Compute liveout and add dependencies.
4982 (build_scops): Force an empty entry block.
4983 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
4985 (sese_build_liveouts): Declare.
4986 (sese_trivially_empty_bb_p): Likewise.
4987 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
4988 compute liveout and debug_liveout.
4989 (sese_bad_liveouts_use): Remove.
4990 (sese_reset_debug_liveouts_bb): Likewise.
4991 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
4992 (sese_build_liveouts): Build liveout and debug_liveout and store
4994 (new_sese_info): Adjust.
4995 (free_sese_info): Likewise.
4996 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
4997 do not build liveout here.
4998 (move_sese_in_condition): Adjust region entry.
4999 (scev_analyzable_p): Match up with chrec_apply requirements.
5000 (sese_trivially_empty_bb_p): New.
5001 * tree-into-ssa.c (get_reaching_def): Properly support generating
5002 default-defs for incremental rewrite of anonymous names.
5004 2017-10-06 Richard Biener <rguenther@suse.de>
5006 * graphite-sese-to-poly.c (extract_affine): For casts increasing
5007 precision do not perform modulo reduction.
5009 2017-10-06 Richard Biener <rguenther@suse.de>
5011 PR tree-optimization/82436
5012 * tree-vect-slp.c (vect_supported_load_permutation_p): More
5013 conservatively choose the vectorization factor when checking
5014 whether we can perform the required load permutation.
5015 (vect_transform_slp_perm_load): Assert when we may not fail.
5017 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
5019 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
5020 message for incompatible -msdata=* and -mcall-* options.
5022 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5024 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
5025 rate for post-reload scheduling.
5027 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5029 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
5031 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5033 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
5034 to improve monte carlo in scimark.
5036 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5038 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
5039 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
5040 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
5041 pentium4_cost, nocona_cost): Set reassociation width to 1.
5042 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
5043 width to 2 for fp operations and 1 otherwise.
5044 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
5046 (atom_cost): Set reassociation width to 2.
5047 (slm_cost, generic_cost): Set fp reassociation width
5048 to 2 and 1 otherwise.
5049 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
5050 (core_cost): Set fp reassociation width to 4 and vector to 2.
5051 (ix86_reassociation_width): Rewrite using cost table; special case
5052 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
5053 and TARGET_AVX128_OPTIMAL.
5054 * config/i386/i386.h (processor_costs): Add
5055 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
5056 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
5057 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
5058 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
5059 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
5060 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
5062 2017-10-05 Nathan Sidwell <nathan@acm.org>
5064 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
5066 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5068 * config/arm/arm.c (arm_test_fpu_data): New.
5069 (arm_run_selftests): Call arm_test_fpu_data.
5071 2017-10-04 Nathan Sidwell <nathan@acm.org>
5073 * toplev.c (toplev::main): Remove excess parens on pretty_printer
5075 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
5077 2017-10-04 Sudakshina Das <sudi.das@arm.com>
5079 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
5080 check type for aarch64_simd_valid_immediate.
5081 (aarch64_output_simd_mov_immediate): Update prototype.
5082 (aarch64_simd_valid_immediate): Update prototype.
5083 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
5084 support for ORR-immediate.
5085 (and<mode>3): modified pattern to add support for BIC-immediate.
5086 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
5087 now checks for valid immediate for BIC and ORR based on new enum
5089 (aarch64_output_simd_mov_immediate): Function now used to output
5090 BIC/ORR imm as well based on new enum argument.
5091 * config/aarch64/constraints.md (Do): New vector immediate constraint.
5093 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
5094 (aarch64_reg_or_bic_imm): Likewise.
5096 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5098 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
5099 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
5101 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5105 PR rtl-optimization/82396
5106 * haifa-sched.c (autopref_multipass_init): Simplify
5108 (autopref_rank_data): Simplify sort order.
5109 * sched-int.h (autopref_multipass_data_): Remove
5110 multi_mem_insn_p, min_offset and max_offset.
5112 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5114 * doc/sourcebuild.texi: Document vect_peeling_profitable.
5116 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5118 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
5121 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5123 * doc/sourcebuild.texi: Document vect_long_mult.
5125 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
5127 PR tree-optimization/82413
5128 * fold-const.c (build_range_check): Use widest_int when comparing
5129 the maximum ETYPE value with HIGH.
5131 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5133 PR rtl-optimization/82396
5134 * haifa-sched.c (autopref_multipass_init): Simplify
5136 (autopref_rank_data): Simplify sort order.
5137 * sched-int.h (autopref_multipass_data_): Remove
5138 multi_mem_insn_p, min_offset and max_offset.
5140 2017-10-04 Jakub Jelinek <jakub@redhat.com>
5142 PR tree-optimization/82381
5143 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
5144 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
5147 PR tree-optimization/82374
5148 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
5149 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
5150 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
5151 current_function_decl to the new decl.
5153 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
5155 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
5156 helper macro for IEEE float128 hardware built-in functions.
5157 (SQRTF128_ODD): Add built-in functions with the round-to-odd
5159 (TRUNCF128_ODD): Likewise.
5160 (ADDF128_ODD): Likewise.
5161 (SUBF128_ODD): Likewise.
5162 (MULF128_ODD): Likewise.
5163 (DIVF128_ODD): Likewise.
5164 (FMAF128_ODD): Likewise.
5165 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
5166 UNSPEC_TRUNC_ROUND_TO_ODD.
5167 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
5168 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
5169 floating point round to odd instructions.
5170 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
5171 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
5172 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
5173 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
5174 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
5175 (trunc<mode>sf2_hw): Change the truncate with round to odd
5176 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
5177 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
5178 to odd hardware instructions.
5179 (sub<mode>3_odd): Likewise.
5180 (mul<mode>3_odd): Likewise.
5181 (div<mode>3_odd): Likewise.
5182 (sqrt<mode>2_odd): Likewise.
5183 (fma<mode>4_odd): Likewise.
5184 (fms<mode>4_odd): Likewise.
5185 (nfma<mode>4_odd): Likewise.
5186 (nfms<mode>4_odd): Likewise.
5187 (trunc<mode>df2_odd): Change the truncate with round to odd
5188 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
5190 * doc/extend.texi (PowerPC built-in functions): Update documentation
5191 for existing IEEE float128-bit built-in functions. Add built-in
5192 functions that generate the IEEE 128-bit floating point round to
5195 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
5197 PR rtl-optimization/77729
5198 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
5199 to (X&(C1&~C2))|C2 transformations.
5201 2017-10-03 Martin Jambor <mjambor@suse.cz>
5203 PR tree-optimization/82363
5204 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
5205 mismatch, mark lacc written regardless of racc.
5207 2017-10-03 Jakub Jelinek <jakub@redhat.com>
5209 PR tree-optimization/82381
5210 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
5211 stmt_to_insert nor wheather SSA_NAMEs are default defs.
5212 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
5213 fallthrough into reassoc_stmt_dominates_stmt_p.
5216 * combine.c (combine_instructions): Don't combine in unreachable
5219 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
5222 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
5223 function to not use the have_cpu variable. Do not set cpu_index,
5224 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
5226 (rs6000_valid_attribute_p): Remove duplicate initializations of
5227 old_optimize and func_optimize.
5228 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
5229 (rs6000_activate_target_options): Make global.
5230 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
5233 2017-10-02 Jakub Jelinek <jakub@redhat.com>
5235 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
5236 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
5237 Set *poffset to 0 and *psize and *pmax_size to -1 if
5238 *poffset + *psize overflows in HOST_WIDE_INT.
5240 PR tree-optimization/82387
5241 PR tree-optimization/82388
5242 PR tree-optimization/82389
5243 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
5244 instead of live_bytes non-NULL.
5246 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
5249 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
5252 2017-10-02 Richard Biener <rguenther@suse.de>
5254 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
5255 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
5256 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
5258 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
5260 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
5261 requested precision matches the type's.
5262 * calls.c (alloc_max_size): Calculate the new candidate size as
5263 a widest_int and use wi::to_widest when comparing it with the
5264 current candidate size.
5265 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
5266 zero rather than integer_zero_node.
5267 * match.pd: Check for a no-op conversion before using wi::add
5268 rather than after. Use tree_to_uhwi when summing small shift
5269 counts into an unsigned int.
5271 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
5272 Alan Hayward <alan.hayward@arm.com>
5273 David Sherwood <david.sherwood@arm.com>
5276 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
5277 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
5278 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
5279 POINTER_AND_FP_REGS.
5281 2017-10-02 Richard Biener <rguenther@suse.de>
5283 PR tree-optimization/82355
5284 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5285 a mapping for the enclosing loop but avoid generating one for
5287 (copy_bb_and_scalar_dependences): Remove premature codegen
5288 error on PHIs in blocks duplicated into multiple places.
5289 * graphite-scop-detection.c
5290 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5291 in the region use it as loop and nest to analyze the DR in.
5292 (try_generate_gimple_bb): Likewise.
5293 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5294 (add_loop_constraints): For blocks in a loop not in the region
5295 create a dimension with a single iteration.
5296 * sese.h (gbb_loop_at_index): Remove assert.
5298 2017-10-01 Kevin Buettner <kevinb@redhat.com>
5300 * omp-expand.c (adjust_context_scope): New function.
5301 (expand_parallel_call): Call adjust_context_scope.
5303 2017-10-01 Jeff Law <law@redhat.com>
5305 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
5306 dom_opt_dom_walker class with direct access to private members.
5307 Add comments. Call test_for_singularity.
5308 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
5309 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
5310 m_dummy_cond anymore.
5311 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
5313 (pass_dominator:execute): Build the dummy_cond here and pass it
5314 to the dom_opt_dom_walker ctor.
5315 (test_for_singularity): New function.
5317 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
5318 Maya Rashish <coypu@sdf.org>
5320 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
5321 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
5322 (alpha*-*-netbsd*) Use nbsd_tm_file.
5323 (arm*-*-netbsdelf*) Likewise.
5324 (i[34567]86-*-netbsdelf*) Likewise.
5325 (x86_64-*-netbsd*) Likewise.
5326 (mips*-*-netbsd*) Likewise.
5327 (powerpc-*-netbsd*) Likewise.
5328 (sh*-*-netbsd*) Likewise.
5329 (sparc-*-netbsdelf*) Likewise.
5330 (sparc64-*-netbsd*) Likewise.
5331 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
5333 (vax-*-netbsdelf*) Likewise.
5334 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
5335 (UINT_FAST8_TYPE) Likewise.
5336 (INT_FAST16_TYPE) Check CHAR_FAST16.
5337 (UINT_FAST16_TYPE) Likewise.
5339 2017-09-30 Jakub Jelinek <jakub@redhat.com>
5342 * config/i386/i386.md
5343 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
5344 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
5345 *divmodsi4_zext_2): New define_insn_and_split.
5346 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
5347 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
5348 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
5349 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
5350 New define_insn_and_split.
5351 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
5352 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
5353 operands[1] having DImode when mode is SImode.
5355 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
5356 always SImode for DIV and MOD in REG_EQUAL notes.
5358 2017-09-29 Yury Gribov <tetra2005@gmail.com>
5361 * match.pd: Fix handling of NaNs in pattern.
5363 2017-09-29 Jeff Law <law@redhat.com>
5365 * sbitmap.c (bitmap_bit_in_range_p): New function.
5366 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
5367 * tree-ssa-dse.c (live_bytes_read): New function.
5368 (dse_classify_store): Ignore reads of dead bytes.
5370 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
5371 typos and whitespace errors.
5372 * config/i386/predicates.md (address_no_seg_operand): Likewise.
5373 * config/s390/s390.c (s390_emit_prologue): Likewise.
5375 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
5378 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
5379 with a symbol for LRA.
5381 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
5383 PR rtl-optimization/82338
5384 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
5386 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
5388 * genmodes.c (calc_wider_mode): Suppress qsort macro.
5389 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
5390 (qsort_chk): Declare.
5391 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
5392 (qsort_chk): New function.
5394 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5396 PR tree-optimization/82337
5397 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
5398 phi definition if the PHI result appears in an abnormal PHI.
5399 (find_basis_for_base_expr): Don't record a basis if the LHS of the
5400 basis appears in an abnormal PHI.
5402 2017-09-29 Richard Biener <rguenther@suse.de>
5404 * graphite-isl-ast-to-gimple.c
5405 (translate_isl_ast_to_gimple::set_codegen_error): New function.
5406 (binary_op_to_tree): Use it.
5407 (get_rename_from_scev): Likewise.
5408 (copy_loop_phi_nodes): Likewise.
5409 (copy_bb_and_scalar_dependences): Likewise.
5410 (translate_pending_phi_nodes): Likewise.
5412 2017-09-29 Jakub Jelinek <jakub@redhat.com>
5415 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
5416 for movabsq $(i32 << shift), r64.
5418 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
5420 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
5421 index when encoding %esp as %rsp to avoid 0x67 prefix.
5423 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
5425 * config/i386/i386.md (*movsf_internal, *movdf_internal):
5426 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
5428 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
5430 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
5431 Extensions with more than 16 double VFP registers.
5432 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
5433 to_clear_mask and all code related to it. Replace the remaining
5434 entry by a sbitmap and adapt code accordingly.
5436 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
5438 * brig-builtins.def: Change pure attributes to const.
5440 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5442 * config.gcc (default_gnu_indirect_function): Default to yes for
5443 sparc*-*-linux* with glibc.
5445 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5447 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
5448 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
5449 when creating .init_array and .fini_array sections with priority
5452 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
5455 * config/aarch64/aarch64.c
5456 (aarch64_builtin_support_vector_misalignment): Always return false
5457 when misalignment is unknown.
5459 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
5461 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
5462 this function to return false if the definition used by the swap
5463 instruction is artificial, or if the memory address from which the
5464 constant value is loaded is not represented by a base address held
5465 in a register or if the base address register is a frame or stack
5466 pointer. Additionally, return false if the base address of the
5467 loaded constant is a SYMBOL_REF but is not considered to be a
5469 (replace_swapped_load_constant): New function.
5470 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
5471 loaded constant vector with a load of a swapped constant vector.
5473 2017-09-27 Carl Love <cel@us.ibm.com>
5475 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
5476 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
5477 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
5480 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
5482 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
5483 first, always call autopref_rank_data otherwise.
5485 2017-09-27 Richard Biener <rguenther@suse.de>
5487 * graphite-scop-detection.c (find_scop_parameters): Move
5488 loop bound handling ...
5489 (gather_bbs::before_dom_children): ... here, avoiding the need
5490 to build scop_info->loop_nest.
5491 (record_loop_in_sese): Remove.
5492 * sese.h (sese_info_t::loop_nest): Remove.
5493 * sese.c (new_sese_info): Do not allocate loop_nest.
5494 (free_sese_info): Do not free loop_nest.
5496 2017-09-27 Jakub Jelinek <jakub@redhat.com>
5499 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
5500 lhs from calls if the lhs has addressable type.
5502 2017-09-27 Richard Biener <rguenther@suse.de>
5504 * graphite.h (scop::max_alias_set): New member.
5505 * graphite-scop-detection.c: Remove references to non-existing
5506 --param in comments.
5507 (build_alias_sets): Record the maximum alias set used for drs.
5508 (build_scops): Support zero as unlimited for
5509 --param graphite-max-arrays-per-scop.
5510 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
5512 (build_poly_sr_1): ... here. Compute alias set based on the
5513 maximum alias set used for drs rather than
5514 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
5516 2017-09-27 Richard Biener <rguenther@suse.de>
5518 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
5519 --param loop-block-tile-size=0 to disable tiling.
5521 2017-09-27 Richard Biener <rguenther@suse.de>
5523 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
5524 (graphite-max-nb-scop-params): Document special value zero.
5525 * domwalk.h (dom_walker::STOP): New symbolical constant.
5526 (dom_walker::dom_walker): Add optional parameter for bb to
5528 (dom_walker::~dom_walker): Declare.
5529 (dom_walker::before_dom_children): Document STOP return value.
5530 (dom_walker::m_user_bb_to_rpo): New member.
5531 (dom_walker::m_bb_to_rpo): Likewise.
5532 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
5533 mapping here if not provided by the user.
5534 (dom_walker::~dom_walker): Free bb to RPO mapping if not
5535 provided by the user.
5536 (dom_walker::STOP): Define.
5537 (dom_walker::walk): Do not compute bb to RPO mapping here.
5538 Support STOP return value from before_dom_children to stop
5540 * graphite-optimize-isl.c (optimize_isl): If the schedule
5541 is the same still generate code if -fgraphite-identity
5542 or -floop-parallelize-all are given.
5543 * graphite-scop-detection.c: Include cfganal.h.
5544 (gather_bbs::gather_bbs): Get and pass through bb to RPO
5546 (gather_bbs::before_dom_children): Return STOP for BBs
5548 (build_scops): Compute bb to RPO mapping and pass it to
5549 the domwalk. Treat --param graphite-max-nb-scop-params=0
5550 as not limiting the number of params.
5551 * graphite.c (graphite_initialize): Remove limit on the
5552 number of basic-blocks in a function.
5553 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
5554 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
5555 default value of 10.
5557 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5559 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
5560 Adjust code to eliminate needing to do the shift right 32-bits
5561 operation after XSCVDPSPN.
5563 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5565 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
5566 ((X / Y) != 0 -> X >= Y): Likewise.
5568 2017-09-26 Carl Love <cel@us.ibm.com>
5570 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
5571 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
5572 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
5573 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
5574 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
5575 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
5576 definitions and overloading.
5577 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
5578 statement for P9V_BUILTIN_XST_LEN_R.
5579 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
5580 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
5581 define_expand and define_insn for the instructions and builtins.
5582 * doc/extend.texi: Update the built-in documentation file for the new
5584 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
5585 define_insn for the instructions
5587 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
5590 * gcc/config/netbsd-protos.h: New file.
5591 * gcc/config/netbsd.c: New file.
5592 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
5593 * gcc/config/t-netbsd: New file.
5594 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
5595 (tmake_file) Add t-netbsd.
5596 (extra_objs) Add netbsd.o.
5598 2017-09-26 Janus Weil <janus@gcc.gnu.org>
5602 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
5604 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5606 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
5607 sign extension from a vector register to a GPR by doing a 32-bit
5608 direct move and then an EXTSW.
5609 (extendsi<mode>2 splitter): Likewise.
5610 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
5611 right or vector extract after doing XSCVDPSPN. Use
5612 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
5614 (movdi_from_sf_zero_ext): Likewise.
5615 (reload_gpr_from_vsxsf): Likewise.
5616 (p8_mfvsrd_4_disf): Delete, no longer used.
5617 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
5618 then needing to move the SFmode to a GPR to use the XSCVDPSP
5619 instruction instead of FRSP and XSCVDPSPN.
5620 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
5621 it is adjacent to the other XSCVSPDP insns.
5622 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
5623 SFmode to be in traditional Altivec registers.
5624 (vsx_xscvdpspn): Eliminate useless alternative constraint.
5625 (vsx_xscvspdpn): Likewise.
5626 (vsx_xscvspdpn_scalar): Likewise.
5628 2017-09-26 Martin Jambor <mjambor@suse.cz>
5630 * tree-sra.c (compare_access_positions): Put integral types first,
5631 stabilize sorting of integral types, remove conditions putting
5632 non-full-precision integers last.
5633 (sort_and_splice_var_accesses): Disable scalarization if a
5634 non-integert would be represented by a non-full-precision integer.
5636 2017-09-26 Joseph Myers <joseph@codesourcery.com>
5638 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
5639 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
5640 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
5641 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
5642 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
5643 conditionals inside the function instead of around it. Call
5644 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
5645 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
5647 2017-09-26 Richard Biener <rguenther@suse.de>
5649 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
5651 (scop_detection::build_scop_breadth): ... this. Removed.
5652 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
5653 (scop_detection::harmful_stmt_in_bb): Likewise.
5654 (scop_detection::graphite_can_represent_stmt): Likewise.
5655 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
5656 (scop_detection::can_represent_loop): Remove recursion, fold in ...
5657 (scop_detection::can_represent_loop_1): ... this. Removed.
5658 (scop_detection::harmful_loop_in_region): Simplify after inlining
5659 the above and remove more quadraticness.
5660 (build_scops): Adjust.
5661 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
5664 2017-09-26 Jakub Jelinek <jakub@redhat.com>
5667 * config/i386/i386.c (ix86_print_operand_address_as): Only test
5668 REGNO (base) == SP_REG if base is a REG.
5671 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
5672 if it is different SSA_NAME.
5673 (optimize_range_tests_cmp_bitwise): New function.
5674 (optimize_range_tests): Call it.
5676 2017-09-26 Richard Biener <rguenther@suse.de>
5678 PR tree-optimization/82321
5679 * graphite.c (canonicalize_loop_closed_ssa): Properly check
5680 for the def being inside the loop.
5682 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5684 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
5686 * config/s390/s390-builtins.def: Fix constraint on op4.
5688 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5690 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
5691 independent expanders.
5692 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
5693 ("vec_ordered", "vec_unordered"): New expanders.
5695 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5697 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
5700 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5702 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
5703 vec_unpacks_lo_v16qi.
5704 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
5706 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5708 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
5709 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
5710 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
5712 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5714 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
5716 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
5718 ("*vec_slb<mode>"): New insn pattern.
5719 ("vec_shr_<mode>"): New expander.
5720 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
5721 and force the shift count operand to V16QImode.
5722 ("vec_srb<mode>"): Set shift count mode to V16QI.
5724 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5726 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
5727 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
5728 ("vec_widen_smult_hi_<mode>"): New expander definitions.
5730 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
5733 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
5735 2017-09-26 Richard Biener <rguenther@suse.de>
5737 PR tree-optimization/82320
5738 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
5741 2017-09-25 Jeff Law <law@redhat.com>
5743 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
5744 prototype for new argument.
5745 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
5746 mostly extracted from rs6000_emit_allocate_stack.
5747 (rs6000_emit_probe_stack_range_stack_clash): New function.
5748 (rs6000_emit_allocate_stack): Call
5749 rs6000_emit_probe_stack_range_stack_clash as needed.
5750 (rs6000_emit_probe_stack_range): Add additional argument
5751 to call to gen_probe_stack_range{si,di}.
5752 (output_probe_stack_range): New.
5753 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
5754 (output_probe_stack_range_stack_clash): New.
5755 (rs6000_emit_prologue): Emit notes into dump file as requested.
5756 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
5757 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
5758 Add additional operand and pass it to output_probe_stack_range.
5760 2017-09-25 Bin Cheng <bin.cheng@arm.com>
5762 PR tree-optimization/82163
5763 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
5764 (checking_verify_loop_closed_ssa): New parameter.
5765 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
5766 (check_loop_closed_ssa_stmt): Delete.
5767 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
5768 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
5769 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
5772 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
5774 * brig-builtins.def: Treat HSAIL barrier builtins as
5775 setjmp/longjump style functions.
5777 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
5779 * target.def (constant_alignment): New hook.
5780 * defaults.h (CONSTANT_ALIGNMENT): Delete.
5781 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
5782 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
5783 * doc/tm.texi: Regenerate.
5784 * targhooks.h (default_constant_alignment): Declare.
5785 (constant_alignment_word_strings): Likewise.
5786 * targhooks.c (default_constant_alignment): New function.
5787 (constant_alignment_word_strings): Likewise.
5788 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
5789 instead of CONSTANT_ALIGNMENT.
5790 * varasm.c (align_variable, get_variable_align, build_constant_desc)
5791 (force_const_mem): Likewise.
5792 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
5793 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
5794 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
5795 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5796 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
5798 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
5799 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5800 constant_alignment_word_strings.
5801 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
5802 (CONSTANT_ALIGNMENT): Likewise.
5803 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5804 (arm_constant_alignment): New function.
5805 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
5806 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5807 constant_alignment_word_strings.
5808 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
5809 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5810 constant_alignment_word_strings.
5811 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
5812 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5813 (cris_constant_alignment): New function.
5814 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
5815 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5816 (epiphany_constant_alignment): New function.
5817 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
5818 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5819 constant_alignment_word_strings.
5820 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
5821 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5822 constant_alignment_word_strings.
5823 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
5824 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5825 constant_alignment_word_strings.
5826 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
5827 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
5828 * config/i386/i386.c (ix86_constant_alignment): Make static.
5829 Use the same interface as the target hook.
5830 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5831 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
5832 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5833 constant_alignment_word_strings.
5834 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
5835 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
5836 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5837 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
5838 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5839 constant_alignment_word_strings.
5840 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
5841 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5842 constant_alignment_word_strings.
5843 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
5844 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5845 constant_alignment_word_strings.
5846 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
5847 * config/microblaze/microblaze.c (microblaze_constant_alignment):
5849 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5850 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
5851 * config/mips/mips.c (mips_constant_alignment): New function.
5852 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5853 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
5854 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
5855 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5856 (mmix_constant_alignment): Make static. Use the same interface
5858 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
5859 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5860 constant_alignment_word_strings.
5861 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
5862 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5863 constant_alignment_word_strings.
5864 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
5865 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5866 constant_alignment_word_strings.
5867 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
5868 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5869 (rs6000_constant_alignment): New function.
5870 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
5871 * config/riscv/riscv.c (riscv_constant_alignment): New function.
5872 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5873 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
5874 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5875 (rs6000_constant_alignment): New function.
5876 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
5877 * config/s390/s390.c (s390_constant_alignment): New function.
5878 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5879 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
5880 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5881 constant_alignment_word_strings.
5882 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
5883 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5884 (sparc_constant_alignment): New function.
5885 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
5886 * config/spu/spu.c (spu_constant_alignment): New function.
5887 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5888 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
5889 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5890 constant_alignment_word_strings.
5891 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
5892 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5893 constant_alignment_word_strings.
5894 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
5895 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5896 constant_alignment_word_strings.
5897 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
5898 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5899 (visium_constant_alignment): New function.
5900 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
5901 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5902 (xtensa_constant_alignment): New function.
5903 * system.h (CONSTANT_ALIGNMENT): Poison.
5905 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
5907 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
5908 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
5909 (rs6000_builtin_valid_without_lhs): New helper function.
5910 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5911 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
5913 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
5915 * target.h (vec_perm_indices): Use unsigned short rather than
5917 (auto_vec_perm_indices): Likewise.
5918 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
5919 Use unsigned int rather than unsigned char.
5920 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
5922 2017-09-25 Richard Biener <rguenther@suse.de>
5924 * cfgloop.h (sort_sibling_loops): Declare.
5925 * cfgloop.c (sort_sibling_loops_cmp): New helper.
5926 (sort_sibling_loops): New function sorting the sibling loop list
5928 * graphite.c (graphite_transform_loops): Sort sibling loops.
5930 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
5932 * target.def (vec_perm_const_ok): Change sel parameter to
5934 * optabs-query.c (can_vec_perm_p): Update accordingly.
5935 * doc/tm.texi: Regenerate.
5936 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
5937 auto_vec_perm_indices and remove separate nelt field.
5938 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
5939 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
5940 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
5941 (aarch64_expand_vec_perm_const): Update accordingly.
5942 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
5943 to vec_perm_indices.
5944 * config/arm/arm.c (expand_vec_perm_d): Change perm to
5945 auto_vec_perm_indices and remove separate nelt field.
5946 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
5947 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
5948 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
5950 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
5951 to vec_perm_indices.
5952 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
5953 sel to vec_perm_indices.
5954 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
5955 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
5956 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
5958 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
5961 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
5964 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
5965 on the FUNCTION_DECL function context if it has a DIE that is a
5968 2017-09-25 Richard Biener <rguenther@suse.de>
5970 PR tree-optimization/82285
5971 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
5974 2017-09-25 Tom de Vries <tom@codesourcery.com>
5978 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
5981 2017-09-25 Richard Biener <rguenther@suse.de>
5983 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
5984 ISL errors other than isl_error_quota happen. Dump if the
5985 schedule is the same.
5986 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
5987 errors instead of aborting inside ISL.
5989 2017-09-25 Iain Sandoe <iain@codesourcery.com>
5992 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
5993 of libgcc_eh for m64.
5994 * config/i386/darwin64.h: Likewise.
5996 2017-09-25 Richard Biener <rguenther@suse.de>
5999 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
6000 attribute for incomplete types nor twice for complete ones.
6002 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
6005 * config/i386/i386.c (ix86_print_operand_address_as): Encode
6006 %esp as %rsp to avoid 0x67 prefix if there is no index or base
6009 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
6012 * config/i386/i386.opt (mprefer-avx256): Use
6013 ix86_target_flags variable.
6014 * config/i386/i386.c (ix86_target_string): Move
6015 -mprefer-avx256 to flag2_opts.
6017 2017-09-22 Jakub Jelinek <jakub@redhat.com>
6020 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
6021 and x != -1 | y != -1 into (x & y) != -1.
6023 2017-09-22 Steve Ellcey <sellcey@cavium.com>
6025 * config.gcc: Add new case statement to set
6026 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
6027 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
6028 s390x-*-linux* case statements. Added aarch64 to the list of
6029 supported architectures.
6031 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6033 PR tree-optimization/82289
6034 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
6035 STMT_VINFO_RELEVANT_P.
6037 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6038 Alan Hayward <alan.hayward@arm.com>
6039 David Sherwood <david.sherwood@arm.com>
6041 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
6042 for VR_RANGE only; don't allow VR_ANTI_RANGE.
6043 (extract_range_from_binary_expr_1): Don't call
6044 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
6046 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6047 Alan Hayward <alan.hayward@arm.com>
6048 David Sherwood <david.sherwood@arm.com>
6050 * target.def (preferred_vector_alignment): New hook.
6051 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
6053 * doc/tm.texi: Regenerate.
6054 * targhooks.h (default_preferred_vector_alignment): Declare.
6055 * targhooks.c (default_preferred_vector_alignment): New function.
6056 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
6058 (DR_TARGET_ALIGNMENT): New macro.
6059 (aligned_access_p): Update commentary.
6060 (vect_known_alignment_in_bytes): New function.
6061 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
6063 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
6064 Calculate the misalignment based on the target alignment rather than
6066 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
6067 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
6068 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
6069 the target alignment, rather than masking the element misalignment
6070 with the number of elements in a vector. Also use the target
6071 alignment when calculating the maximum number of peels.
6072 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
6073 instead of TYPE_ALIGN_UNIT.
6074 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
6075 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
6076 (vect_create_addr_base_for_vector_ref): Update call accordingly.
6077 (vect_create_data_ref_ptr): Likewise.
6078 (vect_setup_realignment): Realign by ANDing with
6079 -DR_TARGET_MISALIGNMENT.
6080 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
6081 the number of peels based on DR_TARGET_ALIGNMENT.
6082 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
6083 with the guaranteed alignment boundary when deciding whether
6085 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
6086 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
6087 (ensure_base_align): Remove stmt_info parameter. Get the
6088 target base alignment from DR_TARGET_ALIGNMENT.
6089 (vectorizable_store): Update call accordingly. Interpret
6090 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
6092 (vectorizable_load): Likewise.
6094 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6095 Alan Hayward <alan.hayward@arm.com>
6096 David Sherwood <david.sherwood@arm.com>
6098 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
6099 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
6100 (vect_enhance_data_refs_alignment): Likewise.
6102 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6104 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
6105 error. Only quit immediately if parsing is complete.
6106 (BEGIN): Initialize fatal_err and parse_done.
6107 (begin fpu, end fpu): Check number of arguments.
6108 (begin arch, end arch): Likewise.
6109 (begin cpu, end cpu): Likewise.
6110 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
6111 (optalias): Likewise.
6113 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6115 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
6116 * config/arm/arm-isa.h: Delete. Move definitions to ...
6117 * arm-cpus.in: ... here. Use new feature and fgroup values.
6118 * config/arm/arm.c (arm_option_override): Use lower case for feature
6120 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
6121 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
6122 * config/arm/parsecpu.awk (END): Add new command 'isa'.
6124 (print_isa_bits_for): New function.
6125 (gen_isa): New function.
6126 (gen_comm_data): Use print_isa_bits_for.
6127 (define feature): New keyword.
6128 (define fgroup): New keyword.
6129 * config/arm/t-arm (TM_H): Remove.
6130 (GTM_H): Add arm-isa.h.
6131 (arm-isa.h): Add rule to generate file.
6132 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
6133 case for feature bit names.
6135 2017-09-22 Richard Biener <rguenther@suse.de>
6137 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
6139 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
6140 print of no dependency loops ...
6141 * graphite.c (graphite_transform_loops): ... here.
6142 (canonicalize_loop_closed_ssa_form): Work from inner to outer
6144 (same_close_phi_node, remove_duplicate_close_phi,
6145 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
6146 (canonicalize_loop_closed_ssa): ... here and simplify.
6147 * graphite-optimize-isl.c: Include tree-vectorizer.h.
6148 (optimize_isl): Use dump_printf_loc to tell when we stopped
6149 optimizing because of an ISL timeout.
6151 2017-09-22 Richard Biener <rguenther@suse.de>
6153 PR tree-optimization/82291
6154 * tree-if-conv.c (predicate_mem_writes): Make sure to
6155 remove writes in blocks predicated with false.
6157 2017-09-22 Richard Biener <rguenther@suse.de>
6159 * sese.c: Include cfganal.h.
6160 (if_region_set_false_region): Remove.
6161 (create_if_region_on_edge): Likewise.
6162 (move_sese_in_condition): Re-implement without destroying
6165 2017-09-22 Richard Biener <rguenther@suse.de>
6167 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
6168 Verify both BBs contain loop PHI nodes before dispatching to
6170 (graphite_regenerate_ast_isl): Do not recompute dominators,
6171 do not verify three times. Restructure for clarity.
6172 * graphite-scop-detection.c (same_close_phi_node,
6173 remove_duplicate_close_phi, make_close_phi_nodes_unique,
6174 defined_in_loop_p, canonicalize_loop_closed_ssa,
6175 canonicalize_loop_closed_ssa_form): Simplify, remove excess
6176 checking and SSA rewrite, move to ...
6177 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
6178 (graphite_initialize): Do not pass in ctx, do not reset the
6179 SCEV cache, compute only dominators.
6180 (graphite_transform_loops): Allocate ISL ctx after
6181 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
6182 Maintain post-dominators only around build_scops.
6183 * sese.c (if_region_set_false_region): Make static. Free
6184 and recompute dominators.
6185 (move_sese_in_condition): Assert we don't get called with
6186 post-dominators computed.
6187 * sese.h (if_region_set_false_region): Remove.
6189 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
6191 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
6192 mode attribute for TARGET_AVX512VL.
6194 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
6196 * config/i386/i386.opt (mprefer-avx256): New option.
6197 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
6199 (ix86_preferred_simd_mode): Return 256-bit AVX modes
6200 for TARGET_PREFER_AVX256.
6201 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
6203 2017-09-21 Jeff Law <law@redhat.com>
6205 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
6206 Fix dump output if the only stack space is for pushed registers.
6208 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6210 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
6213 2017-09-21 Martin Sebor <msebor@redhat.com>
6216 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
6217 code (in C++) or code that triggers warnings.
6219 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
6221 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
6223 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
6225 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
6226 * sched-rgn.c: Ditto.
6227 * sel-sched-ir.c: Ditto.
6229 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
6231 * toplev.h (set_random_seed): Adjust return type.
6232 * toplev.c (init_local_tick): Move eager initialization of random_seed
6233 to get_random_seed. Adjust comment.
6234 (init_random_seed): Inline to get_random_seed, delete.
6235 (get_random_seed): Initialize random_seed lazily.
6236 (set_random_seed): Do not return previous value.
6237 (print_switch_value): Do not call get_random_seed.
6239 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
6241 * cgraph.c (delete_function_version): New, broken out from...
6242 (cgraph_node::delete_function_version): ...here. Rename to
6243 cgraph_node::delete_function_version_by_decl. Update all uses.
6244 (cgraph_node::remove): Call delete_function_version.
6246 2017-09-21 Jakub Jelinek <jakub@redhat.com>
6249 * tree-inline.c (expand_call_inline): Emit clobber stmts for
6250 VAR_DECLs to which addressable non-volatile parameters are mapped
6251 and for id->retvar after the return value assignment. Clear
6252 id->retval and id->retbnd after inlining.
6254 2017-09-21 Richard Biener <rguenther@suse.de>
6256 PR tree-optimization/82276
6257 PR tree-optimization/82244
6258 * tree-vrp.c (build_assert_expr_for): Set
6259 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
6261 (remove_range_assertions): Revert earlier change.
6263 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
6266 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
6268 2017-09-21 Richard Biener <rguenther@suse.de>
6270 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
6271 Restore valid IL after code generation errors.
6272 * graphite.c (graphite_transform_loops): Diagnose code
6273 generation issues as MSG_MISSED_OPTIMIZATION and continue
6274 with processing SCOPs.
6276 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6277 Alan Hayward <alan.hayward@arm.com>
6278 David Sherwood <david.sherwood@arm.com>
6280 * calls.c (compute_argument_addresses): Use simplify_gen_binary
6281 rather than choosing between plus_constant and gen_rtx_<CODE>.
6282 * expr.c (emit_push_insn): Likewise.
6283 (expand_expr_real_2): Likewise.
6285 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6286 Alan Hayward <alan.hayward@arm.com>
6287 David Sherwood <david.sherwood@arm.com>
6289 * loop-unroll.c (split_iv): Call copy_rtx on the step.
6291 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6292 Alan Hayward <alan.hayward@arm.com>
6293 David Sherwood <david.sherwood@arm.com>
6295 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
6296 calling tree_to_uhwi.
6298 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6299 Alan Hayward <alan.hayward@arm.com>
6300 David Sherwood <david.sherwood@arm.com>
6302 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
6303 INTEGER_CST rather than a negative test for ADDR_EXPR.
6305 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6306 Alan Hayward <alan.hayward@arm.com>
6307 David Sherwood <david.sherwood@arm.com>
6309 * tree-vrp.c (extract_range_from_binary_expr_1): Check
6310 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
6312 2017-09-21 Richard Biener <rguenther@suse.de>
6314 PR tree-optimization/71351
6315 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
6316 graphite_create_new_loop_guard): Remove, fold remaining parts
6318 (translate_isl_ast_node_for): ... here and simplify.
6320 2017-09-21 Jakub Jelinek <jakub@redhat.com>
6323 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
6324 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
6325 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
6326 alternative always use QI mode, for -Os imov (=R,R) alternative
6327 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
6330 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6331 Jeff Law <law@redhat.com>
6333 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
6334 (allocate_stack_space): New function, partially extracted from
6336 (s390_emit_prologue): Track offset to most recent stack probe.
6337 Code to allocate space moved into allocate_stack_space.
6338 Dump actions when no stack is allocated.
6339 (s390_prologue_plus_offset): New function.
6340 (s390_emit_stack_probe): Likewise.
6342 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
6344 * common.opt (Wa, Wl, Wp, g, gz=): Add
6346 (gno-column-info): Remove.
6347 (gcolumn-info): Drop RejectNegative.
6349 (gno-record-gcc-switches): Remove.
6350 (grecord-gcc-switches): Drop RejectNegative.
6351 (gno-split-dwarf): Remove.
6352 (gsplit-dwarf): Drop RejectNegative.
6353 (gno-strict-dwarf): Remove.
6354 (gstrict-dwarf): Drop RejectNegative.
6355 * config/darwin.opt (gfull, gused): Add RejectNegative.
6356 * dwarf2out.c (gen_producer_string): Drop
6357 gno-record-gcc-switches handler.
6358 * optc-gen.awk: Add g to prefixes with negative forms.
6359 * opts-common.c (remapping_prefix_p): New.
6360 (find_opt): Check it.
6361 (generate_canonical_option): Test g prefix.
6362 (option_map): Add -gno- mapping.
6363 (add_misspelling_candidates): Check remapping_prefix_p.
6365 2017-09-20 Jeff Law <law@redhat.com>
6367 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
6368 thinko in stack clash protection support.
6370 * explow.c (compute_stack_clash_protection_loop_data): Use
6371 CONST_INT_P instead of explicit test. Verify object is a
6372 CONST_INT_P before looking at INTVAL.
6373 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
6374 instead of explicit test.
6376 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
6379 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
6380 address instead of to r1 and r11.
6382 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
6384 * config.gcc: Support "knm".
6385 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
6386 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
6388 * config/i386/i386.c (m_KNM): Define.
6389 (processor_target_table): Add "knm".
6391 (ix86_option_override_internal): Add "knm".
6392 (ix86_issue_rate): Add PROCESSOR_KNM.
6393 (ix86_adjust_cost): Ditto.
6394 (ia32_multipass_dfa_lookahead): Ditto.
6395 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
6396 (fold_builtin_cpu): Add M_INTEL_KNM.
6397 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
6398 (processor_type): Add PROCESSOR_KNM.
6399 * config/i386/x86-tune.def: Add m_KNM.
6400 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
6402 2017-09-20 Richard Biener <rguenther@suse.de>
6404 PR tree-optimization/80213
6405 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
6406 are allowed in empty BBs as well.
6407 (canonicalize_loop_closed_ssa): Also look for other complex
6409 (scop_detection::get_sese): Include the loop-closed PHI block
6411 (scop_detection::merge_sese): Remove code adding extra blocks.
6412 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
6413 (build_scops): Assert the final returned scop is invalid.
6415 2017-09-20 Richard Biener <rguenther@suse.de>
6417 PR tree-optimization/82264
6418 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
6420 (vn_phi_lookup): Likewise.
6421 (vn_phi_insert): Likewise.
6423 2017-09-20 Jakub Jelinek <jakub@redhat.com>
6425 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
6426 that fits into uhwi or shwi, add DW_AT_const_value regardless
6427 of early_dwarf without going through RTL, using add_AT_unsigned
6430 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
6431 (DEBUG_LTO_ABBREV_SECTION): Likewise.
6432 (DEBUG_LTO_MACINFO_SECTION): Likewise.
6433 (DEBUG_MACRO_SECTION): Likewise.
6434 (DEBUG_LTO_MACRO_SECTION): Likewise.
6435 (DEBUG_STR_DWO_SECTION): Likewise.
6436 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
6437 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
6438 (DEBUG_LTO_DWO_LINE_SECTION): Define.
6439 (DEBUG_LTO_LINE_STR_SECTION): Define.
6440 (init_sections_and_labels): Initialize debug_line_str_section
6441 variable. Initialize debug_loc_section for -gdwarf-5 to
6442 DEBUG_LOCLISTS_SECTION. Formatting fixes.
6444 2017-09-20 Richard Biener <rguenther@suse.de>
6446 * graphite-sese-to-poly.c (extract_affine): Properly handle
6447 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
6449 2017-09-20 Richard Biener <rguenther@suse.de>
6451 PR tree-optimization/81373
6452 * graphite-scop-detection.c (build_cross_bb_scalars_def):
6453 Force SESE live-out defs to be handled even if they are
6456 2017-09-19 Jeff Law <law@redhat.com>
6458 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
6459 nothing for stack adjustments with REG_STACK_CHECK.
6460 * sched-deps.c (parse_add_or_inc): Reject insns with
6461 REG_STACK_CHECK from dependency breaking.
6462 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
6463 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
6464 * reg-notes.def (STACK_CHECK): New note.
6466 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
6467 (ix86_expand_prologue): Dump stack clash info as needed.
6468 Call ix86_adjust_stack_and_probe_stack_clash as needed.
6470 * function.c (dump_stack_clash_frame_info): New function.
6471 * function.h (dump_stack_clash_frame_info): Prototype.
6472 (enum stack_clash_probes): New enum.
6474 * config/alpha/alpha.c (alpha_expand_prologue): Also check
6475 flag_stack_clash_protection.
6476 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
6477 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
6478 (arm_frame_pointer_required): Likewise.
6479 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
6480 (ia64_expand_prologue): Likewise.
6481 * config/mips/mips.c (mips_expand_prologue): Likewise.
6482 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
6483 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6484 (sparc_flat_expand_prologue): Likewise.
6485 * config/spu/spu.c (spu_expand_prologue): Likewise.
6487 * explow.c: Include "params.h".
6488 (anti_adjust_stack_and_probe_stack_clash): New function.
6489 (get_stack_check_protect): Likewise.
6490 (compute_stack_clash_protection_loop_data): Likewise.
6491 (emit_stack_clash_protection_loop_start): Likewise.
6492 (emit_stack_clash_protection_loop_end): Likewise.
6493 (allocate_dynamic_stack_space): Use get_stack_check_protect.
6494 Use anti_adjust_stack_and_probe_stack_clash.
6495 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
6496 (emit_stack_clash_protection_loop_start): Likewise.
6497 (emit_stack_clash_protection_loop_end): Likewise.
6498 * rtl.h (get_stack_check_protect): Prototype.
6499 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
6500 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
6501 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
6503 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
6505 * doc/tm.texi: Rebuilt.
6506 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
6507 get_stack_check_protect.
6508 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
6509 * config/arm/arm.c (arm_expand_prologue): Likewise.
6510 (arm_frame_pointer_required): Likewise.
6511 * config/i386/i386.c (ix86_expand_prologue): Likewise.
6512 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
6513 * config/mips/mips.c (mips_expand_prologue): Likewise.
6514 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
6515 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
6516 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6517 (sparc_flat_expand_prologue): Likewise.
6519 * common.opt (-fstack-clash-protection): New option.
6520 * flag-types.h (enum stack_check_type): Note difference between
6521 -fstack-check= and -fstack-clash-protection.
6522 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
6523 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
6524 * toplev.c (process_options): Issue warnings/errors for cases
6525 not handled with -fstack-clash-protection.
6526 * doc/invoke.texi (-fstack-clash-protection): Document new option.
6527 (-fstack-check): Note additional problem with -fstack-check=generic.
6528 Note that -fstack-check is primarily for Ada and refer users
6529 to -fstack-clash-protection for stack-clash-protection.
6530 Document new params for stack clash protection.
6532 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
6534 * config/i386/i386.c (ix86_split_long_move): Do not handle
6535 address used for LEA in a special way.
6537 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
6539 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
6541 2017-09-19 Martin Sebor <msebor@redhat.com>
6544 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
6545 of incompatible types.
6547 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
6549 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6550 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
6551 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6552 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
6554 2017-09-19 Richard Biener <rguenther@suse.de>
6556 PR tree-optimization/82244
6557 * tree-vrp.c (remove_range_assertions): Do not propagate
6558 a constant to abnormals but replace the assert with a copy.
6560 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
6562 PR rtl-optimization/57878
6563 PR rtl-optimization/68988
6564 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
6565 avoidance test involving non_reload_pseudos. Move frequency test
6566 below the general fragmentation avoidance test.
6568 2017-09-19 Richard Biener <rguenther@suse.de>
6570 PR tree-optimization/69728
6571 * graphite-sese-to-poly.c (schedule_error): New global.
6572 (add_loop_schedule): Handle empty domain by failing the
6574 (build_original_schedule): Handle schedule_error.
6576 2017-09-19 Richard Biener <rguenther@suse.de>
6578 * graphite-scop-detection.c (scop_detection::can_represent_loop):
6579 Do not iterate to sibling loops but only to siblings of inner
6582 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
6585 * config/m68k/m68k.md (moveq feeding equality comparison): Check
6586 that the registers are different.
6588 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
6590 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
6591 to processor_model and "amdfam17h" to arch_names_table.
6592 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
6594 2017-09-18 Jakub Jelinek <jakub@redhat.com>
6597 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
6599 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6600 Alan Hayward <alan.hayward@arm.com>
6601 David Sherwood <david.sherwood@arm.com>
6603 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
6605 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
6607 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
6608 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
6609 vect_schedule_slp_instance.
6610 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
6611 Update call to vect_slp_analyze_node_operations. Simplify return
6613 (vect_slp_analyze_bb_1): Update call accordingly.
6614 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
6615 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
6616 (vect_schedule_slp): Update call accordingly.
6618 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6619 Alan Hayward <alan.hayward@arm.com>
6620 David Sherwood <david.sherwood@arm.com>
6622 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
6623 with types that aren't in fact scalar.
6625 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6627 * tree-vect-slp.c (vect_record_max_nunits): New function,
6629 (vect_build_slp_tree_1): ...here.
6630 (vect_build_slp_tree_2): Call it for phis too.
6632 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6634 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
6635 to vect_get_vec_def_for_operand when getting the mask operand.
6637 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6638 Alan Hayward <alan.hayward@arm.com>
6639 David Sherwood <david.sherwood@arm.com>
6641 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
6644 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6645 Alan Hayward <alan.hayward@arm.com>
6646 David Sherwood <david.sherwood@arm.com>
6648 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
6649 calculation for vector booleans.
6651 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6652 Alan Hayward <alan.hayward@arm.com>
6653 David Sherwood <david.sherwood@arm.com>
6655 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
6657 (vect_transform_stmt): ...here.
6658 (vect_analyze_stmt): Use it instead of calling
6659 vectorizable_live_operation directly.
6661 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
6663 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
6664 non-SIMT targets in acc vector loops.
6666 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
6668 * configure.ac: Add arc and check if assembler supports gdwarf2.
6669 * configure: Regenerate.
6671 2017-09-18 Richard Biener <rguenther@suse.de>
6673 PR tree-optimization/82220
6674 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
6675 epilogue niters from the min_profitable_iters compute.
6677 2017-09-18 Jakub Jelinek <jakub@redhat.com>
6680 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
6681 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
6682 (ix86_init_pic_reg): Revert 2017-09-01 changes.
6684 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
6687 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
6688 switching to a new text section.
6690 2017-09-18 Richard Biener <rguenther@suse.de>
6692 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
6694 (build_alias_set): Reject aliases with no access function.
6696 2017-09-18 Richard Biener <rguenther@suse.de>
6698 PR tree-optimization/79622
6699 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
6701 (build_cross_bb_scalars_use): Likewise.
6703 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
6705 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
6707 2017-09-18 Alan Modra <amodra@gmail.com>
6710 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
6711 stack_pointer_rtx for count 0. Update comments. Break up
6712 large rtl expression.
6714 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
6716 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
6717 Increase to 20 bytes.
6718 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
6719 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
6720 or avx version of the stub.
6722 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
6725 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
6726 compute the minimum stack alignment. Also update preferred stack
6727 boundary for leaf functions.
6729 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
6731 PR tree-optimization/82228
6732 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
6735 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
6737 * common/config/nds32/nds32-common.c
6738 (nds32_option_optimization_table): Refine formatting.
6739 (nds32_option_optimization_table): Use -fsched-pressure and
6740 -fomit-frame-pointer for specific optimization level.
6742 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
6744 * config/nds32/nds32.c: Refine formatting and comments.
6745 * config/nds32/nds32.h: Likewise.
6746 * config/nds32/nds32.md: Likewise.
6747 * config/nds32/nds32-cost.c: Likewise.
6748 * config/nds32/nds32-isr.c: Likewise.
6749 * config/nds32/nds32-md-auxiliary.c: Likewise.
6750 * config/nds32/nds32-multiple.md: Likewise.
6751 * config/nds32/nds32-predicates.c: Likewise.
6753 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
6754 Jakub Jelinek <jakub@redhat.com>
6756 Add support for -std=c++2a.
6757 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
6759 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
6761 2017-09-15 Steve Ellcey <sellcey@cavium.com>
6764 * doc/extend.texi (Common Function Attributes): Add
6765 references to ARM, AArch64, and S/390 specific attributes.
6766 (Function Specific Option Pragmas): Add AArch64 and S/390
6767 to list of back ends that support the target pragma.
6769 2017-09-15 Nathan Sidwell <nathan@acm.org>
6771 * doc/standards.texi: Fix C++17 description. Update URLs for
6774 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
6776 * common.opt (Wcast-align=strict): New warning option.
6777 * doc/invoke.texi: Document -Wcast-align=strict.
6779 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
6781 * cgraph.h (cgraph_thunk_info): Add comments.
6782 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
6783 assert for VIRTUAL_* arguments stricter.
6785 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
6787 PR tree-optimization/71026
6788 * match.pd: Move RDIV patterns from fold-const.c
6789 * fold-const.c (distribute_real_division): Removed.
6790 (fold_binary_loc): Remove calls to distribute_real_divison.
6792 2017-09-15 Jakub Jelinek <jakub@redhat.com>
6794 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
6795 c++1z and gnu++1z as deprecated. Change other references to
6796 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
6797 Change -Wc++1z-compat to -Wc++17-compat.
6798 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
6799 * dwarf2out.c (highest_c_language): Handle C++17.
6800 (gen_compile_unit_die): Likewise.
6802 2017-09-15 Jakub Jelinek <jakub@redhat.com>
6804 PR rtl-optimization/82192
6805 * combine.c (make_extraction): Don't look through non-paradoxical
6806 SUBREGs or TRUNCATE if pos + len is or might be bigger than
6809 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
6810 Alan Hayward <alan.hayward@arm.com>
6811 David Sherwood <david.sherwood@arm.com>
6813 * target.def (function_arg_offset): New hook.
6814 * targhooks.h (default_function_arg_offset): Declare.
6815 * targhooks.c (default_function_arg_offset): New function.
6816 * function.c (locate_and_pad_parm): Use
6817 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
6818 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
6819 (TARGET_FUNCTION_ARG_OFFSET): ...this.
6820 * doc/tm.texi: Regenerate.
6821 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
6822 * config/spu/spu.c (spu_function_arg_offset): New function.
6823 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
6824 * system.h (FUNCTION_ARG_OFFSET): Poison.
6826 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
6827 Alan Hayard <alan.hayward@arm.com>
6828 David Sherwood <david.sherwood@arm.com>
6830 * target.def (truly_noop_truncation): New hook.
6831 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
6832 than TRULY_NOOP_TRUNCATION.
6833 * hooks.h (hook_bool_uint_uint_true): Declare.
6834 * hooks.c (hook_bool_uint_uint_true): New function.
6835 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
6836 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
6837 * doc/tm.texi: Regenerate.
6838 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
6839 rather than TRULY_NOOP_TRUNCATION in comments.
6840 (simplify_comparison): Likewise.
6841 (record_truncated_value): Likewise.
6842 * expmed.c (extract_bit_field_1): Likewise.
6843 (extract_split_bit_field): Likewise.
6844 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
6845 instead of TRULY_NOOP_TRUNCATION.
6846 * function.c (assign_parm_setup_block): Likewise.
6847 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
6848 * rtlhooks.c: Include target.h.
6849 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
6850 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
6851 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
6852 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
6853 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
6854 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
6855 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
6856 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
6857 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
6858 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
6859 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
6860 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
6861 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
6862 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
6863 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
6864 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
6865 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
6866 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
6867 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
6868 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
6869 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
6870 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
6871 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
6872 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
6873 * config/mips/mips.c (mips_truly_noop_truncation): New function.
6874 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
6875 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
6876 rather than TRULY_NOOP_TRUNCATION in comments.
6877 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
6878 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
6879 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
6880 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
6881 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
6882 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
6883 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
6884 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
6885 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
6886 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
6887 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
6888 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
6889 rather than TRULY_NOOP_TRUNCATION in comments.
6890 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
6891 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
6892 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
6893 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
6894 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
6895 TRULY_NOOP_TRUNCATION condition.
6896 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
6897 (TRULY_NOOP_TRUNCATION): Delete.
6898 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
6899 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
6900 * config/spu/spu.c (spu_truly_noop_truncation): New function.
6901 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
6902 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
6903 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
6904 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
6905 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
6906 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
6907 rather than TRULY_NOOP_TRUNCATION in comments.
6908 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
6909 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
6910 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
6911 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
6912 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
6913 * system.h (TRULY_NOOP_TRUNCATION): Poison.
6915 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
6918 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
6919 (*cmp_ior): Likewise.
6920 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
6921 (*ior_scc_scc_cmp): Likewise.
6922 (*and_scc_scc): Likewise.
6923 (*and_scc_scc_cmp): Likewise.
6925 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
6926 Alan Hayard <alan.hayward@arm.com>
6927 David Sherwood <david.sherwood@arm.com>
6929 * target.def (can_change_mode_class): New hook.
6930 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
6931 (hard_regno_nregs): Likewise.
6932 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
6933 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
6934 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
6935 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
6936 (LOAD_EXTEND_OP): Update accordingly.
6937 * doc/tm.texi: Regenerate.
6938 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
6939 CANNOT_CHANGE_MODE_CLASS.
6940 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
6941 (REG_CAN_CHANGE_MODE_P): ...this new macro.
6942 * combine.c (simplify_set): Update accordingly.
6943 * emit-rtl.c (validate_subreg): Likewise.
6944 * recog.c (general_operand): Likewise.
6945 * regcprop.c (mode_change_ok): Likewise.
6946 * reload1.c (choose_reload_regs): Likewise.
6947 (inherit_piecemeal_p): Likewise.
6948 * rtlanal.c (simplify_subreg_regno): Likewise.
6949 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
6950 instead of CANNOT_CHANGE_MODE_CLASS.
6951 (reload_cse_simplify_operands): Likewise.
6952 * reload.c (push_reload): Use targetm.can_change_mode_class
6953 instead of CANNOT_CHANGE_MODE_CLASS.
6954 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
6955 REG_CANNOT_CHANGE_MODE_P.
6956 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6957 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
6958 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6959 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6960 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6961 (arm_can_change_mode_class): New function.
6962 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
6963 than CANNOT_CHANGE_MODE_CLASS in comments.
6964 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6965 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
6966 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
6967 (ix86_can_change_mode_class): ...this new function, inverting the
6968 sense of the return value.
6969 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6970 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6971 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6972 (ia64_can_change_mode_class): New function.
6973 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6974 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
6975 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
6976 (m32c_can_change_mode_class): ...this new function, inverting the
6977 sense of the return value.
6978 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6979 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6980 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
6981 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
6982 (mips_can_change_mode_class): ...this new function, inverting the
6983 sense of the return value.
6984 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6985 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6986 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6987 (msp430_can_change_mode_class): New function.
6988 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6989 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
6990 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6991 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6992 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6993 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
6994 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6995 (pa_cannot_change_mode_class): Replace with...
6996 (pa_can_change_mode_class): ...this new function, inverting the
6997 sense of the return value.
6998 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
6999 than CANNOT_CHANGE_MODE_CLASS in comments.
7000 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7001 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
7002 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7003 (pdp11_cannot_change_mode_class): Replace with...
7004 (pdp11_can_change_mode_class): ...this new function, inverting the
7005 sense of the return value.
7006 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7007 * config/powerpcspe/powerpcspe-protos.h
7008 (rs6000_cannot_change_mode_class_ptr): Delete.
7009 * config/powerpcspe/powerpcspe.c
7010 (rs6000_cannot_change_mode_class_ptr): Delete.
7011 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7012 (rs6000_option_override_internal): Assign to
7013 targetm.can_change_mode_class instead of
7014 rs6000_cannot_change_mode_class_ptr.
7015 (rs6000_cannot_change_mode_class): Replace with...
7016 (rs6000_can_change_mode_class): ...this new function, inverting the
7017 sense of the return value.
7018 (rs6000_debug_cannot_change_mode_class): Replace with...
7019 (rs6000_debug_can_change_mode_class): ...this new function.
7020 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7021 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
7022 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7023 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7024 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
7026 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
7027 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7028 (rs6000_option_override_internal): Assign to
7029 targetm.can_change_mode_class instead of
7030 rs6000_cannot_change_mode_class_ptr.
7031 (rs6000_cannot_change_mode_class): Replace with...
7032 (rs6000_can_change_mode_class): ...this new function, inverting the
7033 sense of the return value.
7034 (rs6000_debug_cannot_change_mode_class): Replace with...
7035 (rs6000_debug_can_change_mode_class): ...this new function.
7036 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7037 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
7038 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
7039 (s390_can_change_mode_class): ...this new function, inverting the
7040 sense of the return value.
7041 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7042 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7043 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
7044 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7045 (sh_cannot_change_mode_class): Replace with...
7046 (sh_can_change_mode_class): ...this new function, inverting the
7047 sense of the return value.
7048 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7049 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7050 (sparc_can_change_mode_class): New function.
7051 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7052 * config/spu/spu.c (spu_can_change_mode_class): New function.
7053 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7054 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7055 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7056 (visium_can_change_mode_class): New function.
7057 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
7059 2017-09-15 Richard Biener <rguenther@suse.de>
7061 PR tree-optimization/82217
7062 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
7063 but not undefined case.
7065 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7068 * postreload.c (reload_cse_simplify_operands): Skip
7069 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
7071 2017-09-15 Richard Biener <rguenther@suse.de>
7073 PR tree-optimization/68823
7074 * graphite-scop-detection.c (build_alias_set): If we have a
7075 possible dependence check whether we can handle them by just
7076 looking at the DRs DR_ACCESS_FNs.
7077 (build_scops): If build_alias_set fails, fail the SCOP.
7079 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
7081 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
7082 to support float128 built-in functions that require the ISA 3.0
7084 (BU_FLOAT128_3_HW): Likewise.
7085 (SQRTF128): Add support for the IEEE 128-bit square root and fma
7087 (FMAF128): Likewise.
7089 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
7090 support for built-in functions that need the ISA 3.0 IEEE 128-bit
7091 floating point instructions.
7092 (rs6000_invalid_builtin): Likewise.
7093 (rs6000_builtin_mask_names): Likewise.
7094 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
7095 (RS6000_BTM_FLOAT128_HW): Likewise.
7096 (RS6000_BTM_COMMON): Likewise.
7097 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
7099 * doc/extend.texi (RS/6000 built-in functions): Document the
7100 IEEE 128-bit floating point square root and fused multiply-add
7103 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
7105 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
7106 reg (r2) isn't in the set of registers defined in the prologue.
7108 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7109 Alan Hayward <alan.hayward@arm.com>
7110 David Sherwood <david.sherwood@arm.com>
7112 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
7113 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
7114 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
7115 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
7116 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
7118 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
7119 max_vectorization_factor.
7120 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
7122 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7123 Alan Hayward <alan.hayward@arm.com>
7124 David Sherwood <david.sherwood@arm.com>
7126 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
7127 (vect_worthwhile_without_simd_p): Declare.
7128 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
7129 (vectorizable_reduction): Use it.
7130 * tree-vect-stmts.c (vectorizable_shift): Likewise.
7131 (vectorizable_operation): Likewise.
7133 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7134 Alan Hayward <alan.hayward@arm.com>
7135 David Sherwood <david.sherwood@arm.com>
7137 * tree-vectorizer.h (vect_get_num_copies): New function.
7138 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
7139 * tree-vect-loop.c (vectorizable_reduction): Likewise.
7140 (vectorizable_induction): Likewise.
7141 (vectorizable_live_operation): Likewise.
7142 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7143 (vectorizable_bswap): Likewise.
7144 (vectorizable_call): Likewise.
7145 (vectorizable_conversion): Likewise.
7146 (vectorizable_assignment): Likewise.
7147 (vectorizable_shift): Likewise.
7148 (vectorizable_operation): Likewise.
7149 (vectorizable_store): Likewise.
7150 (vectorizable_load): Likewise.
7151 (vectorizable_condition): Likewise.
7152 (vectorizable_comparison): Likewise.
7153 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
7155 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7156 Alan Hayward <alan.hayward@arm.com>
7157 David Sherwood <david.sherwood@arm.com>
7159 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
7160 of vect_init_vector.
7162 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7163 Alan Hayward <alan.hayward@arm.com>
7164 David Sherwood <david.sherwood@arm.com>
7166 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
7167 an inline wrapper that provides a location.
7168 (gimple_build_vector): Likewise.
7169 * gimple-fold.c (gimple_build_vector_from_val): New function.
7170 (gimple_build_vector): Likewise.
7171 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
7172 functions to build the initial value. Always return a gimple value.
7173 (get_initial_defs_for_reduction): Likewise. Only compute
7175 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
7176 vect_init_vector on the results from get_initial_def(s)_for_reduction.
7177 (vectorizable_induction): Use gimple_build_vector rather than
7180 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7181 Alan Hayward <alan.hayward@arm.com>
7182 David Sherwood <david.sherwood@arm.com>
7184 * target.h (vec_perm_indices): New typedef.
7185 (auto_vec_perm_indices): Likewise.
7186 * optabs-query.h: Include target.h
7187 (can_vec_perm_p): Take a vec_perm_indices *.
7188 * optabs-query.c (can_vec_perm_p): Likewise.
7189 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
7190 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7191 * tree-vect-generic.c (lower_vec_perm): Likewise.
7192 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
7193 (vect_grouped_load_supported): Likewise.
7194 (vect_shift_permute_load_chain): Likewise.
7195 (vect_permute_store_chain): Use auto_vec_perm_indices.
7196 (vect_permute_load_chain): Likewise.
7197 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
7198 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
7199 Update uses of can_vec_perm_p.
7200 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
7201 mode with a number of elements. Take a vec_perm_indices *.
7202 (vect_create_epilog_for_reduction): Update accordingly.
7203 Use auto_vec_perm_indices.
7204 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
7205 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
7206 (vect_transform_slp_perm_load): Likewise.
7207 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
7208 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
7209 (vect_gen_perm_mask_checked): Likewise.
7210 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
7211 (vect_gen_perm_mask_checked): Likewise.
7212 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
7213 (vectorizable_store): Likewise.
7214 (vectorizable_load): Likewise.
7215 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
7216 (vectorizable_bswap): Likewise.
7218 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7219 Alan Hayward <alan.hayward@arm.com>
7220 David Sherwood <david.sherwood@arm.com>
7222 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
7223 * tree.c (build_vector): Likewise.
7224 (build_vector_from_ctor): Update accordingly.
7225 (build_vector_from_val): Likewise.
7226 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
7227 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7228 * tree-vect-generic.c (add_rshift): Likewise.
7229 (expand_vector_divmod): Likewise.
7230 (optimize_vector_constructor): Likewise.
7231 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
7232 (vect_transform_slp_perm_load): Likewise.
7233 (vect_schedule_slp_instance): Likewise.
7234 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
7235 (vectorizable_call): Likewise.
7236 (vect_gen_perm_mask_any): Likewise. Add elements in order.
7237 * expmed.c (make_tree): Likewise.
7238 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
7239 a vector passed to build_vector.
7240 (fold_convert_const): Likewise.
7241 (exact_inverse): Likewise.
7242 (fold_ternary_loc): Likewise.
7243 (fold_relational_const): Likewise.
7244 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
7245 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
7246 (const_unop): Likewise. Store the reduction accumulator in a
7247 variable rather than an array.
7248 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
7249 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
7250 the new vector, rather than constructing it after the input arrays.
7251 (native_interpret_vector): Use auto_vec<tree> when building
7252 a vector passed to build_vector. Add elements in order.
7253 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
7254 auto_vec<tree> when building a vector passed to build_vector.
7255 (vect_create_epilog_for_reduction): Likewise.
7256 (vectorizable_induction): Likewise.
7257 (get_initial_def_for_reduction): Likewise. Fix indentation of
7259 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
7261 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
7262 passed to build_vector.
7264 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7265 Alan Hayward <alan.hayward@arm.com>
7266 David Sherwood <david.sherwood@arm.com>
7268 * tree-core.h (tree_base::u): Add an "nelts" field.
7269 (tree_vector): Use VECTOR_CST_NELTS as the length.
7270 * tree.c (tree_size): Likewise.
7271 (make_vector): Initialize VECTOR_CST_NELTS.
7272 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
7273 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
7274 TYPE_VECTOR_SUBPARTS.
7275 * expr.c (const_vector_mask_from_tree): Consistently use "units"
7276 as the number of units, setting it from VECTOR_CST_NELTS.
7277 (const_vector_from_tree): Likewise.
7278 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
7279 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
7280 (fold_negate_expr_1): Likewise.
7281 (fold_convert_const): Likewise.
7282 (const_binop): Likewise. Differentiate the number of output and
7284 (const_unop): Likewise.
7285 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
7286 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
7287 in cases that did the opposite.
7289 2017-09-14 Richard Biener <rguenther@suse.de>
7291 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
7294 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
7296 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
7298 2017-09-14 Jakub Jelinek <jakub@redhat.com>
7301 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
7302 if and where to split a bb, except for splitting before debug insn
7303 sequences followed by non-label real insn. Delete debug insns
7304 in between basic blocks.
7306 * combine.c (make_compound_operation_int): Formatting fixes.
7308 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
7309 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7310 * config/netbsd.h (LINK_EH_SPEC): Likewise.
7311 * config/sol2.h (LINK_EH_SPEC): Likewise.
7312 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7313 * config/s390/linux.h (LINK_SPEC): Likewise.
7314 * config/freebsd.h (LINK_EH_SPEC): Likewise.
7315 * config/openbsd.h (LINK_EH_SPEC): Likewise.
7316 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7317 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
7318 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
7319 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7320 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
7321 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
7323 2017-09-13 Jakub Jelinek <jakub@redhat.com>
7325 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
7327 (ENDFILE_LINUX_SPEC): Likewise.
7328 (LINK_EH_SPEC): Likewise.
7329 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
7330 (LINK_OS_LINUX_SPEC32): Likewise.
7331 (LINK_OS_LINUX_SPEC64): Likewise.
7332 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
7333 (LINK_OS_LINUX_SPEC): Likewise.
7335 2017-09-13 Martin Liska <mliska@suse.cz>
7338 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
7339 CASE_HIGH is NULL_TREE.
7341 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
7342 Alan Hayward <alan.hayward@arm.com>
7343 David Sherwood <david.sherwood@arm.com>
7345 * target.def (secondary_memory_needed): New hook.
7346 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
7347 instead of SECONDARY_MEMORY_NEEDED.
7348 (secondary_memory_needed_mode): Likewise.
7349 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
7350 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
7351 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
7352 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
7353 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
7354 * doc/tm.texi: Regenerate.
7355 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
7356 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
7357 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7358 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
7359 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
7360 * config/i386/i386.c (inline_secondary_memory_needed): Put the
7361 mode argument first and change the reg_class arguments to reg_class_t.
7362 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
7363 Make static. Update the call to inline_secondary_memory_needed.
7364 (ix86_register_move_cost): Update the call to
7365 inline_secondary_memory_needed.
7366 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7367 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
7369 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
7370 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7372 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
7373 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
7374 * config/mips/mips.c (mips_secondary_memory_needed): Make static
7375 and match hook interface. Add comment from mips.h.
7376 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7377 * config/mmix/mmix.md (truncdfsf2): Refer to
7378 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7380 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
7381 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
7382 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7383 (pa_secondary_memory_needed): New function.
7384 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
7385 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
7386 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7387 (pdp11_secondary_memory_needed): Make static and match hook interface.
7388 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
7389 * config/powerpcspe/powerpcspe-protos.h
7390 (rs6000_secondary_memory_needed_ptr): Delete.
7391 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
7393 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7394 (rs6000_option_override_internal): Assign to
7395 targetm.secondary_memory_needed rather than
7396 rs6000_secondary_memory_needed_ptr.
7397 (rs6000_secondary_memory_needed): Match hook interface.
7398 (rs6000_debug_secondary_memory_needed): Likewise.
7399 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
7400 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
7401 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
7402 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7403 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
7404 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
7406 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
7407 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7408 (rs6000_option_override_internal): Assign to
7409 targetm.secondary_memory_needed rather than
7410 rs6000_secondary_memory_needed_ptr.
7411 (rs6000_secondary_memory_needed): Match hook interface.
7412 (rs6000_debug_secondary_memory_needed): Likewise.
7413 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
7414 * config/s390/s390.c (s390_secondary_memory_needed): New function.
7415 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7416 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
7417 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7418 (sparc_secondary_memory_needed): New function.
7419 * lra-constraints.c (check_and_process_move): Refer to
7420 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7422 (curr_insn_transform): Likewise.
7423 (process_alt_operands): Use targetm.secondary_memory_needed
7424 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7425 (check_secondary_memory_needed_p): Likewise.
7426 (choose_split_class): Likewise.
7427 * reload.c: Unconditionally include code that was previously
7428 conditional on SECONDARY_MEMORY_NEEDED.
7429 (push_secondary_reload): Use targetm.secondary_memory_needed
7430 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7431 (push_reload): Likewise.
7432 * reload1.c: Unconditionally include code that was previously
7433 conditional on SECONDARY_MEMORY_NEEDED.
7434 (choose_reload_regs): Use targetm.secondary_memory_needed
7435 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7436 (gen_reload): Likewise.
7437 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
7439 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
7440 Alan Hayward <alan.hayward@arm.com>
7441 David Sherwood <david.sherwood@arm.com>
7443 * target.def (secondary_memory_needed_mode): New hook:
7444 * targhooks.c (default_secondary_memory_needed_mode): Declare.
7445 * targhooks.h (default_secondary_memory_needed_mode): New function.
7446 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
7447 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
7448 * doc/tm.texi: Regenerate.
7449 * lra-constraints.c (check_and_process_move): Use
7450 targetm.secondary_memory_needed_mode instead of
7451 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
7452 (curr_insn_transform): Likewise.
7453 * reload.c (get_secondary_mem): Likewise.
7454 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7455 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
7457 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7458 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7459 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
7460 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7461 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
7463 * config/powerpcspe/powerpcspe-protos.h
7464 (rs6000_secondary_memory_needed_mode): Delete.
7465 * config/powerpcspe/powerpcspe.c
7466 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7467 (rs6000_secondary_memory_needed_mode): Make static.
7468 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7469 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
7471 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7473 (rs6000_secondary_memory_needed_mode): Make static.
7474 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7475 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
7476 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7477 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7478 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7480 (sparc_secondary_memory_needed_mode): New function.
7481 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
7483 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
7485 * config/aarch64/constraints.md (Umq): New constraint.
7486 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
7488 (mov<mode>): Update condition.
7490 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7492 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
7493 when bitposition is the same.
7495 2017-09-13 Richard Biener <rguenther@suse.de>
7497 * dwarf2out.c (output_die_symbol): Remove.
7498 (output_die): Do not output a DIEs symbol.
7500 2017-09-13 Richard Biener <rguenther@suse.de>
7503 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
7504 default-def to avoid breaking iterator update with the weird
7505 interaction with cgraph_update_edges_for_call_stmt_node.
7507 2017-09-13 Richard Biener <rguenther@suse.de>
7509 * tree-cfg.c (verify_gimple_assign_binary): Add verification
7510 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
7511 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
7512 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
7514 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
7516 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
7517 Disable pc relative literal load irrespective of
7518 TARGET_FIX_ERR_A53_84341 for default.
7520 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
7522 * config/sparc/sparc.c (output_return): Output the source location of
7523 the insn in the delay slot, if any.
7524 (output_sibcall): Likewise.
7526 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
7529 * common.opt (-static-pie): New alias.
7530 (shared): Negate static-pie.
7531 (-no-pie): Update help text.
7533 (static-pie): New option.
7534 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
7535 -static-pie support.
7536 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
7537 (LINK_EH_SPEC): Likewise.
7538 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7539 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7540 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7541 * gcc.c (LINK_COMMAND_SPEC): Likewise.
7542 (init_gcc_specs): Likewise.
7543 (init_spec): Likewise.
7544 (display_help): Update help message for -pie.
7545 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
7548 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
7550 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
7551 (movdi_aarch64): Likewise.
7552 (movti_aarch64): Likewise.
7554 2017-09-12 Simon Wright <simon@pushface.org>
7557 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
7558 calculation of the minor version, always output as 0.
7560 2017-09-12 Jakub Jelinek <jakub@redhat.com>
7563 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
7564 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
7565 on it early, rather than manual conversion late. For
7566 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
7567 instead of performing manual conversion.
7569 2017-09-12 Carl Love <cel@us.ibm.com>
7571 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
7572 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
7574 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7575 VMULOSW): Add definitions.
7576 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7577 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7578 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7579 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7580 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7582 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
7584 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
7586 (movti_aarch64): Likewise.
7587 (movdf_aarch64): Likewise.
7588 (movtf_aarch64): Likewise.
7589 (load_pairdi): Likewise.
7590 (store_pairdi): Likewise.
7591 (load_pairdf): Likewise.
7592 (store_pairdf): Likewise.
7593 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
7594 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
7595 (ldr_got_small_<mode>): Likewise.
7596 (ldr_got_small_28k_<mode>): Likewise.
7597 (ldr_got_tiny): Likewise.
7598 * config/aarch64/iterators.md (ldst_sz): New.
7599 (ldpstp_sz): Likewise.
7600 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
7602 (thunderx_load): Split load_8 to load_16.
7603 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
7605 (thunderx2t99_storepair_basic): Split store_8 to store_16.
7606 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
7607 (xgene1_store_pair): Split store_8 to store_16.
7608 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
7609 (falkor_st_0_st_sd): Split store_8 to store_16.
7611 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
7613 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
7614 and store1/2/3/4 to store_4/8/12/16.
7615 * config/aarch64/aarch64.md: Update for rename.
7616 * config/arm/arm.md: Likewise.: Likewise.
7617 * config/arm/arm.c: Likewise.
7618 * config/arm/thumb1.md: Likewise.
7619 * config/arm/thumb2.md: Likewise.
7620 * config/arm/vfp.md: Likewise.
7621 * config/arm/arm-generic.md: Likewise.
7622 * config/arm/arm1020e.md: Likewise.
7623 * config/arm/arm1026ejs.md: Likewise.
7624 * config/arm/arm1136jfs.md: Likewise.
7625 * config/arm/arm926ejs.md: Likewise.
7626 * config/arm/cortex-a15.md: Likewise.
7627 * config/arm/cortex-a17.md: Likewise.
7628 * config/arm/cortex-a5.md: Likewise.
7629 * config/arm/cortex-a53.md: Likewise.
7630 * config/arm/cortex-a57.md: Likewise.
7631 * config/arm/cortex-a7.md: Likewise.
7632 * config/arm/cortex-a8.md: Likewise.
7633 * config/arm/cortex-a9.md: Likewise.
7634 * config/arm/cortex-m4.md: Likewise.
7635 * config/arm/cortex-m7.md: Likewise.
7636 * config/arm/cortex-r4.md: Likewise.
7637 * config/arm/exynos-m1.md: Likewise.
7638 * config/arm/fa526.md: Likewise.
7639 * config/arm/fa606te.md: Likewise.
7640 * config/arm/fa626te.md: Likewise.
7641 * config/arm/fa726te.md: Likewise.
7642 * config/arm/fmp626.md: Likewise.
7643 * config/arm/iwmmxt.md: Likewise.
7644 * config/arm/ldmstm.md: Likewise.
7645 * config/arm/marvell-pj4.md: Likewise.
7646 * config/arm/xgene1.md: Likewise.
7647 * config/aarch64/thunderx.md: Likewise.
7648 * config/aarch64/thunderx2t99.md: Likewise.
7649 * config/aarch64/falkor.md: Likewise.
7651 2017-09-12 Martin Liska <mliska@suse.cz>
7653 * attribs.c (private_lookup_attribute): New function.
7654 * attribs.h (private_lookup_attribute): Declared here.
7655 (lookup_attribute): Called from this place.
7657 2017-09-12 Richard Biener <rguenther@suse.de>
7659 PR tree-optimization/82157
7660 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
7661 stmts with side-effects.
7663 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7664 Alan Hayward <alan.hayward@arm.com>
7665 David Sherwood <david.sherwood@arm.com>
7667 * target.def (hard_regno_nregs): New hook.
7668 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
7669 * targhooks.h (default_hard_regno_nregs): Declare.
7670 * targhooks.c (default_hard_regno_nregs): New function.
7671 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
7672 (TARGET_HARD_REGNO_NREGS): ...this hook.
7673 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
7674 (CLASS_MAX_NREGS): Likewise.
7675 * doc/tm.texi: Regenerate.
7676 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
7677 instead of HARD_REGNO_NREGS.
7678 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
7679 HARD_REGNO_NREGS in the comment.
7680 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
7681 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
7682 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
7683 Return an unsigned int.
7684 (TARGET_HARD_REGNO_NREGS): Redefine.
7685 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
7686 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
7687 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
7688 (arc_hard_regno_nregs): New function.
7689 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
7690 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
7691 (arm_hard_regno_nregs): New function.
7692 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
7693 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
7694 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
7695 (TARGET_HARD_REGNO_NREGS): Redefine.
7696 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
7697 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
7698 (HARD_REGNO_NREGS): Delete.
7699 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
7700 (cr16_hard_regno_nregs): New function.
7701 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
7702 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
7703 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
7704 (cris_hard_regno_nregs): New function.
7705 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
7706 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
7707 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
7708 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
7709 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
7710 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
7711 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
7712 (frv_hard_regno_nregs): Make static. Take and return an
7714 (frv_class_max_nregs): Remove outdated copy of documentation.
7715 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
7716 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
7717 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
7718 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
7719 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
7720 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
7721 (TARGET_HARD_REGNO_NREGS): Redefine.
7722 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
7723 (CLASS_MAX_NREGS): Update comment.
7724 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
7725 (ia64_hard_regno_nregs): New function.
7726 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
7727 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
7728 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
7729 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
7730 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
7732 (m32c_hard_regno_nregs): Likewise. Make static.
7733 (TARGET_HARD_REGNO_NREGS): Redefine.
7734 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
7735 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
7736 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
7737 (m68k_hard_regno_nregs): New function.
7738 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
7739 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
7740 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
7741 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
7742 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
7743 Take and return an unsigned int.
7744 (TARGET_HARD_REGNO_NREGS): Redefine.
7745 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
7746 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
7747 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
7748 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
7749 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
7750 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
7751 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
7752 (msp430_hard_regno_nregs): Make static. Take and return an
7754 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
7755 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
7756 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
7757 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
7758 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
7759 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
7760 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
7761 (TARGET_HARD_REGNO_NREGS): Redefine.
7762 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
7763 (PA_HARD_REGNO_NREGS): ...this.
7764 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
7765 (PA_HARD_REGNO_NREGS): ...this.
7766 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
7767 (pa_hard_regno_nregs): New function.
7768 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
7769 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
7770 (pdp11_hard_regno_nregs): New function.
7771 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
7772 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
7773 (rs6000_hard_regno_nregs_hook): New function.
7774 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
7775 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
7776 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
7777 Take and return an unsigned int. Move earlier in file.
7778 (TARGET_HARD_REGNO_NREGS): Redefine.
7779 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
7780 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
7781 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
7782 (rl78_hard_regno_nregs): Make static. Take and return an
7784 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
7785 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
7786 (rs6000_hard_regno_nregs_hook): New function.
7787 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
7788 * config/rx/rx.c (rx_hard_regno_nregs): New function.
7789 (TARGET_HARD_REGNO_NREGS): Redefine.
7790 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
7791 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
7792 instead of HARD_REGNO_NREGS.
7793 (s390_hard_regno_nregs): New function.
7794 (s390_hard_regno_mode_ok): Add comment from s390.h.
7795 (TARGET_HARD_REGNO_NREGS): Redefine.
7796 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
7797 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
7798 (sh_hard_regno_nregs): New function.
7799 (sh_pass_in_reg_p): Use it.
7800 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
7801 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
7802 (sparc_hard_regno_nregs): New function.
7803 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
7804 * config/spu/spu.c (spu_hard_regno_nregs): New function.
7805 (spu_function_arg_advance): Use it, supplying a valid register number.
7806 (TARGET_HARD_REGNO_NREGS): Redefine.
7807 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
7808 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
7809 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
7810 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
7811 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
7812 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
7813 (CLASS_MAX_NREGS): Remove copy of old documentation.
7814 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
7815 (visium_hard_regno_nregs): New function.
7816 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
7817 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
7818 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
7819 xtensa_hard_regno_nregs): New function.
7820 * system.h (HARD_REGNO_NREGS): Poison.
7822 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7824 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
7825 hard_regno_nregs instead of HARD_REGNO_NREGS.
7826 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
7827 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
7828 (c6x_expand_epilogue): Likewise.
7829 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
7830 (frv_read_iacc_argument): Likewise.
7831 * config/sh/sh.c: Include regs.h.
7832 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
7833 (regs_used): Likewise.
7834 (output_stack_adjust): Likewise.
7835 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
7836 * expmed.c: Include regs.h.
7837 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
7838 * ree.c: Include regs.h.
7839 (combine_reaching_defs): Use hard_regno_nregs instead of
7841 (add_removable_extension): Likewise.
7843 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7845 * regs.h (hard_regno_nregs): Turn into a function.
7846 (end_hard_regno): Update accordingly.
7847 * caller-save.c (setup_save_areas): Likewise.
7848 (save_call_clobbered_regs): Likewise.
7849 (replace_reg_with_saved_mem): Likewise.
7850 (insert_restore): Likewise.
7851 (insert_save): Likewise.
7852 * combine.c (can_change_dest_mode): Likewise.
7853 (move_deaths): Likewise.
7854 (distribute_notes): Likewise.
7855 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
7856 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
7857 (rs6000_split_multireg_move): Likewise.
7858 (rs6000_register_move_cost): Likewise.
7859 (rs6000_memory_move_cost): Likewise.
7860 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
7861 (rs6000_split_multireg_move): Likewise.
7862 (rs6000_register_move_cost): Likewise.
7863 (rs6000_memory_move_cost): Likewise.
7864 * cselib.c (cselib_reset_table): Likewise.
7865 (cselib_lookup_1): Likewise.
7866 * emit-rtl.c (set_mode_and_regno): Likewise.
7867 * function.c (aggregate_value_p): Likewise.
7868 * ira-color.c (setup_profitable_hard_regs): Likewise.
7869 (check_hard_reg_p): Likewise.
7870 (calculate_saved_nregs): Likewise.
7871 (assign_hard_reg): Likewise.
7872 (improve_allocation): Likewise.
7873 (calculate_spill_cost): Likewise.
7874 * ira-emit.c (modify_move_list): Likewise.
7875 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
7876 (ira_hard_reg_in_set_p): Likewise.
7877 * ira.c (setup_reg_mode_hard_regset): Likewise.
7878 (clarify_prohibited_class_mode_regs): Likewise.
7879 (check_allocation): Likewise.
7880 * lra-assigns.c (find_hard_regno_for_1): Likewise.
7881 (lra_setup_reg_renumber): Likewise.
7882 (setup_try_hard_regno_pseudos): Likewise.
7883 (spill_for): Likewise.
7884 (assign_hard_regno): Likewise.
7885 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
7886 * lra-constraints.c (in_class_p): Likewise.
7887 (lra_constraint_offset): Likewise.
7888 (simplify_operand_subreg): Likewise.
7889 (lra_constraints): Likewise.
7890 (split_reg): Likewise.
7891 (split_if_necessary): Likewise.
7892 (invariant_p): Likewise.
7893 (inherit_in_ebb): Likewise.
7894 * lra-lives.c (process_bb_lives): Likewise.
7895 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
7896 (get_hard_regs): Likewise.
7897 (do_remat): Likewise.
7898 * lra-spills.c (assign_spill_hard_regs): Likewise.
7899 * mode-switching.c (create_pre_exit): Likewise.
7900 * postreload.c (reload_combine_recognize_pattern): Likewise.
7901 * recog.c (peep2_find_free_register): Likewise.
7902 * regcprop.c (kill_value_regno): Likewise.
7903 (set_value_regno): Likewise.
7904 (copy_value): Likewise.
7905 (maybe_mode_change): Likewise.
7906 (find_oldest_value_reg): Likewise.
7907 (copyprop_hardreg_forward_1): Likewise.
7908 * regrename.c (check_new_reg_p): Likewise.
7909 (regrename_do_replace): Likewise.
7910 * reload.c (push_reload): Likewise.
7911 (combine_reloads): Likewise.
7912 (find_dummy_reload): Likewise.
7913 (operands_match_p): Likewise.
7914 (find_reloads): Likewise.
7915 (find_equiv_reg): Likewise.
7916 (reload_adjust_reg_for_mode): Likewise.
7917 * reload1.c (count_pseudo): Likewise.
7918 (count_spilled_pseudo): Likewise.
7919 (find_reg): Likewise.
7920 (clear_reload_reg_in_use): Likewise.
7921 (free_for_value_p): Likewise.
7922 (allocate_reload_reg): Likewise.
7923 (choose_reload_regs): Likewise.
7924 (reload_adjust_reg_for_temp): Likewise.
7925 (emit_reload_insns): Likewise.
7926 (delete_output_reload): Likewise.
7927 * rtlanal.c (subreg_get_info): Likewise.
7928 * sched-deps.c (sched_analyze_reg): Likewise.
7929 * sel-sched.c (init_regs_for_mode): Likewise.
7930 (mark_unavailable_hard_regs): Likewise.
7931 (choose_best_reg_1): Likewise.
7932 (verify_target_availability): Likewise.
7933 * valtrack.c (dead_debug_insert_temp): Likewise.
7934 * var-tracking.c (track_loc_p): Likewise.
7935 (emit_note_insn_var_location): Likewise.
7936 * varasm.c (make_decl_rtl): Likewise.
7937 * reginfo.c (choose_hard_reg_mode): Likewise.
7938 (init_reg_modes_target): Refer directly to
7939 this_target_regs->x_hard_regno_nregs.
7941 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7943 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
7944 instead of hard_regno_nregs.
7946 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7948 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
7949 end_hard_regno instead of hard_regno_nregs.
7950 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
7951 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
7952 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
7953 * ira-color.c (improve_allocation): Likewise.
7954 * lra-assigns.c (find_hard_regno_for_1): Likewise.
7955 * lra-lives.c (mark_regno_live): Likewise.
7956 (mark_regno_dead): Likewise.
7957 * lra-remat.c (operand_to_remat): Likewise.
7958 * lra.c (collect_non_operand_hard_regs): Likewise.
7959 * postreload.c (reload_combine_note_store): Likewise.
7960 (move2add_valid_value_p): Likewise.
7961 * reload.c (regno_clobbered_p): Likewise.
7963 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7965 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
7967 * config/v850/v850.c (v850_reorg): Likewise.
7968 * reload.c (refers_to_regno_for_reload_p): Likewise.
7969 (find_equiv_reg): Likewise.
7970 * reload1.c (reload_reg_reaches_end_p): Likewise.
7972 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7974 * caller-save.c (add_used_regs): Use REG_NREGS instead of
7976 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
7977 * config/arm/arm.c (output_move_neon): Likewise.
7978 (arm_attr_length_move_neon): Likewise.
7979 (neon_split_vcombine): Likewise.
7980 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
7981 (c6x_mark_reg_written): Likewise.
7982 (c6x_dwarf_register_span): Likewise.
7983 * config/i386/i386.c (ix86_save_reg): Likewise.
7984 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
7985 (rws_access_reg): Likewise.
7986 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
7987 * mode-switching.c (create_pre_exit): Likewise.
7988 * ree.c (combine_reaching_defs): Likewise.
7989 (add_removable_extension): Likewise.
7990 * regcprop.c (find_oldest_value_reg): Likewise.
7991 (copyprop_hardreg_forward_1): Likewise.
7992 * reload.c (reload_inner_reg_of_subreg): Likewise.
7993 (push_reload): Likewise.
7994 (combine_reloads): Likewise.
7995 (find_dummy_reload): Likewise.
7996 (reload_adjust_reg_for_mode): Likewise.
7997 * reload1.c (find_reload_regs): Likewise.
7998 (forget_old_reloads_1): Likewise.
7999 (reload_reg_free_for_value_p): Likewise.
8000 (reload_adjust_reg_for_temp): Likewise.
8001 (emit_reload_insns): Likewise.
8002 (delete_output_reload): Likewise.
8003 * sel-sched.c (choose_best_reg_1): Likewise.
8004 (choose_best_pseudo_reg): Likewise.
8006 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8007 Alan Hayward <alan.hayward@arm.com>
8008 David Sherwood <david.sherwood@arm.com>
8010 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
8011 * target.def (slow_unaligned_access): New hook.
8012 * targhooks.h (default_slow_unaligned_access): Declare.
8013 * targhooks.c (default_slow_unaligned_access): New function.
8014 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
8015 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
8016 * doc/tm.texi: Regenerate.
8017 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
8018 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
8019 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
8021 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
8022 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
8024 (rs6000_slow_unaligned_access): New function.
8025 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8026 (expand_block_compare): Likewise.
8027 (expand_strn_compare): Likewise.
8028 (rs6000_rtx_costs): Likewise.
8029 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
8030 (riscv_slow_unaligned_access): Likewise.
8031 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
8032 (riscv_slow_unaligned_access_p): ...this and make static.
8033 (riscv_option_override): Update accordingly.
8034 (riscv_slow_unaligned_access): New function.
8035 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8036 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
8037 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8038 (rs6000_slow_unaligned_access): New function.
8039 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8040 (rs6000_rtx_costs): Likewise.
8041 * config/rs6000/rs6000-string.c (expand_block_compare)
8042 (expand_strn_compare): Use targetm.slow_unaligned_access instead
8043 of SLOW_UNALIGNED_ACCESS.
8044 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
8045 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
8046 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
8047 of SLOW_UNALIGNED_ACCESS.
8048 * expmed.c (simple_mem_bitfield_p): Likewise.
8049 * expr.c (alignment_for_piecewise_move): Likewise.
8050 (emit_group_load_1): Likewise.
8051 (emit_group_store): Likewise.
8052 (copy_blkmode_from_reg): Likewise.
8053 (emit_push_insn): Likewise.
8054 (expand_assignment): Likewise.
8055 (store_field): Likewise.
8056 (expand_expr_real_1): Likewise.
8057 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
8058 * lra-constraints.c (simplify_operand_subreg): Likewise.
8059 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
8060 * gimple-ssa-store-merging.c: Likewise in block comment at start
8062 * tree-ssa-strlen.c: Include target.h.
8063 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
8064 of SLOW_UNALIGNED_ACCESS.
8065 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
8067 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8069 PR rtl-optimization/82185
8070 * expmed.c (emit_store_flag_int): Only test tem if it has been
8073 2017-09-12 Richard Biener <rguenther@suse.de>
8076 * match.pd ((FTYPE) N CMP CST): Fix typo.
8078 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
8080 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
8082 (mips_near_type_p): Add 'short_call' attribute as a synonym
8084 * doc/extend.texi (short_call): Document new function attribute.
8086 2017-09-12 Jakub Jelinek <jakub@redhat.com>
8089 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
8090 assertion check that in the condition.
8091 (get_atomic_generic_size): Likewise. Before testing if parameter
8092 has pointer type, if it has array type, call for C++
8093 default_conversion to perform array-to-pointer conversion.
8095 2017-09-12 Richard Biener <rguenther@suse.de>
8097 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
8098 for operations we cannot scalarize.
8100 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
8102 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
8103 vectors heap vectors. Clean up comments.
8104 Make visited_bbs a reference.
8105 (profitable_jump_thread_path): Make GC
8106 vectors heap vectors. Clean up comments.
8108 (convert_and_register_jump_thread_path): Make GC vectors heap
8110 (check_subpath_and_update_thread_path): Same. Clean up comments.
8111 Make visited_bbs a reference.
8112 (handle_phi): Abstract common code to to
8113 register_jump_thread_path_if_profitable.
8114 Rename VAR_BB to DEF_BB.
8116 Make GC vectors heap vectors.
8117 Make visited_bbs a reference.
8118 (handle_assignment): Same.
8119 (register_jump_thread_path_if_profitable): New.
8120 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
8122 Make GC vectors heap vectors. Clean up comments.
8123 Make visited_bbs a reference.
8124 (find_jump_threads_backwards): Make visited_bbs live in the stack.
8125 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
8128 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
8131 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
8132 words of E_DImode object are reachable by xtensa_uimm8x4 access.
8134 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
8136 Revert r251800 and r251799.
8138 2017-09-11 Martin Jambor <mjambor@suse.cz>
8141 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
8142 arguments in advance.
8143 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
8144 use it to find predecessor edges.
8145 (naive_outof_ssa): Collect vector of predecessors.
8147 2017-09-08 Jason Merrill <jason@redhat.com>
8149 PR c++/70029 - ICE with ref-qualifier and -flto
8150 * langhooks.h (struct lang_hooks_for_types): Add
8151 copy_lang_qualifiers.
8152 * attribs.c (build_type_attribute_qual_variant): Use it.
8153 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
8155 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
8156 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
8158 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
8161 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
8162 (*mulsi3_sp64): New instruction.
8163 (mulsi3): New expander.
8165 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
8167 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
8169 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8171 * sancov.c: Include memmodel.h.
8173 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
8176 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
8179 2017-09-07 Carl Love <cel@us.ibm.com>
8181 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
8182 the sldi instruction.
8184 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
8186 * sancov.c: Include tm_p.h.
8188 2017-09-07 Jakub Jelinek <jakub@redhat.com>
8191 * output.h (switch_to_other_text_partition): New declaration.
8192 * varasm.c (switch_to_other_text_partition): New function.
8193 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
8194 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
8195 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
8196 to the other text partition before emitting LCL label and switch back
8197 after emitting the word after it.
8199 2017-09-07 Richard Biener <rguenther@suse.de>
8201 * passes.def (pass_split_crit_edges): Remove instance before PRE.
8202 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
8203 critical edges here, after loop init.
8204 (pass_data_pre): Remove PROP_no_crit_edges flags.
8205 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
8206 for valueization of call args to avoid leaking VN_TOP.
8207 (visit_use): Assert we do not visit default defs.
8208 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
8209 Use error_mark_node to more easily detect leaking VN_TOP.
8210 All default-defs are varying, not VN_TOP. Mark them visited.
8211 (run_scc_vn): Make code match comment.
8213 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8215 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
8216 OPTION_MASK_FLOAT128_KEYWORD.
8217 (POWERPC_MASKS): Likewise.
8218 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
8219 support for the -mfloat128-type option, and make -mfloat128
8220 default on PowerPC Linux systems. Define or undefine
8221 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
8222 Define __float128 to be __ieee128 if IEEE 128-bit support is
8223 enabled, or undefine it.
8224 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
8225 Delete defining __FLOAT128_TYPE__.
8226 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
8227 -mfloat128-type option and make -mfloat128 default on PowerPC
8229 (TARGET_FLOAT128_TYPE): Likewise.
8230 (-mfloat128-type): Likewise.
8231 * config/rs6000/rs6000.c (rs6000_option_override_internal):
8232 Delete the -mfloat128-type option and make -mfloat128 default on
8233 PowerPC Linux systems. Always use __ieee128 to be the keyword for
8234 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
8235 128-bit floating point is enabled. Change tests from using
8236 -mfloat128-type to -mfloat128.
8237 (rs6000_mangle_type): Use the correct mangling for the __float128
8238 type even if normal long double is restricted to 64-bits.
8239 (floatn_mode): Enable the _Float128 type by default on VSX Linux
8241 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
8242 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
8243 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
8245 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
8246 documentation for -mfloat128.
8248 2017-09-06 Olivier Hainque <hainque@adacore.com>
8250 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
8252 2017-09-06 Wish Wu <wishwu007@gmail.com>
8253 Jakub Jelinek <jakub@redhat.com>
8255 * asan.c (initialize_sanitizer_builtins): Add
8256 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
8257 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
8258 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
8259 BT_FN_VOID_UINT64_PTR variables.
8260 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
8261 (BT_FN_VOID_UINT16_UINT16): Likewise.
8262 (BT_FN_VOID_UINT32_UINT32): Likewise.
8263 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
8264 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
8265 (BT_FN_VOID_UINT64_PTR): Likewise.
8266 * common.opt (flag_sanitize_coverage): New variable.
8267 (fsanitize-coverage=trace-pc): Remove.
8268 (fsanitize-coverage=): Add.
8269 * flag-types.h (enum sanitize_coverage_code): New enum.
8270 * fold-const.c (fold_range_test): Disable non-short-circuit
8271 optimization if flag_sanitize_coverage.
8272 (fold_truth_andor): Likewise.
8273 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
8274 * opts.c (COVERAGE_SANITIZER_OPT): Define.
8275 (coverage_sanitizer_opts): New array.
8276 (get_closest_sanitizer_option): Add OPTS argument, handle also
8277 OPT_fsanitize_coverage_.
8278 (parse_sanitizer_options): Adjusted to also handle
8279 OPT_fsanitize_coverage_.
8280 (common_handle_option): Add OPT_fsanitize_coverage_.
8281 * sancov.c (instrument_comparison, instrument_switch): New function.
8282 (sancov_pass): Add trace-cmp support.
8283 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
8284 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
8285 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
8286 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
8287 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
8288 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
8289 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
8290 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
8291 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
8292 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
8294 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
8296 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
8297 error. Only quit immediately if parsing is complete.
8298 (BEGIN): Initialize fatal_err and parse_done.
8299 (begin fpu, end fpu): Check number of arguments.
8300 (begin arch, end arch): Likewise.
8301 (begin cpu, end cpu): Likewise.
8302 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
8303 (optalias): Likewise.
8305 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
8307 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
8308 * config/arm/arm-isa.h: Delete. Move definitions to ...
8309 * arm-cpus.in: ... here. Use new feature and fgroup values.
8310 * config/arm/arm.c (arm_option_override): Use lower case for feature
8312 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
8313 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
8314 * config/arm/parsecpu.awk (END): Add new command 'isa'.
8316 (print_isa_bits_for): New function.
8317 (gen_isa): New function.
8318 (gen_comm_data): Use print_isa_bits_for.
8319 (define feature): New keyword.
8320 (define fgroup): New keyword.
8321 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
8322 (arm-isa.h): Add rule to generate file.
8323 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
8324 case for feature bit names.
8326 2017-09-06 Richard Biener <rguenther@suse.de>
8328 * tree-ssa-pre.c (NECESSARY): Remove.
8329 (create_expression_by_pieces): Do not touch pass-local flags.
8330 (insert_into_preds_of_block): Likewise.
8331 (do_pre_regular_insertion): Likewise.
8332 (eliminate_insert): Likewise.
8333 (eliminate_dom_walker::before_dom_children): Likewise.
8334 (fini_eliminate): Do not look at inserted_exprs.
8335 (mark_operand_necessary): Remove.
8336 (remove_dead_inserted_code): Replace with simple work-list
8337 algorithm based on inserted_exprs and SSA uses.
8338 (pass_pre::execute): Re-order fini_eliminate and
8339 remove_dead_inserted_code.
8341 2017-09-06 Olivier Hainque <hainque@adacore.com>
8343 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
8344 for VxWorks 7. Adjust surrounding comments.
8346 2017-09-06 Richard Biener <rguenther@suse.de>
8348 * gimple-ssa-strength-reduction.c
8349 (find_candidates_dom_walker::before_dom_children): Also allow
8352 2017-09-06 Richard Biener <rguenther@suse.de>
8354 PR tree-optimization/82108
8355 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
8356 for gap in the non-permutation SLP case.
8358 2017-09-06 Martin Jambor <mjambor@suse.cz>
8360 PR tree-optimization/82078
8361 * tree-sra.c (sort_and_splice_var_accesses): Move call to
8362 add_access_to_work_queue...
8363 (build_accesses_from_assign): ...here.
8364 (propagate_all_subaccesses): Make sure racc is the group
8365 representative, if there is one.
8367 2017-09-06 Jakub Jelinek <jakub@redhat.com>
8370 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
8373 2017-09-06 Richard Biener <rguenther@suse.de>
8375 * gimple-ssa-strength-reduction.c
8376 (find_candidates_dom_walker::before_doom_children): Use a
8377 type and not a mode check.
8379 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
8382 * config/arm/predicates.md (arm_general_adddi_operand): Create new
8384 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
8386 2017-09-05 Jeff Law <law@redhat.com>
8388 PR tree-optimization/64910
8389 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
8390 cases where we have 3 or more operands.
8392 2017-09-05 Jakub Jelinek <jakub@redhat.com>
8395 * omp-low.c (lower_omp_for): Recompute tree invariant if
8396 gimple_omp_for_initial/final is ADDR_EXPR.
8399 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
8400 into gimple val before gimplification fo the COND_EXPR.
8402 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
8404 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
8405 REGION_COPY argument.
8406 (thread_through_all_blocks): Remove unused argument to
8407 duplicate_thread_path.
8409 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8410 Alan Hayward <alan.hayward@arm.com>
8411 David Sherwood <david.sherwood@arm.com>
8413 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
8414 Take a scalar_mode rather than a machine_mode.
8415 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8416 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
8417 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8418 (aarch64_gen_adjusted_ldpstp): Likewise.
8419 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
8421 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8422 Alan Hayward <alan.hayward@arm.com>
8423 David Sherwood <david.sherwood@arm.com>
8425 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
8426 Take a scalar_int_mode instead of a machine_mode.
8427 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8428 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8429 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8430 (aarch64_simd_attr_length_rglist): Delete.
8431 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
8432 a scalar_int_mode instead of a machine_mode.
8433 (aarch64_add_offset): Likewise.
8434 (aarch64_internal_mov_immediate): Likewise
8435 (aarch64_add_constant_internal): Likewise.
8436 (aarch64_add_constant): Likewise.
8437 (aarch64_movw_imm): Likewise.
8438 (aarch64_rtx_arith_op_extract_p): Likewise.
8439 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8440 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8441 Remove assert that the mode isn't a vector.
8442 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8443 (aarch64_expand_mov_immediate): Update calls after above changes.
8444 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
8445 (aarch64_and_bitmask_imm): Check for scalar integer modes.
8446 (aarch64_move_imm): Likewise.
8447 (aarch64_can_const_movi_rtx_p): Likewise.
8448 (aarch64_strip_extend): Likewise.
8449 (aarch64_extr_rtx_p): Likewise.
8450 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
8451 a CONST_INT when the mode parameter is VOIDmode.
8452 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
8454 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8456 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
8457 * stor-layout.c (bitwise_mode_for_mode): Likewise.
8458 (bitwise_type_for_mode): Update accordingly.
8460 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8462 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
8463 * stor-layout.c (mode_for_size_tree): Likewise.
8464 (mode_for_array): Update accordingly.
8465 (layout_decl): Likewise.
8466 (compute_record_mode): Likewise. Only set the mode once.
8468 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8470 * target.def (get_mask_mode): Change return type to opt_mode.
8472 * doc/tm.texi: Regenerate.
8473 * targhooks.h (default_get_mask_mode): Return an opt_mode.
8474 * targhooks.c (default_get_mask_mode): Likewise.
8475 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
8476 * optabs-query.c (can_vec_mask_load_store_p): Update use of
8477 targetm.get_mask_mode.
8478 * tree.c (build_truth_vector_type): Likewise.
8480 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8482 * machmode.h (mode_for_vector): Return an opt_mode.
8483 * stor-layout.c (mode_for_vector): Likewise.
8484 (mode_for_int_vector): Update accordingly.
8485 (layout_type): Likewise.
8486 * config/i386/i386.c (emit_memmov): Likewise.
8487 (ix86_expand_set_or_movmem): Likewise.
8488 (ix86_expand_vector_init): Likewise.
8489 (ix86_get_mask_mode): Likewise.
8490 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
8492 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
8493 * expmed.c (extract_bit_field_1): Likewise.
8494 * expr.c (expand_expr_real_2): Likewise.
8495 * optabs-query.c (can_vec_perm_p): Likewise.
8496 (can_vec_mask_load_store_p): Likewise.
8497 * optabs.c (expand_vec_perm): Likewise.
8498 * targhooks.c (default_get_mask_mode): Likewise.
8499 * tree-vect-stmts.c (vectorizable_store): Likewise.
8500 (vectorizable_load): Likewise.
8501 (get_vectype_for_scalar_type_and_size): Likewise.
8503 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8505 * machmode.h (mode_for_int_vector): New function.
8506 * stor-layout.c (mode_for_int_vector): Likewise.
8507 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
8508 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
8509 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
8510 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
8511 (s390_expand_vcond): Likewise.
8513 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8515 * machmode.h (opt_machine_mode): New type.
8516 (opt_mode<T>): Allow construction from anything that can be
8518 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
8519 (mode_for_size): Return an opt_machine_mode.
8520 * stor-layout.c (mode_for_size): Likewise.
8521 (mode_for_size_tree): Update call accordingly.
8522 (bitwise_mode_for_mode): Likewise.
8523 (make_fract_type): Likewise.
8524 (make_accum_type): Likewise.
8525 * caller-save.c (replace_reg_with_saved_mem): Update call
8527 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8528 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8529 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8530 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8531 * expmed.c (extract_bit_field_1): Likewise.
8532 * reload.c (get_secondary_mem): Likewise.
8533 * varasm.c (assemble_integer): Likewise.
8534 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
8537 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8539 * machmode.h (decimal_float_mode_for_size): New function.
8540 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
8541 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
8542 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
8543 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
8544 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
8545 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
8547 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8549 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
8550 (get_builtin_sync_mode): Likewise.
8551 (expand_ifn_atomic_compare_exchange): Likewise.
8552 (expand_builtin_atomic_clear): Likewise.
8553 (expand_builtin_atomic_test_and_set): Likewise.
8554 (fold_builtin_atomic_always_lock_free): Likewise.
8555 * calls.c (compute_argument_addresses): Likewise.
8556 (emit_library_call_value_1): Likewise.
8557 (store_one_arg): Likewise.
8558 * combine.c (combine_instructions): Likewise.
8559 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
8560 * config/arm/arm.c (arm_function_value): Likewise.
8561 (aapcs_allocate_return_reg): Likewise.
8562 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
8563 * config/i386/i386.c (construct_container): Likewise.
8564 (ix86_gimplify_va_arg): Likewise.
8565 (ix86_expand_sse_cmp): Likewise.
8566 (emit_memmov): Likewise.
8567 (emit_memset): Likewise.
8568 (expand_small_movmem_or_setmem): Likewise.
8569 (ix86_expand_pextr): Likewise.
8570 (ix86_expand_pinsr): Likewise.
8571 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
8572 * config/microblaze/microblaze.c (microblaze_block_move_straight):
8574 * config/mips/mips.c (mips_function_value_1) Likewise.
8575 (mips_block_move_straight): Likewise.
8576 (mips_expand_ins_as_unaligned_store): Likewise.
8577 * config/powerpcspe/powerpcspe.c
8578 (rs6000_darwin64_record_arg_advance_flush): Likewise.
8579 (rs6000_darwin64_record_arg_flush): Likewise.
8580 * config/rs6000/rs6000.c
8581 (rs6000_darwin64_record_arg_advance_flush): Likewise.
8582 (rs6000_darwin64_record_arg_flush): Likewise.
8583 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
8584 (sparc_function_value_1): Likewise.
8585 * config/spu/spu.c (adjust_operand): Likewise.
8586 (spu_emit_branch_or_set): Likewise.
8587 (arith_immediate_p): Likewise.
8588 * emit-rtl.c (gen_lowpart_common): Likewise.
8589 * expr.c (expand_expr_real_1): Likewise.
8590 * function.c (assign_parm_setup_block): Likewise.
8591 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
8592 * reload1.c (alter_reg): Likewise.
8593 * stor-layout.c (mode_for_vector): Likewise.
8594 (layout_type): Likewise.
8596 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8598 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
8599 (spu_convert_move): Likewise.
8600 * lower-subreg.c (resolve_simple_move): Likewise.
8602 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8605 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
8606 define_insn to a define_expand.
8607 (altivec_vsum2sws_direct): New define_insn.
8608 (altivec_vsumsws): Convert from a define_insn to a define_expand.
8610 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
8612 * config/arm/arm.c (arm_option_params_internal): Improve setting of
8615 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
8620 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
8621 realign stack if stack alignment needed is less than incoming
8624 2017-09-05 Marek Polacek <polacek@redhat.com>
8627 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
8630 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
8632 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
8634 2017-09-05 Richard Biener <rguenther@suse.de>
8636 PR tree-optimization/82084
8637 * fold-const.c (can_native_encode_string_p): Handle wide characters.
8639 2017-09-05 Richard Biener <rguenther@suse.de>
8641 PR tree-optimization/82102
8642 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
8644 2017-09-05 Martin Liska <mliska@suse.cz>
8646 PR tree-optimization/82032
8647 * tree-cfg.c (generate_range_test): New function.
8648 * tree-cfg.h (generate_range_test): Declared here.
8649 * tree-cfgcleanup.c (convert_single_case_switch): New function.
8650 (cleanup_control_expr_graph): Use it.
8651 * tree-switch-conversion.c (try_switch_expansion): Remove
8653 (emit_case_nodes): Use generate_range_test.
8655 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
8658 * config/i386/i386.md (*<btsc><mode>_mask): Add
8659 TARGET_USE_BT to insn constraint.
8660 (*btr<mode>_mask): Ditto.
8662 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
8664 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
8665 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
8667 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
8670 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
8671 TARGET_NEON and TARGET_IWMMXT.
8672 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
8673 TARGET_NEON and TARGET_IWMMXT.
8674 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
8676 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
8678 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
8679 (ix86_rewrite_tls_address): Ditto.
8680 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
8681 (ix86_rewrite_tls_address_1): Ditto.
8682 (ix86_rewrite_tls_address): Ditto.
8683 * config/i386/predicates.md (tls_address_pattern): New predicate.
8684 * config/i386/i386.md (TLS address splitter): New splitter.
8686 2017-09-04 Richard Biener <rguenther@suse.de>
8688 PR tree-optimization/82084
8689 * fold-const.h (can_native_encode_string_p): Declare.
8690 * fold-const.c (can_native_encode_string_p): Factor out from ...
8691 (native_encode_string): ... here.
8692 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
8693 vectorizing stores from constants we later cannot handle.
8695 2017-09-04 Marek Polacek <polacek@redhat.com>
8698 * doc/invoke.texi: Update -Wtautological-compare documentation.
8700 2017-09-04 Jeff Law <law@redhat.com>
8702 PR tree-optimization/64910
8703 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
8704 swap the first and last operand if the last is a constant.
8706 2017-09-04 Marek Polacek <polacek@redhat.com>
8709 * convert.c (do_narrow): When sanitizing signed integer overflows,
8710 bail out for signed types.
8711 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
8713 2017-09-04 Richard Biener <rguenther@suse.de>
8715 PR tree-optimization/82060
8716 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
8717 Move devirtualization after stmt folding and before EH/AB/noreturn
8718 cleanup to get the stmt refs canonicalized. Use a bool instead
8719 of gimple_modified_p since that doesn't work for NOPs. Schedule
8720 NOPs generated by folding for removal.
8722 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8723 Alan Hayward <alan.hayward@arm.com>
8724 David Sherwood <david.sherwood@arm.com>
8726 * coretypes.h (pad_direction): New enum.
8727 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
8728 (FUNCTION_ARG_PADDING): Likewise.
8729 * target.def (function_arg_padding): New hook.
8730 * targhooks.h (default_function_arg_padding): Declare.
8731 * targhooks.c (default_function_arg_padding): New function.
8732 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
8733 (TARGET_FUNCTION_ARG_PADDING): ...this.
8734 * doc/tm.texi: Regenerate.
8735 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
8736 instead of direction.
8737 (compute_argument_addresses): Likewise.
8738 (load_register_parameters): Likewise.
8739 (emit_library_call_value_1): Likewise.
8740 (store_one_arg): Use targetm.calls.function_arg_padding instead
8741 of FUNCTION_ARG_PADDING.
8742 (must_pass_in_stack_var_size_or_pad): Likewise.
8743 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
8744 (emit_group_store): Likewise.
8745 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
8746 instead of FUNCTION_ARG_PADDING.
8747 (emit_push_insn): Likewise, and propagate enum change throughout
8749 * function.h (direction): Delete.
8750 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
8752 * function.c (assign_parm_find_stack_rtl): Likewise.
8753 (assign_parm_setup_block_p): Likewise.
8754 (assign_parm_setup_block): Likewise.
8755 (gimplify_parameters): Likewise.
8756 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
8757 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
8759 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
8760 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8761 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
8762 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
8763 (aarch64_function_arg_padding): ...this new function.
8764 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
8765 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8766 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
8767 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8768 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
8769 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8770 (arm_pad_arg_upward): Replace with...
8771 (arm_function_arg_padding): ...this new function.
8772 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
8774 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
8775 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
8776 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8777 (ia64_hpux_function_arg_padding): Replace with...
8778 (ia64_function_arg_padding): ...this new function. Use pad_direction
8779 instead of direction. Check for TARGET_HPUX.
8780 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
8781 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8782 (iq2000_function_arg_padding): New function.
8783 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
8784 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
8785 (mips_function_arg_padding): ...this new function.
8786 (mips_pad_reg_upward): Update accordingly.
8787 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8788 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
8789 targetm.calls.function_arg_padding.
8790 (FUNCTION_ARG_PADDING): Delete.
8791 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8792 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
8793 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8794 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
8795 (nios2_block_reg_padding): Return pad_direction instead of direction.
8796 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
8797 instead of direction.
8798 (nios2_function_arg_padding): Likewise. Make static.
8799 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8800 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
8801 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
8802 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
8803 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8804 (pa_function_arg_padding): Make static. Return pad_direction instead
8806 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
8807 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8808 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
8809 instead of direction. Use targetm.calls.function_arg_padding.
8810 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
8811 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
8812 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
8813 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
8814 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
8816 (function_arg_padding): Rename to...
8817 (rs6000_function_arg_padding): ...this. Make static. Return
8818 pad_direction instead of direction.
8819 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
8820 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
8821 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8822 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
8823 instead of direction. Use targetm.calls.function_arg_padding.
8824 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
8825 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
8826 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
8827 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
8828 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8829 (function_arg_padding): Rename to...
8830 (rs6000_function_arg_padding): ...this. Make static. Return
8831 pad_direction instead of direction.
8832 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
8833 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
8834 * config/s390/s390.c (s390_function_arg_padding): New function.
8835 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8836 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
8837 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
8838 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8839 (function_arg_padding): Rename to...
8840 (sparc_function_arg_padding): ...this. Make static. Return
8841 pad_direction instead of direction.
8842 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
8843 * config/spu/spu.c (spu_function_arg_padding): New function.
8844 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8845 * system.h (FUNCTION_ARG_PADDING): Poison.
8847 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8848 Alan Hayward <alan.hayward@arm.com>
8849 David Sherwood <david.sherwood@arm.com>
8851 * target.def (modes_tieable_p): New hook.
8852 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
8853 (TARGET_MODES_TIEABLE_P): ...this.
8854 * doc/tm.texi.in: Regenerate.
8855 * hooks.h (hook_bool_mode_mode_true): Declare.
8856 * hooks.c (hook_bool_mode_mode_true): New function.
8857 * combine.c (subst): Use targetm.modes_tieable_p instead of
8859 * dse.c (find_shift_sequence): Likewise.
8860 * expmed.c (extract_low_bits): Likewise.
8861 * lower-subreg.c: Include target.h.
8862 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
8864 * rtlanal.c (rtx_cost): Likewise.
8865 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
8866 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
8867 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
8868 (TARGET_MODES_TIEABLE_P): Redefine.
8869 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
8870 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
8871 (TARGET_MODES_TIEABLE_P): Redefine.
8872 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
8873 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
8874 (arc_modes_tieable_p): New function.
8875 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
8876 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
8877 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
8878 (arm_modes_tieable_p): Make static.
8879 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
8880 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
8881 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
8882 (TARGET_MODES_TIEABLE_P): Redefine.
8883 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
8884 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
8885 (TARGET_MODES_TIEABLE_P): Redefine.
8886 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
8887 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
8888 (cr16_modes_tieable_p): New function.
8889 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
8890 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
8891 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
8892 (TRULY_NOOP_TRUNCATION): Update comment.
8893 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
8894 (TRULY_NOOP_TRUNCATION): Update comment.
8895 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
8896 (frv_modes_tieable_p): New function.
8897 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
8898 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
8899 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
8900 (TARGET_MODES_TIEABLE_P): Redefine.
8901 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
8902 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
8903 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
8904 (TARGET_MODES_TIEABLE_P): Redefine.
8905 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
8906 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
8907 (ia64_modes_tieable_p): New function.
8908 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
8909 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
8910 (iq2000_modes_tieable_p): New function.
8911 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
8912 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
8913 (lm32_modes_tieable_p): New function.
8914 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
8915 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
8916 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
8917 (TARGET_MODES_TIEABLE_P): Redefine.
8918 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
8919 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
8920 (m32r_modes_tieable_p): New function.
8921 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
8922 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
8923 (m68k_modes_tieable_p): New function.
8924 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
8925 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
8926 (mcore_modes_tieable_p): New function.
8927 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
8928 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
8930 (TARGET_MODES_TIEABLE_P): Redefine.
8931 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
8932 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
8933 * config/mips/mips.c (mips_modes_tieable_p): Make static.
8934 (TARGET_MODES_TIEABLE_P): Redefine.
8935 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
8936 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
8937 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
8938 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
8939 (mn10300_modes_tieable_p): ...this and make static.
8940 (TARGET_MODES_TIEABLE_P): Redefine.
8941 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
8942 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
8943 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
8944 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
8945 (msp430_modes_tieable_p): Make static.
8946 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
8947 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
8948 (TARGET_MODES_TIEABLE_P): Redefine.
8949 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
8950 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
8951 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
8952 (TARGET_MODES_TIEABLE_P): Redefine.
8953 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
8954 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
8955 * config/pa/pa.c (pa_modes_tieable_p): Make static.
8956 (TARGET_MODES_TIEABLE_P): Redefine.
8957 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
8958 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
8959 (pdp11_modes_tieable_p): New function.
8960 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
8961 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
8962 (rs6000_modes_tieable_p): New function.
8963 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
8964 * config/powerpcspe/powerpcspe.md: Update comment.
8965 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
8966 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
8967 (TARGET_MODES_TIEABLE_P): Redefine.
8968 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
8969 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
8970 (rl78_modes_tieable_p): New function.
8971 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
8972 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
8973 (rs6000_modes_tieable_p): New function.
8974 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
8975 * config/rs6000/rs6000.md: Update comment.
8976 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
8977 * config/rx/rx.c (rx_modes_tieable_p): New function.
8978 (TARGET_MODES_TIEABLE_P): Redefine.
8979 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
8980 * config/s390/s390.c (s390_modes_tieable_p): New function.
8981 (TARGET_MODES_TIEABLE_P): Redefine.
8982 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
8983 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
8984 (sh_modes_tieable_p): New function.
8985 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
8986 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
8987 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
8988 (sparc_modes_tieable_p): Make static.
8989 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
8990 * config/spu/spu.c (spu_modes_tieable_p): New function.
8991 (TARGET_MODES_TIEABLE_P): Redefine.
8992 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
8993 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
8994 (TARGET_MODES_TIEABLE_P): Redefine.
8995 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
8996 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
8997 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
8998 * config/v850/v850.c (v850_modes_tieable_p): New function.
8999 (TARGET_MODES_TIEABLE_P): Redefine.
9000 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
9001 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
9002 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
9003 (visium_modes_tieable_p): New function.
9004 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
9005 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
9006 (xtensa_modes_tieable_p): New function.
9007 * system.h (MODES_TIEABLE_P): Poison.
9009 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9010 Alan Hayward <alan.hayward@arm.com>
9011 David Sherwood <david.sherwood@arm.com>
9013 * target.def (hard_regno_mode_ok): New hook.
9014 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
9015 (TARGET_HARD_REGNO_MODE_OK): ...this.
9016 * doc/tm.texi.in: Regenerate.
9017 * hooks.h (hook_bool_uint_mode_true): Declare.
9018 * hooks.c (hook_bool_uint_mode_true): New function.
9019 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
9021 * genpreds.c (write_insn_preds_c): Add an include of target.h.
9022 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
9023 instead of HARD_REGNO_MODE_OK.
9024 * caller-save.c: Include target.h.
9025 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
9027 * combine.c (can_combine_p): Likewise.
9028 (combinable_i3pat): Likewise.
9029 (can_change_dest_mode): Likewise.
9030 * expr.c (init_expr_target): Likewise.
9031 (convert_move): Likewise.
9032 (convert_modes): Likewise.
9033 * ira.c (setup_prohibited_class_mode_regs): Likewise.
9034 (setup_prohibited_mode_move_regs): Likewise.
9035 * ira.h (target_ira): Likewise.
9036 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9037 * lra-constraints.c (process_alt_operands): Likewise.
9038 (split_reg): Likewise.
9039 * recog.c (peep2_find_free_register): Likewise.
9040 * ree.c (combine_reaching_defs): Likewise.
9041 * regcprop.c (maybe_mode_change): Likewise.
9042 * reginfo.c (init_reg_sets_1): Likewise.
9043 (choose_hard_reg_mode): Likewise.
9044 (simplifiable_subregs): Likewise.
9045 * regrename.c (check_new_reg_p): Likewise.
9046 * reload.c (find_valid_class): Likewise.
9047 (find_valid_class_1): Likewise.
9048 (reload_inner_reg_of_subreg): Likewise.
9049 (push_reload): Likewise.
9050 (combine_reloads): Likewise.
9051 (find_dummy_reload): Likewise.
9052 (find_reloads): Likewise.
9053 * reload1.c (find_reg): Likewise.
9054 (set_reload_reg): Likewise.
9055 (allocate_reload_reg): Likewise.
9056 (choose_reload_regs): Likewise.
9057 (reload_adjust_reg_for_temp): Likewise.
9058 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
9059 (simplify_subreg_regno): Likewise.
9060 * sel-sched.c (init_regs_for_mode): Likewise.
9061 * varasm.c (make_decl_rtl): Likewise.
9062 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
9063 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
9065 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
9066 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
9067 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9068 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
9069 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
9070 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9071 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
9072 (arc_mode_class): Delete.
9073 (HARD_REGNO_MODE_OK): Delete.
9074 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9075 (arc_hard_regno_mode_ok): Rename old array to...
9076 (arc_hard_regno_mode_ok_modes): ...this.
9077 (arc_conditional_register_usage): Update accordingly.
9078 (arc_mode_class): Make static.
9079 (arc_hard_regno_mode_ok): New function.
9080 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
9081 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
9082 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9083 (arm_hard_regno_mode_ok): Make static.
9084 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
9086 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
9087 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
9088 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
9090 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9091 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
9092 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
9093 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
9094 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
9095 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9096 * config/bfin/predicates.md (valid_reg_operand): Use
9097 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
9098 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
9099 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
9100 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9101 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
9102 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
9103 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9104 (cr16_hard_regno_mode_ok): Make static and return a bool.
9105 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
9106 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9107 (cris_hard_regno_mode_ok): New function.
9108 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
9109 (epiphany_mode_class): Delete.
9110 (HARD_REGNO_MODE_OK): Delete.
9111 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
9112 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9113 (hard_regno_mode_ok): Rename to...
9114 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
9115 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
9116 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
9118 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
9119 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
9120 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9121 (frv_hard_regno_mode_ok): Make static and return a bool.
9122 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
9124 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
9125 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
9126 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
9127 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
9129 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9130 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
9131 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
9132 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
9134 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9135 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
9136 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9137 (ia64_hard_regno_mode_ok): New function.
9138 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
9139 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9140 (iq2000_hard_regno_mode_ok): New function.
9141 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
9142 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9143 (lm32_hard_regno_mode_ok): New function.
9144 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
9145 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
9146 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
9147 instead of HARD_REGNO_MODE_OK.
9148 (m32c_hard_regno_ok): Rename to...
9149 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
9150 (m32c_cannot_change_mode_class): Update accordingly.
9151 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9152 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
9153 (m32r_mode_class): Delete.
9154 (HARD_REGNO_MODE_OK): Delete.
9155 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9156 (m32r_hard_regno_mode_ok): Rename to...
9157 (m32r_hard_regno_modes): ...this.
9158 (m32r_mode_class): Make static.
9159 (m32r_hard_regno_mode_ok): New function.
9160 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
9161 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
9162 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9163 (m68k_hard_regno_mode_ok): Make static.
9164 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
9165 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9166 (mcore_hard_regno_mode_ok): New function.
9167 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
9168 (HARD_REGNO_MODE_OK): Delete.
9169 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
9171 (microblaze_hard_regno_mode_ok_p): ...this and make static.
9172 (microblaze_hard_regno_mode_ok): New function.
9173 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9174 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
9175 (mips_hard_regno_mode_ok): Delete.
9176 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
9177 (mips_hard_regno_mode_ok_p): ...this and make static.
9178 (mips_hard_regno_mode_ok_p): Rename to...
9179 (mips_hard_regno_mode_ok_uncached): ...this.
9180 (mips_hard_regno_mode_ok): New function.
9181 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
9182 of HARD_REGNO_MODE_OK.
9183 (mips_option_override): Update after above name changes.
9184 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9185 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
9186 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
9187 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
9188 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
9189 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9190 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
9191 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
9192 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
9193 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9194 (msp430_hard_regno_mode_ok): Make static and return a bool.
9195 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
9196 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
9197 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
9199 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9200 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
9201 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
9202 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
9203 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
9204 (PA_HARD_REGNO_MODE_OK): ...this
9205 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
9206 (PA_HARD_REGNO_MODE_OK): ...this.
9207 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9208 (pa_hard_regno_mode_ok): New function.
9209 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
9210 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9211 (pdp11_hard_regno_mode_ok): New function.
9212 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
9213 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
9215 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
9217 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9218 (rs6000_hard_regno_mode_ok): Rename to...
9219 (rs6000_hard_regno_mode_ok_uncached): ...this.
9220 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9221 (rs6000_hard_regno_mode_ok): New function.
9222 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
9223 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
9224 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
9225 (riscv_hard_regno_mode_ok): ...this and make static.
9226 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9227 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
9228 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
9229 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9230 (rl78_hard_regno_mode_ok): Make static and return bool.
9231 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
9232 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
9234 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
9235 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9236 (rs6000_hard_regno_mode_ok): Rename to...
9237 (rs6000_hard_regno_mode_ok_uncached): ...this.
9238 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9239 (rs6000_hard_regno_mode_ok): New function.
9240 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
9241 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
9242 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9243 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
9244 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
9245 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
9246 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9247 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
9248 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
9249 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9250 (sh_hard_regno_mode_ok): Make static.
9251 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
9252 instead of HARD_REGNO_MODE_OK.
9253 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
9254 (sparc_mode_class): Delete.
9255 (HARD_REGNO_MODE_OK): Delete.
9256 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9257 (hard_regno_mode_classes): Make static.
9258 (sparc_mode_class): Likewise.
9259 (sparc_hard_regno_mode_ok): New function.
9260 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
9261 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
9262 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
9264 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9265 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
9266 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
9267 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
9268 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
9269 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9270 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
9271 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
9272 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9273 (visium_hard_regno_mode_ok): New function.
9274 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
9275 instead of HARD_REGNO_MODE_OK.
9276 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
9277 (HARD_REGNO_MODE_OK): Delete.
9278 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
9279 (xtensa_hard_regno_mode_ok_p): ...this and make static.
9280 (xtensa_option_override): Update accordingly.
9281 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9282 (xtensa_hard_regno_mode_ok): New function.
9283 * system.h (HARD_REGNO_MODE_OK): Poison.
9285 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9286 Alan Hayward <alan.hayward@arm.com>
9287 David Sherwood <david.sherwood@arm.com>
9289 * target.def (hard_regno_call_part_clobbered): New hook.
9290 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
9291 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
9292 * doc/tm.texi: Regenerate.
9293 * hooks.h (hook_bool_uint_mode_false): Declare.
9294 * hooks.c (hook_bool_uint_mode_false): New function.
9295 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9296 * cselib.c (cselib_process_insn): Use
9297 targetm.hard_regno_call_part_clobbered instead of
9298 HARD_REGNO_CALL_PART_CLOBBERED.
9299 * ira-conflicts.c (ira_build_conflicts): Likewise.
9300 * ira-costs.c (ira_tune_allocno_costs): Likewise.
9301 * lra-constraints.c (need_for_call_save_p): Likewise.
9302 * lra-lives.c: Include target.h.
9303 (check_pseudos_live_through_calls): Use
9304 targetm.hard_regno_call_part_clobbered instead of
9305 HARD_REGNO_CALL_PART_CLOBBERED.
9306 * regcprop.c: Include target.h.
9307 (copyprop_hardreg_forward_1): Use
9308 targetm.hard_regno_call_part_clobbered instead of
9309 HARD_REGNO_CALL_PART_CLOBBERED.
9310 * reginfo.c (choose_hard_reg_mode): Likewise.
9311 * regrename.c (check_new_reg_p): Likewise.
9312 * reload.c (find_equiv_reg): Likewise.
9313 * reload1.c (emit_reload_insns): Likewise.
9314 * sched-deps.c (deps_analyze_insn): Likewise.
9315 * sel-sched.c (init_regs_for_mode): Likewise.
9316 (mark_unavailable_hard_regs): Likewise.
9317 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
9318 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9319 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
9321 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9322 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9323 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
9325 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
9327 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9328 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9329 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
9331 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9332 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9333 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
9335 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9336 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
9338 * config/powerpcspe/powerpcspe.c
9339 (rs6000_hard_regno_call_part_clobbered): New function.
9340 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9341 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9342 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
9344 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9345 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9346 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
9348 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9349 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9350 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
9352 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9353 Alan Hayward <alan.hayward@arm.com>
9354 David Sherwood <david.sherwood@arm.com>
9356 * rtl.h (subreg_memory_offset): Declare.
9357 * emit-rtl.c (subreg_memory_offset): New function.
9358 * expmed.c (store_bit_field_1): Use it.
9359 * expr.c (undefined_operand_subword_p): Likewise.
9360 * simplify-rtx.c (simplify_subreg): Likewise.
9362 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
9364 PR rtl-optimization/57448
9367 * optabs.c (expand_atomic_load): Place compiler memory barriers if
9368 using atomic_load pattern.
9369 (expand_atomic_store): Likewise.
9371 2017-09-04 Jakub Jelinek <jakub@redhat.com>
9374 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
9375 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
9376 handling. Use replace_call_with_value with NULL instead of
9377 gsi_replace, unlink_stmt_vdef and release_defs.
9379 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
9382 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
9384 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9387 * rtl.h (emit_library_call_value_1): Declare.
9388 (emit_library_call): Replace declaration with a series of overloads.
9389 Remove the parameter count argument.
9390 (emit_library_call_value): Likewise.
9391 * calls.c (emit_library_call_value_1): Make global. Replace varargs
9392 with an "rtx_mode_t *".
9393 (emit_library_call_value): Delete.
9394 (emit_library_call): Likewise.
9395 * asan.c (asan_emit_stack_protection): Update calls accordingly.
9396 (asan_emit_allocas_unpoison): Likewise.
9397 * builtins.c (expand_builtin_powi): Likewise.
9398 (expand_asan_emit_allocas_unpoison): Likewise.
9399 * cfgexpand.c (expand_main_function): Likewise.
9400 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
9401 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
9402 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
9403 * config/arm/arm.c (arm_trampoline_init): Likewise.
9404 (arm_call_tls_get_addr): Likewise.
9405 (arm_expand_divmod_libfunc): Likewise.
9406 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
9407 (smulsi3_highpart): Likewise.
9408 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
9409 (c6x_expand_compare): Likewise.
9410 (c6x_expand_movmem): Likewise.
9411 * config/frv/frv.c (frv_trampoline_init): Likewise.
9412 * config/i386/i386.c (ix86_trampoline_init): Likewise.
9413 (ix86_expand_divmod_libfunc): Likewise.
9414 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
9415 (ia64_expand_compare): Likewise.
9416 (ia64_profile_hook): Likewise.
9417 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
9418 (nonlocal_goto): Likewise.
9419 (restore_stack_nonlocal): Likewise.
9420 * config/m32r/m32r.c (block_move_call): Likewise.
9421 (m32r_trampoline_init): Likewise.
9422 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
9423 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
9424 (m68k_call_m68k_read_tp): Likewise.
9425 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
9426 (microblaze_expand_divide): Likewise.
9427 * config/mips/mips.h (mips_args): Likewise.
9428 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
9429 (MIPS_ICACHE_SYNC): Likewise.
9430 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
9431 (nios2_trampoline_init): Likewise.
9432 * config/pa/pa.c (hppa_tls_call): Likewise.
9433 (pa_trampoline_init): Likewise.
9434 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
9435 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
9436 (expand_strn_compare): Likewise.
9437 (rs6000_generate_compare): Likewise.
9438 (rs6000_expand_float128_convert): Likewise.
9439 (output_profile_hook): Likewise.
9440 (rs6000_trampoline_init): Likewise.
9441 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
9442 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
9443 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
9444 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
9445 (rs6000_generate_compare): Likewise.
9446 (rs6000_expand_float128_convert): Likewise.
9447 (output_profile_hook): Likewise.
9448 (rs6000_trampoline_init): Likewise.
9449 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
9450 * config/sh/sh.c (sh_trampoline_init): Likewise.
9451 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
9452 (sparc_emit_float_lib_cmp): Likewise.
9453 (sparc32_initialize_trampoline): Likewise.
9454 (sparc64_initialize_trampoline): Likewise.
9455 (sparc_profile_hook): Likewise.
9456 * config/spu/spu.c (ea_load_store): Likewise.
9457 * config/spu/spu.md (floatunssidf2): Likewise.
9458 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
9459 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
9460 * config/visium/visium.c (expand_block_move_4): Likewise.
9461 (expand_block_move_2): Likewise.
9462 (expand_block_move_1): Likewise.
9463 (expand_block_set_4): Likewise.
9464 (expand_block_set_2): Likewise.
9465 (expand_block_set_1): Likewise.
9466 (visium_trampoline_init): Likewise.
9467 (visium_profile_hook): Likewise.
9468 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
9469 (xtensa_setup_frame_addresses): Likewise.
9470 (xtensa_trampoline_init): Likewise.
9471 * except.c (sjlj_emit_function_enter): Likewise.
9472 (sjlj_emit_function_exit): Likewise.
9473 * explow.c (allocate_dynamic_stack_space): Likewise.
9474 (probe_stack_range): Likewise.
9475 * expr.c (convert_mode_scalar): Likewise.
9476 * optabs.c (expand_binop): Likewise.
9477 (expand_twoval_binop_libfunc): Likewise.
9478 (expand_unop): Likewise.
9479 (prepare_cmp_insn): Likewise.
9480 (prepare_float_lib_cmp): Likewise.
9481 (expand_float): Likewise.
9482 (expand_fix): Likewise.
9483 (expand_fixed_convert): Likewise.
9484 (maybe_emit_sync_lock_test_and_set): Likewise.
9485 (expand_atomic_compare_and_swap): Likewise.
9486 (expand_mem_thread_fence): Likewise.
9487 (expand_atomic_fetch_op): Likewise.
9489 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
9491 * doc/generic.texi (OpenACC): Adjust URL.
9492 * doc/invoke.texi (C Dialect Options): Ditto.
9494 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
9496 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
9497 predicate for operand 1. Add (m,<S>) constraint.
9498 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
9499 Prevent memory operand 1 with register operand 2.
9501 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
9503 PR rtl-optimization/82024
9504 * combine.c (try_combine): If the combination result is a PARALLEL,
9505 and we only need to retain the SET in there that would be placed
9506 at I2, check that we can place that at I3 instead, before doing so.
9508 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9511 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
9513 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
9514 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
9517 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
9518 Jeff Law <law@redhat.com>
9520 * varasm.c (bss_initializer_p): Do not put constants into .bss
9521 (categorize_decl_for_section): Handle bss_initializer_p returning
9522 false when DECL_INITIAL is NULL.
9524 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9527 * config/s390/s390.c (s390_can_inline_p): New function.
9529 2017-09-01 Jeff Law <law@redhat.com>
9531 PR tree-optimization/82052
9532 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
9533 Always initialize the returned slot after a hash table miss
9534 when INSERT is true.
9536 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
9538 * config/s390/s390.md (mem_signal_fence): Remove.
9539 * doc/md.texi (mem_signal_fence): Remove.
9540 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
9542 * target-insns.def (mem_signal_fence): Remove.
9544 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9547 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
9550 * asan.c (create_odr_indicator): Strip name encoding from assembler
9551 name before appending it after __odr_asan_.
9553 2017-09-01 Martin Liska <mliska@suse.cz>
9555 PR tree-optimization/82059
9556 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
9557 frequency only when an edge is redirected.
9559 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9561 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
9562 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
9563 (arc_conditional_register_usage): Remove ARC600 lp_count
9565 (arc_file_start): Emit Tag_ARC_CPU_variation.
9566 (arc_can_use_doloop_p): New conditions to use ZOLs.
9567 (hwloop_fail): New function.
9568 (hwloop_optimize): Likewise.
9569 (hwloop_pattern_reg): Likewise.
9570 (arc_doloop_hooks): New struct, to be used with reorg_loops.
9571 (arc_reorg_loops): New function, calls reorg_loops.
9572 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
9573 (arc600_corereg_hazard): Remove ZOL checking, case handled by
9575 (arc_loop_hazard): Remove function, functionality moved into
9577 (arc_hazard): Remove arc_loop_hazard call.
9578 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
9579 into hwloop_optimize.
9580 (arc_label_align): Remove ZOL handling.
9581 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
9582 * config/arc/arc.md (doloop_begin): Remove pattern.
9583 (doloop_begin_i): Likewise.
9584 (doloop_end_i): Likewise.
9585 (doloop_fallback): Likewise.
9586 (doloop_fallback_m): Likewise.
9587 (doloop_end): Reimplement expand.
9588 (arc_lp): New pattern for LP instruction.
9589 (loop_end): New pattern.
9590 (loop_fail): Likewise.
9591 (decrement_and_branch_until_zero): Likewise.
9592 * config/arc/arc.opt (mlpc-width): New option.
9593 * doc/invoke.texi (mlpc-width): Document option.
9595 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9597 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
9598 (arc_ccfsm_advance): Fix checking for delay slots.
9599 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
9601 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9603 * config/arc/arc.md (movqi_insn): Add stores to save constant long
9605 (movhi_insn): Update store instruction constraint which are saving
9606 6-bit short immediates.
9607 (movsi_insn): Consider also short scaled load operations.
9608 (zero_extendhisi2_i): Use Usd constraint instead of T.
9609 (extendhisi2_i): Add q constraint.
9610 (arc_clzsi2): Add type and length attributes.
9611 (arc_ctzsi2): Likewise.
9612 * config/arc/constraints.md (Usc): Update constraint, the
9613 assembler can parse two relocations for a single instruction.
9615 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9617 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
9618 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
9620 2017-08-31 Olivier Hainque <hainque@adacore.com>
9622 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
9623 match as powerpc-wrs-vxworks*.
9625 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
9627 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
9628 register constraint for by-element operand.
9629 (aarch64_mls_elt_merge<mode>): Likewise.
9631 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9633 * config/arc/arc.c (arc_can_follow_jump): Check for short
9636 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9638 * config.gcc: Use g.opt for arc.
9639 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
9640 functionality moved to ...
9641 (legitimate_scaled_address_p): New function, ...here.
9642 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
9643 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
9644 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
9646 (arc_override_options): Handle G option.
9647 (arc_output_pic_addr_const): Correct function definition.
9648 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
9649 (arc_decl_anon_ns_mem_p): Delete.
9650 (arc_in_small_data_p): Overhaul this function to take into
9651 consideration the value given via G option.
9652 (arc_rewrite_small_data_1): Renamed and corrected old
9653 arc_rewrite_small_data function.
9654 (arc_rewrite_small_data): New function.
9655 (small_data_pattern): Don't use pic_offset_table_rtx.
9656 * config/arc/arc.h (CC1_SPEC): Recognize G option.
9657 * config/arc/simdext.md (movmisalignv2hi): Use
9658 prepare_move_operands function.
9660 (movmisalign*): Likewise.
9661 * doc/invoke.texi (ARC options): Document -G option.
9663 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9665 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
9667 * config/arc/arc.c (arc_print_operand): Output scalled address for
9668 sdata whenever is possible.
9669 (arc_in_small_data_p): Allow sdata for 64bit datum when double
9670 load/stores are available.
9671 (compact_sda_memory_operand): Check for the alignment required by
9672 code density instructions.
9673 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
9675 * config/arc/constraints.md (Usd): Update constraint.
9676 (Us0): New constraint.
9677 (Usc): Update constraint.
9679 2017-08-31 Richard Biener <rguenther@suse.de>
9682 * dwarf2out.c (dwarf2out_early_global_decl): Process each
9685 2017-08-31 Tamar Christina <tamar.christina@arm.com>
9687 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
9688 Resize type_signature.
9690 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9691 Alan Hayward <alan.hayward@arm.com>
9692 David Sherwood <david.sherwood@arm.com>
9694 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
9695 subregs whose inner modes can be stored in GPRs.
9696 (aarch64_classify_index): Likewise.
9698 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9699 Alan Hayward <alan.hayward@arm.com>
9700 David Sherwood <david.sherwood@arm.com>
9702 * config/aarch64/iterators.md (V_cmp_result): Rename to...
9703 (V_INT_EQUIV): ...this.
9704 (v_cmp_result): Rename to...
9705 (v_int_equiv): ...this.
9706 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
9707 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
9708 (copysign<mode>3): Likewise.
9709 (aarch64_simd_bsl<mode>_internal): Likewise.
9710 (aarch64_simd_bsl<mode>): Likewise.
9711 (vec_cmp<mode><mode>): Likewise.
9712 (vcond<mode><mode>): Likewise.
9713 (vcond<v_cmp_mixed><mode>): Likewise.
9714 (vcondu<mode><v_cmp_mixed>): Likewise.
9715 (aarch64_cm<optab><mode>): Likewise.
9716 (aarch64_cmtst<mode>): Likewise.
9717 (aarch64_fac<optab><mode>): Likewise.
9718 (vec_perm_const<mode>): Likewise.
9719 (vcond_mask_<mode><v_cmp_result>): Rename to...
9720 (vcond_mask_<mode><v_int_equiv>): ...this.
9721 (vec_cmp<mode><v_cmp_result>): Rename to...
9722 (vec_cmp<mode><v_int_equiv>): ...this.
9724 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9725 Alan Hayward <alan.hayward@arm.com>
9726 David Sherwood <david.sherwood@arm.com>
9728 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
9730 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
9731 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
9732 (UNSPEC_LD4_DREG): New unspecs.
9733 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
9734 (aarch64_ld2<mode>_dreg_be): Replace with...
9735 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
9737 (aarch64_ld3<mode>_dreg_le)
9738 (aarch64_ld3<mode>_dreg_be): Replace with...
9739 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
9741 (aarch64_ld4<mode>_dreg_le)
9742 (aarch64_ld4<mode>_dreg_be): Replace with...
9743 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
9746 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9748 PR tree-optimization/81987
9749 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
9750 insert an initializer in a location not dominated by the stride
9753 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
9755 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
9756 on the entire header of the finally block in the fallthru case.
9758 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
9760 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
9762 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
9764 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
9765 rs6000_emit_move_from_cr and call renamed function.
9766 (rs6000_emit_prologue): Call renamed functions.
9767 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
9768 movesi_from_cr, remove volatile CRs.
9770 2017-08-30 Jon Beniston <jon@beniston.com>
9771 Richard Biener <rguenther@suse.de>
9773 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
9774 of VECTOR_MODE_P check.
9775 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
9776 element vector types.
9778 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9780 * df.h (df_read_modify_subreg_p): Remove in favor of...
9781 * rtl.h (read_modify_subreg_p): ...this new function. Take a
9782 const_rtx instead of an rtx.
9783 * cprop.c (local_cprop_find_used_regs): Update accordingly.
9784 * df-problems.c (df_word_lr_mark_ref): Likewise.
9785 * ira-lives.c (mark_pseudo_reg_live): Likewise.
9786 (mark_pseudo_reg_dead): Likewise.
9787 (mark_ref_dead): Likewise.
9788 * reginfo.c (init_subregs_of_mode): Likewise.
9789 * sched-deps.c (sched_analyze_1): Likewise.
9790 * df-scan.c (df_def_record_1): Likewise.
9791 (df_uses_record): Likewise.
9792 (df_read_modify_subreg_p): Remove in favor of...
9793 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
9794 const_rtx instead of an rtx.
9796 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9797 Alan Hayward <alan.hayward@arm.com>
9798 David Sherwood <david.sherwood@arm.com>
9800 * rtl.h (partial_subreg_p): New function.
9801 * caller-save.c (save_call_clobbered_regs): Use it.
9802 * calls.c (expand_call): Likewise.
9803 * combine.c (combinable_i3pat): Likewise.
9804 (simplify_set): Likewise.
9805 (make_extraction): Likewise.
9806 (make_compound_operation_int): Likewise.
9807 (gen_lowpart_or_truncate): Likewise.
9808 (force_to_mode): Likewise.
9809 (make_field_assignment): Likewise.
9810 (reg_truncated_to_mode): Likewise.
9811 (record_truncated_value): Likewise.
9812 (move_deaths): Likewise.
9813 * cse.c (record_jump_cond): Likewise.
9814 (cse_insn): Likewise.
9815 * cselib.c (cselib_lookup_1): Likewise.
9816 * expmed.c (extract_bit_field_using_extv): Likewise.
9817 * function.c (assign_parm_setup_reg): Likewise.
9818 * ifcvt.c (noce_convert_multiple_sets): Likewise.
9819 * ira-build.c (create_insn_allocnos): Likewise.
9820 * lra-coalesce.c (merge_pseudos): Likewise.
9821 * lra-constraints.c (match_reload): Likewise.
9822 (simplify_operand_subreg): Likewise.
9823 (curr_insn_transform): Likewise.
9824 * lra-lives.c (process_bb_lives): Likewise.
9825 * lra.c (new_insn_reg): Likewise.
9826 (lra_substitute_pseudo): Likewise.
9827 * regcprop.c (mode_change_ok): Likewise.
9828 (maybe_mode_change): Likewise.
9829 (copyprop_hardreg_forward_1): Likewise.
9830 * reload.c (push_reload): Likewise.
9831 (find_reloads): Likewise.
9832 (find_reloads_subreg_address): Likewise.
9833 * reload1.c (alter_reg): Likewise.
9834 (eliminate_regs_1): Likewise.
9835 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9837 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
9839 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
9840 back to if statements, including unpack.
9842 2017-08-30 Martin Liska <mliska@suse.cz>
9845 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
9847 (func_checker::compare_asm_inputs_outputs): ... this function.
9848 (func_checker::compare_gimple_asm): Use the function to compare
9849 also ASM constrains.
9850 * ipa-icf-gimple.h: Rename the function.
9852 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9853 Alan Hayward <alan.hayward@arm.com>
9854 David Sherwood <david.sherwood@arm.com>
9856 * coretypes.h (complex_mode): New type.
9857 * gdbhooks.py (build_pretty_printer): Handle it.
9858 * machmode.h (complex_mode): New class.
9859 (complex_mode::includes_p): New function.
9860 (is_complex_int_mode): Likewise.
9861 (is_complex_float_mode): Likewise.
9862 * genmodes.c (get_mode_class): Handle complex mode classes.
9863 * function.c (expand_function_end): Use is_complex_int_mode.
9865 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9866 Alan Hayward <alan.hayward@arm.com>
9867 David Sherwood <david.sherwood@arm.com>
9869 * coretypes.h (scalar_mode_pod): New typedef.
9870 * gdbhooks.py (build_pretty_printer): Handle it.
9871 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
9872 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
9873 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
9874 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
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 * machmode.h (mode_for_vector): Take a scalar_mode instead
9883 * stor-layout.c (mode_for_vector): Likewise.
9884 * explow.c (promote_mode): Use as_a <scalar_mode>.
9885 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
9887 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9888 Alan Hayward <alan.hayward@arm.com>
9889 David Sherwood <david.sherwood@arm.com>
9891 * target.def (preferred_simd_mode): Take a scalar_mode
9892 instead of a machine_mode.
9893 * targhooks.h (default_preferred_simd_mode): Likewise.
9894 * targhooks.c (default_preferred_simd_mode): Likewise.
9895 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
9896 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
9897 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
9898 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
9899 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
9900 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
9901 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
9902 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
9904 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
9905 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
9906 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
9907 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
9908 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
9909 * doc/tm.texi: Regenerate.
9910 * optabs-query.c (can_vec_mask_load_store_p): Return false for
9913 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9914 Alan Hayward <alan.hayward@arm.com>
9915 David Sherwood <david.sherwood@arm.com>
9917 * target.def (scalar_mode_supported_p): Take a scalar_mode
9918 instead of a machine_mode.
9919 * targhooks.h (default_scalar_mode_supported_p): Likewise.
9920 * targhooks.c (default_scalar_mode_supported_p): Likewise.
9921 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
9922 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
9923 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
9924 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
9925 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
9926 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
9927 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
9928 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
9929 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
9930 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
9931 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
9932 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
9934 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
9935 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
9936 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
9937 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
9938 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
9940 * doc/tm.texi: Regenerate.
9942 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9943 Alan Hayward <alan.hayward@arm.com>
9944 David Sherwood <david.sherwood@arm.com>
9946 * coretypes.h (opt_scalar_mode): New typedef.
9947 * gdbhooks.py (build_pretty_printers): Handle it.
9948 * machmode.h (mode_iterator::get_2xwider): Add overload for
9950 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
9952 * expr.c (convert_mode_scalar): Likewise.
9953 * omp-low.c (omp_clause_aligned_alignment): Likewise.
9954 * optabs.c (expand_float): Likewise.
9955 (expand_fix): Likewise.
9956 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
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 * optabs.c (expand_float): Explicitly check for scalars before
9963 using a branching expansion.
9964 (expand_fix): Likewise.
9966 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9967 Alan Hayward <alan.hayward@arm.com>
9968 David Sherwood <david.sherwood@arm.com>
9970 * expr.c (convert_mode): Split scalar handling out into...
9971 (convert_mode_scalar): ...this new function. Treat the modes
9974 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9975 Alan Hayward <alan.hayward@arm.com>
9976 David Sherwood <david.sherwood@arm.com>
9978 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
9980 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
9982 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9983 Alan Hayward <alan.hayward@arm.com>
9984 David Sherwood <david.sherwood@arm.com>
9986 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
9987 rather than a machine_mode.
9988 (fixed_from_string): Likewise.
9989 (fixed_convert): Likewise.
9990 (fixed_convert_from_int): Likewise.
9991 (fixed_convert_from_real): Likewise.
9992 (real_convert_from_fixed): Likewise.
9993 * fixed-value.c (fixed_from_double_int): Likewise.
9994 (fixed_from_string): Likewise.
9995 (fixed_convert): Likewise.
9996 (fixed_convert_from_int): Likewise.
9997 (fixed_convert_from_real): Likewise.
9998 (real_convert_from_fixed): Likewise.
9999 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
10001 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10002 Alan Hayward <alan.hayward@arm.com>
10003 David Sherwood <david.sherwood@arm.com>
10005 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
10006 of separate mode class checks. Do not allow vector modes here.
10007 (immed_wide_int_const): Use as_a <scalar_mode>.
10008 * explow.c (trunc_int_for_mode): Likewise.
10009 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
10010 (wi::shwi): Likewise.
10011 (wi::min_value): Likewise.
10012 (wi::max_value): Likewise.
10013 * dwarf2out.c (loc_descriptor): Likewise.
10014 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
10015 for CONST_WIDE_INT.
10017 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10018 Alan Hayward <alan.hayward@arm.com>
10019 David Sherwood <david.sherwood@arm.com>
10021 * tree.h (SCALAR_TYPE_MODE): New macro.
10022 * expr.c (expand_expr_addr_expr_1): Use it.
10023 (expand_expr_real_2): Likewise.
10024 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
10025 (fold_convert_const_fixed_from_int): Likewise.
10026 (fold_convert_const_fixed_from_real): Likewise.
10027 (native_encode_fixed): Likewise
10028 (native_encode_complex): Likewise
10029 (native_encode_vector): Likewise.
10030 (native_interpret_fixed): Likewise.
10031 (native_interpret_real): Likewise.
10032 (native_interpret_complex): Likewise.
10033 (native_interpret_vector): Likewise.
10034 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
10035 (simd_clone_adjust_argument_types): Likewise.
10036 (simd_clone_init_simd_arrays): Likewise.
10037 (simd_clone_adjust): Likewise.
10038 * stor-layout.c (layout_type): Likewise.
10039 * tree.c (build_minus_one_cst): Likewise.
10040 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
10041 * tree-inline.c (estimate_move_cost): Likewise.
10042 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
10043 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
10044 (vectorizable_reduction): Likewise.
10045 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
10046 (vect_recog_mixed_size_cond_pattern): Likewise.
10047 (check_bool_pattern): Likewise.
10048 (adjust_bool_pattern): Likewise.
10049 (search_type_for_mask_1): Likewise.
10050 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
10051 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10052 (vectorizable_load): Likewise.
10053 (vectorizable_store): Likewise.
10054 * ubsan.c (ubsan_encode_value): Likewise.
10055 * varasm.c (output_constant): Likewise.
10057 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10058 Alan Hayward <alan.hayward@arm.com>
10059 David Sherwood <david.sherwood@arm.com>
10061 * coretypes.h (scalar_mode): New class.
10062 * machmode.h (scalar_mode): Likewise.
10063 (scalar_mode::includes_p): New function.
10064 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
10065 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
10066 * genmodes.c (get_mode_class): Handle remaining scalar modes.
10067 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
10068 * expmed.c (store_bit_field_1): Likewise.
10069 (extract_bit_field_1): Likewise.
10070 * expr.c (write_complex_part): Likewise.
10071 (read_complex_part): Likewise.
10072 (emit_move_complex_push): Likewise.
10073 (expand_expr_real_2): Likewise.
10074 * function.c (assign_parm_setup_reg): Likewise.
10075 (assign_parms_unsplit_complex): Likewise.
10076 * optabs.c (expand_binop): Likewise.
10077 * rtlanal.c (subreg_get_info): Likewise.
10078 * simplify-rtx.c (simplify_immed_subreg): Likewise.
10079 * varasm.c (output_constant_pool_2): Likewise.
10081 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10082 Alan Hayward <alan.hayward@arm.com>
10083 David Sherwood <david.sherwood@arm.com>
10085 * expmed.c (extract_high_half): Use scalar_int_mode and remove
10087 (expmed_mult_highpart_optab): Likewise.
10088 (expmed_mult_highpart): Likewise.
10090 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10091 Alan Hayward <alan.hayward@arm.com>
10092 David Sherwood <david.sherwood@arm.com>
10094 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
10095 instead of a machine_mode.
10096 (builtin_memset_read_str): Likewise.
10097 * builtins.c (c_readstr): Likewise.
10098 (builtin_memcpy_read_str): Likewise.
10099 (builtin_strncpy_read_str): Likewise.
10100 (builtin_memset_read_str): Likewise.
10101 (builtin_memset_gen_str): Likewise.
10102 (expand_builtin_signbit): Use scalar_int_mode for local variables.
10103 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
10104 instead of a machine_mode.
10105 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
10107 (make_extraction): Likewise.
10108 (try_widen_shift_mode): Take and return scalar_int_modes instead
10110 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
10111 a scalar_int_mode instead of a machine_mode.
10112 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
10113 (avr_addr_space_pointer_mode): Likewise.
10114 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
10115 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
10116 (msp430_unwind_word_mode): Likewise.
10117 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
10118 (spu_addr_space_pointer_mode): Likewise.
10119 (spu_addr_space_address_mode): Likewise.
10120 (spu_libgcc_cmp_return_mode): Likewise.
10121 (spu_libgcc_shift_count_mode): Likewise.
10122 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
10123 (rl78_addr_space_pointer_mode): Likewise.
10124 (fl78_unwind_word_mode): Likewise.
10125 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
10127 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
10128 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
10129 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
10130 (mips_valid_pointer_mode): Likewise.
10131 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
10132 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
10133 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
10135 (ft32_addr_space_address_mode): Likewise.
10136 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
10137 scalar_int_mode instead of a machine_mode.
10138 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
10140 (m32c_addr_space_address_mode): Likewise.
10141 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
10142 (rs6000_eh_return_filter_mode): Likewise.
10143 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
10144 (rs6000_eh_return_filter_mode): Likewise.
10145 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
10146 (s390_libgcc_shift_count_mode): Likewise.
10147 (s390_unwind_word_mode): Likewise.
10148 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
10150 * target.def (mode_rep_extended): Likewise.
10151 (valid_pointer_mode): Likewise.
10152 (addr_space.valid_pointer_mode): Likewise.
10153 (eh_return_filter_mode): Return a scalar_int_mode rather than
10155 (libgcc_cmp_return_mode): Likewise.
10156 (libgcc_shift_count_mode): Likewise.
10157 (unwind_word_mode): Likewise.
10158 (addr_space.pointer_mode): Likewise.
10159 (addr_space.address_mode): Likewise.
10160 * doc/tm.texi: Regenerate.
10161 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
10163 (do_jump): Use scalar_int_mode for local variables.
10164 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
10165 rather than a machine_mode.
10166 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
10167 (scompare_loc_descriptor_wide): Likewise.
10168 (scompare_loc_descriptor_narrow): Likewise.
10169 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
10171 * except.c (sjlj_emit_dispatch_table): Likewise.
10172 (expand_builtin_eh_copy_values): Likewise.
10173 * explow.c (convert_memory_address_addr_space_1): Likewise.
10174 Take a scalar_int_mode rather than a machine_mode.
10175 (convert_memory_address_addr_space): Take a scalar_int_mode rather
10176 than a machine_mode.
10177 (memory_address_addr_space): Use scalar_int_mode for local variables.
10178 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
10179 rather than a machine_mode.
10180 * expmed.c (mask_rtx): Likewise.
10181 (init_expmed_one_conv): Likewise.
10182 (expand_mult_highpart_adjust): Likewise.
10183 (extract_high_half): Likewise.
10184 (expmed_mult_highpart_optab): Likewise.
10185 (expmed_mult_highpart): Likewise.
10186 (expand_smod_pow2): Likewise.
10187 (expand_sdiv_pow2): Likewise.
10188 (emit_store_flag_int): Likewise.
10189 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
10191 (extract_low_bits): Likewise.
10192 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
10194 * expr.c (pieces_addr::adjust): Likewise.
10195 (can_store_by_pieces): Likewise.
10196 (store_by_pieces): Likewise.
10197 (clear_by_pieces_1): Likewise.
10198 (expand_expr_addr_expr_1): Likewise.
10199 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
10200 (expand_expr_real_1): Likewise.
10201 (try_casesi): Likewise.
10202 * final.c (shorten_branches): Likewise.
10203 * fold-const.c (fold_convert_const_int_from_fixed): Change the
10204 type of "mode" to machine_mode.
10205 * internal-fn.c (expand_arith_overflow_result_store): Take a
10206 scalar_int_mode rather than a machine_mode.
10207 (expand_mul_overflow): Use scalar_int_mode for local variables.
10208 * loop-doloop.c (doloop_modify): Likewise.
10209 (doloop_optimize): Likewise.
10210 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
10211 than a machine_mode.
10212 (expand_doubleword_shift_condmove): Likewise.
10213 (expand_doubleword_shift): Likewise.
10214 (expand_doubleword_clz): Likewise.
10215 (expand_doubleword_popcount): Likewise.
10216 (expand_doubleword_parity): Likewise.
10217 (expand_absneg_bit): Use scalar_int_mode for local variables.
10218 (prepare_float_lib_cmp): Likewise.
10219 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
10220 rather than a machine_mode.
10221 (convert_memory_address_addr_space): Likewise.
10222 (get_mode_bounds): Likewise.
10223 (get_address_mode): Return a scalar_int_mode rather than a
10225 * rtlanal.c (get_address_mode): Likewise.
10226 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
10227 than a machine_mode.
10228 * targhooks.c (default_mode_rep_extended): Likewise.
10229 (default_valid_pointer_mode): Likewise.
10230 (default_addr_space_valid_pointer_mode): Likewise.
10231 (default_eh_return_filter_mode): Return a scalar_int_mode rather
10232 than a machine_mode.
10233 (default_libgcc_cmp_return_mode): Likewise.
10234 (default_libgcc_shift_count_mode): Likewise.
10235 (default_unwind_word_mode): Likewise.
10236 (default_addr_space_pointer_mode): Likewise.
10237 (default_addr_space_address_mode): Likewise.
10238 * targhooks.h (default_eh_return_filter_mode): Likewise.
10239 (default_libgcc_cmp_return_mode): Likewise.
10240 (default_libgcc_shift_count_mode): Likewise.
10241 (default_unwind_word_mode): Likewise.
10242 (default_addr_space_pointer_mode): Likewise.
10243 (default_addr_space_address_mode): Likewise.
10244 (default_mode_rep_extended): Take a scalar_int_mode rather than
10246 (default_valid_pointer_mode): Likewise.
10247 (default_addr_space_valid_pointer_mode): Likewise.
10248 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
10250 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
10251 rather than a machine_mode.
10252 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
10253 for local variables.
10254 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10255 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
10256 than a machine_mode.
10258 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10259 Alan Hayward <alan.hayward@arm.com>
10260 David Sherwood <david.sherwood@arm.com>
10262 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
10263 the mode argument to scalar_int_mode.
10264 (do_jump_by_parts_zero_rtx): Likewise.
10265 (do_jump_by_parts_equality_rtx): Likewise.
10266 (do_jump_by_parts_greater): Take a mode argument.
10267 (do_jump_by_parts_equality): Likewise.
10268 (do_jump_1): Update calls accordingly.
10270 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10271 Alan Hayward <alan.hayward@arm.com>
10272 David Sherwood <david.sherwood@arm.com>
10274 * is-a.h (safe_dyn_cast): New function.
10275 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
10276 (jump_table_for_label): Likewise.
10277 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
10278 instead of an rtx_insn *.
10279 (shorten_branches): Use dyn_cast instead of LABEL_P and
10280 JUMP_TABLE_DATA_P. Use jump_table_for_label and
10281 rtx_jump_table_data::get_data_mode.
10282 (final_scan_insn): Likewise.
10284 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10285 Alan Hayward <alan.hayward@arm.com>
10286 David Sherwood <david.sherwood@arm.com>
10288 * combine.c (try_combine): Use is_a <scalar_int_mode> when
10289 trying to combine a full-register integer set with a subreg
10292 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10293 Alan Hayward <alan.hayward@arm.com>
10294 David Sherwood <david.sherwood@arm.com>
10296 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
10297 that is always either address_mode or pointer_mode.
10299 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10300 Alan Hayward <alan.hayward@arm.com>
10301 David Sherwood <david.sherwood@arm.com>
10303 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
10304 when the two are known to be equal.
10306 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10307 Alan Hayward <alan.hayward@arm.com>
10308 David Sherwood <david.sherwood@arm.com>
10310 * simplify-rtx.c (simplify_const_unary_operation): Use
10311 is_a <scalar_int_mode> instead of checking for a nonzero
10312 precision. Forcibly convert op_mode to a scalar_int_mode
10313 in that case. More clearly differentiate the operand and
10314 result modes and use the former when deciding what the value
10315 of a count-bits operation should be. Use is_int_mode instead
10316 of checking for a MODE_INT. Remove redundant check for whether
10317 this mode has a zero precision.
10319 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10320 Alan Hayward <alan.hayward@arm.com>
10321 David Sherwood <david.sherwood@arm.com>
10323 * optabs.c (widen_leading): Change the type of the mode argument
10324 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
10325 (widen_bswap): Likewise.
10326 (expand_parity): Likewise.
10327 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
10328 (expand_ffs): Likewise.
10329 (epand_unop): Check for scalar integer modes before calling the
10332 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10333 Alan Hayward <alan.hayward@arm.com>
10334 David Sherwood <david.sherwood@arm.com>
10336 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
10338 (expand_expr_real_1): Update call accordingly.
10340 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10341 Alan Hayward <alan.hayward@arm.com>
10342 David Sherwood <david.sherwood@arm.com>
10344 * expmed.c (store_bit_field_using_insv): Add op0_mode and
10345 value_mode arguments. Use scalar_int_mode internally.
10346 (store_bit_field_1): Rename the new integer mode from imode
10347 to op0_mode and use it instead of GET_MODE (op0). Update calls
10348 to store_split_bit_field, store_bit_field_using_insv and
10349 store_fixed_bit_field.
10350 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
10351 Use scalar_int_mode internally. Use a bit count rather than a mode
10352 when calculating the largest bit size for get_best_mode.
10353 Update calls to store_split_bit_field and store_fixed_bit_field_1.
10354 (store_fixed_bit_field_1): Add mode and value_mode arguments.
10355 Remove assertion that OP0 has a scalar integer mode.
10356 (store_split_bit_field): Add op0_mode and value_mode arguments.
10357 Update calls to extract_fixed_bit_field.
10358 (extract_bit_field_using_extv): Add an op0_mode argument.
10359 Use scalar_int_mode internally.
10360 (extract_bit_field_1): Rename the new integer mode from imode to
10361 op0_mode and use it instead of GET_MODE (op0). Update calls to
10362 extract_split_bit_field, extract_bit_field_using_extv and
10363 extract_fixed_bit_field.
10364 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
10365 to extract_split_bit_field and extract_fixed_bit_field_1.
10366 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
10367 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
10368 on the target mode.
10369 (extract_split_bit_field): Add an op0_mode argument. Update call
10370 to extract_fixed_bit_field.
10372 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10373 Alan Hayward <alan.hayward@arm.com>
10374 David Sherwood <david.sherwood@arm.com>
10376 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
10377 * explow.c (hard_function_value): Likewise.
10378 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
10379 convert_to_mode call outside the loop.
10380 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
10381 for the mode iterator. Require the mode specified by max_pieces
10383 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
10385 (copy_blkmode_to_reg): Likewise.
10386 (set_storage_via_setmem): Likewise.
10387 * optabs.c (prepare_cmp_insn): Likewise.
10388 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
10389 * stor-layout.c (finish_bitfield_representative): Likewise.
10391 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10392 Alan Hayward <alan.hayward@arm.com>
10393 David Sherwood <david.sherwood@arm.com>
10395 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
10396 * expr.c (convert_move): Use them.
10397 (convert_modes): Likewise.
10398 (store_expr_with_bounds): Likewise.
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 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
10405 parameter for the mode of "x". Remove the "known_x", "known_mode"
10406 and "known_ret" arguments. Change the type of the mode argument
10407 to scalar_int_mode.
10408 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
10409 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
10410 (reg_num_sign_bit_copies_for_combine): Likewise.
10411 * rtlanal.c (nonzero_bits1): Likewise.
10412 (num_sign_bit_copies1): Likewise.
10413 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
10414 (reg_num_sign_bit_copies_general): Likewise.
10415 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
10416 (reg_nonzero_bits_general): Likewise.
10418 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10419 Alan Hayward <alan.hayward@arm.com>
10420 David Sherwood <david.sherwood@arm.com>
10422 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
10423 than in subroutines. Return 1 for non-integer modes.
10424 (cached_num_sign_bit_copies): Change the type of the mode parameter
10425 to scalar_int_mode.
10426 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
10427 classes. Handle CONST_INT_P first and then check whether X also
10428 has a scalar integer mode. Check the same thing for inner registers
10429 of a SUBREG and for values that are being extended or truncated.
10431 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10432 Alan Hayward <alan.hayward@arm.com>
10433 David Sherwood <david.sherwood@arm.com>
10435 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
10436 in subroutines. Return the mode mask for non-integer modes.
10437 (cached_nonzero_bits): Change the type of the mode parameter
10438 to scalar_int_mode.
10439 (nonzero_bits1): Likewise. Remove early exit for other mode
10440 classes. Handle CONST_INT_P first and then check whether X
10441 also has a scalar integer mode.
10443 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10444 Alan Hayward <alan.hayward@arm.com>
10445 David Sherwood <david.sherwood@arm.com>
10447 * expr.c (widest_int_mode_for_size): Make the comment match the code.
10448 Return a scalar_int_mode and assert that the size is greater than
10450 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
10451 (op_by_pieces_d::op_by_pieces_d): Likewise.
10452 (op_by_pieces_d::run): Likewise.
10453 (can_store_by_pieces): Likewise.
10455 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10456 Alan Hayward <alan.hayward@arm.com>
10457 David Sherwood <david.sherwood@arm.com>
10459 * combine.c (extract_left_shift): Add a mode argument and update
10461 (make_compound_operation_int): Change the type of the mode parameter
10462 to scalar_int_mode and update the call to extract_left_shift.
10464 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10465 Alan Hayward <alan.hayward@arm.com>
10466 David Sherwood <david.sherwood@arm.com>
10468 * combine.c (simplify_and_const_int): Change the type of the mode
10469 parameter to scalar_int_mode.
10470 (simplify_and_const_int_1): Likewise. Update recursive call.
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 * combine.c (simplify_compare_const): Check that the mode is a
10477 scalar_int_mode (rather than VOIDmode) before testing its
10479 (simplify_comparison): Move COMPARISON_P handling out of the
10480 loop and restrict the latter part of the loop to scalar_int_modes.
10481 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
10482 and when considering SUBREG_REGs. Use is_int_mode instead of
10483 checking GET_MODE_CLASS against MODE_INT.
10485 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10486 Alan Hayward <alan.hayward@arm.com>
10487 David Sherwood <david.sherwood@arm.com>
10489 * combine.c (try_widen_shift_mode): Move check for equal modes to...
10490 (simplify_shift_const_1): ...here. Use scalar_int_mode for
10491 shift_unit_mode and for modes involved in scalar shifts.
10493 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10494 Alan Hayward <alan.hayward@arm.com>
10495 David Sherwood <david.sherwood@arm.com>
10497 * combine.c (force_int_to_mode): New function, split out from...
10498 (force_to_mode): ...here. Keep xmode up-to-date and use it
10499 instead of GET_MODE (x).
10501 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10502 Alan Hayward <alan.hayward@arm.com>
10503 David Sherwood <david.sherwood@arm.com>
10505 * optabs-query.h (extraction_insn::struct_mode): Change type to
10506 opt_scalar_int_mode and update comment.
10507 (extraction_insn::field_mode): Change type to scalar_int_mode.
10508 (extraction_insn::pos_mode): Likewise.
10509 * combine.c (make_extraction): Update accordingly.
10510 * optabs-query.c (get_traditional_extraction_insn): Likewise.
10511 (get_optab_extraction_insn): Likewise.
10512 * recog.c (simplify_while_replacing): Likewise.
10513 * expmed.c (narrow_bit_field_mem): Change the type of the mode
10514 parameter to opt_scalar_int_mode.
10516 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10517 Alan Hayward <alan.hayward@arm.com>
10518 David Sherwood <david.sherwood@arm.com>
10520 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
10521 to a scalar_int_mode instead of a machine_mode.
10522 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
10523 (get_best_mode): Return a boolean and use a pointer argument to store
10524 the selected mode. Replace the limit mode parameter with a bit limit.
10525 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
10526 for the values returned by bit_field_mode_iterator::next_mode.
10527 (store_bit_field): Update call to get_best_mode.
10528 (store_fixed_bit_field): Likewise.
10529 (extract_fixed_bit_field): Likewise.
10530 * expr.c (optimize_bitfield_assignment_op): Likewise.
10531 * fold-const.c (optimize_bit_field_compare): Likewise.
10532 (fold_truth_andor_1): Likewise.
10533 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
10534 Update for new type of m_mode.
10535 (get_best_mode): As above.
10537 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10538 Alan Hayward <alan.hayward@arm.com>
10539 David Sherwood <david.sherwood@arm.com>
10541 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
10542 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
10543 (store_bit_field): Check is_a <scalar_int_mode> before calling
10544 strict_volatile_bitfield_p.
10545 (extract_bit_field): Likewise.
10547 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10548 Alan Hayward <alan.hayward@arm.com>
10549 David Sherwood <david.sherwood@arm.com>
10551 * target.def (cstore_mode): Return a scalar_int_mode.
10552 * doc/tm.texi: Regenerate.
10553 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
10554 * targhooks.h (default_cstore_mode): Likewise.
10555 * targhooks.c (default_cstore_mode): Likewise, using a forced
10557 * expmed.c (emit_cstore): Expect the target of the cstore to be
10560 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10561 Alan Hayward <alan.hayward@arm.com>
10562 David Sherwood <david.sherwood@arm.com>
10564 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
10566 (niter_desc): Likewise mode.
10567 (iv_analyze): Add a mode parameter.
10569 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
10570 and change its type to scalar_int_mode.
10571 * loop-iv.c: Update commentary at head of file.
10572 (iv_constant): Pass the mode paraeter before the rtx it describes
10573 and change its type to scalar_int_mode. Remove VOIDmode handling.
10574 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
10575 (iv_extend): Likewise.
10576 (shorten_into_mode): Likewise.
10577 (iv_add): Use scalar_int_mode.
10578 (iv_mult): Likewise.
10579 (iv_shift): Likewise.
10580 (canonicalize_iv_subregs): Likewise.
10581 (get_biv_step_1): Pass the outer_mode parameter before the rtx
10582 it describes and change its mode to scalar_int_mode. Also change
10583 the type of the returned inner_mode to scalar_int_mode.
10584 (get_biv_step): Likewise, turning outer_mode from a pointer
10585 into a direct parameter. Update call to get_biv_step_1.
10586 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
10587 iv_constant and get_biv_step.
10588 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
10589 and change its type to scalar_int_mode. Don't initialise iv->mode
10590 to VOIDmode and remove later checks for its still being VOIDmode.
10591 Update calls to iv_analyze_op and iv_analyze_expr. Check
10592 is_a <scalar_int_mode> when changing the mode under consideration.
10593 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
10594 Update call to iv_analyze_expr.
10595 (iv_analyze_op): Add a mode parameter. Reject subregs whose
10596 inner register is not also a scalar_int_mode. Update call to
10598 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
10599 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
10600 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
10601 separate mode class checks. Update calls to iv_analyze. Remove
10602 fix-up of VOIDmodes after iv_analyze_biv.
10603 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
10604 don't have a scalar_int_mode. Update call to biv_p.
10606 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10607 Alan Hayward <alan.hayward@arm.com>
10608 David Sherwood <david.sherwood@arm.com>
10610 * cfgexpand.c (convert_debug_memory_address): Use
10611 as_a <scalar_int_mode>.
10612 * combine.c (expand_compound_operation): Likewise.
10613 (make_extraction): Likewise.
10614 (change_zero_ext): Likewise.
10615 (simplify_comparison): Likewise.
10616 * cse.c (cse_insn): Likewise.
10617 * dwarf2out.c (minmax_loc_descriptor): Likewise.
10618 (mem_loc_descriptor): Likewise.
10619 (loc_descriptor): Likewise.
10620 * expmed.c (init_expmed_one_mode): Likewise.
10621 (synth_mult): Likewise.
10622 (emit_store_flag_1): Likewise.
10623 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
10624 of a comparison with size.
10625 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
10626 (reduce_to_bit_field_precision): Likewise.
10627 * function.c (expand_function_end): Likewise.
10628 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
10629 * loop-doloop.c (doloop_modify): Likewise.
10630 * optabs.c (expand_binop): Likewise.
10631 (expand_unop): Likewise.
10632 (expand_copysign_absneg): Likewise.
10633 (prepare_cmp_insn): Likewise.
10634 (maybe_legitimize_operand): Likewise.
10635 * recog.c (const_scalar_int_operand): Likewise.
10636 * rtlanal.c (get_address_mode): Likewise.
10637 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10638 (simplify_cond_clz_ctz): Likewise.
10639 * tree-nested.c (get_nl_goto_field): Likewise.
10640 * tree.c (build_vector_type_for_mode): Likewise.
10641 * var-tracking.c (use_narrower_mode): Likewise.
10643 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10644 Alan Hayward <alan.hayward@arm.com>
10645 David Sherwood <david.sherwood@arm.com>
10647 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
10648 * builtins.c (expand_builtin_signbit): Use it.
10649 * cfgexpand.c (expand_debug_expr): Likewise.
10650 * dojump.c (do_jump): Likewise.
10651 (do_compare_and_jump): Likewise.
10652 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
10653 * expmed.c (make_tree): Likewise.
10654 * expr.c (expand_expr_real_2): Likewise.
10655 (expand_expr_real_1): Likewise.
10656 (try_casesi): Likewise.
10657 * fold-const-call.c (fold_const_call_ss): Likewise.
10658 * fold-const.c (unextend): Likewise.
10659 (extract_muldiv_1): Likewise.
10660 (fold_single_bit_test): Likewise.
10661 (native_encode_int): Likewise.
10662 (native_encode_string): Likewise.
10663 (native_interpret_int): Likewise.
10664 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
10665 * internal-fn.c (expand_addsub_overflow): Likewise.
10666 (expand_neg_overflow): Likewise.
10667 (expand_mul_overflow): Likewise.
10668 (expand_arith_overflow): Likewise.
10669 * match.pd: Likewise.
10670 * stor-layout.c (layout_type): Likewise.
10671 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
10672 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
10673 * tree-ssanames.c (get_range_info): Likewise.
10674 * tree-switch-conversion.c (array_value_type) Likewise.
10675 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
10676 (vect_recog_divmod_pattern): Likewise.
10677 (vect_recog_mixed_size_cond_pattern): Likewise.
10678 * tree-vrp.c (extract_range_basic): Likewise.
10679 (simplify_float_conversion_using_ranges): Likewise.
10680 * tree.c (int_fits_type_p): Likewise.
10681 * ubsan.c (instrument_bool_enum_load): Likewise.
10682 * varasm.c (mergeable_string_section): Likewise.
10683 (narrowing_initializer_constant_valid_p): Likewise.
10684 (output_constant): Likewise.
10686 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10687 Alan Hayward <alan.hayward@arm.com>
10688 David Sherwood <david.sherwood@arm.com>
10690 * machmode.h (NARROWEST_INT_MODE): New macro.
10691 * expr.c (alignment_for_piecewise_move): Use it instead of
10692 GET_CLASS_NARROWEST_MODE (MODE_INT).
10693 (push_block): Likewise.
10694 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
10696 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10698 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10699 Alan Hayward <alan.hayward@arm.com>
10700 David Sherwood <david.sherwood@arm.com>
10702 * postreload.c (move2add_valid_value_p): Change the type of the
10703 mode parameter to scalar_int_mode.
10704 (move2add_use_add2_insn): Add a mode parameter and use it instead
10706 (move2add_use_add3_insn): Likewise.
10707 (reload_cse_move2add): Update accordingly.
10709 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10710 Alan Hayward <alan.hayward@arm.com>
10711 David Sherwood <david.sherwood@arm.com>
10713 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
10715 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
10716 * optabs.c (expand_unop): Likewise.
10718 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10719 Alan Hayward <alan.hayward@arm.com>
10720 David Sherwood <david.sherwood@arm.com>
10722 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
10723 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
10724 (popcount_loc_descriptor): Likewise.
10725 (bswap_loc_descriptor): Likewise.
10726 (rotate_loc_descriptor): Likewise.
10727 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
10728 calling the functions above.
10730 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10731 Alan Hayward <alan.hayward@arm.com>
10732 David Sherwood <david.sherwood@arm.com>
10734 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
10736 (try_combine): Likewise.
10737 (simplify_if_then_else): Likewise.
10738 * cse.c (cse_insn): Likewise.
10739 * dwarf2out.c (mem_loc_descriptor): Likewise.
10740 * emit-rtl.c (gen_lowpart_common): Likewise.
10741 * simplify-rtx.c (simplify_truncation): Likewise.
10742 (simplify_binary_operation_1): Likewise.
10743 (simplify_const_relational_operation): Likewise.
10744 (simplify_ternary_operation): Likewise.
10745 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
10747 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10748 Alan Hayward <alan.hayward@arm.com>
10749 David Sherwood <david.sherwood@arm.com>
10751 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
10752 * reload.c (push_reload): Likewise.
10753 (find_reloads): Likewise.
10755 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10756 Alan Hayward <alan.hayward@arm.com>
10757 David Sherwood <david.sherwood@arm.com>
10759 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
10760 (make_compound_operation_int): Likewise.
10761 (change_zero_ext): Likewise.
10762 * expr.c (convert_move): Likewise.
10763 (convert_modes): Likewise.
10764 * fwprop.c (forward_propagate_subreg): Likewise.
10765 * loop-iv.c (get_biv_step_1): Likewise.
10766 * optabs.c (widen_operand): Likewise.
10767 * postreload.c (move2add_valid_value_p): Likewise.
10768 * recog.c (simplify_while_replacing): Likewise.
10769 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10770 (simplify_binary_operation_1): Likewise. Remove redundant
10771 mode equality check.
10773 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10774 Alan Hayward <alan.hayward@arm.com>
10775 David Sherwood <david.sherwood@arm.com>
10777 * combine.c (combine_simplify_rtx): Add checks for
10778 is_a <scalar_int_mode>.
10779 (simplify_if_then_else): Likewise.
10780 (make_field_assignment): Likewise.
10781 (simplify_comparison): Likewise.
10782 * ifcvt.c (noce_try_bitop): Likewise.
10783 * loop-invariant.c (canonicalize_address_mult): Likewise.
10784 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10786 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10787 Alan Hayward <alan.hayward@arm.com>
10788 David Sherwood <david.sherwood@arm.com>
10790 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
10791 is_a <scalar_int_mode> instead of != BLKmode.
10793 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10794 Alan Hayward <alan.hayward@arm.com>
10795 David Sherwood <david.sherwood@arm.com>
10797 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
10798 instead of != VOIDmode.
10799 * combine.c (if_then_else_cond): Likewise.
10800 (change_zero_ext): Likewise.
10801 * dwarf2out.c (mem_loc_descriptor): Likewise.
10802 (loc_descriptor): Likewise.
10803 * rtlanal.c (canonicalize_condition): Likewise.
10804 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
10806 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10807 Alan Hayward <alan.hayward@arm.com>
10808 David Sherwood <david.sherwood@arm.com>
10810 * simplify-rtx.c (simplify_binary_operation_1): Use
10811 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
10813 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10814 Alan Hayward <alan.hayward@arm.com>
10815 David Sherwood <david.sherwood@arm.com>
10817 * wide-int.h (int_traits<unsigned char>) New class.
10818 (int_traits<unsigned short>) Likewise.
10819 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
10820 Use GET_MODE_UNIT_PRECISION and remove redundant test for
10822 * combine.c (set_nonzero_bits_and_sign_copies): Use
10823 is_a <scalar_int_mode>.
10824 (find_split_point): Likewise.
10825 (combine_simplify_rtx): Likewise.
10826 (simplify_logical): Likewise.
10827 (expand_compound_operation): Likewise.
10828 (expand_field_assignment): Likewise.
10829 (make_compound_operation): Likewise.
10830 (extended_count): Likewise.
10831 (change_zero_ext): Likewise.
10832 (simplify_comparison): Likewise.
10833 * dwarf2out.c (scompare_loc_descriptor): Likewise.
10834 (ucompare_loc_descriptor): Likewise.
10835 (minmax_loc_descriptor): Likewise.
10836 (mem_loc_descriptor): Likewise.
10837 (loc_descriptor): Likewise.
10838 * expmed.c (init_expmed_one_mode): Likewise.
10839 * lra-constraints.c (lra_constraint_offset): Likewise.
10840 * optabs.c (prepare_libcall_arg): Likewise.
10841 * postreload.c (move2add_note_store): Likewise.
10842 * reload.c (operands_match_p): Likewise.
10843 * rtl.h (load_extend_op): Likewise.
10844 * rtlhooks.c (gen_lowpart_general): Likewise.
10845 * simplify-rtx.c (simplify_truncation): Likewise.
10846 (simplify_unary_operation_1): Likewise.
10847 (simplify_binary_operation_1): Likewise.
10848 (simplify_const_binary_operation): Likewise.
10849 (simplify_const_relational_operation): Likewise.
10850 (simplify_subreg): Likewise.
10851 * stor-layout.c (bitwise_mode_for_mode): Likewise.
10852 * var-tracking.c (adjust_mems): Likewise.
10853 (prepare_call_arguments): Likewise.
10855 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10856 Alan Hayward <alan.hayward@arm.com>
10857 David Sherwood <david.sherwood@arm.com>
10859 * machmode.h (is_int_mode): New fuction.
10860 * combine.c (find_split_point): Use it.
10861 (combine_simplify_rtx): Likewise.
10862 (simplify_if_then_else): Likewise.
10863 (simplify_set): Likewise.
10864 (simplify_shift_const_1): Likewise.
10865 (simplify_comparison): Likewise.
10866 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
10867 * cse.c (notreg_cost): Likewise.
10868 (cse_insn): Likewise.
10869 * cselib.c (cselib_lookup_1): Likewise.
10870 * dojump.c (do_jump_1): Likewise.
10871 (do_compare_rtx_and_jump): Likewise.
10872 * dse.c (get_call_args): Likewise.
10873 * dwarf2out.c (rtl_for_decl_init): Likewise.
10874 (native_encode_initializer): Likewise.
10875 * expmed.c (emit_store_flag_1): Likewise.
10876 (emit_store_flag): Likewise.
10877 * expr.c (convert_modes): Likewise.
10878 (store_field): Likewise.
10879 (expand_expr_real_1): Likewise.
10880 * fold-const.c (fold_read_from_constant_string): Likewise.
10881 * gimple-ssa-sprintf.c (get_format_string): Likewise.
10882 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
10883 * optabs.c (expand_binop): Likewise.
10884 (expand_unop): Likewise.
10885 (expand_abs_nojump): Likewise.
10886 (expand_one_cmpl_abs_nojump): Likewise.
10887 * simplify-rtx.c (mode_signbit_p): Likewise.
10888 (val_signbit_p): Likewise.
10889 (val_signbit_known_set_p): Likewise.
10890 (val_signbit_known_clear_p): Likewise.
10891 (simplify_relational_operation_1): Likewise.
10892 * tree.c (vector_type_mode): Likewise.
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 * machmode.h (smallest_mode_for_size): Fix formatting.
10899 (smallest_int_mode_for_size): New function.
10900 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
10901 instead of smallest_mode_for_size.
10902 * combine.c (make_extraction): Likewise.
10903 * config/arc/arc.c (arc_expand_movmem): Likewise.
10904 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
10905 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
10906 * config/s390/s390.c (s390_expand_insv): Likewise.
10907 * config/sparc/sparc.c (assign_int_registers): Likewise.
10908 * config/spu/spu.c (spu_function_value): Likewise.
10909 (spu_function_arg): Likewise.
10910 * coverage.c (get_gcov_type): Likewise.
10911 (get_gcov_unsigned_t): Likewise.
10912 * dse.c (find_shift_sequence): Likewise.
10913 * expmed.c (store_bit_field_1): Likewise.
10914 * expr.c (convert_move): Likewise.
10915 (store_field): Likewise.
10916 * internal-fn.c (expand_arith_overflow): Likewise.
10917 * optabs-query.c (get_best_extraction_insn): Likewise.
10918 * optabs.c (expand_twoval_binop_libfunc): Likewise.
10919 * stor-layout.c (layout_type): Likewise.
10920 (initialize_sizetypes): Likewise.
10921 * targhooks.c (default_get_mask_mode): Likewise.
10922 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
10924 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10925 Alan Hayward <alan.hayward@arm.com>
10926 David Sherwood <david.sherwood@arm.com>
10928 * machmode.h (opt_mode::else_blk): New function.
10929 (int_mode_for_mode): Declare.
10930 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
10931 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
10933 * cfgexpand.c (expand_debug_expr): Likewise.
10934 * combine.c (gen_lowpart_or_truncate): Likewise.
10935 (gen_lowpart_for_combine): Likewise.
10936 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
10937 * config/avr/avr.c (avr_to_int_mode): Likewise.
10938 (avr_out_plus_1): Likewise.
10939 (avr_out_plus): Likewise.
10940 (avr_out_round): Likewise.
10941 * config/i386/i386.c (ix86_split_to_parts): Likewise.
10942 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
10943 (s390_expand_vcond): Likewise.
10944 * config/spu/spu.c (spu_split_immediate): Likewise.
10945 (spu_expand_mov): Likewise.
10946 * dse.c (get_stored_val): Likewise.
10947 * expmed.c (store_bit_field_1): Likewise.
10948 (convert_extracted_bit_field): Use int_mode_for_mode instead of
10950 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
10951 (extract_low_bits): Likewise.
10952 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
10953 handling rather than repeating the check.
10954 (emit_group_store): Likewise.
10955 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
10956 * optabs.c (expand_absneg_bit): Likewise.
10957 (expand_copysign_absneg): Likewise.
10958 (expand_copysign_bit): Likewise.
10959 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
10960 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
10961 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
10962 * var-tracking.c (prepare_call_arguments): Likewise.
10963 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
10964 int_mode_for_mode instead of mode_for_size.
10965 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
10967 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10968 Alan Hayward <alan.hayward@arm.com>
10969 David Sherwood <david.sherwood@arm.com>
10971 * machmode.h (int_mode_for_size): New function.
10972 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
10973 instead of mode_for_size.
10974 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
10976 * combine.c (expand_field_assignment): Use int_mode_for_size
10977 instead of mode_for_size.
10978 (make_extraction): Likewise.
10979 (simplify_shift_const_1): Likewise.
10980 (simplify_comparison): Likewise.
10981 * dojump.c (do_jump): Likewise.
10982 * dwarf2out.c (mem_loc_descriptor): Likewise.
10983 * emit-rtl.c (init_derived_machine_modes): Likewise.
10984 * expmed.c (flip_storage_order): Likewise.
10985 (convert_extracted_bit_field): Likewise.
10986 * expr.c (copy_blkmode_from_reg): Likewise.
10987 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
10988 * internal-fn.c (expand_mul_overflow): Likewise.
10989 * lower-subreg.c (simple_move): Likewise.
10990 * optabs-libfuncs.c (init_optabs): Likewise.
10991 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10992 * tree.c (vector_type_mode): Likewise.
10993 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
10994 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
10995 * tree-vect-generic.c (expand_vector_parallel): Likewise.
10996 * tree-vect-stmts.c (vectorizable_load): Likewise.
10997 (vectorizable_store): Likewise.
10999 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11000 Alan Hayward <alan.hayward@arm.com>
11001 David Sherwood <david.sherwood@arm.com>
11003 * coretypes.h (pod_mode): New type.
11004 (scalar_int_mode_pod): New typedef.
11005 * machmode.h (pod_mode): New class.
11006 (int_n_data_t::m): Change type to scalar_int_mode_pod.
11007 * genmodes.c (emit_mode_int_n): Update accordingly.
11008 * lower-subreg.h (target_lower_subreg): Change type to
11009 scalar_int_mode_pod.
11010 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
11011 scalar_int_mode_pod.
11013 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11014 Alan Hayward <alan.hayward@arm.com>
11015 David Sherwood <david.sherwood@arm.com>
11017 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
11018 machine_mode to scalar_int_mode.
11019 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
11020 (rs6000_option_override_internal): Remove cast to int.
11021 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
11022 machine_mode to scalar_int_mode.
11023 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
11024 (rs6000_option_override_internal): Remove cast to int.
11025 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
11026 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
11028 * config/s390/s390.c (s390_expand_builtin): Likewise.
11029 * coretypes.h (scalar_int_mode): New type.
11030 (opt_scalar_int_mode): New typedef.
11031 * machmode.h (scalar_int_mode): New class.
11032 (scalar_int_mode::includes_p): New function.
11033 (byte_mode): Change type to scalar_int_mode.
11034 (word_mode): Likewise.
11035 (ptr_mode): Likewise.
11036 * emit-rtl.c (byte_mode): Likewise.
11037 (word_mode): Likewise.
11038 (ptr_mode): Likewise.
11039 (init_derived_machine_modes): Update accordingly.
11040 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
11041 and MODE_PARTIAL_INT.
11042 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
11043 opt_scalar_int_mode.
11045 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11046 Alan Hayward <alan.hayward@arm.com>
11047 David Sherwood <david.sherwood@arm.com>
11049 * target.def (libgcc_floating_mode_supported_p): Take a
11051 * doc/tm.texi: Regenerate.
11052 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
11054 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
11055 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
11058 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11059 Alan Hayward <alan.hayward@arm.com>
11060 David Sherwood <david.sherwood@arm.com>
11062 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
11063 * doc/tm.texi: Regenerate.
11064 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
11065 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
11066 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
11067 * targhooks.h (default_floatn_mode): Likewise.
11068 * targhooks.c (default_floatn_mode): Likewise.
11069 * tree.c (build_common_tree_nodes): Update accordingly.
11071 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11072 Alan Hayward <alan.hayward@arm.com>
11073 David Sherwood <david.sherwood@arm.com>
11075 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
11076 (mode_iterator::iterate_p): Likewise.
11077 (mode_iterator::get_wider): Likewise.
11078 * expr.c (init_expr_target): Use opt_scalar_float_mode.
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 (opt_scalar_float_mode): New typedef.
11085 * machmode.h (float_mode_for_size): New function.
11086 * emit-rtl.c (double_mode): Delete.
11087 (init_emit_once): Use float_mode_for_size.
11088 * stor-layout.c (layout_type): Likewise.
11089 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
11091 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11092 Alan Hayward <alan.hayward@arm.com>
11093 David Sherwood <david.sherwood@arm.com>
11095 * output.h (assemble_real): Take a scalar_float_mode.
11096 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
11097 * config/arm/arm.md (consttable_4): Likewise.
11098 (consttable_8): Likewise.
11099 (consttable_16): Likewise.
11100 * config/mips/mips.md (consttable_float): Likewise.
11101 * config/s390/s390.c (s390_output_pool_entry): Likewise.
11102 * varasm.c (assemble_real): Take a scalar_float_mode.
11103 (output_constant_pool_2): Update accordingly.
11104 (output_constant): Likewise.
11106 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11107 Alan Hayward <alan.hayward@arm.com>
11108 David Sherwood <david.sherwood@arm.com>
11110 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
11111 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
11112 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
11113 (native_encode_real): Likewise.
11114 (native_interpret_real): Likewise.
11115 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
11116 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11118 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11119 Alan Hayward <alan.hayward@arm.com>
11120 David Sherwood <david.sherwood@arm.com>
11122 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
11123 <scalar_float_mode>. Simplify.
11124 (gen_extend_conv_libfunc): Likewise.
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 * coretypes.h (scalar_float_mode): New type.
11131 * machmode.h (mode_traits::from_int): Use machine_mode if
11132 USE_ENUM_MODES is defined.
11133 (is_a): New function.
11135 (dyn_cast): Likewise.
11136 (scalar_float_mode): New class.
11137 (scalar_float_mode::includes_p): New function.
11138 (is_float_mode): Likewise.
11139 * gdbhooks.py (MachineModePrinter): New class.
11140 (build_pretty_printer): Use it for scalar_float_mode.
11141 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
11142 (format_helper::format_helper): Turn into a template.
11143 * genmodes.c (get_mode_class): New function.
11144 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
11145 or machine_mode if none.
11146 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
11147 as_a <scalar_float_mode>.
11148 * dwarf2out.c (mem_loc_descriptor): Likewise.
11149 (insert_float): Likewise.
11150 (add_const_value_attribute): Likewise.
11151 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11152 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
11153 (expand_unop): Update accordingly.
11154 (expand_abs_nojump): Likewise.
11155 (expand_copysign_absneg): Take a scalar_float_mode.
11156 (expand_copysign_bit): Likewise.
11157 (expand_copysign): Update accordingly.
11159 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11160 Alan Hayward <alan.hayward@arm.com>
11161 David Sherwood <david.sherwood@arm.com>
11163 * coretypes.h (opt_mode): New class.
11164 * machmode.h (opt_mode): Likewise.
11165 (opt_mode::else_void): New function.
11166 (opt_mode::require): Likewise.
11167 (opt_mode::exists): Likewise.
11168 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
11169 (GET_MODE_2XWIDER_MODE): Likewise.
11170 (mode_iterator::get_wider): Update accordingly.
11171 (mode_iterator::get_2xwider): Likewise.
11172 (mode_iterator::get_known_wider): Likewise, turning into a template.
11173 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
11174 forcing a wider mode to exist.
11175 * config/cr16/cr16.h (LONG_REG_P): Likewise.
11176 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11177 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
11178 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
11179 * lower-subreg.c (init_lower_subreg): Likewise.
11180 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
11181 on the final iteration.
11182 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
11183 a wider mode exists before asking for a move pattern.
11184 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
11185 forcing a wider mode to exist.
11186 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
11187 returning false if no such mode exists.
11188 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
11189 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
11190 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
11191 Avoid checking for a MODE_INT if we already know the mode is not a
11193 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
11194 forcing a wider mode to exist.
11195 (expmed_mult_highpart_optab): Likewise.
11196 (expmed_mult_highpart): Likewise.
11197 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
11199 * lto-streamer-in.c (lto_input_mode_table): Likewise.
11200 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
11201 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
11202 * internal-fn.c (expand_mul_overflow): Update use of
11203 GET_MODE_2XWIDER_MODE.
11204 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11205 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
11206 GET_MODE_WIDER_MODE.
11207 (convert_plusminus_to_widen): Likewise.
11208 * tree-switch-conversion.c (array_value_type): Likewise.
11209 * var-tracking.c (emit_note_insn_var_location): Likewise.
11210 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11211 Return false inside rather than outside the loop if no wider mode
11213 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
11214 and GET_MODE_2XWIDER_MODE
11215 (can_compare_p): Use else_void.
11216 * gdbhooks.py (OptMachineModePrinter): New class.
11217 (build_pretty_printer): Use it for opt_mode.
11219 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11220 Alan Hayward <alan.hayward@arm.com>
11221 David Sherwood <david.sherwood@arm.com>
11223 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
11224 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
11226 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11227 Alan Hayward <alan.hayward@arm.com>
11228 David Sherwood <david.sherwood@arm.com>
11230 * machmode.h (mode_traits): New structure.
11231 (get_narrowest_mode): New function.
11232 (mode_iterator::start): Likewise.
11233 (mode_iterator::iterate_p): Likewise.
11234 (mode_iterator::get_wider): Likewise.
11235 (mode_iterator::get_known_wider): Likewise.
11236 (mode_iterator::get_2xwider): Likewise.
11237 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
11238 (FOR_EACH_MODE): Likewise.
11239 (FOR_EACH_MODE_FROM): Likewise.
11240 (FOR_EACH_MODE_UNTIL): Likewise.
11241 (FOR_EACH_WIDER_MODE): Likewise.
11242 (FOR_EACH_2XWIDER_MODE): Likewise.
11243 * builtins.c (expand_builtin_strlen): Use new mode iterators.
11244 * combine.c (simplify_comparison): Likewise
11245 * config/i386/i386.c (type_natural_mode): Likewise.
11246 * cse.c (cse_insn): Likewise.
11247 * dse.c (find_shift_sequence): Likewise.
11248 * emit-rtl.c (init_derived_machine_modes): Likewise.
11249 (init_emit_once): Likewise.
11250 * explow.c (hard_function_value): Likewise.
11251 * expmed.c (extract_fixed_bit_field_1): Likewise.
11252 (extract_bit_field_1): Likewise.
11253 (expand_divmod): Likewise.
11254 (emit_store_flag_1): Likewise.
11255 * expr.c (init_expr_target): Likewise.
11256 (convert_move): Likewise.
11257 (alignment_for_piecewise_move): Likewise.
11258 (widest_int_mode_for_size): Likewise.
11259 (emit_block_move_via_movmem): Likewise.
11260 (copy_blkmode_to_reg): Likewise.
11261 (set_storage_via_setmem): Likewise.
11262 (compress_float_constant): Likewise.
11263 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11264 * optabs-query.c (get_best_extraction_insn): Likewise.
11265 * optabs.c (expand_binop): Likewise.
11266 (expand_twoval_unop): Likewise.
11267 (expand_twoval_binop): Likewise.
11268 (widen_leading): Likewise.
11269 (widen_bswap): Likewise.
11270 (expand_parity): Likewise.
11271 (expand_unop): Likewise.
11272 (prepare_cmp_insn): Likewise.
11273 (prepare_float_lib_cmp): Likewise.
11274 (expand_float): Likewise.
11275 (expand_fix): Likewise.
11276 (expand_sfix_optab): Likewise.
11277 * postreload.c (move2add_use_add2_insn): Likewise.
11278 * reg-stack.c (reg_to_stack): Likewise.
11279 * reginfo.c (choose_hard_reg_mode): Likewise.
11280 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11281 * stor-layout.c (mode_for_size): Likewise.
11282 (smallest_mode_for_size): Likewise.
11283 (mode_for_vector): Likewise.
11284 (finish_bitfield_representative): Likewise.
11285 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
11286 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
11287 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11288 * var-tracking.c (prepare_call_arguments): Likewise.
11290 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11291 Alan Hayward <alan.hayward@arm.com>
11292 David Sherwood <david.sherwood@arm.com>
11294 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
11295 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
11296 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
11297 * machmode.h (mode_size): Move earlier in file.
11298 (mode_precision): Likewise.
11299 (mode_inner): Likewise.
11300 (mode_nunits): Likewise.
11301 (mode_unit_size): Likewise.
11302 (unit_unit_precision): Likewise.
11303 (mode_wider): Likewise.
11304 (mode_2xwider): Likewise.
11305 (machine_mode): New class.
11306 (mode_to_bytes): New function.
11307 (mode_to_bits): Likewise.
11308 (mode_to_precision): Likewise.
11309 (mode_to_inner): Likewise.
11310 (mode_to_unit_size): Likewise.
11311 (mode_to_unit_precision): Likewise.
11312 (mode_to_nunits): Likewise.
11313 (GET_MODE_SIZE): Use mode_to_bytes.
11314 (GET_MODE_BITSIZE): Use mode_to_bits.
11315 (GET_MODE_PRECISION): Use mode_to_precision.
11316 (GET_MODE_INNER): Use mode_to_inner.
11317 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
11318 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
11319 (GET_MODE_NUNITS): Use mode_to_nunits.
11320 * system.h (ALWAYS_INLINE): New macro.
11321 * config/powerpcspe/powerpcspe-c.c
11322 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
11323 int for arg1_mode and arg2_mode.
11325 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11326 Alan Hayward <alan.hayward@arm.com>
11327 David Sherwood <david.sherwood@arm.com>
11329 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
11330 Prefix mode names with E_ in case statements.
11331 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11332 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
11333 (aarch64_split_simd_move): Likewise.
11334 (aarch64_gen_storewb_pair): Likewise.
11335 (aarch64_gen_loadwb_pair): Likewise.
11336 (aarch64_gen_store_pair): Likewise.
11337 (aarch64_gen_load_pair): Likewise.
11338 (aarch64_get_condition_code_1): Likewise.
11339 (aarch64_constant_pool_reload_icode): Likewise.
11340 (get_rsqrte_type): Likewise.
11341 (get_rsqrts_type): Likewise.
11342 (get_recpe_type): Likewise.
11343 (get_recps_type): Likewise.
11344 (aarch64_gimplify_va_arg_expr): Likewise.
11345 (aarch64_simd_container_mode): Likewise.
11346 (aarch64_emit_load_exclusive): Likewise.
11347 (aarch64_emit_store_exclusive): Likewise.
11348 (aarch64_expand_compare_and_swap): Likewise.
11349 (aarch64_gen_atomic_cas): Likewise.
11350 (aarch64_emit_bic): Likewise.
11351 (aarch64_emit_atomic_swap): Likewise.
11352 (aarch64_emit_atomic_load_op): Likewise.
11353 (aarch64_evpc_trn): Likewise.
11354 (aarch64_evpc_uzp): Likewise.
11355 (aarch64_evpc_zip): Likewise.
11356 (aarch64_evpc_ext): Likewise.
11357 (aarch64_evpc_rev): Likewise.
11358 (aarch64_evpc_dup): Likewise.
11359 (aarch64_gen_ccmp_first): Likewise.
11360 (aarch64_gen_ccmp_next): Likewise.
11361 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11362 (alpha_emit_xfloating_libcall): Likewise.
11363 (emit_insxl): Likewise.
11364 (alpha_arg_type): Likewise.
11365 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
11366 (arc_preferred_simd_mode): Likewise.
11367 (arc_secondary_reload): Likewise.
11368 (get_arc_condition_code): Likewise.
11369 (arc_print_operand): Likewise.
11370 (arc_legitimate_constant_p): Likewise.
11371 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11372 * config/arc/arc.md (casesi_load): Likewise.
11373 (casesi_compact_jump): Likewise.
11374 * config/arc/predicates.md (proper_comparison_operator): Likewise.
11375 (cc_use_register): Likewise.
11376 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11377 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
11378 (arm_init_iwmmxt_builtins): Likewise.
11379 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
11380 (neon_expand_vector_init): Likewise.
11381 (arm_attr_length_move_neon): Likewise.
11382 (maybe_get_arm_condition_code): Likewise.
11383 (arm_emit_vector_const): Likewise.
11384 (arm_preferred_simd_mode): Likewise.
11385 (arm_output_iwmmxt_tinsr): Likewise.
11386 (thumb1_output_casesi): Likewise.
11387 (thumb2_output_casesi): Likewise.
11388 (arm_emit_load_exclusive): Likewise.
11389 (arm_emit_store_exclusive): Likewise.
11390 (arm_expand_compare_and_swap): Likewise.
11391 (arm_evpc_neon_vuzp): Likewise.
11392 (arm_evpc_neon_vzip): Likewise.
11393 (arm_evpc_neon_vrev): Likewise.
11394 (arm_evpc_neon_vtrn): Likewise.
11395 (arm_evpc_neon_vext): Likewise.
11396 (arm_validize_comparison): Likewise.
11397 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
11398 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
11399 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
11400 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
11401 (c6x_preferred_simd_mode): Likewise.
11402 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
11403 (epiphany_rtx_costs): Likewise.
11404 * config/epiphany/predicates.md (proper_comparison_operator):
11406 * config/frv/frv.c (condexec_memory_operand): Likewise.
11407 (frv_emit_move): Likewise.
11408 (output_move_single): Likewise.
11409 (output_condmove_single): Likewise.
11410 (frv_hard_regno_mode_ok): Likewise.
11411 (frv_matching_accg_mode): Likewise.
11412 * config/h8300/h8300.c (split_adds_subs): Likewise.
11413 (h8300_rtx_costs): Likewise.
11414 (h8300_print_operand): Likewise.
11415 (compute_mov_length): Likewise.
11416 (output_logical_op): Likewise.
11417 (compute_logical_op_length): Likewise.
11418 (compute_logical_op_cc): Likewise.
11419 (h8300_shift_needs_scratch_p): Likewise.
11420 (output_a_shift): Likewise.
11421 (compute_a_shift_length): Likewise.
11422 (compute_a_shift_cc): Likewise.
11423 (expand_a_rotate): Likewise.
11424 (output_a_rotate): Likewise.
11425 * config/i386/i386.c (classify_argument): Likewise.
11426 (function_arg_advance_32): Likewise.
11427 (function_arg_32): Likewise.
11428 (function_arg_64): Likewise.
11429 (function_value_64): Likewise.
11430 (ix86_gimplify_va_arg): Likewise.
11431 (ix86_legitimate_constant_p): Likewise.
11432 (put_condition_code): Likewise.
11433 (split_double_mode): Likewise.
11434 (ix86_avx256_split_vector_move_misalign): Likewise.
11435 (ix86_expand_vector_logical_operator): Likewise.
11436 (ix86_split_idivmod): Likewise.
11437 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
11438 (ix86_build_const_vector): Likewise.
11439 (ix86_build_signbit_mask): Likewise.
11440 (ix86_match_ccmode): Likewise.
11441 (ix86_cc_modes_compatible): Likewise.
11442 (ix86_expand_branch): Likewise.
11443 (ix86_expand_sse_cmp): Likewise.
11444 (ix86_expand_sse_movcc): Likewise.
11445 (ix86_expand_int_sse_cmp): Likewise.
11446 (ix86_expand_vec_perm_vpermi2): Likewise.
11447 (ix86_expand_vec_perm): Likewise.
11448 (ix86_expand_sse_unpack): Likewise.
11449 (ix86_expand_int_addcc): Likewise.
11450 (ix86_split_to_parts): Likewise.
11451 (ix86_vectorize_builtin_gather): Likewise.
11452 (ix86_vectorize_builtin_scatter): Likewise.
11453 (avx_vpermilp_parallel): Likewise.
11454 (inline_memory_move_cost): Likewise.
11455 (ix86_tieable_integer_mode_p): Likewise.
11456 (x86_maybe_negate_const_int): Likewise.
11457 (ix86_expand_vector_init_duplicate): Likewise.
11458 (ix86_expand_vector_init_one_nonzero): Likewise.
11459 (ix86_expand_vector_init_one_var): Likewise.
11460 (ix86_expand_vector_init_concat): Likewise.
11461 (ix86_expand_vector_init_interleave): Likewise.
11462 (ix86_expand_vector_init_general): Likewise.
11463 (ix86_expand_vector_set): Likewise.
11464 (ix86_expand_vector_extract): Likewise.
11465 (emit_reduc_half): Likewise.
11466 (ix86_emit_i387_round): Likewise.
11467 (ix86_mangle_type): Likewise.
11468 (ix86_expand_round_sse4): Likewise.
11469 (expand_vec_perm_blend): Likewise.
11470 (canonicalize_vector_int_perm): Likewise.
11471 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
11472 (expand_vec_perm_1): Likewise.
11473 (expand_vec_perm_interleave3): Likewise.
11474 (expand_vec_perm_even_odd_pack): Likewise.
11475 (expand_vec_perm_even_odd_1): Likewise.
11476 (expand_vec_perm_broadcast_1): Likewise.
11477 (ix86_vectorize_vec_perm_const_ok): Likewise.
11478 (ix86_expand_vecop_qihi): Likewise.
11479 (ix86_expand_mul_widen_hilo): Likewise.
11480 (ix86_expand_sse2_abs): Likewise.
11481 (ix86_expand_pextr): Likewise.
11482 (ix86_expand_pinsr): Likewise.
11483 (ix86_preferred_simd_mode): Likewise.
11484 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
11485 * config/i386/sse.md (*andnot<mode>3): Likewise.
11486 (<mask_codefor><code><mode>3<mask_name>): Likewise.
11487 (*<code><mode>3): Likewise.
11488 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
11489 (ia64_expand_atomic_op): Likewise.
11490 (ia64_arg_type): Likewise.
11491 (ia64_mode_to_int): Likewise.
11492 (ia64_scalar_mode_supported_p): Likewise.
11493 (ia64_vector_mode_supported_p): Likewise.
11494 (expand_vec_perm_broadcast): Likewise.
11495 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
11496 (iq2000_function_arg_advance): Likewise.
11497 (iq2000_function_arg): Likewise.
11498 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
11499 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
11500 (m68k_libcall_value): Likewise.
11501 (m68k_function_value): Likewise.
11502 (sched_attr_op_type): Likewise.
11503 * config/mcore/mcore.c (mcore_output_move): Likewise.
11504 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
11506 (microblaze_function_arg): Likewise.
11507 * config/mips/mips.c (mips16_build_call_stub): Likewise.
11508 (mips_print_operand): Likewise.
11509 (mips_mode_ok_for_mov_fmt_p): Likewise.
11510 (mips_vector_mode_supported_p): Likewise.
11511 (mips_preferred_simd_mode): Likewise.
11512 (mips_expand_vpc_loongson_even_odd): Likewise.
11513 (mips_expand_vec_unpack): Likewise.
11514 (mips_expand_vi_broadcast): Likewise.
11515 (mips_expand_vector_init): Likewise.
11516 (mips_expand_vec_reduc): Likewise.
11517 (mips_expand_msa_cmp): Likewise.
11518 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
11519 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
11520 (cc_flags_for_mode): Likewise.
11521 * config/msp430/msp430.c (msp430_print_operand): Likewise.
11522 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
11523 (nds32_output_casesi_pc_relative): Likewise.
11524 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11525 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
11526 (nvptx_gen_unpack): Likewise.
11527 (nvptx_gen_pack): Likewise.
11528 (nvptx_gen_shuffle): Likewise.
11529 (nvptx_gen_wcast): Likewise.
11530 (nvptx_preferred_simd_mode): Likewise.
11531 * config/pa/pa.c (pa_secondary_reload): Likewise.
11532 * config/pa/predicates.md (base14_operand): Likewise.
11533 * config/powerpcspe/powerpcspe-c.c
11534 (altivec_resolve_overloaded_builtin): Likewise.
11535 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
11537 (rs6000_preferred_simd_mode): Likewise.
11538 (output_vec_const_move): Likewise.
11539 (rs6000_expand_vector_extract): Likewise.
11540 (rs6000_split_vec_extract_var): Likewise.
11541 (reg_offset_addressing_ok_p): Likewise.
11542 (rs6000_legitimate_offset_address_p): Likewise.
11543 (rs6000_legitimize_address): Likewise.
11544 (rs6000_emit_set_const): Likewise.
11545 (rs6000_const_vec): Likewise.
11546 (rs6000_emit_move): Likewise.
11547 (spe_build_register_parallel): Likewise.
11548 (rs6000_darwin64_record_arg_recurse): Likewise.
11549 (swap_selector_for_mode): Likewise.
11550 (spe_init_builtins): Likewise.
11551 (paired_init_builtins): Likewise.
11552 (altivec_init_builtins): Likewise.
11553 (do_load_for_compare): Likewise.
11554 (rs6000_generate_compare): Likewise.
11555 (rs6000_expand_float128_convert): Likewise.
11556 (emit_load_locked): Likewise.
11557 (emit_store_conditional): Likewise.
11558 (rs6000_output_function_epilogue): Likewise.
11559 (rs6000_handle_altivec_attribute): Likewise.
11560 (rs6000_function_value): Likewise.
11561 (emit_fusion_gpr_load): Likewise.
11562 (emit_fusion_p9_load): Likewise.
11563 (emit_fusion_p9_store): Likewise.
11564 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
11565 (fusion_gpr_mem_load): Likewise.
11566 (fusion_addis_mem_combo_load): Likewise.
11567 (fusion_addis_mem_combo_store): Likewise.
11568 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
11569 (fusion_gpr_mem_load): Likewise.
11570 (fusion_addis_mem_combo_load): Likewise.
11571 (fusion_addis_mem_combo_store): Likewise.
11572 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11574 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
11575 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
11576 (rs6000_preferred_simd_mode): Likewise.
11577 (output_vec_const_move): Likewise.
11578 (rs6000_expand_vector_extract): Likewise.
11579 (rs6000_split_vec_extract_var): Likewise.
11580 (reg_offset_addressing_ok_p): Likewise.
11581 (rs6000_legitimate_offset_address_p): Likewise.
11582 (rs6000_legitimize_address): Likewise.
11583 (rs6000_emit_set_const): Likewise.
11584 (rs6000_const_vec): Likewise.
11585 (rs6000_emit_move): Likewise.
11586 (rs6000_darwin64_record_arg_recurse): Likewise.
11587 (swap_selector_for_mode): Likewise.
11588 (paired_init_builtins): Likewise.
11589 (altivec_init_builtins): Likewise.
11590 (rs6000_expand_float128_convert): Likewise.
11591 (emit_load_locked): Likewise.
11592 (emit_store_conditional): Likewise.
11593 (rs6000_output_function_epilogue): Likewise.
11594 (rs6000_handle_altivec_attribute): Likewise.
11595 (rs6000_function_value): Likewise.
11596 (emit_fusion_gpr_load): Likewise.
11597 (emit_fusion_p9_load): Likewise.
11598 (emit_fusion_p9_store): Likewise.
11599 * config/rx/rx.c (rx_gen_move_template): Likewise.
11600 (flags_from_mode): Likewise.
11601 * config/s390/predicates.md (s390_alc_comparison): Likewise.
11602 (s390_slb_comparison): Likewise.
11603 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
11604 (s390_vector_mode_supported_p): Likewise.
11605 (s390_cc_modes_compatible): Likewise.
11606 (s390_match_ccmode_set): Likewise.
11607 (s390_canonicalize_comparison): Likewise.
11608 (s390_emit_compare_and_swap): Likewise.
11609 (s390_branch_condition_mask): Likewise.
11610 (s390_rtx_costs): Likewise.
11611 (s390_secondary_reload): Likewise.
11612 (__SECONDARY_RELOAD_CASE): Likewise.
11613 (s390_expand_cs): Likewise.
11614 (s390_preferred_simd_mode): Likewise.
11615 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
11616 * config/sh/sh.c (sh_print_operand): Likewise.
11617 (dump_table): Likewise.
11618 (sh_secondary_reload): Likewise.
11619 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11620 * config/sh/sh.md (casesi_worker_1): Likewise.
11621 (casesi_worker_2): Likewise.
11622 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
11623 (fcc_comparison_operator): Likewise.
11624 * config/sparc/sparc.c (sparc_expand_move): Likewise.
11625 (emit_soft_tfmode_cvt): Likewise.
11626 (sparc_preferred_simd_mode): Likewise.
11627 (output_cbranch): Likewise.
11628 (sparc_print_operand): Likewise.
11629 (sparc_expand_vec_perm_bmask): Likewise.
11630 (vector_init_bshuffle): Likewise.
11631 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11632 (spu_vector_mode_supported_p): Likewise.
11633 (spu_expand_insv): Likewise.
11634 (spu_emit_branch_or_set): Likewise.
11635 (spu_handle_vector_attribute): Likewise.
11636 (spu_builtin_splats): Likewise.
11637 (spu_builtin_extract): Likewise.
11638 (spu_builtin_promote): Likewise.
11639 (spu_expand_sign_extend): Likewise.
11640 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11641 (tilegx_simd_int): Likewise.
11642 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
11643 (tilepro_simd_int): Likewise.
11644 * config/v850/v850.c (const_double_split): Likewise.
11645 (v850_print_operand): Likewise.
11646 (ep_memory_offset): Likewise.
11647 * config/vax/vax.c (vax_rtx_costs): Likewise.
11648 (vax_output_int_move): Likewise.
11649 (vax_output_int_add): Likewise.
11650 (vax_output_int_subtract): Likewise.
11651 * config/visium/predicates.md (visium_branch_operator): Likewise.
11652 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
11653 (visium_print_operand_address): Likewise.
11654 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11655 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
11656 (xtensa_expand_conditional_branch): Likewise.
11657 (xtensa_copy_incoming_a7): Likewise.
11658 (xtensa_output_literal): Likewise.
11659 * dfp.c (decimal_real_maxval): Likewise.
11660 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
11662 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11663 Alan Hayward <alan.hayward@arm.com>
11664 David Sherwood <david.sherwood@arm.com>
11666 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
11667 (mode_nunits_inline): Likewise.
11668 (mode_inner_inline): Likewise.
11669 (mode_unit_size_inline): Likewise.
11670 (mode_unit_precision_inline): Likewise.
11671 (emit_insn_modes_h): Likewise. Also emit a #define of the
11673 (emit_mode_wider): Add an E_ prefix to mode names.
11674 (emit_mode_complex): Likewise.
11675 (emit_mode_inner): Likewise.
11676 (emit_mode_adjustments): Likewise.
11677 (emit_mode_int_n): Likewise.
11678 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
11679 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
11680 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
11681 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
11682 (CRC32_BUILTIN, ENTRY): Likewise.
11683 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
11684 (aarch64_pop_regs): Likewise.
11685 (aarch64_process_components): Likewise.
11686 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
11687 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
11688 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
11689 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
11690 * config/arm/arm.c (arm_init_libfuncs): Likewise.
11691 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
11693 * config/i386/i386-builtin.def (pcmpestr): Likewise.
11694 (pcmpistr): Likewise.
11695 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
11696 * config/mmix/mmix.c (mmix_output_condition): Likewise.
11697 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
11699 * config/rl78/rl78.c (mduc_regs): Likewise.
11700 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
11701 (htm_expand_builtin): Likewise.
11702 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
11703 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
11704 * config/xtensa/xtensa.c (print_operand): Likewise.
11705 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
11706 (NUM_MODE_VECTOR_INT): Likewise.
11707 * genoutput.c (null_operand): Likewise.
11708 (output_operand_data): Likewise.
11709 * genrecog.c (print_parameter_value): Likewise.
11710 * lra.c (debug_operand_data): Likewise.
11712 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11714 * dwarf2out.c (scompare_loc_descriptor_wide)
11715 (scompare_loc_descriptor_narrow): New functions, split out from...
11716 (scompare_loc_descriptor): ...here.
11717 * expmed.c (emit_store_flag_int): New function, split out from...
11718 (emit_store_flag): ...here.
11720 2017-08-30 Richard Biener <rguenther@suse.de>
11722 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
11723 (dwarf2out_early_finish): Move setting of AT_pubnames from
11724 early debug output to early finish.
11726 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11728 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
11729 and -mdata-region to the assembler.
11731 2017-08-30 Richard Biener <rguenther@suse.de>
11733 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
11735 (gen_subprogram_die): Add DW_AT_object_pointer only early.
11736 (dwarf2out_early_global_decl): Only generate a DIE for the
11737 abstract origin if it doesn't already exist or is a declaration DIE.
11738 (resolve_addr): Do not add the linkage name twice when
11739 generating a stub DIE for the DW_TAG_GNU_call_site target.
11741 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11743 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11744 Use machine_mode rather than int for arg1_mode.
11746 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
11749 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
11750 that the second argument of the built-in functions to unpack
11751 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
11752 switch statement instead a lot of if statements.
11753 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
11754 Allow 64-bit values to be in Altivec registers as well as
11755 traditional floating point registers.
11756 (pack<mode>, FMOVE128_VSX iterator): Likewise.
11758 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
11760 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
11761 MAX_REGS_PER_ADDRESS == 1.
11763 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
11765 * config/i386/i386.opt (flag_fentry): Do not init to -1.
11766 * config/i386/i386.c (ix86_option_override_internal): Simplify
11767 setting of opts->x_flag_entry.
11769 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11770 Jakub Jelinek <jakub@redhat.com>
11771 Richard Biener <rguenther@suse.de>
11773 PR tree-optimization/81503
11774 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
11775 folded constant fits in the target type; reorder tests for clarity.
11777 2017-08-29 Martin Liska <mliska@suse.cz>
11779 * passes.def: Include pass_lower_switch.
11780 * stmt.c (dump_case_nodes): Remove and move to
11781 tree-switch-conversion.
11782 (case_values_threshold): Likewise.
11783 (expand_switch_as_decision_tree_p): Likewise.
11784 (emit_case_decision_tree): Likewise.
11785 (expand_case): Likewise.
11786 (balance_case_nodes): Likewise.
11787 (node_has_low_bound): Likewise.
11788 (node_has_high_bound): Likewise.
11789 (node_is_bounded): Likewise.
11790 (emit_case_nodes): Likewise.
11791 (struct simple_case_node): New struct.
11792 (add_case_node): Remove.
11793 (emit_case_dispatch_table): Use vector instead of case_list.
11794 (reset_out_edges_aux): Remove.
11795 (compute_cases_per_edge): Likewise.
11796 (expand_case): Build list of simple_case_node.
11797 (expand_sjlj_dispatch_table): Use it.
11798 * tree-switch-conversion.c (struct case_node): Moved from
11799 stmt.c and adjusted.
11800 (emit_case_nodes): Likewise.
11801 (node_has_low_bound): Likewise.
11802 (node_has_high_bound): Likewise.
11803 (node_is_bounded): Likewise.
11804 (case_values_threshold): Likewise.
11805 (reset_out_edges_aux): Likewise.
11806 (compute_cases_per_edge): Likewise.
11807 (add_case_node): Likewise.
11808 (dump_case_nodes): Likewise.
11809 (balance_case_nodes): Likewise.
11810 (expand_switch_as_decision_tree_p): Likewise.
11811 (emit_jump): Likewise.
11812 (emit_case_decision_tree): Likewise.
11813 (try_switch_expansion): Likewise.
11814 (do_jump_if_equal): Likewise.
11815 (emit_cmp_and_jump_insns): Likewise.
11816 (fix_phi_operands_for_edge): New function.
11817 (record_phi_operand_mapping): Likewise.
11818 (class pass_lower_switch): New pass.
11819 (pass_lower_switch::execute): New function.
11820 (make_pass_lower_switch): Likewise.
11821 (conditional_probability):
11822 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
11823 * tree-pass.h: Add make_pass_lower_switch.
11825 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11828 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
11831 2017-08-29 Richard Biener <rguenther@suse.de>
11833 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
11834 we do not add a DW_AT_inline attribute twice.
11835 (gen_subprogram_die): Remove code setting DW_AT_inline on
11836 DECL_ABSTRACT_P nodes.
11838 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
11840 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
11841 calls to internal functions.
11842 (gimplify_modify_expr): Likewise.
11843 * tree-call-cdce.c (use_internal_fn): Likewise.
11844 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
11845 (convert_to_divmod): Set the nothrow flag.
11846 * tree-if-conv.c (predicate_mem_writes): Likewise.
11847 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
11848 (vectorizable_call): Likewise.
11849 (vectorizable_store): Likewise.
11850 (vectorizable_load): Likewise.
11851 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
11852 (vect_recog_mask_conversion_pattern): Likewise.
11854 2017-08-29 Martin Liska <mliska@suse.cz>
11857 * gcc.c (driver_handle_option): Add new argument.
11858 * opts-common.c (handle_option): Pass
11859 target_option_override_hook.
11860 * opts-global.c (lang_handle_option): Add new option.
11861 (set_default_handlers): Add new argument.
11862 (decode_options): Likewise.
11863 * opts.c (target_handle_option): Likewise.
11864 (common_handle_option): Call target_option_override_hook.
11865 * opts.h (struct cl_option_handler_func): Add hook for
11866 target option override.
11867 (struct cl_option_handlers): Likewise.
11868 (set_default_handlers): Add new argument.
11869 (decode_options): Likewise.
11870 (common_handle_option): Likewise.
11871 (target_handle_option): Likewise.
11872 * toplev.c (toplev::main): Pass targetm.target_option.override
11875 2017-08-29 Richard Biener <rguenther@suse.de>
11876 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
11878 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
11879 life to the active subtree.
11881 2017-08-28 Jeff Law <law@redhat.com>
11883 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
11884 derive_equivalences.
11885 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
11886 Code moved into....
11887 (edge_info::derive_equivalences): New private member function
11889 * tree-ssa-dom.c (class edge_info): Changed from a struct
11890 to a class. Add ctor/dtor, methods and data members.
11891 (edge_info::edge_info): Renamed from allocate_edge_info.
11892 Initialize additional members.
11893 (edge_info::~edge_info): New.
11894 (free_dom_edge_info): Delete the edge info.
11895 (record_edge_info): Use new class & associated member functions.
11896 Tighten forms for testing for edge equivalences.
11897 (record_temporary_equivalences): Iterate over the simple
11898 equivalences rather than assuming there's only one per edge.
11899 (cprop_into_successor_phis): Iterate over the simple
11900 equivalences rather than assuming there's only one per edge.
11901 (optimize_stmt): Use operand_equal_p rather than pointer
11902 equality for mini-DSE code.
11904 2017-08-28 Nathan Sidwell <nathan@acm.org>
11906 * gcc.c (execute): Fold SIGPIPE handling into switch
11907 statement. Adjust internal error message.
11909 2017-08-28 Richard Biener <rguenther@suse.de>
11912 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
11913 Do nothing for removed DIEs.
11915 2017-08-28 Richard Biener <rguenther@suse.de>
11917 PR tree-optimization/81977
11918 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
11921 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
11924 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
11925 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
11926 using targetm.gen_mem_thread_fence.
11928 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
11931 * config/i386/i386.md (*<btsc><mode>): Change operand 2
11932 predicate to register_operand. Reorder operands.
11933 (*btr<mode>): Ditto.
11934 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
11935 (*btr<mode>_mask): Ditto.
11937 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
11939 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
11940 * config/rs6000/xmmintrin.h: New file.
11941 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
11943 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11946 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
11947 parameter and_insn and return it.
11948 (recombine_lvx_pattern): Insert a copy to ensure availability of
11949 the base register of the copied masking operation at the point of
11950 the instruction replacement.
11951 (recombine_stvx_pattern): Likewise.
11953 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
11955 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
11956 undocumented switches.
11957 (-mpower9-dform-vector): Likewise.
11958 (-mpower9-dform): Likewise.
11959 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
11960 comments to delete references to -mpower9-dform* switches.
11961 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
11962 Delete reference to -mpower9-dform* switches, test for
11963 -mpower9-vector instead.
11964 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
11965 (OTHER_P9_VECTOR_MASKS): Likewise.
11966 (POWERPC_MASKS): Likewise.
11967 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
11968 tests against -mpower9-dform* to -mpower9-vector. Delete code
11969 that checked for -mpower9-dform* consistancy with other options.
11970 Add test for -mpower9-misc to enable other power9 switches.
11971 (rs6000_init_hard_regno_mode_ok): Likewise.
11972 (rs6000_option_override_internal): Likewise.
11973 (rs6000_emit_prologue): Likewise.
11974 (rs6000_emit_epilogue): Likewise.
11975 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
11976 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
11977 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
11979 (emit_fusion_p9_store): Likewise.
11980 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
11981 resetting these macros if the assembler does not support ISA 3.0
11983 (TARGET_P9_DFORM_VECTOR): Likewise.
11984 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
11985 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
11987 2017-08-25 Alan Modra <amodra@gmail.com>
11989 PR rtl-optimization/81747
11990 * cse.c (cse_extended_basic_block): Don't attempt to record
11991 equivalences for degenerate conditional branches that branch
11992 to their fall-through.
11994 2017-08-24 Martin Sebor <msebor@redhat.com>
11996 PR middle-end/81908
11997 * gimple-fold.c (size_must_be_zero_p): New function.
11998 (gimple_fold_builtin_memory_op): Call it.
12000 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
12002 * config/rs6000/mm_malloc.h: New file.
12004 2017-08-24 Bin Cheng <bin.cheng@arm.com>
12006 PR tree-optimization/81913
12007 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
12008 analysis when either IVs in condition can wrap.
12010 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
12012 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
12013 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
12015 2017-08-24 Richard Biener <rguenther@suse.de>
12018 * targhooks.c (default_target_can_inline_p): Properly
12019 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
12020 is present and always compare.
12021 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
12022 infer -mfpmath=sse from TARGET_SSE_P.
12023 (ix86_can_inline_p): Properly use target_option_default_node when
12024 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
12026 2017-08-24 Richard Biener <rguenther@suse.de>
12029 * dwarf2out.c (output_die): Handle flag_generate_offload like
12031 (output_comp_unit): Likewise.
12032 (gen_array_type_die): Likewise.
12033 (dwarf2out_early_finish): Likewise.
12034 (note_variable_value_in_expr): Likewise.
12035 (dwarf2out_finish): Likewise. Adjust assert.
12036 * cgraphunit.c (symbol_table::compile): Move setting of
12037 flag_generate_offload earlier ...
12038 (symbol_table::finalize_compilation_unit): ... here, before
12039 early debug finalization.
12041 2017-08-24 Richard Biener <rguenther@suse.de>
12043 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
12044 and ipa-fnsummary.h.
12045 (ix86_can_inline_p): When ix86_fpmath flags do not match
12046 check whether the callee uses FP math at all.
12048 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
12050 PR middle-end/81931
12051 * tree-ssanames.c (get_nonzero_bits): Use element_precision
12052 instead of TYPE_PRECISION.
12054 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
12055 Alan Hayward <alan.hayward@arm.com>
12056 David Sherwood <david.sherwood@arm.com>
12058 * combine.c (make_extraction): Use subreg_offset_from_lsb.
12060 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
12062 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
12064 (ix86_frame::stack_realign_allocate): New field.
12065 (struct machine_frame_state): Modify comments.
12066 (machine_frame_state::sp_realigned_fp_end): New field.
12067 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
12068 layout calculation.
12069 (sp_valid_at): Add assertion to assure no attempt to access invalid
12070 offset of a realigned stack.
12071 (fp_valid_at): Likewise.
12072 (choose_baseaddr): Modify comments.
12073 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
12074 ix86_expand_prologue.
12075 (ix86_expand_prologue): Modify stack realignment and allocation.
12076 (ix86_expand_epilogue): Modify comments.
12077 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
12078 avx2_runtime, avx512f, and avx512f_runtime.
12080 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
12082 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
12083 (mstackrealign): Do not init to -1.
12084 * config/i386/i386.c (ix86_option_override_internal):
12085 Check opts_set, not opts when setting default value of
12086 opts->x_ix86_force_align_arg_pointer.
12088 2017-08-23 Richard Biener <rguenther@suse.de>
12090 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
12091 lang_hooks.decl_printable_name.
12092 * print-rtl-function.c (print_rtx_function): Likewise.
12093 * tree-pretty-print.c (dump_function_header): Likewise.
12095 2017-08-23 Richard Biener <rguenther@suse.de>
12098 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
12099 -g0 at compile-time.
12101 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12103 PR middle-end/19706
12104 * doc/sourcebuild.texi (Other hardware attributes):
12107 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12109 PR middle-end/19706
12110 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
12111 Fix single-use check.
12113 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
12115 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
12117 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12119 * doc/install.texi: Modify to add more details on running selected
12122 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12124 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
12125 is combined with -mabi=ms.
12126 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
12129 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12131 PR tree-optimization/81488
12132 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
12133 and cached_basis fields.
12134 (MAX_SPREAD): New constant.
12135 (alloc_cand_and_find_basis): Initialize new fields.
12136 (clear_visited): New function.
12137 (create_phi_basis_1): Rename from create_phi_basis, set visited
12138 and cached_basis fields.
12139 (create_phi_basis): New wrapper function.
12140 (phi_add_costs_1): Rename from phi_add_costs, add spread
12141 parameter, set visited field, short-circuit when limits reached.
12142 (phi_add_costs): New wrapper function.
12143 (record_phi_increments_1): Rename from record_phi_increments, set
12145 (record_phi_increments): New wrapper function.
12146 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
12147 (phi_incr_cost): New wrapper function.
12148 (all_phi_incrs_profitable_1): Rename from
12149 all_phi_incrs_profitable, set visited field.
12150 (all_phi_incrs_profitable): New wrapper function.
12152 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12153 Alan Hayward <alan.hayward@arm.com>
12154 David Sherwood <david.sherwood@arm.com>
12156 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
12157 that takes the outer and inner modes.
12158 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
12159 comparison as the canonical test for a paradoxical subreg.
12160 * combine.c (simplify_set): Use paradoxical_subreg_p.
12161 (make_extraction): Likewise.
12162 (force_to_mode): Likewise.
12163 (rtx_equal_for_field_assignment_p): Likewise.
12164 (gen_lowpart_for_combine): Likewise.
12165 (simplify_comparison): Likewise.
12166 * cse.c (equiv_constant): Likewise.
12167 * expmed.c (store_bit_field_1): Likewise.
12168 * final.c (alter_subreg): Likewise.
12169 * fwprop.c (propagate_rtx): Likewise.
12170 (forward_propagate_subreg): Likewise.
12171 * ira-conflicts.c (ira_build_conflicts): Likewise.
12172 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
12173 * lra-constraints.c (curr_insn_transform): Likewise.
12174 (split_reg): Likewise.
12175 * lra-eliminations.c (move_plus_up): Likewise.
12176 (lra_eliminate_regs_1): Likewise.
12177 * recog.c (general_operand): Likewise.
12178 * ree.c (combine_reaching_defs): Likewise.
12179 * reload.c (push_reload): Likewise.
12180 (find_reloads): Likewise.
12181 * reload1.c (elimination_effects): Likewise.
12182 (compute_reload_subreg_offset): Likewise.
12183 (choose_reload_regs): Likewise.
12184 * rtlanal.c (subreg_lsb_1): Likewise.
12185 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12186 (simplify_subreg): Likewise.
12187 * var-tracking.c (track_loc_p): Likewise.
12188 * emit-rtl.c (byte_lowpart_offset): Likewise.
12189 (paradoxical_subreg_p): Delete out-of-line definition.
12191 2017-08-22 Jeff Law <law@redhat.com>
12193 PR tree-optimization/81741
12194 PR tree-optimization/71947
12195 * tree-ssa-dom.c: Include tree-inline.h.
12196 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
12197 equivalences if one is more expensive to compute than the other.
12198 * tree-ssa-scopedtables.h (class const_or_copies): Make
12199 record_const_or_copy_raw method private.
12200 (class avail_exprs_stack): New method simplify_binary_operation.
12201 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
12202 avail_exprs_stack::simplify_binary_operation as needed.
12203 (avail_exprs_stack::simplify_binary_operation): New function.
12205 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
12207 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
12208 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
12209 (DOT_SYMBOLS): Likewise.
12210 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
12211 (RELOCATABLE_NEEDS_FIXUP): Likewise.
12212 (RS6000_ABI_NAME): Likewise.
12213 (TARGET_CMODEL): Likewise.
12214 (TOC_SECTION_ASM_OP): Likewise.
12215 (SET_CMODEL): New macro.
12216 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
12218 2017-08-22 Richard Biener <rguenther@suse.de>
12220 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
12221 to allow for free-lang-data replacements similar to verify_type_variant.
12223 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12224 Alan Hayward <alan.hayward@arm.com>
12225 David Sherwood <david.sherwood@arm.com>
12227 * config/aarch64/aarch64.md (casesi): Use DImode rather than
12228 VOIDmode for the LABEL_REF.
12230 2017-08-22 Richard Biener <rguenther@suse.de>
12232 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
12234 2017-08-22 Richard Biener <rguenther@suse.de>
12236 * common.opt (feliminate-dwarf2-dups): Ignore.
12237 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
12238 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
12239 same_die_p_wrap, compute_section_prefix,
12240 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
12241 (comdat_symbol_id, comdat_symbol_number): Likewise.
12242 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
12244 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
12245 (output_die): Mark unreachable path unreachable.
12246 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
12247 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
12248 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
12249 (dwarf2out_early_finish): Likewise.
12251 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
12253 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
12255 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
12258 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
12259 not VAR_P. Filter attribute warnings with OPT_Wattributes.
12260 (avr_attribute_table) <io, io_low, address>: Initialize
12261 .decl_required with true.
12263 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
12265 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
12266 undocumented debugging options.
12267 (-mvsx-scalar-double): Likewise.
12268 (-mallow-df-permute): Likewise.
12269 (-mvectorize-builtins): Likewise.
12270 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12271 (rs6000_builtin_vectorized_function): Likewise.
12272 (rs6000_builtin_md_vectorized_function): Likewise.
12273 (rs6000_opt_vars): Likewise.
12275 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
12278 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
12279 (*btrq_imm): Rename from *btrq.
12280 (*btcq_imm): Rename from *btcq.
12281 (btsc): New code attribute.
12282 (*<btsc><mode>): New insn pattern.
12283 (*btr<mode>): Ditto.
12284 (*<btsc><mode>_mask): New insn_and_split pattern.
12285 (*btr<mode>_mask): Ditto.
12287 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12288 Alan Hayward <alan.hayward@arm.com>
12289 David Sherwood <david.sherwood@arm.com>
12291 * function.c (pad_below): Simplify padding calculation.
12293 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12294 Alan Hayward <alan.hayward@arm.com>
12295 David Sherwood <david.sherwood@arm.com>
12297 * target.def (function_prologue): Remove frame size argument.
12298 (function_epilogue): Likewise.
12299 * doc/tm.texi: Regenerate.
12300 * final.c (final_start_function): Update call to function_prologue.
12301 (final_end_function): Update call to function_epilogue.
12302 (default_function_pro_epilogue): Remove frame size argument.
12303 * output.h (default_function_pro_epilogue): Likewise.
12304 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
12305 (arm_output_function_prologue): Likewise.
12306 * config/frv/frv.c (frv_function_prologue): Likewise.
12307 (frv_function_epilogue): Likewise.
12308 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
12309 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
12310 (ia64_output_function_epilogue): Likewise.
12311 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
12312 (m32r_output_function_epilogue): Likewise.
12313 * config/microblaze/microblaze.c (microblaze_function_prologue)
12314 (microblaze_function_epilogue): Likewise.
12315 * config/mips/mips.c (mips_output_function_prologue): Likewise.
12316 (mips_output_function_epilogue): Likewise.
12317 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
12318 (mmix_target_asm_function_epilogue): Likewise.
12319 * config/msp430/msp430.c (msp430_start_function): Likewise.
12320 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
12321 (nds32_asm_function_epilogue): Likewise.
12322 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
12323 * config/pa/pa.c (pa_output_function_prologue): Likewise.
12324 (pa_output_function_epilogue): Likewise.
12325 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
12326 (rs6000_output_function_epilogue): Likewise.
12327 * config/rl78/rl78.c (rl78_start_function): Likewise.
12328 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
12329 (rs6000_output_function_epilogue): Likewise.
12330 * config/rx/rx.c (rx_output_function_prologue): Likewise.
12331 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
12332 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
12333 (sparc_asm_function_epilogue): Likewise.
12335 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12337 * tree.h (type_has_mode_precision_p): New function.
12338 * convert.c (convert_to_integer_1): Use it.
12339 * expr.c (expand_expr_real_2): Likewise.
12340 (expand_expr_real_1): Likewise.
12341 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
12342 * match.pd: Likewise.
12343 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
12344 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
12345 * tree-tailcall.c (process_assignment): Likewise.
12346 * tree-vect-loop.c (vectorizable_reduction): Likewise.
12347 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
12348 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
12349 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12350 (vectorizable_assignment): Likewise.
12351 (vectorizable_shift): Likewise.
12352 (vectorizable_operation): Likewise.
12353 * tree-vrp.c (register_edge_assert_for_2): Likewise.
12355 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
12357 * match.pd: Add pow (C, x) simplification.
12359 2017-08-21 Richard Biener <rguenther@suse.de>
12361 PR tree-optimization/81900
12362 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
12363 for blocks with abnormal predecessors.
12364 (compute_antic): Do not set visited flag prematurely.
12366 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
12369 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
12371 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12373 * stor-layout.h (vector_type_mode): Move to...
12374 * tree.h (vector_type_mode): ...here.
12375 * stor-layout.c (vector_type_mode): Move to...
12376 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
12378 2017-08-21 Richard Biener <rguenther@suse.de>
12380 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
12381 register_external_die hooks.
12382 (debug_false_tree_charstarstar_uhwistar): Declare.
12383 (debug_nothing_tree_charstar_uhwi): Likewise.
12384 * debug.c (do_nothing_debug_hooks): Adjust.
12385 (debug_false_tree_charstarstar_uhwistar): New do nothing.
12386 (debug_nothing_tree_charstar_uhwi): Likewise.
12387 * dbxout.c (dbx_debug_hooks): Adjust.
12388 (xcoff_debug_hooks): Likewise.
12389 * sdbout.c (sdb_debug_hooks): Likewise.
12390 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
12391 * dwarf2out.c (macinfo_label_base): New global.
12392 (dwarf2out_register_external_die): New function for the
12393 register_external_die hook.
12394 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
12395 (dwarf2_debug_hooks): Use them.
12396 (dwarf2_lineno_debug_hooks): Adjust.
12397 (struct die_struct): Add with_offset flag.
12398 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
12399 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
12400 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
12401 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
12402 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
12403 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
12404 defining section names for the early LTO debug variants.
12405 (reset_indirect_string): New helper.
12406 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
12407 (print_dw_val): Add support for offsetted symbol references.
12408 (get_ultimate_context): Split out from is_cxx.
12409 (is_cxx): Use get_ultimate_context.
12410 (is_fortran): Add decl overload.
12411 (compute_comp_unit_symbol): Split out worker from
12412 compute_section_prefix.
12413 (compute_section_prefix): Call compute_comp_unit_symbol and
12414 set comdat_type_p here.
12415 (output_die): Skip DIE symbol output for the LTO added one.
12416 Handle DIE symbol references with offset.
12417 (output_comp_unit): Guard section name mangling properly.
12418 For LTO debug sections emit a symbol at the section beginning
12419 which we use to refer to its DIEs.
12420 (add_abstract_origin_attribute): For DIEs registered via
12421 dwarf2out_register_external_die directly refer to the early
12422 DIE rather than indirectly through the shadow one we created.
12423 Remove obsolete call to dwarf2out_abstract_function for
12424 non-function/block origins.
12425 (gen_array_type_die): When generating early LTO debug do
12426 not emit DW_AT_string_length.
12427 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
12428 late when in LTO. As suggested place a gcc_unreachable for
12429 the DECL_ABSTRACT_P case.
12430 (gen_subprogram_die): Avoid another specification DIE
12431 for early built declarations/definitions for the late LTO case.
12432 (gen_variable_die): Add type references for late duplicated VLA dies
12434 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
12435 we have the abstract instance already.
12436 (process_scope_var): Adjust decl DIE contexts in LTO which
12437 first puts them in limbo.
12438 (gen_decl_die): Do not generate type DIEs late apart from
12439 types for VLAs or for decls we do not yet have a DIE. Do not
12440 call dwarf2out_abstract_function late.
12441 (dwarf2out_early_global_decl): Make sure to create DIEs
12442 for abstract instances of a decl first.
12443 (dwarf2out_late_global_decl): Adjust comment.
12444 (output_macinfo_op): With multiple macro sections use
12445 macinfo_label_base to distinguish labels.
12446 (output_macinfo): Likewise. Update macinfo_label_base.
12447 Pass in the line info label.
12448 (note_variable_value_in_expr): When generating LTO resolve
12449 all variable values here by generating DIEs as needed.
12450 (init_sections_and_labels): Add early LTO debug flag parameter
12451 and generate different sections and names if set. Add generation
12452 counter for the labels so we can have multiple of them.
12453 (reset_dies): Helper to allow DIEs to be output multiple times.
12454 (dwarf2out_finish): When outputting DIEs to the fat part of an
12455 LTO object first reset DIEs.
12456 (dwarf2out_early_finish): Output early DIEs when generating LTO.
12457 (modified_type_die): Check for decl_ultimate_origin being self
12459 (gen_type_die_with_usage): Likewise.
12460 (gen_typedef_die): Allow decl_ultimate_origin being self.
12461 (set_decl_abstract_flags): Remove.
12462 (set_block_abstract_flags): Likewise.
12463 (dwarf2out_abstract_function): Treat the early generated DIEs
12464 as the abstract copy and only add DW_AT_inline and
12465 DW_AT_artificial here and call set_decl_origin_self.
12466 If the DIE has an abstract origin don't do anything.
12467 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
12468 if we have none yet (Go fails to build one, PR78628).
12469 (variably_modified_type_p): Prevent endless recursion for Ada
12470 cyclic pointer types.
12471 * lto-streamer-in.c: Include debug.h.
12472 (dref_queue): New global.
12473 (lto_read_tree_1): Stream in DIE references.
12474 (lto_input_tree): Register DIE references.
12475 (input_function): Stream DECL_DEBUG_ARGS.
12476 * lto-streamer-out.c: Include debug.h.
12477 (lto_write_tree_1): Output DIE references.
12478 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
12479 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
12480 (output_function): Stream DECL_DEBUG_ARGS.
12481 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
12482 Stream DECL_ABSTRACT_ORIGIN.
12483 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
12484 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
12485 DECL_CONTEXT for file-scope decls.
12486 * lto-streamer.h (struct dref_entry): Declare.
12487 (dref_queue): Likewise.
12488 * cfgexpand.c (pass_expand::execute): Do not call the
12489 outlining_inline_function hook here.
12490 * lto-wrapper.c (debug_obj): New global.
12491 (tool_cleanup): Unlink it if required.
12492 (debug_objcopy): New function.
12493 (run_gcc): Handle early debug sections in the IL files by
12494 extracting them to separate files, partially linkin them and
12495 feeding the result back as result to the linker.
12496 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
12497 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
12498 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
12499 sections into a separate segment.
12500 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
12502 (darwin_asm_dwarf_section): Likewise.
12503 (darwin_asm_output_dwarf_offset): Likewise.
12504 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
12506 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12507 Alan Hayward <alan.hayward@arm.com>
12508 David Sherwood <david.sherwood@arm.com>
12510 * read-md.h (md_reader::record_potential_iterator_use): Replace
12511 pointer argument with an rtx and an index.
12512 * read-rtl.c (iterator_group::apply_iterator): Likewise.
12513 (apply_mode_iterator): Likewise.
12514 (apply_code_iterator): Likewise.
12515 (apply_int_iterator): Likewise.
12516 (apply_subst_iterator): Likewise.
12517 (record_iterator_use): Likewise.
12518 (record_attribute_use): Likewise.
12519 (md_reader::record_potential_iterator_use): Likewise. Update calls
12520 to record_iterator_use and apply_iterator.
12521 (iterator_use): Replace ptr with x and index.
12522 (attribute_use): Likewise.
12523 (apply_attribute_uses): Update calls to apply_iterator.
12524 (apply_iterators): Likewise. Update initialization of iterator_use.
12525 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
12526 and record_potential_iterator_use.
12527 (rtx_reader::read_rtx_operand): Likewise.
12529 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12530 Alan Hayward <alan.hayward@arm.com>
12531 David Sherwood <david.sherwood@arm.com>
12533 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
12536 2017-08-21 Richard Biener <rguenther@suse.de>
12538 PR middle-end/81884
12539 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
12540 at struct end conservatively when comparing common bases.
12542 2017-08-21 Richard Biener <rguenther@suse.de>
12544 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
12545 (mem_ref_in_stmt): Remove.
12546 (determine_max_movement): Use ref index to get at the reference.
12547 (invariantness_dom_walker::before_dom_children): Deal with
12548 lim data already initialized.
12549 (gather_mem_refs_stmt): Initialize lim data and record ref index.
12551 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12553 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
12554 (TARGET_ISA_ROUND): Ditto.
12555 (TARGET_ROUND): Ditto.
12556 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
12557 * config/i386/i386.md: Ditto.
12558 * config/i386/sse.md: Ditto.
12559 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
12560 with OPTION_MASK_ISA_SSE4_1.
12562 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12565 * doc/extend.texi (x86 Built-in Functions): Correct the name of
12566 __builtin_ia32_lzcnt_u16.
12568 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
12571 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
12572 (rs6000_set_current_function): Rewrite function to use it.
12574 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
12577 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
12578 and TYPE_WARN_IF_NOT_ALIGN.
12579 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
12580 (handle_warn_if_not_align): New.
12581 (place_union_field): Call handle_warn_if_not_align.
12582 (place_field): Call handle_warn_if_not_align.
12583 Copy TYPE_WARN_IF_NOT_ALIGN.
12584 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
12585 (layout_type): Likewise.
12586 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
12588 (tree_decl_common): Add warn_if_not_align.
12589 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
12590 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
12591 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
12592 (DECL_WARN_IF_NOT_ALIGN): Likewise.
12593 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
12594 * doc/extend.texi: Document warn_if_not_aligned attribute.
12595 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
12597 2017-08-17 Martin Liska <mliska@suse.cz>
12600 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
12601 (get_data_dependence): Use it as pointer type.
12602 (distribute_loop): Likewise.
12604 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12606 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
12607 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
12608 (p8_vmrgow_<mode>_direct): New define_insn.
12609 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
12610 handle endianness for vmrgew and vmrgow permute patterns.
12612 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
12614 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
12615 * config/rs6000/rs6000-cpus.def: Remove comment.
12616 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
12617 (POWERPC_MASKS): Likewise.
12618 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
12619 use of TARGET_VSX_TIMODE.
12620 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
12621 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
12622 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
12623 (rs6000_option_override_internal): Remove dead code.
12624 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
12625 (rs6000_legitimize_reload_address): Likewise.
12626 (rs6000_legitimate_address_p): Likewise.
12627 (rs6000_opt_masks): Delete "vsx-timode".
12628 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
12629 from function comment.
12630 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
12631 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
12632 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
12634 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
12635 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
12636 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
12637 (VSX_TI): Likewise.
12639 (define_peephole2): Likewise.
12641 2017-08-17 Martin Sebor <msebor@redhat.com>
12644 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
12645 past the end of an array.
12646 (test_pp_format): Add test cases.
12648 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
12650 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
12651 missing ECF_NOTHROW flags.
12653 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
12656 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
12657 operands residing in integer registers.
12658 (*vsx_le_perm_load_<mode>): Likewise.
12659 (*vsx_le_perm_store_<mode>): Likewise.
12660 (define_peephole2): Add peepholes to optimize the above.
12662 2017-08-17 Marek Polacek <polacek@redhat.com>
12664 PR middle-end/81814
12665 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
12666 to mimic what shorten_compare did. Change the return type to bool.
12667 (fold_cond_expr_with_comparison): Update call to
12668 operand_equal_for_comparison_p.
12669 (fold_ternary_loc): Likewise.
12671 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
12673 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
12675 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
12677 2017-08-17 Richard Biener <rguenther@suse.de>
12679 * tree-ssa-structalias.c (solve_graph): When propagating
12680 to successors update the graphs succ edges and avoid duplicate work.
12682 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
12685 * config/i386/i386.c (ix86_option_override_internal): Save target
12686 specific options after ix86_stack_protector_guard_reg was changed.
12688 2017-08-17 Richard Biener <rguenther@suse.de>
12690 PR tree-optimization/81827
12691 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
12692 (new_var_info): Initialize it conservatively.
12693 (get_call_vi): Mark register vars.
12694 (new_scalar_tmp_constraint_exp): Likewise.
12695 (handle_rhs_call): Likewise.
12696 (handle_const_call): Likewise.
12697 (create_function_info_for): Likewise.
12698 (solve_constraints): Sort varinfos to separate register from
12699 non-register vars to pack points-to solution bitmaps during
12702 2017-08-17 Marek Polacek <polacek@redhat.com>
12704 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
12706 2017-08-17 Richard Biener <rguenther@suse.de>
12708 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
12709 to true when overflow is undefined and we saturated the result.
12711 2017-08-17 Alan Modra <amodra@gmail.com>
12714 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
12715 Don't use store multiple if only one reg needs saving.
12716 (interesting_frame_related_regno): New function.
12717 (rs6000_frame_related): Don't emit frame info for regs that
12719 (rs6000_emit_epilogue): Likewise.
12721 2017-08-16 Nathan Sidwell <nathan@acm.org>
12723 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
12724 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
12725 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
12726 * tree.c (free_lang_data_in_type): Use else-if chain. Always
12727 clear TYPE_LANG_1. Remove obsolete member-function stripping.
12728 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
12729 (verify_type): Adjust for TYPE_BINFO move.
12730 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
12731 process TYPE_BINFO directly.
12732 (hash_tree): Likewise.
12733 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
12735 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
12738 2017-08-16 David Malcolm <dmalcolm@redhat.com>
12740 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
12742 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
12745 * config/i386/i386.md (*anddi_1_btr): Change predicates of
12746 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
12747 Add ix86_binary_operator_ok to insn constraint.
12748 (*iordi_1_bts): Ditto.
12749 (*xordi_1_btc): Ditto.
12750 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
12751 Update corresponding peephole2 pattern.
12755 2017-08-16 Bin Cheng <bin.cheng@arm.com>
12757 PR tree-optimization/81832
12758 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
12759 copy loop header which has IFN_LOOP_DIST_ALIAS call.
12761 2017-08-16 Marek Polacek <polacek@redhat.com>
12764 * fold-const.c (fold_indirect_ref_1): Restore original behavior
12765 regarding size_zero_node.
12767 2017-08-16 Martin Liska <mliska@suse.cz>
12770 * config.gcc: Respect previously set extra_objs in case
12773 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
12775 PR tree-optimization/81835
12776 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
12777 the phi SSA_NAME. Check that the condition in a COND_EXPR does
12778 not depend on the phi.
12780 2017-08-16 Alan Modra <amodra@gmail.com>
12782 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
12785 2017-08-16 Alan Modra <amodra@gmail.com>
12787 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
12788 (save_reg_p): ..into this. Update all callers.
12789 (first_reg_to_save): Simplify.
12791 2017-08-16 Alan Modra <amodra@gmail.com>
12793 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
12796 2017-08-15 Joseph Myers <joseph@codesourcery.com>
12800 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
12801 constant count if that count is less than 32.
12803 2017-08-15 Nathan Sidwell <nathan@acm.org>
12805 * gcc.c (execute): Emit friendlier message if inferior is killed
12806 by an external cause.
12808 2017-08-15 Richard Biener <rguenther@suse.de>
12810 PR tree-optimization/81790
12811 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
12812 CONSTRUCTORs from simplifying and VN.
12814 2017-08-14 Martin Sebor <msebor@redhat.com>
12816 * builtin-attrs.def: Add comments.
12818 2017-08-14 Martin Sebor <msebor@redhat.com>
12821 * doc/extend.texi (attribute nonstring): Document new attribute.
12823 2017-08-14 Martin Sebor <msebor@redhat.com>
12826 * tree-diagnostic.c (default_tree_printer): Handle %G.
12827 * gimple-pretty-print.h (percent_G_format): Declare new function.
12828 * gimple-pretty-print.c (percent_G_format): Define.
12829 * tree-pretty-print.c (percent_K_format): Add argument.
12831 2017-08-14 Martin Sebor <msebor@redhat.com>
12833 PR translation/79998
12834 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
12835 Remove a stray space.
12837 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
12840 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
12841 (*iordi_1_bts): Ditto.
12842 (*xordi_1_btc): Ditto.
12844 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12847 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
12848 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12850 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
12851 quoted strings, and make more translator-friendly.
12852 (darwin_rs6000_override_options): Likewise.
12853 (rs6000_option_override_internal): Likewise.
12854 (rs6000_return_in_memory): Fix overlong line.
12855 (init_cmulative_args): Use quoted strings, and make more
12856 translator-friendly.
12857 (rs6000_pass_by_reference): Fix overlong line.
12858 (def_builtin): Use quoted strings.
12859 (altivec_expand_predicate_builtin): Use quoted strings, and make
12860 more translator-friendly.
12861 (htm_expand_builtin): Use quoted strings.
12862 (cpu_expand_builtin): Use quoted strings, and make more
12863 translator-friendly.
12864 (altivec_expand_builtin): Likewise.
12865 (paired_expand_predicate_builtin): Likewise.
12866 (rs6000_invalid_builtin): Likewise.
12867 (builtin_function_type): Use quoted strings.
12868 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
12869 more translator-friendly.
12870 (rs6000_trampoline_init): Likewise.
12871 (rs6000_handle_altivec_attribute): Likewise.
12872 (rs6000_inner_target_options): Use quoted strings.
12873 (rs6000_disable_incompatible_switches): Likewise.
12874 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
12875 strings, and make more translator-friendly.
12876 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
12878 2017-08-14 Bin Cheng <bin.cheng@arm.com>
12880 PR tree-optimization/81799
12881 * tree-loop-distribution.c (version_loop_by_alias_check): Force
12882 cond_expr to simple gimple operand.
12884 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
12886 PR middle-end/46932
12887 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
12889 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
12893 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
12894 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
12895 TARGET_GASISR_PROLOGUES.
12896 * config/avr/avr.c (avr_option_override): Same.
12897 (avr_pass_pre_proep::execute): Same.
12899 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
12902 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
12903 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
12905 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
12907 * config/i386/i386.md (*load_tp_<mode>): Redefine as
12908 define_insn_and_split. Split to a memory load from 0 in
12909 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
12910 using PTR mode iterator.
12911 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
12912 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
12913 (*add_tp_<mode>): Redefine as define_insn_and_split.
12914 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
12915 address space. Merge with *add_tp_x32 using PTR mode iterator.
12916 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
12917 Split to an add with a memory load from 0 in
12918 DEFAULT_TLS_SEG_REG address space.
12920 2017-08-12 Andrew Pinski <apinski@cavium.com>
12922 * config/aarch64/aarch64-option-extensions.def (rdma):
12923 Fix feature string to what Linux prints out in /proc/cpuinfo.
12925 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
12928 * dwarf2out.c (modified_type_die): For C typedef types that have
12929 an ultimate origin, process the ultimate origin instead of the
12931 (gen_typedef_die): Assert that input DECLs have no ultimate
12933 (gen_type_die_with_usage): For typedef variants that have an
12934 ultimate origin, just call gen_decl_die on the original DECL.
12935 (process_scope_var): Avoid creating DIEs for local typedefs and
12936 concrete static variables.
12938 2017-08-12 Alan Modra <amodra@gmail.com>
12942 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
12943 match gnu-user.h startfile.
12944 (ENDFILE_LINUX_SPEC): Similarly.
12946 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
12949 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
12951 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
12953 2017-08-11 Tamar Christina <tamar.christina@arm.com>
12954 * config/aarch64/aarch64.md (mov<mode>): Change.
12955 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
12956 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
12957 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
12959 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
12961 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
12962 for storage order barriers.
12964 2017-08-11 Martin Liska <mliska@suse.cz>
12967 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
12968 variables of void type.
12970 2017-08-11 Martin Liska <mliska@suse.cz>
12972 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
12973 TARGET_SUPPORTS_ALIASES.
12974 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
12975 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
12976 (optimize_weakref): Likewise.
12977 * symtab.c (symtab_node::noninterposable_alias): Likewise.
12978 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
12979 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
12981 2017-08-11 Martin Liska <mliska@suse.cz>
12984 * config/i386/i386.c (make_resolver_func): Do complete
12985 refactoring of the function.
12987 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
12990 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
12991 * config/i386/i386.c (ix86_stack_protect_guard): Use
12992 ix86_stack_protect_guard_symbol_str to generate varible declaration.
12993 * doc/invoke.texi (x86 Options): Document
12994 -mstack-protector-guard-symbol= option.
12996 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
12998 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
12999 * config/i386/i386.c (ix86_split_stack_guard): New function.
13000 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
13001 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
13002 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
13003 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
13004 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
13005 (split_stack_space_check): Call ix86_split_stack_guard.
13007 2017-08-10 Martin Sebor <msebor@redhat.com>
13009 * print-tree.c (print_node): Print location using the established
13011 Replace spaces with colons.
13012 (debug_raw, debug): Ditto.
13014 2017-08-10 Martin Sebor <msebor@redhat.com>
13017 * pretty-print.c (pp_format): Correct the handling of %s precision.
13019 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
13022 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13024 (ix86_finalize_stack_frame_flags): This. Also clear
13025 frame_pointer_needed if -fno-omit-frame-pointer is used without
13027 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13028 with ix86_finalize_stack_frame_flags.
13029 (ix86_expand_epilogue): Likewise.
13030 (ix86_expand_split_stack_prologue): Likewise.
13031 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
13033 2017-08-10 Martin Liska <mliska@suse.cz>
13036 * c-attribs.c (handle_target_attribute):
13037 Report warning for an empty string argument of target attribute.
13039 2017-08-09 Jakub Jelinek <jakub@redhat.com>
13042 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
13044 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
13045 or DECL_NONLOCAL labels.
13046 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
13047 or DECL_NONLOCAL labels here.
13049 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
13051 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
13052 to indicate when early gimple folding has been disabled.
13053 (rs6000_gimple_fold_builtin): Add debug content.
13054 (rs6000_invalid_builtin): Fix whitespace.
13055 (rs6000_expand_builtin): Fix whitespace.
13056 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
13058 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
13061 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
13062 SAVE_MULTIPLE if not all the registers that saves, should be saved.
13064 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
13066 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
13067 (qdf24xx): Likewise.
13068 * config/aarch64/aarch64.md: Include falkor.md.
13069 * config/aarch64/falkor.md: New.
13071 2017-08-09 Marek Polacek <polacek@redhat.com>
13074 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
13075 * diagnostic.c (emit_diagnostic): Add a comment.
13076 (emit_diagnostic_valist): New function.
13078 2017-08-09 Marek Polacek <polacek@redhat.com>
13081 * input.c (make_location): New overload.
13082 * input.h (make_location): Declare.
13084 2017-08-08 Alan Modra <amodra@gmail.com>
13085 H.J. Lu <hongjiu.lu@intel.com>
13088 * gcc.c (NO_PIE_SPEC): Delete.
13089 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
13091 (LINK_PIE_SPEC): ..to here.
13092 (LINK_COMMAND_SPEC): Support -no-pie.
13093 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
13094 chain of crtbegin*.o selection, update for PIE_SPEC changes and
13096 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
13097 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
13098 (ENDFILE_CRTEND_SPEC): Similarly.
13100 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
13103 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
13104 (mstack-protector-guard-offset=): Ditto.
13105 * config/i386/i386.c (ix86_option_override): Handle
13106 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
13108 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
13109 ix86_stack_protect_guard_offset variables.
13110 (TARGET_STACK_PROTECT_GUARD): Always define.
13111 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
13112 and -mstack-protector-guard-offset= options.
13114 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13116 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
13117 boundary case for the last candidate.
13119 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13121 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
13122 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
13124 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13126 PR middle-end/19706
13127 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
13128 * config/aarch64/aarch64-builtins.c
13129 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
13130 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
13131 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
13133 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13134 Andrew Pinski <pinskia@gmail.com>
13136 PR middle-end/19706
13137 * internal-fn.def (XORSIGN): New.
13138 * optabs.def (xorsign_optab): New.
13139 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
13140 (convert_expand_mult_copysign): New.
13141 (pass_optimize_widening_mul::execute): Call
13142 convert_expand_mult_copysign.
13144 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13146 PR tree-optimization/81354
13147 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
13148 Insert on edges rather than explicitly creating landing pads.
13149 (analyze_candidates_and_replace): Commit edge inserts.
13151 2017-08-08 Richard Biener <rguenther@suse.de>
13153 PR middle-end/81719
13154 * tree-ssa-loop-niter.c: Include tree-dfa.h.
13155 (expand_simple_operations): Also look through ADDR_EXPRs with
13156 MEM_REF bases treating them as POINTER_PLUS_EXPR.
13158 2017-08-08 Richard Biener <rguenther@suse.de>
13160 PR tree-optimization/81723
13161 * tree-vect-slp.c (struct bst_traits): New hash traits.
13162 (bst_fail): New global.
13163 (vect_build_slp_tree_2): New worker, split out from ...
13164 (vect_build_slp_tree): ... this now wrapping it with using
13165 bst_fail set to cache SLP tree build fails. Properly handle
13167 (vect_analyze_slp_instance): Allocate and free bst_fail.
13169 2017-08-08 Martin Liska <mliska@suse.cz>
13172 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
13173 LABEL_DECLs that can be from a different function.
13175 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13177 PR tree-optimization/81744
13178 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
13179 loop's number of iterations.
13181 2017-08-08 Martin Liska <mliska@suse.cz>
13183 * asan.c: Include header files.
13184 * attribs.c (build_decl_attribute_variant): New function moved
13186 (build_type_attribute_qual_variant): Likewise.
13187 (cmp_attrib_identifiers): Likewise.
13188 (simple_cst_list_equal): Likewise.
13189 (omp_declare_simd_clauses_equal): Likewise.
13190 (attribute_value_equal): Likewise.
13191 (comp_type_attributes): Likewise.
13192 (build_type_attribute_variant): Likewise.
13193 (lookup_ident_attribute): Likewise.
13194 (remove_attribute): Likewise.
13195 (merge_attributes): Likewise.
13196 (merge_type_attributes): Likewise.
13197 (merge_decl_attributes): Likewise.
13198 (merge_dllimport_decl_attributes): Likewise.
13199 (handle_dll_attribute): Likewise.
13200 (attribute_list_equal): Likewise.
13201 (attribute_list_contained): Likewise.
13202 * attribs.h (lookup_attribute): New function moved from tree.[ch].
13203 (lookup_attribute_by_prefix): Likewise.
13204 * bb-reorder.c: Include header files.
13205 * builtins.c: Likewise.
13206 * calls.c: Likewise.
13207 * cfgexpand.c: Likewise.
13208 * cgraph.c: Likewise.
13209 * cgraphunit.c: Likewise.
13210 * convert.c: Likewise.
13211 * dwarf2out.c: Likewise.
13212 * final.c: Likewise.
13213 * fold-const.c: Likewise.
13214 * function.c: Likewise.
13215 * gimple-expr.c: Likewise.
13216 * gimple-fold.c: Likewise.
13217 * gimple-pretty-print.c: Likewise.
13218 * gimple.c: Likewise.
13219 * gimplify.c: Likewise.
13220 * hsa-common.c: Likewise.
13221 * hsa-gen.c: Likewise.
13222 * internal-fn.c: Likewise.
13223 * ipa-chkp.c: Likewise.
13224 * ipa-cp.c: Likewise.
13225 * ipa-devirt.c: Likewise.
13226 * ipa-fnsummary.c: Likewise.
13227 * ipa-inline.c: Likewise.
13228 * ipa-visibility.c: Likewise.
13230 * lto-cgraph.c: Likewise.
13231 * omp-expand.c: Likewise.
13232 * omp-general.c: Likewise.
13233 * omp-low.c: Likewise.
13234 * omp-offload.c: Likewise.
13235 * omp-simd-clone.c: Likewise.
13236 * opts-global.c: Likewise.
13237 * passes.c: Likewise.
13238 * predict.c: Likewise.
13239 * sancov.c: Likewise.
13240 * sanopt.c: Likewise.
13241 * symtab.c: Likewise.
13242 * toplev.c: Likewise.
13243 * trans-mem.c: Likewise.
13244 * tree-chkp.c: Likewise.
13245 * tree-eh.c: Likewise.
13246 * tree-into-ssa.c: Likewise.
13247 * tree-object-size.c: Likewise.
13248 * tree-parloops.c: Likewise.
13249 * tree-profile.c: Likewise.
13250 * tree-ssa-ccp.c: Likewise.
13251 * tree-ssa-live.c: Likewise.
13252 * tree-ssa-loop.c: Likewise.
13253 * tree-ssa-sccvn.c: Likewise.
13254 * tree-ssa-structalias.c: Likewise.
13255 * tree-ssa.c: Likewise.
13256 * tree-streamer-in.c: Likewise.
13257 * tree-vectorizer.c: Likewise.
13258 * tree-vrp.c: Likewise.
13259 * tsan.c: Likewise.
13260 * ubsan.c: Likewise.
13261 * varasm.c: Likewise.
13262 * varpool.c: Likewise.
13263 * tree.c: Remove functions moved to attribs.[ch].
13264 * tree.h: Likewise.
13265 * config/aarch64/aarch64.c: Add attrs.h header file.
13266 * config/alpha/alpha.c: Likewise.
13267 * config/arc/arc.c: Likewise.
13268 * config/arm/arm.c: Likewise.
13269 * config/avr/avr.c: Likewise.
13270 * config/bfin/bfin.c: Likewise.
13271 * config/c6x/c6x.c: Likewise.
13272 * config/cr16/cr16.c: Likewise.
13273 * config/cris/cris.c: Likewise.
13274 * config/darwin.c: Likewise.
13275 * config/epiphany/epiphany.c: Likewise.
13276 * config/fr30/fr30.c: Likewise.
13277 * config/frv/frv.c: Likewise.
13278 * config/ft32/ft32.c: Likewise.
13279 * config/h8300/h8300.c: Likewise.
13280 * config/i386/winnt.c: Likewise.
13281 * config/ia64/ia64.c: Likewise.
13282 * config/iq2000/iq2000.c: Likewise.
13283 * config/lm32/lm32.c: Likewise.
13284 * config/m32c/m32c.c: Likewise.
13285 * config/m32r/m32r.c: Likewise.
13286 * config/m68k/m68k.c: Likewise.
13287 * config/mcore/mcore.c: Likewise.
13288 * config/microblaze/microblaze.c: Likewise.
13289 * config/mips/mips.c: Likewise.
13290 * config/mmix/mmix.c: Likewise.
13291 * config/mn10300/mn10300.c: Likewise.
13292 * config/moxie/moxie.c: Likewise.
13293 * config/msp430/msp430.c: Likewise.
13294 * config/nds32/nds32-isr.c: Likewise.
13295 * config/nds32/nds32.c: Likewise.
13296 * config/nios2/nios2.c: Likewise.
13297 * config/nvptx/nvptx.c: Likewise.
13298 * config/pa/pa.c: Likewise.
13299 * config/pdp11/pdp11.c: Likewise.
13300 * config/powerpcspe/powerpcspe.c: Likewise.
13301 * config/riscv/riscv.c: Likewise.
13302 * config/rl78/rl78.c: Likewise.
13303 * config/rx/rx.c: Likewise.
13304 * config/s390/s390.c: Likewise.
13305 * config/sh/sh.c: Likewise.
13306 * config/sol2.c: Likewise.
13307 * config/sparc/sparc.c: Likewise.
13308 * config/spu/spu.c: Likewise.
13309 * config/stormy16/stormy16.c: Likewise.
13310 * config/tilegx/tilegx.c: Likewise.
13311 * config/tilepro/tilepro.c: Likewise.
13312 * config/v850/v850.c: Likewise.
13313 * config/vax/vax.c: Likewise.
13314 * config/visium/visium.c: Likewise.
13315 * config/xtensa/xtensa.c: Likewise.
13317 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13320 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
13321 constraints since the -mupper-regs-* switches have been
13323 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
13324 into a vector from a double word element that was extracted from
13325 another vector, and eliminate extra XXPERMDI instructions.
13326 (vsx_concat_<mode>_2): Likewise.
13327 (vsx_concat_<mode>_3): Likewise.
13328 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
13329 concat to allow optimizing inserts from previous extracts.
13331 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
13333 * config/i386/i386.c (ix86_stack_protect_guard): Generate
13334 memory reference to a SSP offset in TLS address space.
13335 (ix86_print_operand) <case '@'>: Remove.
13336 (ix86_print_operand_punct_valid_p): Remove '@' code.
13337 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
13338 UNSPEC_SP_TLS_TEST.
13339 (stack_tls_protect_set_<mode>): Remove.
13340 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
13341 (stack_tls_protect_test_<mode>): Remove.
13342 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
13344 2017-08-07 Olivier Hainque <hainque@adacore.com>
13347 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
13349 2017-08-07 Douglas Rupp <rupp@adacore.com>
13351 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
13352 variable was referenced as multidir in command.
13354 2017-08-07 Jakub Jelinek <jakub@redhat.com>
13357 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
13360 2017-08-07 Martin Liska <mliska@suse.cz>
13362 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
13363 * config/rl78/rl78.c: Add include of attribs.h.
13364 * config/sh/sh.c: Likewise.
13365 * config/v850/v850.c: Likewise.
13367 2017-08-07 Tom de Vries <tom@codesourcery.com>
13369 PR middle-end/78266
13370 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
13372 2017-08-07 Martin Liska <mliska@suse.cz>
13374 * config/mips/mips.c: Include attribs.h.
13376 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
13379 * doc/invoke.texi: Document change in behvaior for -Ofast for
13382 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
13384 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
13386 (aarch64_pop_regs): Likewise.
13387 (aarch64_gen_load_pair): Likewise.
13388 (aarch64_save_callee_saves): Likewise.
13389 (aarch64_restore_callee_saves): Likewise.
13391 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13393 * config/i386/i386.c: Revert the last change.
13395 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13398 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13400 (ix86_finalize_stack_frame_flags): This. Also clear
13401 frame_pointer_needed if -fno-omit-frame-pointer is used without
13403 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13404 with ix86_finalize_stack_frame_flags.
13405 (ix86_expand_epilogue): Likewise.
13406 (ix86_expand_split_stack_prologue): Likewise.
13408 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13411 * config/i386/i386.c (get_builtin_code_for_version): Set priority
13412 to P_AES for Westmere.
13414 2017-08-07 Jonathan Yong <10walls@gmail.com>
13416 * config/i386/mingw.opt (fset-stack-executable): Removed.
13417 * config/i386/cygming.opt (fset-stack-executable): Moved
13419 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
13421 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
13423 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
13425 2017-08-07 Marek Polacek <polacek@redhat.com>
13427 PR middle-end/81737
13428 * fold-const.c (fold_indirect_ref_1): Check type_domain.
13430 2017-08-07 Martin Liska <mliska@suse.cz>
13432 * attribs.h (canonicalize_attr_name): New function.
13433 (cmp_attribs): Move from c-format.c and adjusted.
13434 (is_attribute_p): Moved from tree.h.
13435 * tree-inline.c: Add new includes.
13436 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
13437 (private_is_attribute_p): Remove.
13438 (private_lookup_attribute): Likewise.
13439 (private_lookup_attribute_by_prefix): Simplify.
13440 (remove_attribute): Use is_attribute_p.
13441 * tree.h: Remove removed declarations.
13443 2017-08-07 Jakub Jelinek <jakub@redhat.com>
13445 PR middle-end/81698
13446 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
13447 instead of computing it in the function. Formatting fix.
13448 (expand_case): Don't rely on default_edge being the first edge,
13449 clear it if removing it, pass default_edge to
13450 emit_case_dispatch_table.
13451 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
13454 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
13456 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
13457 insn in the function, emit NOP after the insn.
13459 2017-08-06 Tom de Vries <tom@codesourcery.com>
13461 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
13464 2017-08-06 Tom de Vries <tom@codesourcery.com>
13466 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
13469 2017-08-04 Yury Gribov <tetra2005@gmail.com>
13471 PR tree-optimization/57371
13472 * match.pd: New pattern.
13474 2017-08-04 Marek Polacek <polacek@redhat.com>
13476 PR middle-end/81695
13477 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
13478 perform the computation in offset_int.
13480 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13482 PR tree-optimization/81136
13483 * tree-vectorizer.h: Include tree-hash-traits.h.
13484 (vec_base_alignments): New typedef.
13485 (vec_info): Add a base_alignments field.
13486 (vect_record_base_alignments): Declare.
13487 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
13489 (DR_IS_CONDITIONAL_IN_STMT): New macro.
13490 (create_data_ref): Add an is_conditional_in_stmt argument.
13491 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
13492 the is_conditional_in_stmt field.
13493 (data_ref_loc): Add an is_conditional_in_stmt field.
13494 (get_references_in_stmt): Set the is_conditional_in_stmt field.
13495 (find_data_references_in_stmt): Update call to create_data_ref.
13496 (graphite_find_data_references_in_stmt): Likewise.
13497 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
13498 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
13499 (vect_record_base_alignment): New function.
13500 (vect_record_base_alignments): Likewise.
13501 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
13502 for nested statements even if we fail to compute a misalignment.
13503 Use pooled base alignments for unconditional references.
13504 (vect_find_same_alignment_drs): Compare base addresses instead
13506 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
13507 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
13509 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13511 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
13512 Add an explicit name for the enum. Use auto_vec for slp_instances
13513 and grouped_stores.
13514 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
13516 (_bb_vec_info): Add a constructor and destructor.
13517 (vinfo_for_stmt): Return NULL for uids of -1 as well.
13518 (destroy_loop_vec_info): Delete.
13519 (vect_destroy_datarefs): Likewise.
13520 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
13521 (vec_info::vec_info): New function.
13522 (vec_info::~vec_info): Likewise.
13523 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
13524 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
13525 destroy_loop_vec_info.
13526 * tree-vect-loop.c (new_loop_vec_info): Replace with...
13527 (_loop_vec_info::_loop_vec_info): ...this.
13528 (destroy_loop_vec_info): Replace with...
13529 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
13530 the stmt_vec_infos. Leave handling of vec_info information to its
13531 destructor. Remove explicit vector releases.
13532 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
13533 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
13534 * tree-vect-slp.c (new_bb_vec_info): Replace with...
13535 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
13536 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
13537 (destroy_bb_vec_info): Replace with...
13538 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
13539 information to its destructor.
13540 (vect_slp_analyze_bb_1): Use new and delete instead of
13541 new_bb_vec_info and destroy_bb_vec_info.
13542 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
13545 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13547 * tree-data-ref.h (subscript): Add access_fn field.
13548 (data_dependence_relation): Add could_be_independent_p.
13549 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
13550 (same_access_functions): Move to tree-data-ref.c.
13551 * tree-data-ref.c (ref_contains_union_access_p): New function.
13552 (access_fn_component_p): Likewise.
13553 (access_fn_components_comparable_p): Likewise.
13554 (dr_analyze_indices): Add a reference to access_fn_component_p.
13555 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
13557 (constant_access_functions): Likewise.
13558 (add_other_self_distances): Likewise.
13559 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
13560 (initialize_data_dependence_relation): Use XCNEW and remove
13561 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
13562 of access functions that have the same type. Allow the
13563 subsequence to end with different bases in some circumstances.
13564 Record the chosen access functions in SUB_ACCESS_FN.
13565 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
13566 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
13567 (subscript_dependence_tester_1): Likewise dra and drb.
13568 (build_classic_dist_vector): Update calls accordingly.
13569 (subscript_dependence_tester): Likewise.
13570 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
13571 DDR_COULD_BE_INDEPENDENT_P.
13572 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
13573 comp_alias_ddrs instead of may_alias_ddrs.
13574 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
13576 (vect_analyze_data_ref_dependence): Use it if
13577 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
13578 distance vectors if that fails.
13579 (dependence_distance_ge_vf): New function.
13580 (vect_prune_runtime_alias_test_list): Use it. Don't clear
13581 LOOP_VINFO_MAY_ALIAS_DDRS.
13583 2017-08-04 Richard Biener <rguenther@suse.de>
13585 PR middle-end/81705
13586 * fold-const.c (fold_binary_loc): Properly restrict
13587 minus_var0 && minus_var1 case when associating undefined overflow
13590 2017-08-04 Tom de Vries <tom@codesourcery.com>
13592 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
13594 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13596 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13597 Don't start diagnostic messages with a capital letter.
13598 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13600 (rs6000_invalid_builtin): Likewise.
13601 (rs6000_trampoline_init): Likewise.
13603 2017-08-03 Jakub Jelinek <jakub@redhat.com>
13606 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
13607 after setting changeable df flags.
13609 2017-08-03 Richard Biener <rguenther@suse.de>
13611 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
13612 up if the use is in USE - X.
13614 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
13616 * toplev.c (dumpfile.h): New include.
13617 (internal_error_reentered): New static function. Use it...
13618 (internal_error_function): ...here to handle reentered internal_error.
13620 2017-08-03 Richard Biener <rguenther@suse.de>
13622 PR middle-end/81148
13623 * fold-const.c (split_tree): Add minus_var and minus_con
13624 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
13625 here but always use minus_*.
13626 (associate_trees): Assert we never associate with MINUS_EXPR
13627 and NULL first operand. Do not recurse for PLUS_EXPR operands
13628 when associating as MINUS_EXPR either.
13629 (fold_binary_loc): Track minus_var and minus_con.
13631 2017-08-03 Tom de Vries <tom@codesourcery.com>
13634 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
13635 ACCEL_COMPILER, apply finish_options on
13636 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
13638 2017-08-03 Tom de Vries <tom@codesourcery.com>
13641 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
13642 function_entry_patch_area_size > 0.
13644 2017-08-03 Jakub Jelinek <jakub@redhat.com>
13647 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
13648 instead of 10??LU, perform unit multiplication in wide_int,
13649 don't change alloc_object_size_limit if the limit is larger
13652 PR tree-optimization/81655
13653 PR tree-optimization/81588
13654 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
13655 the case when ranges[i].low and high are 1 for unsigned type with
13658 PR middle-end/81052
13659 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
13660 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
13662 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13664 * tree-vrp.h: Add include guard.
13666 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
13669 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
13670 (ud2): New insn pattern.
13671 * config/i386/i386.c (ix86_expand_epilogue):
13672 For naked functions, generate ud2 instead of trap insn.
13674 2017-08-02 Marek Polacek <polacek@redhat.com>
13677 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
13679 2017-08-02 Tom de Vries <tom@codesourcery.com>
13680 Cesar Philippidis <cesar@codesourcery.com>
13682 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
13683 Add missing edge probabilities.
13685 2017-08-02 Tamar Christina <tamar.christina@arm.com>
13687 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
13688 Correct endianness.
13690 2017-08-02 Jakub Jelinek <jakub@redhat.com>
13692 PR middle-end/79499
13693 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
13694 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
13695 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
13696 prologue_seq sequences - if any.
13698 2017-08-02 Richard Biener <rguenther@suse.de>
13700 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
13701 via vectors if supported, integer extracts via punning if supported
13702 or otherwise vector extracts.
13704 2017-08-02 Richard Biener <rguenther@suse.de>
13706 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
13708 (bitmap_insert_into_set): ... this.
13710 2017-08-02 Richard Biener <rguenther@suse.de>
13712 PR tree-optimization/81633
13714 2015-08-17 Alan Hayward <alan.hayward@arm.com>
13716 PR tree-optimization/71752
13717 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
13719 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
13721 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
13722 (machine_function::call_ms2sysv_pad_out): Remove field.
13723 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
13724 (ix86_compute_frame_layout): Likewise.
13726 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
13729 * config/i386/i386.c (ix86_set_func_type): Disallow naked
13730 attribute with interrupt attribute.
13732 2017-08-01 Andrew Pinski <apinski@cavium.com>
13734 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
13735 BIT_INSERT_EXPR's operand 1
13736 to see if the types precision matches.
13738 2017-08-01 Martin Liska <mliska@suse.cz>
13740 PR middle-end/70140
13741 * builtins.c (expand_builtin_memcpy_args): Remove.
13742 (expand_builtin_memcpy): Call newly added function
13743 expand_builtin_memory_copy_args.
13744 (expand_builtin_memcpy_with_bounds): Likewise.
13745 (expand_builtin_mempcpy): Remove last argument.
13746 (expand_builtin_mempcpy_with_bounds): Likewise.
13747 (expand_builtin_memory_copy_args): New function created from
13748 expand_builtin_mempcpy_args with small modifications.
13749 (expand_builtin_mempcpy_args): Remove.
13750 (expand_builtin_stpcpy): Remove unused argument.
13751 (expand_builtin): Likewise.
13752 (expand_builtin_with_bounds): Likewise.
13754 2017-08-01 Martin Liska <mliska@suse.cz>
13757 Make mempcpy more optimal (PR middle-end/70140).
13759 2017-08-01 Jakub Jelinek <jakub@redhat.com>
13762 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
13763 __builtin_vec_cmpne verify both arguments are compatible vectors
13764 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
13765 verify arg1_type is a pointer or array type. For __builtin_vec_st,
13766 move computation of aligned to after checking the argument types.
13770 * config/rs6000/vsx.md (vextract_fp_from_shorth,
13771 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
13774 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
13775 Doug Rupp <rupp@adacore.com>
13776 Olivier Hainque <hainque@adacore.com>
13778 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
13779 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
13781 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
13782 for TARGET_OS_CPP_BUILTIN.
13783 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
13784 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
13786 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
13787 passing required abi options to the assembler for EABI configurations.
13788 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
13789 of .text.hot and .text.unlikely sections for kernel modules when
13790 using ARM style exceptions.
13791 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
13792 options. Add EXTRA_CC1_SPEC.
13793 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
13795 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
13797 (ARM_TARGET2_DWARF_FORMAT): Define.
13798 * config/arm/t-vxworks: Adjust multilib control to removal of the
13799 Diab command line options.
13801 2017-08-01 Martin Liska <mliska@suse.cz>
13803 PR gcov-profile/81561
13804 * gcov.c (unblock): Make unblocking safe as we need to preserve
13805 index correspondence of blocks and block_lists.
13807 2017-08-01 Richard Biener <rguenther@suse.de>
13809 PR tree-optimization/81181
13810 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
13811 (compute_antic): ... end of iteration here.
13813 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
13815 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
13816 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
13817 (ftree-slp-vectorize): Likewise.
13818 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
13819 can no longer be set independent of flag_tree_loop_vectorize.
13820 * omp-general.c (emp_max_vf): Likewise.
13821 * opts.c (enable_fdo_optimizations): Remove references to
13822 flag_tree_vectorize, these are now implicit.
13823 (common_handle_option): Remove handling for OPT_ftree_vectorize,
13824 and leave it for the options machinery.
13826 2017-08-01 Martin Liska <mliska@suse.cz>
13828 PR middle-end/70140
13829 * builtins.c (expand_builtin_memcpy_args): Remove.
13830 (expand_builtin_memcpy): Call newly added function
13831 expand_builtin_memory_copy_args.
13832 (expand_builtin_memcpy_with_bounds): Likewise.
13833 (expand_builtin_mempcpy): Remove last argument.
13834 (expand_builtin_mempcpy_with_bounds): Likewise.
13835 (expand_builtin_memory_copy_args): New function created from
13836 expand_builtin_mempcpy_args with small modifications.
13837 (expand_builtin_mempcpy_args): Remove.
13838 (expand_builtin_stpcpy): Remove unused argument.
13839 (expand_builtin): Likewise.
13840 (expand_builtin_with_bounds): Likewise.
13842 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
13845 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
13846 print "ds:" only for immediates in generic address space.
13848 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
13851 * config/i386/i386.c (ix86_funciton_naked): New prototype.
13852 (ix86_function_ok_for_sibcall): Return false for naked functions.
13854 2017-08-01 Richard Biener <rguenther@suse.de>
13856 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
13857 (compute_antic): Seed worklist with exit block predecessors.
13858 * cfganal.c (dfs_find_deadend): For a cycle return the source
13859 of the edge closing it.
13861 2017-08-01 Tamar Christina <tamar.christina@arm.com>
13863 * config/aarch64/aarch64.c
13864 (aarch64_can_const_movi_rtx_p): Move 0 check.
13866 2017-08-01 Bin Cheng <bin.cheng@arm.com>
13868 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
13869 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
13871 * match.pd: Ditto in address comparison pattern.
13873 2017-08-01 Bin Cheng <bin.cheng@arm.com>
13875 PR tree-optimization/81627
13876 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
13877 closed ssa form for store-store chain.
13879 2017-08-01 Bin Cheng <bin.cheng@arm.com>
13881 PR tree-optimization/81620
13882 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
13883 for store-store chain.
13885 2017-08-01 Jakub Jelinek <jakub@redhat.com>
13887 PR tree-optimization/81588
13888 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
13889 ranges[i].in_p, invert comparison code ccode. For >/>=,
13890 swap rhs1 and rhs2 and comparison code unconditionally,
13891 for </<= don't do that. Don't swap rhs1/rhs2 again if
13892 ranges[i].in_p, instead invert comparison code ccode if
13893 opcode or oe->rank is BIT_IOR_EXPR.
13896 * optabs.def (vec_extract_optab, vec_init_optab): Change from
13897 a direct optab to conversion optab.
13898 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
13899 with GET_MODE_INNER as last argument instead of optab_handler.
13900 * expmed.c (extract_bit_field_1): Likewise. Use vector from
13901 vector extraction if possible and optab is available.
13902 * expr.c (store_constructor): Use convert_optab_handler instead
13903 of optab_handler. Use vector initialization from smaller
13904 vectors if possible and optab is available.
13905 * tree-vect-stmts.c (vectorizable_load): Likewise.
13906 * doc/md.texi (vec_extract, vec_init): Document that the optabs
13907 now have two modes.
13908 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
13909 of vec_init from half-sized vectors with the same element mode.
13910 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
13911 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
13912 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
13913 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
13914 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
13915 after mode in gen_vec_extract* calls.
13916 (vec_extract<mode>): Renamed to ...
13917 (vec_extract<mode><ssescalarmodelower>): ... this.
13918 (vec_extract<mode><ssehalfvecmodelower>): New expander.
13919 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
13920 element mode after mode in gen_vec_init* calls.
13921 (VEC_INIT_HALF_MODE): New mode iterator.
13922 (vec_init<mode>): Renamed to ...
13923 (vec_init<mode><ssescalarmodelower>): ... this.
13924 (vec_init<mode><ssehalfvecmodelower>): New expander.
13925 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
13926 (vec_extractv2sfsf): ... this.
13927 (vec_initv2sf): Renamed to ...
13928 (vec_initv2sfsf): ... this.
13929 (vec_extractv2si): Renamed to ...
13930 (vec_extractv2sisi): ... this.
13931 (vec_initv2si): Renamed to ...
13932 (vec_initv2sisi): ... this.
13933 (vec_extractv4hi): Renamed to ...
13934 (vec_extractv4hihi): ... this.
13935 (vec_initv4hi): Renamed to ...
13936 (vec_initv4hihi): ... this.
13937 (vec_extractv8qi): Renamed to ...
13938 (vec_extractv8qiqi): ... this.
13939 (vec_initv8qi): Renamed to ...
13940 (vec_initv8qiqi): ... this.
13941 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
13942 (vec_init<mode>): Renamed to ...
13943 (vec_init<mode><VEC_base_l>): ... this.
13944 (vec_extract<mode>): Renamed to ...
13945 (vec_extract<mode><VEC_base_l>): ... this.
13946 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
13947 (vec_initv2sfsf): ... this.
13948 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
13949 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
13950 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
13951 element mode after mode in gen_vec_init* calls.
13952 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
13953 (vec_init<mode><Vel>): ... this.
13954 (vec_extract<mode>): Renamed to ...
13955 (vec_extract<mode><Vel>): ... this.
13956 * config/aarch64/iterators.md (Vel): New mode attribute.
13957 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
13958 Add element mode after mode in gen_vec_extract* calls.
13959 * config/s390/vector.md (non_vec_l): New mode attribute.
13960 (vec_extract<mode>): Renamed to ...
13961 (vec_extract<mode><non_vec_l>): ... this.
13962 (vec_init<mode>): Renamed to ...
13963 (vec_init<mode><non_vec_l>): ... this.
13964 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
13965 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
13967 * config/arm/iterators.md (V_elem_l): New mode attribute.
13968 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
13969 (vec_extract<mode><V_elem_l>): ... this.
13970 (vec_extractv2di): Renamed to ...
13971 (vec_extractv2didi): ... this.
13972 (vec_init<mode>): Renamed to ...
13973 (vec_init<mode><V_elem_l>): ... this.
13974 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
13975 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
13976 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
13977 Add element mode after gen_vec_extract* calls.
13978 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
13979 (vec_init<mode><unitmode>): ... this.
13980 (vec_extract<mode>): Renamed to ...
13981 (vec_extract<mode><unitmode>): ... this.
13982 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
13983 (vec_init<mode><unitmode>): ... this.
13984 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
13985 (vec_initv2sfsf): ... this.
13986 (vec_extractv2sf): Renamed to ...
13987 (vec_extractv2sfsf): ... this.
13988 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
13989 Add element mode after gen_vec_extract* calls.
13990 * config/mips/mips.md (unitmode): New mode iterator.
13991 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
13992 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
13993 * config/spu/spu.md (inner_l): New mode attribute.
13994 (vec_init<mode>): Renamed to ...
13995 (vec_init<mode><inner_l>): ... this.
13996 (vec_extract<mode>): Renamed to ...
13997 (vec_extract<mode><inner_l>): ... this.
13998 * config/sparc/sparc.md (veltmode): New mode iterator.
13999 (vec_init<VMALL:mode>): Renamed to ...
14000 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
14001 * config/ia64/vect.md (vec_initv2si): Renamed to ...
14002 (vec_initv2sisi): ... this.
14003 (vec_initv2sf): Renamed to ...
14004 (vec_initv2sfsf): ... this.
14005 (vec_extractv2sf): Renamed to ...
14006 (vec_extractv2sfsf): ... this.
14007 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
14008 (vec_init<mode>): Renamed to ...
14009 (vec_init<mode><VEC_base_l>): ... this.
14010 (vec_extract<mode>): Renamed to ...
14011 (vec_extract<mode><VEC_base_l>): ... this.
14012 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
14013 (vec_initv2sfsf): ... this.
14014 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
14015 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14016 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
14017 gen_vec_init* calls.
14019 2017-08-01 Richard Biener <rguenther@suse.de>
14021 PR tree-optimization/81297
14022 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
14023 TREE_OVERFLOW from INTEGER_CSTs.
14025 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
14027 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
14029 2017-07-31 Carl Love <cel@us.ibm.com>
14031 * config/rs6000/rs6000-c: Add support for built-in functions
14032 vector signed char vec_xl_be (signed long long, signed char *);
14033 vector unsigned char vec_xl_be (signed long long, unsigned char *);
14034 vector signed int vec_xl_be (signed long long, signed int *);
14035 vector unsigned int vec_xl_be (signed long long, unsigned int *);
14036 vector signed long long vec_xl_be (signed long long, signed long long *);
14037 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
14038 vector signed short vec_xl_be (signed long long, signed short *);
14039 vector unsigned short vec_xl_be (signed long long, unsigned short *);
14040 vector double vec_xl_be (signed long long, double *);
14041 vector float vec_xl_be (signed long long, float *);
14042 * config/rs6000/altivec.h (vec_xl_be): Add #define.
14043 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
14044 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
14046 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
14047 (altivec_expand_builtin): Add switch statement to call
14048 altivec_expand_xl_be for each builtin.
14049 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
14050 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
14051 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
14052 __builtin_vsx_le_be_v16qi.
14053 * doc/extend.texi: Update the built-in documentation file for the
14054 new built-in functions.
14056 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14059 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
14061 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
14063 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14065 * config.gcc: Add z14.
14066 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
14067 CPU model numbers for z13s and z14.
14068 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
14070 * config/s390/s390-opts.h (enum processor_type): Rename
14071 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14072 * config/s390/s390.c (processor_table): Add field for CPU name to
14073 be passed to Binutils.
14074 (s390_asm_output_machine_for_arch): Use the new field in
14075 processor_table for Binutils.
14076 (s390_expand_builtin): Replace arch12 with z14.
14077 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14078 (s390_get_sched_attrmask): Likewise.
14079 (s390_get_unit_mask): Likewise.
14080 * config/s390/s390.opt: Add z14 to processor_type enum.
14082 2017-07-31 Martin Jambor <mjambor@suse.cz>
14085 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
14086 regardless of optimization level.
14088 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
14089 Martin Liska <mliska@suse.cz>
14091 * predict.def: Remove old comment and adjust probability.
14092 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
14093 PREDICT statements.
14095 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14098 * config/i386/i386.c (ix86_function_naked): New function.
14099 (ix86_can_use_return_insn_p): Return false for naked functions.
14100 (ix86_expand_prologue): Skip prologue for naked functions.
14101 (ix86_expand_epilogue): Skip epilogue for naked functions
14102 and emit trap instruction.
14103 (ix86_warn_func_return): New function.
14104 (ix86_attribute_table): Add "naked" attribute specification.
14105 (TARGET_WARN_FUNC_RETURN): Define.
14106 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
14108 2017-07-31 Martin Liska <mliska@suse.cz>
14110 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
14111 (dump_gimple_bb_header): Always dump BB info.
14112 (pp_cfg_jump): Do not append info about BB when dumping a jump.
14114 2017-07-31 Martin Liska <mliska@suse.cz>
14117 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
14118 also with current_function_decl non-null equality.
14120 2017-07-31 Jakub Jelinek <jakub@redhat.com>
14123 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
14124 change type to the element type, instead add eltype variable and
14125 use it where we are interested in the element type.
14127 PR tree-optimization/81603
14128 * ipa-polymorphic-call.c
14129 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
14130 offset arithmetic in offset_int, bail out if the resulting bit offset
14131 doesn't fit into shwi.
14133 2017-07-31 Martin Liska <mliska@suse.cz>
14135 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
14136 (gimplify_save_expr): Fix comment.
14138 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
14141 * config/i386/i386.c (ix86_function_arg): Update arguments for
14143 (ix86_compute_frame_layout): Set the initial stack offset to
14144 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
14145 INCOMING_FRAME_SP_OFFSET.
14146 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
14147 stack before exception handler returns.
14148 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
14149 the 'ERROR_CODE' for exception handler.
14151 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
14153 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
14154 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
14155 (ASM_OUTPUT_REG_POP): Ditto.
14156 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
14157 instead of asm_fprintf to output pure string.
14159 2017-07-29 Jakub Jelinek <jakub@redhat.com>
14161 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
14162 to imported_module_or_decl hook.
14163 (debug_nothing_tree_tree_tree_bool): Remove.
14164 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
14165 * debug.c (do_nothing_debug_hooks): Use
14166 debug_nothing_tree_tree_tree_bool_bool instead of
14167 debug_nothing_tree_tree_tree_bool.
14168 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14169 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
14170 * sdbout.c (sdb_debug_hooks): Likewise.
14171 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
14172 (gen_namespace_die): Add DW_AT_export_symbols attribute if
14174 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
14175 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
14176 attribute, don't add anything.
14178 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14180 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
14181 (fold_build2_stat_loc): Likewise.
14182 (fold_build3_stat_loc): Likewise.
14183 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
14184 (fold_build1_loc): Remove macro.
14185 (fold_build2_loc): Likewise.
14186 (fold_build3_loc): Likewise.
14188 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14190 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
14191 (gimple_build_debug_bind_source_stat): Likewise.
14192 * gimple.h (gimple_build_debug_bind): Remove macro.
14193 (gimple_build_debug_bind_source): Likewise.
14195 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14197 * bitmap.c (bitmap_alloc): Adjust.
14198 (bitmap_gc_alloc): Likewise.
14199 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
14201 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14203 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
14204 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
14205 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
14206 (bitmap_gc_alloc_stat): Likewise.
14207 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
14209 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14211 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
14212 * rtl.h (shallow_copy_rtx): Remove macro.
14214 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14216 * emit-rtl.c (gen_raw_REG): Adjust.
14217 * gengenrtl.c (gendef): Likewise.
14218 * rtl.c (rtx_alloc_stat): Remove _stat from name.
14219 * rtl.h (rtx_alloc): Remove macro.
14221 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14223 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
14224 (build_tree_list_stat): Likewise.
14225 * tree.h (build_tree_list): Remove macro.
14226 (build_tree_list_vec): Likewise.
14228 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14230 * tree.c (make_vector_stat): Remove _stat from name.
14231 (build_vector_stat): Likewise.
14232 * tree.h (make_vector_stat): Remove macro.
14233 (build_vector_stat): Likewise.
14235 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14237 * tree.h (build_var_debug_value): Remove prototype.
14239 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14241 * tree.c (tree_cons_stat): Remove _stat from name.
14242 * tree.h (tree_cons): Remove macro.
14244 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14246 * tree.c (build_vl_exp_stat): Remove _stat from name.
14247 * tree.h (build_vl_exp): Remove macro.
14249 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14251 * tree.c (build_decl_stat): Remove _stat from name.
14252 * tree.h (build_decl): Remove macro.
14254 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14256 * gimple.c (gimple_build_with_ops_stat): Adjust.
14257 (gimple_alloc_stat): Remove _stat from name.
14258 * gimple.h (gimple_alloc): Remove macro.
14260 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14262 * tree.c (make_tree_vec_stat): Remove _stat from name.
14263 (grow_tree_vec_stat): Likewise.
14264 * tree.h (make_tree_vec_stat): Adjust prototype.
14265 (grow_tree_vec_stat): Likewise.
14266 (make_tree_vec): Remove macro.
14267 (grow_tree_vec): Likewise.
14269 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14271 * fold-const.c (fold_build1_stat_loc): Adjust.
14272 (fold_build2_stat_loc): Likewise.
14273 (fold_build3_stat_loc): Likewise.
14274 * tree.c (build0_stat): Remove _stat from name.
14275 (build1_stat): Likewise.
14276 (build2_stat): Likewise.
14277 (build3_stat): Likewise.
14278 (build4_stat): Likewise.
14279 (build5_stat): Likewise.
14280 * tree.h (build1_loc): Remove macro, and rename _stat function
14282 (build2_loc): Likewise.
14283 (build3_loc): Likewise.
14284 (build4_loc): Likewise.
14285 (build5_loc): Likewise.
14287 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14289 * tree.c (make_int_cst_stat): Remove _stat from name.
14290 * tree.h (make_int_cst_stat): Adjust prototype.
14291 (make_int_cst): Remove macro.
14293 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14295 * tree.c (make_tre_binfo_stat): Remove _stat from name.
14296 * tree.h (make_tree_binfo_stat): Adjust prototype.
14297 (make_tree_binfo): Remove.
14299 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14301 * tree.c (copy_node_stat): Rename to copy_node.
14302 (build_distinct_type_copy): Adjust.
14303 * tree.h (copy_node_stat): Adjust prototype.
14304 (copy_node): Remove macro.
14306 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14308 * tree.c (make_node_stat): rename to make_node.
14309 (build_tree_list_stat): Adjust.
14310 (build0_stat): Likewise.
14311 (build2_stat): Likewise.
14312 (build3_stat): Likewise.
14313 (build4_stat): Likewise.
14314 (build5_stat): Likewise.
14315 (build_decl_stat): Likewise.
14316 * tree.h (make_node_stat): Adjust prototype.
14317 (make_node): remove macro.
14319 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
14321 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
14322 (PPC_FEATURE2_SCV): Likewise.
14323 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
14325 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14327 * config/aarch64/aarch64.c
14328 (aarch64_internal_mov_immediate): Add new special pattern.
14329 * config/aarch64/aarch64.md (*movdi_aarch64):
14330 Add reg/32bit const mov case.
14332 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14333 Richard Sandiford <richard.sandiford@linaro.org>
14335 * config/aarch64/aarch64.md (mov<mode>): Generalize.
14336 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14337 Add integer and movi cases.
14338 (movi-split-hf-df-sf split, fp16): New.
14339 (enabled): Added TARGET_FP_F16INST.
14340 * config/aarch64/iterators.md (GPF_HF): New.
14341 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
14343 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14345 * config/aarch64/aarch64.c
14346 (aarch64_simd_container_mode): Add prototype.
14347 (aarch64_expand_mov_immediate): Add HI support.
14348 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
14349 (aarch64_can_const_movi_rtx_p): New.
14350 (aarch64_preferred_reload_class):
14351 Remove restrictions of using FP registers for certain SIMD operations.
14352 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
14353 (aarch64_valid_floating_const): Add integer move validation.
14354 (aarch64_simd_imm_scalar_p): Remove.
14355 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
14356 (aarch64_legitimate_constant_p): Expand list of supported cases.
14357 * config/aarch64/aarch64-protos.h
14358 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
14359 (aarch64_reinterpret_float_as_int): New.
14360 (aarch64_simd_imm_scalar_p): Remove.
14361 * config/aarch64/constraints.md (Uvi): New.
14362 (Dd): Split into Ds and new Dd.
14363 * config/aarch64/aarch64.md (*movsi_aarch64):
14365 (*movdi_aarch64): Add SIMD mov case.
14367 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14369 * tree-predcom.c: (struct chain): Handle store-store chain in which
14370 stores for elimination only store loop invariant values.
14371 (execute_pred_commoning_chain): Ditto.
14372 (prepare_initializers_chain_store_elim): Ditto.
14373 (prepare_finalizers): Ditto.
14374 (is_inv_store_elimination_chain): New function.
14375 (initialize_root_vars_store_elim_1): New function.
14377 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14379 * tree-predcom.c: Revise general description of the pass.
14380 (enum chain_type): New enum type for store elimination.
14381 (struct chain): New field supporting store elimination.
14382 (struct component): Ditto.
14383 (dump_chain): Dump store-stores chain.
14384 (release_chain): Release resources.
14385 (split_data_refs_to_components): Compute and create component
14386 contains only stores for elimination.
14387 (get_chain_last_ref_at): New function.
14388 (make_invariant_chain): Initialization.
14389 (make_rooted_chain): Specify chain type in parameter and record it.
14390 (add_looparound_copies): Skip for store-stores chain.
14391 (determine_roots_comp): Compute type of chain and pass it to
14393 (initialize_root_vars_store_elim_2): New function.
14394 (finalize_eliminated_stores): New function.
14395 (remove_stmt): Handle store for elimination.
14396 (execute_pred_commoning_chain): Execute predictive commoning on
14397 store-store chains.
14398 (determine_unroll_factor): Skip unroll for store-stores chain.
14399 (prepare_initializers_chain_store_elim): New function.
14400 (prepare_initializers_chain): Hanlde store-store chain.
14401 (prepare_finalizers_chain, prepare_finalizers): New function.
14402 (tree_predictive_commoning_loop): Return integer value indicating
14403 if loop is unrolled or lcssa form is corrupted.
14404 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
14406 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14408 * tree-predcom.c (initialize_root): Delete.
14409 (execute_pred_commoning_chain): Initialize root vars and replace
14410 reference of non-combined chain directly, rather than call above
14413 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14415 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
14416 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
14418 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14420 * tree-predcom.c (struct chain): New field init_seq.
14421 (release_chain): Release init_seq.
14422 (prepare_initializers_chain): Record intialization stmts in above
14424 (insert_init_seqs): New function.
14425 (tree_predictive_commoning_loop): Call insert_init_seqs.
14427 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14429 * tree-predcom.c (determine_roots_comp): Skip trivial components.
14431 2017-07-28 Richard Biener <rguenther@suse.de>
14433 * match.pd: Remove superfluous :c.
14434 * genmatch.c (simplify::id): Add member.
14435 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
14437 (current_id): New global.
14438 (dt_node::parent): Move from ...
14439 (dt_operand::parent): ... here. Add for_id member.
14440 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
14441 (decision_tree::find_node): Relax order requirement when
14442 merging DT_TRUE nodes to ones inbetween the current simplify
14443 and the one we try to merge with. Add diagnostic whenever
14444 we need to enforce pattern order by not merging.
14445 (decision_tree::insert): Set current_id.
14446 (decision_tree::print_node): Dump parent node and for_id.
14447 (parser::last_id): Add member.
14448 (parser::push_simplify): Assign unique id.
14449 (parser::parser): Initialize last_id.
14451 2017-07-28 Martin Liska <mliska@suse.cz>
14454 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
14455 gimple_build_debug_bind.
14457 2017-07-28 Richard Biener <rguenther@suse.de>
14459 PR tree-optimization/81502
14460 * match.pd: Add pattern combining BIT_INSERT_EXPR with
14462 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
14464 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
14465 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
14466 for BIT_FIELD_REF args.
14467 * fold-const.c (make_bit_field_ref): Likewise.
14468 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
14470 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14473 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
14474 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
14475 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
14476 Or it into SANITIZER_UNDEFINED.
14477 * ubsan.c: Include gimple-fold.h and varasm.h.
14478 (ubsan_expand_ptr_ifn): New function.
14479 (instrument_pointer_overflow): New function.
14480 (maybe_instrument_pointer_overflow): New function.
14481 (instrument_object_size): Formatting fix.
14482 (pass_ubsan::execute): Call instrument_pointer_overflow
14483 and maybe_instrument_pointer_overflow.
14484 * internal-fn.c (expand_UBSAN_PTR): New function.
14485 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
14486 * sanitizer.def (__ubsan_handle_pointer_overflow,
14487 __ubsan_handle_pointer_overflow_abort): New builtins.
14488 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
14489 * internal-fn.def (UBSAN_PTR): New internal function.
14490 * opts.c (sanitizer_opts): Add pointer-overflow.
14491 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
14492 * fold-const.c (build_range_check): Compute pointer range check in
14493 integral type if pointer arithmetics would be needed. Formatting
14496 2017-07-28 Martin Liska <mliska@suse.cz>
14499 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
14500 parameters that are of a variable-length.
14502 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
14504 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
14506 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
14507 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
14508 (CRT_CALL_STATIC_FUNCTION): Likewise.
14509 (ASM_DEFAULT_SPEC): New define.
14510 (ASM_SPEC32): Likewise.
14511 (ASM_SPEC64): Likewise.
14512 (ASM_SPEC_COMMON): Likewise.
14513 (ASM_SPEC): Likewise.
14514 (INVALID_64BIT): Likewise.
14515 (LINK_OS_DEFAULT_SPEC): Likewise.
14516 (LINK_OS_SPEC32): Likewise.
14517 (LINK_OS_SPEC64): Likewise.
14518 (POWERPC_LINUX): Likewise.
14519 (PTRDIFF_TYPE): Likewise.
14520 (RESTORE_FP_PREFIX): Likewise.
14521 (RESTORE_FP_SUFFIX): Likewise.
14522 (SAVE_FP_PREFIX): Likewise.
14523 (SAVE_FP_SUFFIX): Likewise.
14524 (SIZE_TYPE): Likewise.
14525 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
14526 (TARGET_64BIT): Likewise.
14527 (TARGET_64BIT): Likewise.
14528 (TARGET_AIX): Likewise.
14529 (WCHAR_TYPE_SIZE): Likewise.
14530 (WCHAR_TYPE): Undefine.
14531 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
14532 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
14533 (CPP_OS_RTEMS_SPEC): Delete.
14534 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
14535 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
14537 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
14539 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14541 PR tree-optimization/81578
14542 * tree-parloops.c (build_new_reduction): Bail out if
14543 reduction_code isn't one of the standard OpenMP reductions.
14544 Move the details printing after that decision.
14546 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14548 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
14549 related to reload_in_progress.
14550 (splat_input_operand): Likewise.
14551 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
14553 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
14555 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
14556 (TARGET_INSTANTIATE_DECLS): Likewise.
14557 (legitimate_indexed_address_p): Delete reload_in_progress code.
14558 (rs6000_debug_legitimate_address_p): Likewise.
14559 (rs6000_eliminate_indexed_memrefs): Likewise.
14560 (rs6000_emit_le_vsx_store): Likewise.
14561 (rs6000_emit_move_si_sf_subreg): Likewise.
14562 (rs6000_emit_move): Likewise.
14563 (register_to_reg_type): Likewise.
14564 (rs6000_pre_atomic_barrier): Likewise.
14565 (rs6000_machopic_legitimize_pic_address): Likewise.
14566 (rs6000_allocate_stack_temp): Likewise.
14567 (rs6000_address_for_fpconvert): Likewise.
14568 (rs6000_address_for_altivec): Likewise.
14569 (rs6000_secondary_memory_needed_rtx): Delete function.
14570 (rs6000_check_sdmode): Likewise.
14571 (rs6000_alloc_sdmode_stack_slot): Likewise.
14572 (rs6000_instantiate_decls): Likewise.
14573 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
14574 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
14575 Delete reload_in_progress.
14576 (*vec_reload_and_plus_<mptrsize>): Likewise.
14577 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
14578 (vsx_div_v2di): Likewise.
14579 (vsx_udiv_v2di): Likewise.
14581 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14583 * config/rs6000/rs6000.opt (mlra): Replace with stub.
14584 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
14585 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
14586 (rs6000_debug_reg_global): Delete print of LRA status.
14587 (rs6000_option_override_internal): Delete dead LRA related code.
14588 (rs6000_lra_p): Delete function.
14589 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
14591 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
14593 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
14594 * config/riscv/rtems.h: New file.
14596 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14597 Sudakshina Das <sudi.das@arm.com>
14599 * config/aarch64/aarch64.md
14600 (define_split for and<mode>3nr_compare): Move
14601 non aarch64_logical_operand to a register.
14602 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
14603 register immediate operand to a register.
14604 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
14606 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14608 PR middle-end/81564
14609 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
14611 2017-07-27 Richard Biener <rguenther@suse.de>
14613 PR tree-optimization/81573
14614 PR tree-optimization/81494
14615 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
14616 multi defuse cycle case.
14618 2017-07-27 Richard Biener <rguenther@suse.de>
14620 PR tree-optimization/81571
14621 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
14624 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
14626 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
14627 earlier and only if MASK_FPU is set. Adjust formatting.
14629 2017-07-27 Martin Liska <mliska@suse.cz>
14631 * opt-functions.awk: Add validation of value of Init.
14632 * optc-gen.awk: Pass new argument.
14634 2017-07-27 Martin Liska <mliska@suse.cz>
14636 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
14637 Fix wrong condition.
14639 2017-07-27 Martin Liska <mliska@suse.cz>
14641 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
14642 BBs and edges seen by autoFDO.
14644 2017-07-27 Richard Biener <rguenther@suse.de>
14646 PR tree-optimization/81502
14647 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
14648 with incompatible but same sized type.
14649 (execute_update_addresses_taken): Likewise.
14651 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
14653 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
14654 flag_tree_loop_vectorize rather than flag_tree_vectorize.
14656 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14659 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
14660 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
14661 Change s_operand to memory_operand.
14663 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
14665 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
14666 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
14667 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
14668 Emit instructions rather than returning an expression. Handle TFmode
14669 and KFmode by casting to TImode.
14670 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
14671 (rs6000_emit_le_vsx_store): Likewise.
14672 * config/rs6000/vsx.md (VSX_TI): New iterator.
14673 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
14674 (*vsx_le_undo_permute_<mode>): Likewise.
14675 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
14676 emit the split sequence.
14677 (*vsx_le_perm_store_<mode>): Likewise.
14679 2017-07-27 Jakub Jelinek <jakub@redhat.com>
14681 PR tree-optimization/81555
14682 PR tree-optimization/81556
14683 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
14684 if true, force CHANGED for the recursive invocation.
14685 (reassociate_bb): Remember original length of ops array, pass
14686 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
14688 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
14689 attributes for noipa attribute. For naked attribute use
14690 lookup_attribute first before lookup_attribute_spec.
14691 * final.c (rest_of_handle_final): Disable IPA RA for functions with
14693 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
14694 for functions with noipa attribute.
14695 (cgraph_externally_visible_p): Return true for functions with noipa
14697 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
14698 for functions with noipa attribute.
14699 * doc/extend.texi: Document noipa function attribute.
14700 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
14701 also for functions with noipa attribute.
14702 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
14704 2017-07-26 Andrew Pinski <apinski@cavium.com>
14706 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
14707 vec_unalign_load_cost and vec_unalign_store_cost.
14709 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14711 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
14712 -mvsx-small-integer option.
14713 (ISA_3_0_MASKS_IEEE): Likewise.
14714 (OTHER_VSX_VECTOR_MASKS): Likewise.
14715 (POWERPC_MASKS): Likewise.
14716 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
14717 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
14718 code, only testing for DImode being allowed in non-VSX floating
14720 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
14721 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
14723 (rs6000_option_override_internal): Delete -mvsx-small-integer.
14724 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
14725 TARGET_P8_VECTOR test.
14726 (rs6000_secondary_reload_simple_move): Likewise.
14727 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
14728 since TARGET_P9_VECTOR was already tested.
14729 (rs6000_opt_masks): Remove -mvsx-small-integer.
14730 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
14731 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14733 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
14734 test for TARGET_VEXTRACTUB was used, and that uses
14736 (p9 extract splitter): Likewise.
14737 (vsx_extract_<mode>_di_p9): Likewise.
14738 (vsx_extract_<mode>_store_p9): Likewise.
14739 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
14740 for TARGET_P9_VECTOR was used. Delete code that is now dead with
14741 the elimination of TARGET_VSX_SMALL_INTEGER.
14742 (vsx_extract_<mode>_p8): Likewise.
14743 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
14744 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
14745 (vsx_set_<mode>_p9): Likewise.
14746 (vsx_set_v4sf_p9): Likewise.
14747 (vsx_set_v4sf_p9_zero): Likewise.
14748 (vsx_insert_extract_v4sf_p9): Likewise.
14749 (vsx_insert_extract_v4sf_p9_2): Likewise.
14750 * config/rs6000/rs6000.md (sign extend splitter): Change
14751 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
14752 (floatsi<mode>2_lfiwax_mem): Likewise.
14753 (floatunssi<mode>2_lfiwzx_mem): Likewise.
14754 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
14755 since a test for TARGET_P9_VECTOR was used.
14756 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14757 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
14758 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14759 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
14760 TARGET_P8_VECTOR test.
14761 (fix_trunc<mode>si2_stfiwx): Likewise.
14762 (fix_trunc<mode>si2_internal): Likewise.
14763 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
14764 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14766 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
14767 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
14768 TARGET_P8_VECTOR test.
14769 (fixuns_trunc<mode>si2_stfiwx): Likewise.
14770 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
14771 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14773 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
14774 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
14775 since a test for TARGET_P9_VECTOR was used.
14776 (splitter for loading small constants): Likewise.
14778 2017-07-26 Andrew Pinski <apinski@cavium.com>
14780 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
14783 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
14786 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
14787 (fp_valid_at): Likewise.
14789 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
14791 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
14792 (qdf24xx_addrcost_table): Likewise.
14793 (cortexa57_tunings): Update to use generic_branch_cost.
14794 (cortexa72_tunings): Likewise.
14795 (cortexa73_tunings): Likewise.
14796 (qdf24xx_tunings): Likewise.
14798 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
14800 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
14801 (thunderx2t99_branch_cost): Likewise.
14802 (cortexa35_tunings): Update to use generic_branch_cost.
14803 (cortexa53_tunings): Likewise.
14804 (cortexa57_tunings): Likewise.
14805 (cortexa72_tunings): Likewise.
14806 (cortexa73_tunings): Likewise.
14807 (thunderx2t99_tunings): Likewise.
14809 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
14811 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
14812 (sparc_option_override): Honour MASK_FSMULD.
14813 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
14814 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
14815 * config/sparc/sparc.opt (mfsmuld): New option.
14816 * doc/invoke.texi (mfsmuld): Document option.
14818 2017-07-26 Marek Polacek <polacek@redhat.com>
14820 PR middle-end/70992
14821 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
14823 2017-07-26 Richard Biener <rguenther@suse.de>
14825 * gimple-match-head.c (do_valueize): Return OP if valueize
14827 (get_def): New helper to get at the def stmt of a SSA name
14828 if valueize allows.
14829 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
14830 do_valueize to get at the def stmt.
14831 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
14833 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
14835 PR middle-end/46932
14836 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
14838 2017-07-26 Martin Liska <mliska@suse.cz>
14841 * function.c (expand_function_start): Make expansion of
14842 nonlocal_goto_save_area after parm_birth_insn.
14844 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
14846 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
14847 from all CPU target flags enable members.
14849 2017-07-26 Richard Biener <rguenther@suse.de>
14851 * genmatch.c (dt_simplify::gen): Make iterator vars const.
14852 (decision_tree::gen): Make 'type' const.
14853 (write_predicate): Likewise.
14855 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
14857 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
14858 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
14859 (rs6000_option_override_internal): Likewise.
14860 (rs6000_expand_vector_set): Likewise.
14861 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
14862 (TARGET_UPPER_REGS_SF): Likewise.
14863 (TARGET_UPPER_REGS_DI): Likewise.
14864 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
14865 (TARGET_DIRECT_MOVE_64BIT): Likewise.
14866 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
14867 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14868 (Splitters for DI constants in Altivec registers): Likewise.
14869 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
14870 (vsx_set_v4sf_p9): Likewise.
14871 (vsx_set_v4sf_p9_zero): Likewise.
14872 (vsx_insert_extract_v4sf_p9): Likewise.
14873 (vsx_insert_extract_v4sf_p9_2): Likewise.
14875 2017-07-25 Carl Love <cel@us.ibm.com>
14877 * doc/extend.texi: Update the built-in documentation file for the
14878 existing built-in functions
14879 vector signed char vec_cnttz (vector signed char);
14880 vector unsigned char vec_cnttz (vector unsigned char);
14881 vector signed short vec_cnttz (vector signed short);
14882 vector unsigned short vec_cnttz (vector unsigned short);
14883 vector signed int vec_cnttz (vector signed int);
14884 vector unsigned int vec_cnttz (vector unsigned int);
14885 vector signed long long vec_cnttz (vector signed long long);
14886 vector unsigned long long vec_cnttz (vector unsigned long long);
14888 2017-07-25 Andrew Pinski <apinski@cavium.com>
14890 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
14891 accesses where the use is for the first operand of a BIT_INSERT.
14893 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
14896 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
14897 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
14899 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
14901 * config/i386/gstabs.h: Delete.
14902 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
14904 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
14906 * config/i386/i386.c (ix86_decompose_address): Do not check for
14907 register RTX when looking at index_reg or base_reg.
14908 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
14910 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
14912 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
14913 to update EH info here.
14915 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
14917 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
14919 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
14921 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
14923 2017-07-25 Torsten Duwe <duwe@suse.de>
14925 * common.opt: Introduce -fpatchable-function-entry
14926 command line option, and its variables function_entry_patch_area_size
14927 and function_entry_patch_area_start.
14928 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
14929 including a two-value parser.
14930 * target.def (print_patchable_function_entry): New target hook.
14931 * targhooks.h (default_print_patchable_function_entry): New function.
14932 * targhooks.c (default_print_patchable_function_entry): Likewise.
14933 * toplev.c (process_options): Switch off IPA-RA if
14934 patchable function entries are being generated.
14935 * varasm.c (assemble_start_function): Look at the
14936 patchable-function-entry command line switch and current
14937 function attributes and maybe generate NOP instructions by
14938 calling the print_patchable_function_entry hook.
14939 * doc/extend.texi: Document patchable_function_entry attribute.
14940 * doc/invoke.texi: Document -fpatchable_function_entry
14941 command line option.
14942 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
14944 * doc/tm.texi: Re-generate.
14946 2017-07-25 Jakub Jelinek <jakub@redhat.com>
14949 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
14950 TARGET_AVX512DQ rather than TARGET_AVX512BW.
14952 2017-07-25 Tamar Christina <tamar.christina@arm.com>
14954 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
14956 2017-07-25 Richard Biener <rguenther@suse.de>
14958 PR tree-optimization/81455
14959 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
14960 not walk in cycles when looking for guards.
14962 2017-07-25 Richard Biener <rguenther@suse.de>
14964 PR tree-optimization/81529
14965 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
14966 when optimizing backedge uses.
14968 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
14970 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
14972 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
14973 to dl_section_ref. On AIX, append an expression to subtract
14974 the size of the section length to dl_section_ref.
14976 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
14978 * configure.ac: If any of the config.* scripts fail, exit 1.
14979 * configure: Regenerate.
14981 2017-07-25 Richard Biener <rguenther@suse.de>
14983 PR middle-end/81546
14984 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
14985 of immediate uses, be more verbose on errors.
14987 2017-07-25 Richard Biener <rguenther@suse.de>
14989 PR tree-optimization/81510
14990 * tree-vect-loop.c (vect_is_simple_reduction): When the
14991 reduction stmt is not inside the loop bail out.
14993 2017-07-25 Richard Biener <rguenther@suse.de>
14995 PR tree-optimization/81303
14996 * tree-vect-loop-manip.c (vect_loop_versioning): Build
14997 profitability check against LOOP_VINFO_NITERSM1.
14999 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15001 * domwalk.c (cmp_bb_postorder): Simplify.
15002 (sort_bbs_postorder): New function. Use it...
15003 (dom_walker::walk): ...here to optimize common cases.
15005 2017-07-25 Martin Liska <mliska@suse.cz>
15008 * ipa-visibility.c (function_and_variable_visibility): Make the
15009 redirection just on target that supports aliasing.
15010 Fix GNU coding style.
15012 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
15015 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
15017 * config/arm/rtems.h: Likewise.
15018 * config/bfin/rtems.h: Likewise.
15019 * config/i386/rtemself.h: Likewise.
15020 * config/lm32/rtems.h: Likewise.
15021 * config/m32c/rtems.h: Likewise.
15022 * config/m68k/rtemself.h: Likewise.
15023 * config/microblaze/rtems.h: Likewise.
15024 * config/mips/rtems.h: Likewise.
15025 * config/moxie/rtems.h: Likewise.
15026 * config/nios2/rtems.h: Likewise.
15027 * config/powerpcspe/rtems.h: Likewise.
15028 * config/rs6000/rtems.h: Likewise.
15029 * config/rtems.h: Likewise.
15030 * config/sh/rtems.h: Likewise.
15031 * config/sh/rtemself.h: Likewise.
15032 * config/sparc/rtemself.h: Likewise.
15034 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15037 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
15038 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
15039 * tree-ssa-structalias.c (alias_get_name): Same.
15041 2017-07-25 Bin Cheng <bin.cheng@arm.com>
15044 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
15045 instructions if no du chain is found.
15047 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15049 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
15051 2017-07-25 Richard Biener <rguenther@suse.de>
15053 PR middle-end/81505
15054 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
15057 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
15059 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
15060 upper-regs options.
15061 (ISA_2_7_MASKS_SERVER): Likewise.
15062 (ISA_3_0_MASKS_IEEE): Likewise.
15063 (OTHER_P8_VECTOR_MASKS): Likewise.
15064 (OTHER_VSX_VECTOR_MASKS): Likewise.
15065 (POWERPC_MASKS): Likewise.
15066 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
15067 duplicate list of options.
15068 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
15069 explicit -mupper-regs options.
15070 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
15071 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
15072 alias for -mupper-regs-df.
15073 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
15074 (rs6000_init_hard_regno_mode_ok): Likewise.
15075 (rs6000_option_override_internal): Likewise.
15076 (rs6000_opt_masks): Likewise.
15077 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
15078 options in terms of whether -mvsx or -mpower8-vector was used.
15079 (TARGET_UPPER_REGS_DI): Likewise.
15080 (TARGET_UPPER_REGS_SF): Likewise.
15081 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
15082 -mupper-regs-* options.
15084 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15086 * passes.c (emergency_dump_function): Print some empty lines and a
15087 header before the RTL dump.
15089 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15091 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
15093 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
15096 * config/aarch64/aarch64.c (aarch64_classify_symbol):
15097 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
15099 2017-07-24 Carl Love <cel@us.ibm.com>
15101 * config/rs6000/rs6000-c.c: Add support for built-in functions
15102 vector float vec_extract_fp32_from_shorth (vector unsigned short);
15103 vector float vec_extract_fp32_from_shortl (vector unsigned short);
15104 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
15105 vec_extract_fp_from_shortl): Add defines for the two builtins.
15106 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
15107 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
15109 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
15110 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
15111 * doc/extend.texi: Update the built-in documentation file for the
15112 new built-in function.
15114 2017-07-24 Jakub Jelinek <jakub@redhat.com>
15117 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
15119 * doc/generic.texi: Likewise.
15120 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
15121 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
15123 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
15125 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
15126 (aarch64_mls_elt_merge<mode>): Likewise.
15128 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
15130 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
15131 having __cxa_atexit.
15133 2017-07-23 Michael Collison <michael.collison@arm.com>
15135 * config/arm/arm.c (arm_option_override): Deprecate
15136 use of -mstructure-size-boundary.
15137 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
15138 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
15140 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15143 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
15144 Reduce cost estimate for direct moves.
15146 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
15149 * config/i386/i386.c (ix86_option_override_internal): Disable
15150 BMI, BMI2 and TBM instructions for -m16.
15152 2017-07-21 Carl Love <cel@us.ibm.com>
15154 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15155 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15156 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15157 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15158 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15159 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15160 VMULOSW): New enum "unspec" values.
15161 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15162 altivec_vmulosw): New patterns.
15163 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15164 VMULOSW): Add definitions.
15166 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
15168 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
15169 (qdf24xx): Likewise.
15170 * config/aarch64/aarch64-options-extensions.def (rdma); New.
15171 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
15172 (AARCH64_FL_V8_1): Renumber.
15173 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
15174 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
15175 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
15176 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
15177 rdma to feature modifiers list.
15179 2017-07-21 Yury Gribov <tetra2005@gmail.com>
15181 PR middle-end/56727
15182 * ipa-visibility (function_and_variable_visibility): Convert
15183 recursive PLT call to direct call if appropriate.
15185 2017-07-21 Andrew Pinski <apinski@cavium.com>
15187 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
15188 operand 1 to see if the types precision matches.
15189 * fold-const.c (operand_equal_p): Likewise.
15191 2017-07-21 Richard Biener <rguenther@suse.de>
15193 PR tree-optimization/81303
15194 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
15195 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
15196 (vect_peeling_hash_get_lowest_cost): Adjust.
15197 (vect_enhance_data_refs_alignment): Likewise. Use
15198 vect_get_peeling_costs_all_drs to compute the penalty for no
15199 peeling to match up costs.
15201 2017-07-21 Richard Biener <rguenther@suse.de>
15203 PR tree-optimization/81500
15204 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
15205 we didn't identify a reduction path.
15207 2017-07-21 Tom de Vries <tom@codesourcery.com>
15208 Cesar Philippidis <cesar@codesourcery.com>
15210 PR gcov-profile/81442
15211 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
15214 2017-07-21 Tom de Vries <tom@codesourcery.com>
15217 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
15219 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
15220 nvptx_override_options_after_change.
15222 2017-07-21 Ulrich Drepper <drepper@redhat.com>
15224 * dwarf2out.c (output_file_names): Avoid double testing for
15225 dwarf_version >= 5.
15227 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
15229 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
15231 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15233 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
15235 (try_crossjump_to_edge): Do not punt on partitioned functions.
15237 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15239 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
15240 Put all BBs reachable only via paths crossing cold region to cold
15242 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
15244 2016-07-21 Richard Biener <rguenther@suse.de>
15246 PR tree-optimization/81303
15247 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
15248 into account prologue and epilogue iterations when raising
15249 min_profitable_iters to sth at least covering one vector iteration.
15251 2017-07-21 Tamar Christina <tamar.christina@arm.com>
15253 * config/arm/arm.c (arm_test_cpu_arch_dat):
15256 2017-07-20 Nathan Sidwell <nathan@acm.org>
15258 Remove TYPE_METHODS.
15259 * tree.h (TYPE_METHODS): Delete.
15260 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
15261 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
15262 (dbxout_type_methods): Scan TYPE_FIELDS.
15263 (dbxout_type): Don't check TYPE_METHODS here.
15264 * function.c (use_register_for_decl): Always ignore register for
15265 class types when not optimizing.
15266 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
15267 * tree.c (free_lang_data_in_type): Stitch out member functions and
15268 templates from TYPE_FIELDS.
15269 (build_distinct_type_copy, verify_type_variant,
15270 verify_type): Member fns are on TYPE_FIELDS.
15271 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
15272 * tree-pretty-print.c (dump_generic_node): Likewise.
15274 2017-07-20 Jakub Jelinek <jakub@redhat.com>
15277 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
15278 V2TImode and V4TImode.
15279 (ix86_expand_vector_extract): Likewise.
15280 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
15281 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
15282 (ssescalarmode): Handle V4TImode and V2TImode.
15283 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
15284 (*vec_extractv2ti, *vec_extractv4ti): New insns.
15285 (VEXTRACTI128_MODE): New mode iterator.
15286 (splitter for *vec_extractv?ti first element): New.
15287 (VEC_INIT_MODE): New mode iterator.
15288 (vec_init<mode>): Consolidate 3 expanders into one using
15289 VEC_INIT_MODE mode iterator.
15291 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
15293 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
15294 non_spilled_static_chain_regno_p.
15296 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
15298 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
15300 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
15302 * bb-reorder.c (connect_traces): Allow copying of blocks within
15305 2017-07-20 Richard Biener <rguenther@suse.de>
15307 * gimple.h (gimple_phi_result): Add gphi * overload.
15308 (gimple_phi_result_ptr): Likewise.
15309 (gimple_phi_arg): Likewise. Adjust index assert to only
15310 allow actual argument accesses rather than all slots available
15312 (gimple_phi_arg_def): Add gphi * overload.
15313 * tree-phinodes.c (make_phi_node): Initialize only actual
15315 (resize_phi_node): Clear memory not covered by old node,
15316 do not initialize excess argument slots.
15317 (reserve_phi_args_for_new_edge): Initialize new argument slot
15320 2017-07-20 Bin Cheng <bin.cheng@arm.com>
15322 PR tree-optimization/81388
15324 2016-07-21 Bin Cheng <bin.cheng@arm.com>
15326 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
15327 by removing computation of may_be_zero.
15329 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15330 Tom de Vries <tom@codesourcery.com>
15332 PR middle-end/81030
15333 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
15334 when gimple level profile disagrees with what RTL expander did.
15336 2017-07-20 Richard Biener <rguenther@suse.de>
15338 PR tree-optimization/61171
15339 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
15340 (vect_analyze_stmt): Add slp instance parameter.
15341 (vectorizable_reduction): Likewise.
15342 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
15343 (vect_is_simple_reduction): Deal with chains not detected
15344 as SLP reduction chain, specifically not properly associated
15345 chains containing a mix of plus/minus.
15346 (get_reduction_op): Remove.
15347 (get_initial_defs_for_reduction): Simplify, pass in whether
15348 this is a reduction chain, pass in the SLP node for the PHIs.
15349 (vect_create_epilog_for_reduction): Get the SLP instance as
15351 (vectorizable_reduction): Get the SLP instance as arg.
15352 During analysis remember the SLP node with the PHIs in the
15353 instance. Simplify getting at the vectorized reduction PHIs.
15354 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
15355 through SLP instance.
15356 (vect_slp_analyze_operations): Likewise.
15357 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
15358 (vect_transform_stmt): Likewise.
15360 2017-07-20 Tom de Vries <tom@codesourcery.com>
15362 PR tree-optimization/81489
15363 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
15364 read of phi arg location to before loop that modifies phi.
15366 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15368 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
15371 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15373 PR middle-end/81331
15374 * except.c (execute): Fix ordering issue.
15376 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
15378 PR rtl-optimization/81423
15379 * combine.c (make_compound_operation_int): Don't try to optimize
15380 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
15382 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
15384 PR rtl-optimization/81423
15385 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
15386 with a constant that is -1 in the truncated to mode.
15388 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15390 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
15391 (determine_unlikely_bbs): ... here.
15392 * predict.h (propagate_unlikely_bbs_forward): Declare.
15393 * cfgexpand.c (pass_expand::execute): Use it.
15394 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
15396 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
15397 propagate_unlikely_bbs_forward.
15399 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15401 PR middle-end/81331
15402 * except.c (maybe_add_nop_after_section_switch): New function.
15405 2017-07-19 Tom de Vries <tom@codesourcery.com>
15407 * gimple.h (gimple_phi_set_arg): Make assert more strict.
15409 2017-07-19 Tom de Vries <tom@codesourcery.com>
15411 * gimple.h (gimple_phi_arg): Make assert more strict.
15413 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
15415 * config.gcc (powerpc*-*-*): Add mmintrin.h.
15416 * config/rs6000/mmintrin.h: New file.
15417 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
15419 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15421 PR tree-optimization/81346
15422 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
15424 2017-07-19 Tom de Vries <tom@codesourcery.com>
15426 * config/nvptx/nvptx.md (VECIM): Add V2DI.
15428 2017-07-19 Tom de Vries <tom@codesourcery.com>
15430 * config/nvptx/nvptx-modes.def: Add V2DImode.
15431 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
15432 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
15433 (nvptx_output_mov_insn): Handle lack of mov.b128.
15434 (nvptx_print_operand): Handle 'H' and 'L' codes.
15435 (nvptx_vector_mode_supported): Allow V2DImode.
15436 (nvptx_preferred_simd_mode): New function.
15437 (nvptx_data_alignment): New function.
15438 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
15439 nvptx_preferred_simd_mode.
15440 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
15442 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
15444 2017-07-19 Tom de Vries <tom@codesourcery.com>
15446 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
15447 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
15448 (nvptx_vector_mode_supported): New function. Allow V2SImode.
15449 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
15450 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
15451 (mov<VECIM>_insn): New define_insn.
15452 (define_expand "mov<VECIM>): New define_expand.
15454 2017-07-19 Tom de Vries <tom@codesourcery.com>
15456 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
15458 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15460 PR tree-optimization/81346
15461 * fold-const.h (fold_div_compare, range_check_type): Declare.
15462 * fold-const.c (range_check_type): New function.
15463 (build_range_check): Use range_check_type.
15464 (fold_div_compare): No longer static, rewritten into
15465 a match.pd helper function.
15466 (fold_comparison): Don't call fold_div_compare here.
15467 * match.pd (X / C1 op C2): New optimization using fold_div_compare
15468 as helper function.
15470 2017-07-19 Nathan Sidwell <nathan@acm.org>
15472 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
15473 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
15474 * tree.c (find_decls_types_r, verify_type): Use
15475 TYPE_{MIN,MAX}_VALUE_RAW.
15476 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
15477 (hash_tree): Likewise.
15478 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
15480 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
15483 2017-07-18 Tom de Vries <tom@codesourcery.com>
15485 PR middle-end/81464
15486 * omp-expand.c (expand_omp_for_static_chunk): Handle
15487 equal-argument loop exit phi.
15489 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
15492 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
15493 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
15494 operand 2 predicate.
15495 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
15496 operand 2 predicate.
15497 (ror,rol -> rorx splitters): Use const_int_operand as
15498 operand 2 predicate.
15500 2017-06-18 Richard Biener <rguenther@suse.de>
15502 PR tree-optimization/81410
15503 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
15504 the gap in the ! slp_perm SLP case after each group.
15506 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15508 PR middle-end/81463
15509 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
15512 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15514 PR middle-end/81462
15515 * predict.c (set_even_probabilities): Cleanup; do not affect
15516 probabilities that are already known.
15517 (combine_predictions_for_bb): Call even when count is set.
15519 2017-07-18 Nathan Sidwell <nathan@acm.org>
15521 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
15524 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15527 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
15528 optimization for loop niter analysis.
15530 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
15533 * config/avr/avr.c (avr_optimize_casesi): Don't use
15534 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
15536 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
15538 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
15539 body_cost_vec from _vect_peel_extended_info.
15540 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
15541 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
15544 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15546 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
15548 2017-07-18 Richard Biener <rguenther@suse.de>
15550 PR tree-optimization/80620
15551 PR tree-optimization/81403
15552 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
15553 info when re-using a VN table entry.
15555 2017-07-18 Richard Biener <rguenther@suse.de>
15557 PR tree-optimization/81418
15558 * tree-vect-loop.c (vectorizable_reduction): Properly compute
15559 vectype_in. Verify that with lane-reducing reduction operations
15560 we have a single def-use cycle.
15562 2017-07-17 Carl Love <cel@us.ibm.com>
15564 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
15566 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15567 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15568 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15569 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15570 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15571 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15572 VMULOSW): New enum "unspec" values.
15573 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
15574 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
15575 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15576 altivec_vmulosw): New patterns.
15577 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15578 VMULOSW): Add definitions.
15579 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
15581 * config/alpha/alpha.c: Include predict.h.
15583 2017-07-17 Yury Gribov <tetra2005@gmail.com>
15585 * tree-vrp.c (compare_assert_loc): Fix comparison function
15586 to return predictable results.
15588 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15590 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
15592 (subdi3): Likewise.
15593 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
15594 * doc/invoke.texi (mexpand-adddi): Update text.
15596 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15598 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
15599 that also clobbers the CC register. The old expand code is moved
15601 (*arc_clzsi2): ... here.
15602 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
15603 the CC register. The old expand code is moved to ...
15604 (arc_ctzsi2): ... here.
15606 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15608 * config/arc/arc.opt (mindexed-loads): Use initial value
15609 TARGET_INDEXED_LOADS_DEFAULT.
15610 (mauto-modify-reg): Use initial value
15611 TARGET_AUTO_MODIFY_REG_DEFAULT.
15612 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
15613 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
15614 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
15615 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
15617 2017-07-17 Martin Liska <mliska@suse.cz>
15620 * opts.c (finish_options): Do not allow -fgnu-tm
15621 w/ -fsanitize={kernel-,}address. Say sorry.
15623 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15626 * tree-loop-distribution.c (classify_partition): Only assert on
15627 numer of iterations.
15628 (merge_dep_scc_partitions): Delete prameter. Update function call.
15629 (distribute_loop): Remove code handling loop with unknown niters.
15630 (pass_loop_distribution::execute): Skip loop with unknown niters.
15632 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15635 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
15636 function sort_partitions_by_post_order.
15638 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15640 PR tree-optimization/81374
15641 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
15642 the max index of basic blocks, rather than number of basic blocks.
15644 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15646 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
15648 (arc_legitimate_pic_operand_p): Likewise.
15649 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
15651 (arc_needs_pcl_p): Likewise.
15652 (arc_legitimate_pc_offset_p): Likewise.
15653 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
15654 function is also used in constrains.md.
15655 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
15656 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
15657 PLUS. Only return true/false in known cases, otherwise assert.
15658 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
15659 is already called in arc_legitimate_constant_p.
15660 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
15662 (LEGITIMATE_PIC_OPERAND_P): Use
15663 arc_raw_symbolic_reference_mentioned_p function.
15664 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
15669 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15670 Andrew Burgess <andrew.burgess@embecosm.com>
15672 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
15673 (arc_return_address_register): New function.
15674 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
15675 (arc_handle_fndecl_attribute): Add naked attribute.
15676 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
15677 (TARGET_WARN_FUNC_RETURN): Likewise.
15678 (arc_allocate_stack_slots_for_args): New function.
15679 (arc_warn_func_return): Likewise.
15680 (machine_function): Change type fn_type.
15681 (arc_compute_function_type): Consider new naked function type,
15682 change function return type.
15683 (arc_must_save_register): Adapt to handle new
15684 arc_compute_function_type's return type.
15685 (arc_expand_prologue): Likewise.
15686 (arc_expand_epilogue): Likewise.
15687 (arc_return_address_regs): Delete.
15688 (arc_return_address_register): New function.
15689 (arc_epilogue_uses): Use above function.
15690 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
15691 (arc_function_type): Change encoding, add naked type.
15692 (ARC_INTERRUPT_P): Change to handle the new encoding.
15693 (ARC_FAST_INTERRUPT_P): Likewise.
15694 (ARC_NORMAL_P): Define.
15695 (ARC_NAKED_P): Likewise.
15696 (arc_compute_function_type): Delete prototype.
15697 * config/arc/arc.md (in_ret_delay_slot): Use
15698 arc_return_address_register function.
15699 (simple_return): Likewise.
15700 (p_return_i): Likewise.
15702 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15704 PR tree-optimization/81428
15705 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
15706 can't be built for those types.
15708 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
15710 Remove stuff dead since r239246.
15712 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
15713 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
15714 (avr_inform_devices): Remove dead stuff.
15716 2017-07-17 Tamar Christina <tamar.christina@arm.com>
15718 * config/arm/arm_neon.h: Fix softp typo.
15720 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15722 PR tree-optimization/81365
15723 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
15724 aggregate moves onto bb predecessor edges, make sure there are no
15725 loads that could alias the lhs in between the start of bb and the
15728 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
15731 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
15732 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
15733 [LSHIFTRT, outer_code = TRUNCATE]: Same.
15735 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15737 PR tree-optimization/81396
15738 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
15739 (init_symbolic_number): Initialize it to 1.
15740 (perform_symbolic_merge): Add n_ops from both operands into the new
15742 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
15743 without base_addr as useless if they need more than one operation.
15744 (bswap_replace): Handle !bswap case for NULL base_addr.
15746 2017-07-17 Tom de Vries <tom@codesourcery.com>
15749 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
15752 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
15754 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
15755 conditional builtin define __FIX_LEON3FT_B2BST.
15757 2017-07-17 Daniel Cederman <cederman@gaisler.com>
15759 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
15760 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
15763 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
15765 PR rtl-optimization/81424
15766 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
15767 to remove potential trapping from operands if -fnon-call-exceptions.
15769 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15771 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
15772 profile_proability for scalling.
15773 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
15775 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15777 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
15779 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15781 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
15782 fixpoint arithmetics.
15784 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15786 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
15787 fixpoint arithmetics.
15789 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15791 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
15792 fixpoint arithmetics.
15794 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15796 * profile-count.h (profile_probability::from_reg_br_prob_note,
15797 profile_probability::to_reg_br_prob_note): New functions.
15798 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
15799 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
15800 * predict.c (probability_reliable_p): Update.
15801 (edge_probability_reliable_p): Update.
15802 (br_prob_note_reliable_p): Update.
15803 (invert_br_probabilities): Update.
15804 (add_reg_br_prob_note): New function.
15805 (combine_predictions_for_insn): Update.
15806 * asan.c (asan_clear_shadow): Update.
15807 * cfgbuild.c (compute_outgoing_frequencies): Update.
15808 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
15809 (update_br_prob_note): Update.
15810 (rtl_verify_edges): Update.
15811 (purge_dead_edges): Update.
15812 (fixup_reorder_chain): Update.
15813 * emit-rtl.c (try_split): Update.
15814 * ifcvt.c (cond_exec_process_insns): Update.
15815 (cond_exec_process_if_block): Update.
15816 (dead_or_predicable): Update.
15817 * internal-fn.c (expand_addsub_overflow): Update.
15818 (expand_neg_overflow): Update.
15819 (expand_mul_overflow): Update.
15820 * loop-doloop.c (doloop_modify): Update.
15821 * loop-unroll.c (compare_and_jump_seq): Update.
15822 * optabs.c (emit_cmp_and_jump_insn_1): Update.
15823 * predict.h: Update.
15824 * reorg.c (mostly_true_jump): Update.
15826 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
15827 * config/alpha/alpha.c (emit_unlikely_jump): Update.
15828 * config/arc/arc.c: (emit_unlikely_jump): Update.
15829 * config/arm/arm.c: (emit_unlikely_jump): Update.
15830 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
15831 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
15832 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
15833 (ix86_print_operand): Update.
15834 (ix86_split_fp_branch): Update.
15835 (predict_jump): Update.
15836 * config/ia64/ia64.c (ia64_print_operand): Update.
15837 * config/mmix/mmix.c (mmix_print_operand): Update.
15838 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
15839 (rs6000_expand_split_stack_prologue): Update.
15840 * config/rs6000/rs6000.c: Update.
15841 * config/s390/s390.c (s390_expand_vec_strlen): Update.
15842 (s390_expand_vec_movstr): Update.
15843 (s390_expand_cs_tdsi): Update.
15844 (s390_expand_split_stack_prologue): Update.
15845 * config/sh/sh.c (sh_print_operand): Update.
15846 (expand_cbranchsi4): Update.
15847 (expand_cbranchdi4): Update.
15848 * config/sparc/sparc.c (output_v9branch): Update.
15849 * config/spu/spu.c (get_branch_target): Update.
15850 (ea_load_store_inline): Update.
15851 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
15852 * config/tilepro/tilepro.c: Update.
15854 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
15856 * gimplify.c (mostly_copy_tree_r): Revert latest change.
15857 (gimplify_save_expr): Likewise.
15859 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15861 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
15863 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15865 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
15867 * timevar.def (TV_IPA_FNSUMMARY): Define.
15869 2017-07-16 Daniel Cederman <cederman@gaisler.com>
15871 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
15872 to back store errata sensitive sequence from being generated.
15873 (sqrtdf2_fix): Likewise.
15875 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15877 * tree-ssa-threadupdate.c (compute_path_counts,
15878 update_joiner_offpath_counts): Use profile_probability.
15880 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
15883 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15885 * config/arm/arm-c.c (arm_cpu_builtins): Define
15886 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
15888 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
15890 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15891 array entries to represent __ieee128 versions of the
15892 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
15893 scalar_extract_sig, and scalar_insert_exp built-in functions.
15894 (altivec_resolve_overloaded_builtin): Add special case handling
15895 for the __builtin_scalar_insert_exp function, as represented by
15896 the P9V_BUILTIN_VEC_VSIEDP constant.
15897 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
15898 exponent support for __ieee128 argument.
15899 (VSESQP): Add scalar extract signature support for __ieee128
15901 (VSTDCNQP): Add scalar test negative support for __ieee128
15903 (VSIEQP): Add scalar insert exponent support for __int128 argument
15904 with __ieee128 result.
15905 (VSIEQPF): Add scalar insert exponent support for __ieee128
15906 argument with __ieee128 result.
15907 (VSTDCQP): Add scalar test data class support for __ieee128
15909 (VSTDCNQP): Add overload support for scalar test negative with
15910 __ieee128 argument.
15911 (VSTDCQP): Add overload support for scalar test data class
15912 __ieee128 argument.
15913 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
15914 UNSPEC_VSX_SXSIGDP.
15915 (UNSPEC_VSX_SIEXPQP): New constant.
15916 (xsxexpqp): New insn for VSX scalar extract exponent quad
15918 (xsxsigqp): New insn for VSX scalar extract significand quad
15920 (xsiexpqpf): New insn for VSX scalar insert exponent quad
15921 precision with floating point argument.
15922 (xststdcqp): New expand for VSX scalar test data class quad
15924 (xststdcnegqp): New expand for VSX scalar test negative quad
15926 (xststdcqp): New insn to match expansions for VSX scalar test data
15927 class quad precision and VSX scalar test negative quad precision.
15928 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
15929 special case operand checking to enforce that second operand of
15930 VSX scalar test data class with quad precision argument is a 7-bit
15932 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
15933 prototypes and descriptions of __ieee128 versions of
15934 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
15935 scalar_test_data_class, and scalar_test_neg built-in functions.
15937 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15939 PR tree-optimization/81162
15940 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
15941 replace a negate with an add.
15943 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
15945 * doc/invoke.texi (arm/-mcpu): Document +crypto.
15947 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15949 * config/arm/arm-c.c (arm_cpu_builtins): Define
15950 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
15952 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15954 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
15955 (armv8-r): Set ARM Cortex-R52 as default CPU.
15956 * config/arm/arm-tables.opt: Regenerate.
15957 * config/arm/arm-tune.md: Regenerate.
15958 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
15960 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
15961 extension for -mcpu=cortex-r52.
15963 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15965 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
15966 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
15967 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
15968 (fp-armv8): Define it as FP_ARMv8 only.
15969 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
15970 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
15972 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
15973 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
15974 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
15975 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
15976 than TARGET_FPU_ARMV8.
15977 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
15978 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
15979 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
15981 * config/arm/neon.md (neon_vrint): Likewise.
15982 (neon_vcvt): Likewise.
15983 (neon_<fmaxmin_op><mode>): Likewise.
15984 (<fmaxmin><mode>3): Likewise.
15985 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
15986 * config/arm/predicates.md (arm_cond_move_operator): Check against
15987 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
15989 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
15991 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
15992 to top of function.
15994 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15996 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
15997 loop in comment with memset.
15999 2017-07-14 Martin Liska <mliska@suse.cz>
16001 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
16002 * dwarf2out.c (is_java): Remove the function.
16003 (output_pubname): Remove usage of the function.
16004 (lower_bound_default): Remove usage of DW_LANG_Java.
16005 (gen_compile_unit_die): Likewise.
16006 * gcc.c: Remove compiler defaults for .java and .zip files.
16007 * gimple-expr.c (remove_suffix): Change as there's no longer
16008 extension than 4-letter one.
16009 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
16010 (gimplify_save_expr): Likewise.
16011 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
16012 as it's possible even for other languages than Java.
16013 * langhooks.h (struct lang_hooks): Remove Java from a comment.
16014 * lto-opts.c (lto_write_options): Remove reference to Java.
16015 * opts.c (strip_off_ending): Update file extension handling.
16016 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
16017 * tree-eh.c (lower_resx): Likewise.
16018 * tree.c (free_lang_data_in_type): Remove dead code.
16019 (find_decls_types_r): Likewise.
16020 (build_common_builtin_nodes): Remove Java from a comment.
16021 (verify_type): Remove dead code.
16022 * varasm.c (assemble_external): Remove Java from a comment.
16024 2017-07-14 Martin Liska <mliska@suse.cz>
16026 * opts.c (finish_options): Add quotes.
16027 (common_handle_option): Likewise.
16029 2017-07-14 Martin Liska <mliska@suse.cz>
16031 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
16032 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
16033 Remove N_SO_PASCAL.
16034 * dwarf2out.c (lower_bound_default): Do not handle
16036 (gen_compile_unit_die): Likewise.
16037 * gcc.c: Remove default extension binding for GNU Pascal.
16038 * stmt.c: Remove Pascal language from a comment.
16039 * xcoffout.c: Likewise.
16041 2017-07-13 David Malcolm <dmalcolm@redhat.com>
16044 * diagnostic-show-locus.c (fixit_cmp): New function.
16045 (layout::layout): Sort m_fixit_hints.
16046 (column_range::column_range): Assert that the values are valid.
16047 (struct char_span): New struct.
16048 (correction::overwrite): New method.
16049 (struct source_line): New struct.
16050 (line_corrections::add_hint): Add assertions. Reimplement memcpy
16051 calls in terms of classes source_line and char_span, and
16052 correction::overwrite.
16053 (selftest::test_overlapped_fixit_printing_2): New function.
16054 (selftest::diagnostic_show_locus_c_tests): Call it.
16056 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
16058 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
16059 early if there is no lhs.
16061 2017-07-13 Martin Liska <mliska@suse.cz>
16063 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
16064 (gen_reference_type_die): Likewise.
16065 * stor-layout.c: Remove Pascal-related comment.
16067 2017-07-13 Martin Liska <mliska@suse.cz>
16069 * opts.c (finish_options): Add quotes to error messages.
16070 (parse_sanitizer_options): Likewise.
16072 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
16074 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
16076 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
16078 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
16080 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
16082 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
16084 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
16086 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
16089 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
16090 provides the hardware capability bits, define the macro
16091 __BUILTIN_CPU_SUPPORTS__.
16092 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
16093 if GLIBC does not provide the hardware capability bits. Add a
16094 gcc_unreachable call if the built-in cpu function is neither
16095 __builtin_cpu_is nor __builtin_cpu_supports.
16096 (rs6000_get_function_versions_dispatcher): Change the warning
16097 that an old GLIBC is used which does not export the capability
16098 bits to be an error.
16099 * doc/extend.texi (target_clones attribute): Document the
16100 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
16101 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
16102 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
16103 the macros defined by GCC if the newer GLIBC is available.
16105 2017-07-12 Jeff Law <law@redhat.com>
16107 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
16108 remaining includes slightly.
16109 * config/riscv/riscv-builtins.c: Include profile-count.h.
16111 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16114 * config/avr/avr.c (avr_set_current_function): In diagnostic
16115 messages: Quote keywords and (parts of) identifiers.
16116 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
16119 2017-07-12 Carl Love <cel@us.ibm.com>
16121 * config/rs6000/rs6000-c.c: Add support for built-in functions
16122 vector bool char vec_revb (vector bool char);
16123 vector bool short vec_revb (vector short char);
16124 vector bool int vec_revb (vector bool int);
16125 vector bool long long vec_revb (vector bool long long);
16126 * doc/extend.texi: Update the built-in documentation file for the
16127 new built-in functions.
16129 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16131 * config/s390/s390.md: Remove movcc splitter.
16133 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16135 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
16136 load/store on condition.
16138 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16141 * config/avr/avr.c (avr_encode_section_info)
16142 [progmem && !TREE_READONLY]: Error if progmem object needs
16145 2017-07-11 Michael Collison <michael.collison@arm.com>
16147 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
16150 2017-07-11 Carl Love <cel@us.ibm.com>
16152 * config/rs6000/rs6000-c.c: Add support for builtins
16153 vector unsigned int vec_parity_lsbb (vector signed int);
16154 vector unsigned int vec_parity_lsbb (vector unsigned int);
16155 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
16156 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
16157 vector unsigned long long vec_parity_lsbb (vector signed long long);
16158 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
16159 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
16160 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
16161 * doc/extend.texi: Update the built-in documentation file for the
16162 new built-in functions.
16164 2017-07-11 David Malcolm <dmalcolm@redhat.com>
16166 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
16167 (layout::m_primary_loc): New field.
16168 (layout::layout): Initialize new field. Move location filtering
16169 logic from here to...
16170 (layout::maybe_add_location_range): ...this new method. Add
16171 support for filtering to just the lines already specified by other
16173 (layout::will_show_line_p): New method.
16174 (gcc_rich_location::add_location_if_nearby): New method.
16175 (selftest::test_add_location_if_nearby): New test function.
16176 (selftest::diagnostic_show_locus_c_tests): Call it.
16177 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
16180 2017-07-11 Tom de Vries <tom@codesourcery.com>
16182 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
16183 (bb_first_real_insn): New function.
16184 (nvptx_single): Add extra initialization of broadcasted condition
16187 2017-07-11 Nathan Sidwell <nathan@acm.org>
16189 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
16191 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
16193 * doc/extend.texi (AVR Function Attributes): Remove weblink to
16194 Binutils doc as TEXI will mess them up.
16195 * doc/invoke.texi (AVR Options): Same here.
16197 2017-07-11 Daniel Cederman <cederman@gaisler.com>
16199 * config/sparc/sparc.opt (mfix-ut700): New option.
16200 (mfix-gr712rc): Likewise.
16201 (sparc_fix_b2bst): New variable.
16202 * doc/invoke.texi (SPARC options): Document them.
16203 (ARM options): Fix warnings.
16204 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
16205 instructions to prevent sequences that can trigger the store-store
16206 errata for certain LEON3FT processors.
16207 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
16208 (sparc_option_override): Set sparc_fix_b2bst appropriately.
16209 * config/sparc/sparc.md (fix_b2bst): New attribute.
16210 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
16212 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
16215 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
16217 (*rsqrtsf2_sse): Ditto.
16219 (div<mode>3): Macroize insn from divdf3 and divsf3
16220 using MODEF mode iterator.
16222 2017-07-10 Martin Sebor <msebor@redhat.com>
16224 PR tree-optimization/80397
16225 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
16226 instead of testing for equality to INTEGER_TYPE.
16228 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
16230 * config.gcc: Remove uclibc from arc target spec.
16232 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
16234 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
16236 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16239 * lto-wrapper.c (remove_option): New function.
16240 (merge_and_complain): Merge PIC/PIE options more realistically.
16242 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16244 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
16248 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
16249 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
16250 * config.in: Regenerate.
16251 * configure: Regenerate.
16252 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
16253 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
16254 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
16255 (TARGET_GASISR_PROLOGUES): ...target mask.
16256 * common/config/avr/avr-common.c
16257 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
16258 Set -mgas-isr-prologues.
16259 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
16260 INSERT_PASS_BEFORE for it.
16261 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
16262 * config/avr/avr.c (avr_option_override)
16263 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
16264 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
16265 (avr_attribute_table) <no_gccisr>: Add new function attribute.
16266 (avr_set_current_function) <is_no_gccisr>: Init machine field.
16267 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
16269 (make_avr_pass_pre_proep): New function.
16270 (emit_push_sfr) <treg>: Add argument to function and use it
16271 instead of TMP_REG.
16272 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
16273 and set machine->gasisr.yes.
16274 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
16275 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
16276 __gcc_isr.n_pushed to .L__stack_usage.
16277 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
16278 (avr_asm_final_postscan_insn): ...this new static function.
16279 * config/avr/avr.h (machine_function)
16280 <is_no_gccisr, use_L__stack_usage>: New fields.
16281 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
16282 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
16283 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
16284 (gasisr, *gasisr): New expander and insn.
16285 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
16286 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
16287 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
16289 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
16291 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
16294 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16296 Move jump-tables out of .text again.
16299 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
16300 (ASM_OUTPUT_ADDR_VEC): New function.
16301 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
16302 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
16303 INSN_ADDRESSes as asm comment.
16304 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
16305 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
16306 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
16307 * config/avr/avr.md (*tablejump): Adjust comment.
16308 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
16309 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
16311 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
16312 (avr_output_addr_vec): New proto.
16313 (avr_log_t) <insn_addresses>: New field.
16315 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
16318 * config/i386/i386.c (ix86_function_arg_advance): Set
16319 outgoing_args_on_stack to true if there are outgoing arguments
16321 (ix86_function_arg): Likewise.
16322 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
16323 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
16324 * config/i386/i386.h (machine_function): Add
16325 outgoing_args_on_stack.
16327 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
16329 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
16330 supporting pthreds.
16331 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
16333 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
16335 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
16336 (REAL_H): Remove $(MACHMODE_H).
16337 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
16339 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
16340 $(MACHMODE_H) and double-int.h.
16341 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
16343 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
16346 2017-07-07 Andrew Pinski <apinski@cavium.com>
16348 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
16349 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
16351 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16353 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
16354 Add warning if GCC was not configured to link against a GLIBC that
16355 exports the hardware capability bits.
16356 (make_resolver_func): Make resolver function private and not a
16357 COMDAT function. Create the name with clone_function_name instead
16358 of make_unique_name.
16361 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
16362 correct operand in doing the split.
16364 2017-07-07 Carl Love <cel@us.ibm.com>
16366 * config/rs6000/rs6000-c: Add support for built-in function
16367 vector unsigned short vec_pack_to_short_fp32 (vector float,
16369 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
16370 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
16371 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
16372 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
16373 (convert_4f32_8i16): Add define_expand.
16374 * doc/extend.texi: Update the built-in documentation file for the
16375 new built-in function.
16377 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16379 * config/sparc/m8.md: New file.
16380 * config/sparc/sparc.md: Include m8.md.
16382 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16384 * config/sparc/sparc.opt: New option -mvis4b.
16385 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
16386 (sparc_option_override): Handle VIS4B.
16387 (enum sparc_builtins): Define
16388 SPARC_BUILTIN_DICTUNPACK{8,16,32},
16389 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
16390 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
16391 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
16392 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
16393 (check_constant_argument): New function.
16394 (sparc_vis_init_builtins): Define builtins
16395 __builtin_vis_dictunpack{8,16,32},
16396 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
16397 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
16398 __builtin_vis_fpcmpde{8,16,32}shl and
16399 __builtin_vis_fpcmpur{8,16,32}shl.
16400 (sparc_expand_builtin): Check that the constant operands to
16401 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
16402 constant and in range.
16403 * config/sparc/sparc-c.c (sparc_target_macros): Handle
16405 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
16406 (SPARC_IMM5_P): Likewise.
16407 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
16408 (enabled): Handle vis4b.
16409 (UNSPEC_DICTUNPACK): New unspec.
16410 (UNSPEC_FPCMPSHL): Likewise.
16411 (UNSPEC_FPUCMPSHL): Likewise.
16412 (UNSPEC_FPCMPDESHL): Likewise.
16413 (UNSPEC_FPCMPURSHL): Likewise.
16414 (cpu_feature): New CPU feature `vis4b'.
16415 (dictunpack{8,16,32}): New insns.
16416 (FPCSMODE): New mode iterator.
16417 (fpcscond): New code iterator.
16418 (fpcsucond): Likewise.
16419 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
16420 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
16421 (fpcmpde{8,16,32}{si,di}shl): Likewise.
16422 (fpcmpur{8,16,32}{si,di}shl): Likewise.
16423 * config/sparc/constraints.md: Define constraints `q' for unsigned
16424 2-bit integer constants and `t' for unsigned 5-bit integer
16426 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
16428 (imm5_operand_dictunpack16): Likewise.
16429 (imm5_operand_dictunpack32): Likewise.
16430 (imm2_operand): Likewise.
16431 * doc/invoke.texi (SPARC Options): Document -mvis4b.
16432 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
16433 ditunpack* and fpcmp*shl builtins.
16435 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16437 * config.gcc: Handle m8 in --with-{cpu,tune} options.
16438 * config.in: Add HAVE_AS_SPARC6 define.
16439 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
16441 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
16443 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
16444 (CPP_CPU_SPEC): Handle m8.
16445 (ASM_CPU_SPEC): Likewise.
16446 * config/sparc/sparc-opts.h (enum processor_type): Add
16448 * config/sparc/sparc.c (m8_costs): New struct.
16449 (sparc_option_override): Handle TARGET_CPU_m8.
16450 (sparc32_initialize_trampoline): Likewise.
16451 (sparc64_initialize_trampoline): Likewise.
16452 (sparc_issue_rate): Likewise.
16453 (sparc_register_move_cost): Likewise.
16454 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
16455 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
16456 (ASM_CPU64_DEFAULT_SPEC): Likewise.
16457 (CPP_CPU_SPEC): Handle M8.
16458 (ASM_CPU_SPEC): Likewise.
16459 (AS_M8_FLAG): Define.
16460 * config/sparc/sparc.md: Add m8 to the cpu attribute.
16461 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
16462 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
16464 * configure: Regenerate.
16465 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
16468 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16470 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
16472 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
16473 ("*movdi_insn_sp32"): Do not set v3pipe.
16474 ("*movsi_insn"): Likewise.
16475 ("*movdi_insn_sp64"): Likewise.
16476 ("*movsf_insn"): Likewise.
16477 ("*movdf_insn_sp32"): Likewise.
16478 ("*movdf_insn_sp64"): Likewise.
16479 ("*zero_extendsidi2_insn_sp64"): Likewise.
16480 ("*sign_extendsidi2_insn"): Likewise.
16481 ("*mov<VM32:mode>_insn"): Likewise.
16482 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16483 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16484 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16485 ("<vlop:code><VL:mode>3"): Likewise.
16486 ("*not_<vlop:code><VL:mode>3"): Likewise.
16487 ("*nand<VL:mode>_vis"): Likewise.
16488 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
16489 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
16490 ("one_cmpl<VL:mode>2"): Likewise.
16491 ("faligndata<VM64:mode>_vis"): Likewise.
16492 ("alignaddrsi_vis"): Likewise.
16493 ("alignaddrdi_vis"): Likweise.
16494 ("alignaddrlsi_vis"): Likewise.
16495 ("alignaddrldi_vis"): Likewise.
16496 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16497 ("bmaskdi_vis"): Likewise.
16498 ("bmasksi_vis"): Likewise.
16499 ("bshuffle<VM64:mode>_vis"): Likewise.
16500 ("cmask8<P:mode>_vis"): Likewise.
16501 ("cmask16<P:mode>_vis"): Likewise.
16502 ("cmask32<P:mode>_vis"): Likewise.
16503 ("pdistn<P:mode>_vis"): Likewise.
16504 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16506 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16508 * config/sparc/sparc.md ("subtype"): New insn attribute.
16509 ("*wrgsr_sp64"): Set insn subtype.
16510 ("*rdgsr_sp64"): Likewise.
16511 ("alignaddrsi_vis"): Likewise.
16512 ("alignaddrdi_vis"): Likewise.
16513 ("alignaddrlsi_vis"): Likewise.
16514 ("alignaddrldi_vis"): Likewise.
16515 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16516 ("fexpand_vis"): Likewise.
16517 ("fpmerge_vis"): Likewise.
16518 ("faligndata<VM64:mode>_vis"): Likewise.
16519 ("bshuffle<VM64:mode>_vis"): Likewise.
16520 ("cmask8<P:mode>_vis"): Likewise.
16521 ("cmask16<P:mode>_vis"): Likewise.
16522 ("cmask32<P:mode>_vis"): Likewise.
16523 ("fchksm16_vis"): Likewise.
16524 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
16525 ("fmean16_vis"): Likewise.
16526 ("fp<plusminus_insn>64_vis"): Likewise.
16527 ("<plusminus_insn>v8qi3"): Likewise.
16528 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16529 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
16530 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
16531 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
16532 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
16533 ("*movqi_insn"): Likewise.
16534 ("*movhi_insn"): Likewise.
16535 ("*movsi_insn"): Likewise.
16536 ("movsi_pic_gotdata_op"): Likewise.
16537 ("*movdi_insn_sp32"): Likewise.
16538 ("*movdi_insn_sp64"): Likewise.
16539 ("movdi_pic_gotdata_op"): Likewise.
16540 ("*movsf_insn"): Likewise.
16541 ("*movdf_insn_sp32"): Likewise.
16542 ("*movdf_insn_sp64"): Likewise.
16543 ("*zero_extendhisi2_insn"): Likewise.
16544 ("*zero_extendqihi2_insn"): Likewise.
16545 ("*zero_extendqisi2_insn"): Likewise.
16546 ("*zero_extendqidi2_insn"): Likewise.
16547 ("*zero_extendhidi2_insn"): Likewise.
16548 ("*zero_extendsidi2_insn_sp64"): Likewise.
16549 ("ldfsr"): Likewise.
16550 ("prefetch_64"): Likewise.
16551 ("prefetch_32"): Likewise.
16552 ("tie_ld32"): Likewise.
16553 ("tie_ld64"): Likewise.
16554 ("*tldo_ldub_sp32"): Likewise.
16555 ("*tldo_ldub1_sp32"): Likewise.
16556 ("*tldo_ldub2_sp32"): Likewise.
16557 ("*tldo_ldub_sp64"): Likewise.
16558 ("*tldo_ldub1_sp64"): Likewise.
16559 ("*tldo_ldub2_sp64"): Likewise.
16560 ("*tldo_ldub3_sp64"): Likewise.
16561 ("*tldo_lduh_sp32"): Likewise.
16562 ("*tldo_lduh1_sp32"): Likewise.
16563 ("*tldo_lduh_sp64"): Likewise.
16564 ("*tldo_lduh1_sp64"): Likewise.
16565 ("*tldo_lduh2_sp64"): Likewise.
16566 ("*tldo_lduw_sp32"): Likewise.
16567 ("*tldo_lduw_sp64"): Likewise.
16568 ("*tldo_lduw1_sp64"): Likewise.
16569 ("*tldo_ldx_sp64"): Likewise.
16570 ("*mov<VM32:mode>_insn"): Likewise.
16571 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16572 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16574 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16576 * config/sparc/sparc.md ("type"): New insn type viscmp.
16577 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
16579 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
16580 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
16581 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16582 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
16584 ("n7_vis_logical_11cycle"): Likewise.
16585 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
16586 * config/sparc/niagara2.md ("niag3_vis": Likewise.
16587 * config/sparc/niagara.md ("niag_vis"): Likewise.
16588 * config/sparc/ultra3.md ("us3_fga"): Likewise.
16589 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
16591 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16593 * config/sparc/sparc.md: New instruction type `bmask'.
16594 (bmaskdi_vis): Use the `bmask' type.
16595 (bmasksi_vis): Likewise.
16596 * config/sparc/ultra3.md (us3_array): Likewise.
16597 * config/sparc/niagara7.md (n7_array): Likewise.
16598 * config/sparc/niagara4.md (n4_array): Likewise.
16599 * config/sparc/niagara2.md (niag2_vis): Likewise.
16600 (niag3_vis): Likewise.
16601 * config/sparc/niagara.md (niag_vis): Likewise.
16603 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16605 * ipa-comdats.c: Remove optimize check from gate.
16606 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
16607 for functions not optimized.
16608 (ipa_fn_summary_read): Skip optimize check.
16609 (ipa_fn_summary_write): Likewise.
16610 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
16612 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
16614 (can_inline_edge_p): Check flag_pcc_struct_return for match.
16615 (check_callers): Give up on caller which is not optimized.
16616 (inline_small_functions): Likewise.
16617 (ipa_inline): Do not give up when not optimizing.
16618 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
16619 away unoptimizes cdtors.
16620 (whole_program_function_and_variable_visibility): Do
16621 ipa_discover_readonly_nonaddressable_vars in LTO mode.
16622 * ipa.c (process_references): Do not check optimize.
16623 (symbol_table::remove_unreachable_nodes): Update optimize check.
16624 (set_writeonly_bit): Update optimize check.
16625 (pass_ipa_cdtor_merge::gate): Do not check optimize.
16626 (pass_ipa_single_use::gate): Remove.
16628 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16630 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
16631 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
16632 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
16633 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
16634 permute_load, permute_store, adjust_extract, adjust_splat,
16635 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
16636 replace_swap_with_copy, dump_swap_insn_table,
16637 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
16638 recombine_lvx_pattern, recombine_stvx_pattern,
16639 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
16640 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
16641 to file rs6000-p8swap.c.
16642 * config/rs6000/rs6000-p8swap.c: New file.
16643 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
16644 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
16645 and rs6000*-*-* targets.
16647 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16649 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
16651 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16653 * lto-wrapper.c (merge_and_complain): Do not merge
16654 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
16655 fsigned_zeros, ftrapping_math, fwrapv.
16656 (append_compiler_options): Do not track these options.
16657 (append_linker_options): Likewie
16659 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16661 * cgraphunit.c (cgraph_node::finalize_function): When
16662 !flag_toplevel_reorde set no_reorder flag.
16663 (varpool_node::finalize_decl): Likewise.
16664 (symbol_table::compile): Drop no toplevel reorder path.
16666 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16668 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
16669 edges; zero probability is not better than uninitialized.
16671 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
16673 * asan.h (asan_sanitize_allocas_p): Declare.
16674 * asan.c (asan_sanitize_allocas_p): New function.
16675 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
16676 (handle_builtin_alloca): Likewise.
16677 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
16678 if !asan_sanitize_allocas_p.
16679 * params.def (asan-instrument-allocas): Add new option.
16680 * params.h (ASAN_PROTECT_ALLOCAS): Define.
16681 * opts.c (common_handle_option): Disable allocas sanitization for
16684 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
16686 * asan.c: Include gimple-fold.h.
16687 (get_last_alloca_addr): New function.
16688 (handle_builtin_stackrestore): Likewise.
16689 (handle_builtin_alloca): Likewise.
16690 (asan_emit_allocas_unpoison): Likewise.
16691 (get_mem_refs_of_builtin_call): Add new parameter, remove const
16692 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
16693 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
16694 (instrument_builtin_call): Pass gimple iterator to
16695 get_mem_refs_of_builtin_call.
16696 (last_alloca_addr): New global.
16697 * asan.h (asan_emit_allocas_unpoison): Declare.
16698 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
16699 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
16700 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
16701 if function calls alloca.
16702 * gimple-fold.c (replace_call_with_value): Remove static keyword.
16703 * gimple-fold.h (replace_call_with_value): Declare.
16704 * internal-fn.c: Include asan.h.
16705 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
16706 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
16708 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16710 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
16711 (C_SELFTEST_FLAGS): New.
16712 (CPP_SELFTEST_FLAGS): New.
16713 (SELFTEST_DEPS): New, from deps of s-selftest.
16714 (C_SELFTEST_DEPS): New, from deps of s-selftest.
16715 (CPP_SELFTEST_DEPS): New.
16716 (selftest): Add dependency on s-selftest-c++.
16717 (s-selftest): Rename to...
16718 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
16719 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
16720 than SELFTEST_FLAGS.
16721 (selftest-gdb): Rename to...
16722 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
16724 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
16725 (selftest-valgrind): Rename to...
16726 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
16728 (selftest-valgrind): Reintroduce as an alias for
16729 selftest-c-valgrind.
16730 (s-selftest-c++): New.
16731 (selftest-c++-gdb): New.
16732 (selftest-c++-valgrind): New.
16734 2017-07-06 Olivier Hainque <hainque@adacore.com>
16736 * gcc.c (process_command): When deciding if undefined variables
16737 should be ignored when processing specs, accept "gcc -v" as well.
16739 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16741 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
16742 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
16744 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
16746 * config/arm/arm-cpus.in (armv8-r): Add new entry.
16747 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
16748 * config/arm/arm-tables.opt: Regenerate.
16749 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
16751 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
16753 2017-07-06 Carl Love <cel@us.ibm.com>
16755 * ChangeLog: Clean up from mid air collision
16757 2017-07-06 Carl Love <cel@us.ibm.com>
16759 * config/rs6000/rs6000-c.c: Add support for built-in functions
16760 vector signed int vec_subc (vector signed int, vector signed int);
16761 vector signed __int128 vec_subc (vector signed __int128,
16762 vector signed __int128);
16763 vector unsigned __int128 vec_subc (vector unsigned __int128,
16764 vector unsigned __int128);
16765 vector signed int vec_sube (vector signed int, vector signed int,
16766 vector signed int);
16767 vector unsigned int vec_sube (vector unsigned int,
16768 vector unsigned int,
16769 vector unsigned int);
16770 vector signed __int128 vec_sube (vector signed __int128,
16771 vector signed __int128,
16772 vector signed__int128);
16773 vector unsigned __int128 vec_sube (vector unsigned __int128,
16774 vector unsigned __int128,
16775 vector unsigned __int128);
16776 vector signed int vec_subec (vector signed int, vector signed int,
16777 vector signed int);
16778 vector unsigned int vec_subec (vector unsigned int,
16779 vector unsigned int,
16780 vector unsigned int);
16781 vector signed __int128 vec_subec (vector signed __int128,
16782 vector signed __int128,
16783 vector signed__int128);
16784 vector unsigned __int128 vec_subec (vector unsigned __int128,
16785 vector unsigned __int128,
16786 vector unsigned __int128);
16787 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
16788 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
16789 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
16790 BU_ALTIVEC_OVERLOAD_X definitions.
16791 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
16792 * doc/extend.texi: Update the built-in documentation file for the new
16793 built-in functions.
16795 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16798 * diagnostic-show-locus.c (layout::layout): Use start and finish
16799 spelling location for the start and finish of each range.
16800 * genmatch.c (linemap_client_expand_location_to_spelling_point):
16801 Add unused aspect param.
16802 * input.c (expand_location_1): Add "aspect" param, and use it
16803 to access the correct part of the location.
16804 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
16806 (expand_location_to_spelling_point): Likewise.
16807 (linemap_client_expand_location_to_spelling_point): Add "aspect"
16808 param, and pass it to expand_location_1.
16810 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
16812 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
16813 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
16814 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
16815 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
16816 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
16817 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
16818 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
16819 _mm_maskz_getmant_ss): New intrinsics.
16820 (__builtin_ia32_getexpss128_mask): Changed to ...
16821 __builtin_ia32_getexpss128_round ... this.
16822 (__builtin_ia32_getexpsd128_mask): Changed to ...
16823 __builtin_ia32_getexpsd128_round ... this.
16824 * config/i386/i386-builtin-types.def
16825 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
16826 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
16827 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
16828 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
16829 __builtin_ia32_getmantss_mask_round): New builtins.
16830 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
16831 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
16832 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
16833 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
16834 * config/i386/sse.md
16835 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
16836 avx512f_sgetexp<mode><mask_scalar_name>
16837 <round_saeonly_scalar_name> ... this.
16838 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
16839 %0, %1, %2<round_saeonly_op3>}): Changed to ...
16840 vgetexp<ssescalarmodesuffix>
16841 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16842 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
16843 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
16844 avx512f_vgetmant<mode><mask_scalar_name>
16845 <round_saeonly_scalar_name> ... this.
16846 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
16847 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
16848 vgetmant<ssescalarmodesuffix>
16849 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
16850 %0<mask_scalar_operand4>, %1, %2
16851 <round_saeonly_scalar_mask_op4>, %3} ... this.
16852 * config/i386/subst.md (mask_scalar_operand4,
16853 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
16854 round_saeonly_scalar_nimm_predicate): New subst attributes.
16856 2017-07-06 Julia Koval <julia.koval@intel.com>
16858 * config/i386/i386.c (ix86_erase_embedded_rounding):
16859 Remove code for old rounding pattern.
16861 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
16863 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
16865 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
16867 * doc/sourcebuild.texi (Test Directives, Variants of
16868 dg-require-support): Add documentation for dg-require-stack-check.
16870 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
16872 * config/i386/subst.md (mask_scalar, round_scalar,
16873 round_saeonly_scalar): New meta-templates.
16874 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
16875 round_scalar_mask_operand3, round_scalar_mask_op3,
16876 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
16877 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
16878 round_saeonly_scalar_constraint,
16879 round_saeonly_scalar_prefix): New subst attribute.
16880 * config/i386/sse.md
16881 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
16882 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
16883 <round_scalar_name> ... this.
16884 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
16885 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
16886 <round_scalar_name> ... this.
16887 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
16888 <sse>_vm<code><mode>3<mask_scalar_name>
16889 <round_saeonly_scalar_name> ... this.
16890 (v<plusminus_mnemonic><ssescalarmodesuffix>
16891 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
16892 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
16893 v<plusminus_mnemonic><ssescalarmodesuffix>
16894 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16895 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
16896 (v<multdiv_mnemonic><ssescalarmodesuffix>
16897 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
16898 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
16899 v<multdiv_mnemonic><ssescalarmodesuffix>
16900 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16901 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
16902 (v<maxmin_float><ssescalarmodesuffix>
16903 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
16904 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
16905 v<maxmin_float><ssescalarmodesuffix>
16906 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16907 %0<mask_scalar_operand3>, %1, %<iptr>2
16908 <round_saeonly_scalar_mask_op3>} ... this.
16910 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
16912 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
16913 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
16915 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
16916 Alan Hayward <alan.hayward@arm.com>
16917 David Sherwood <david.sherwood@arm.com>
16919 * combine.c (simplify_if_then_else): Remove "enum" before
16921 * compare-elim.c (can_eliminate_compare): Likewise.
16922 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
16924 (aarch64_lookup_simd_builtin_type): Likewise.
16925 (aarch64_simd_builtin_type): Likewise.
16926 (aarch64_init_simd_builtin_types): Likewise.
16927 (aarch64_simd_expand_args): Likewise.
16928 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
16930 (aarch64_reverse_mask): Likewise.
16931 (aarch64_simd_emit_reg_reg_move): Likewise.
16932 (aarch64_gen_adjusted_ldpstp): Likewise.
16933 (aarch64_ccmp_mode_to_code): Likewise.
16934 (aarch64_operands_ok_for_ldpstp): Likewise.
16935 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16936 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
16938 (aarch64_min_divisions_for_recip_mul): Likewise.
16939 (aarch64_reassociation_width): Likewise.
16940 (aarch64_get_condition_code_1): Likewise.
16941 (aarch64_simd_emit_reg_reg_move): Likewise.
16942 (aarch64_simd_attr_length_rglist): Likewise.
16943 (aarch64_reverse_mask): Likewise.
16944 (aarch64_operands_ok_for_ldpstp): Likewise.
16945 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16946 (aarch64_gen_adjusted_ldpstp): Likewise.
16947 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
16949 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
16950 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
16951 (arm_lookup_simd_builtin_type): Likewise.
16952 (arm_simd_builtin_type): Likewise.
16953 (arm_init_simd_builtin_types): Likewise.
16954 (arm_expand_builtin_args): Likewise.
16955 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
16956 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
16957 (ft32_setup_incoming_varargs): Likewise.
16958 (ft32_function_arg): Likewise.
16959 (ft32_function_arg_advance): Likewise.
16960 (ft32_pass_by_reference): Likewise.
16961 (ft32_arg_partial_bytes): Likewise.
16962 (ft32_valid_pointer_mode): Likewise.
16963 (ft32_addr_space_pointer_mode): Likewise.
16964 (ft32_addr_space_legitimate_address_p): Likewise.
16965 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
16967 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
16968 (ix86_emit_outlined_ms2sysv_restore): Likewise.
16969 (iamcu_alignment): Likewise.
16970 (canonicalize_vector_int_perm): Likewise.
16971 (ix86_noce_conversion_profitable_p): Likewise.
16972 (ix86_mpx_bound_mode): Likewise.
16973 (ix86_operands_ok_for_move_multiple): Likewise.
16974 * config/microblaze/microblaze-protos.h
16975 (microblaze_expand_conditional_branch_reg): Likewise.
16976 * config/microblaze/microblaze.c
16977 (microblaze_expand_conditional_branch_reg): Likewise.
16978 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
16980 (rs6000_reassociation_width): Likewise.
16981 (rs6000_invalid_binary_op): Likewise.
16982 (fusion_p9_p): Likewise.
16983 (emit_fusion_p9_load): Likewise.
16984 (emit_fusion_p9_store): Likewise.
16985 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
16987 (riscv_hard_regno_mode_ok_p): Likewise.
16988 (riscv_address_insns): Likewise.
16989 (riscv_split_symbol): Likewise.
16990 (riscv_legitimize_move): Likewise.
16991 (riscv_function_value): Likewise.
16992 (riscv_hard_regno_nregs): Likewise.
16993 (riscv_expand_builtin): Likewise.
16994 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
16995 (riscv_build_integer): Likewise.
16996 (riscv_split_integer): Likewise.
16997 (riscv_legitimate_constant_p): Likewise.
16998 (riscv_cannot_force_const_mem): Likewise.
16999 (riscv_regno_mode_ok_for_base_p): Likewise.
17000 (riscv_valid_base_register_p): Likewise.
17001 (riscv_valid_offset_p): Likewise.
17002 (riscv_valid_lo_sum_p): Likewise.
17003 (riscv_classify_address): Likewise.
17004 (riscv_legitimate_address_p): Likewise.
17005 (riscv_address_insns): Likewise.
17006 (riscv_load_store_insns): Likewise.
17007 (riscv_force_binary): Likewise.
17008 (riscv_split_symbol): Likewise.
17009 (riscv_force_address): Likewise.
17010 (riscv_legitimize_address): Likewise.
17011 (riscv_move_integer): Likewise.
17012 (riscv_legitimize_const_move): Likewise.
17013 (riscv_legitimize_move): Likewise.
17014 (riscv_address_cost): Likewise.
17015 (riscv_subword): Likewise.
17016 (riscv_output_move): Likewise.
17017 (riscv_canonicalize_int_order_test): Likewise.
17018 (riscv_emit_int_order_test): Likewise.
17019 (riscv_function_arg_boundary): Likewise.
17020 (riscv_pass_mode_in_fpr_p): Likewise.
17021 (riscv_pass_fpr_single): Likewise.
17022 (riscv_pass_fpr_pair): Likewise.
17023 (riscv_get_arg_info): Likewise.
17024 (riscv_function_arg): Likewise.
17025 (riscv_function_arg_advance): Likewise.
17026 (riscv_arg_partial_bytes): Likewise.
17027 (riscv_function_value): Likewise.
17028 (riscv_pass_by_reference): Likewise.
17029 (riscv_setup_incoming_varargs): Likewise.
17030 (riscv_print_operand): Likewise.
17031 (riscv_elf_select_rtx_section): Likewise.
17032 (riscv_save_restore_reg): Likewise.
17033 (riscv_for_each_saved_reg): Likewise.
17034 (riscv_register_move_cost): Likewise.
17035 (riscv_hard_regno_mode_ok_p): Likewise.
17036 (riscv_hard_regno_nregs): Likewise.
17037 (riscv_class_max_nregs): Likewise.
17038 (riscv_memory_move_cost): Likewise.
17039 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
17040 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
17041 (rl78_addr_space_address_mode): Likewise.
17042 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17044 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17045 (rs6000_reassociation_width): Likewise.
17046 (rs6000_invalid_binary_op): Likewise.
17047 (fusion_p9_p): Likewise.
17048 (emit_fusion_p9_load): Likewise.
17049 (emit_fusion_p9_store): Likewise.
17050 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
17051 (ok_for_simple_move_operands): Likewise.
17052 (ok_for_simple_move_strict_operands): Likewise.
17053 (ok_for_simple_arith_logic_operands): Likewise.
17054 (visium_legitimize_reload_address): Likewise.
17055 (visium_select_cc_mode): Likewise.
17056 (output_cbranch): Likewise.
17057 (visium_split_double_move): Likewise.
17058 (visium_expand_copysign): Likewise.
17059 (visium_expand_int_cstore): Likewise.
17060 (visium_expand_fp_cstore): Likewise.
17061 * config/visium/visium.c (visium_pass_by_reference): Likewise.
17062 (visium_function_arg): Likewise.
17063 (visium_function_arg_advance): Likewise.
17064 (visium_libcall_value): Likewise.
17065 (visium_setup_incoming_varargs): Likewise.
17066 (visium_legitimate_constant_p): Likewise.
17067 (visium_legitimate_address_p): Likewise.
17068 (visium_legitimize_address): Likewise.
17069 (visium_secondary_reload): Likewise.
17070 (visium_register_move_cost): Likewise.
17071 (visium_memory_move_cost): Likewise.
17072 (prepare_move_operands): Likewise.
17073 (ok_for_simple_move_operands): Likewise.
17074 (ok_for_simple_move_strict_operands): Likewise.
17075 (ok_for_simple_arith_logic_operands): Likewise.
17076 (visium_function_value_1): Likewise.
17077 (rtx_ok_for_offset_p): Likewise.
17078 (visium_legitimize_reload_address): Likewise.
17079 (visium_split_double_move): Likewise.
17080 (visium_expand_copysign): Likewise.
17081 (visium_expand_int_cstore): Likewise.
17082 (visium_expand_fp_cstore): Likewise.
17083 (visium_split_cstore): Likewise.
17084 (visium_select_cc_mode): Likewise.
17085 (visium_split_cbranch): Likewise.
17086 (output_cbranch): Likewise.
17087 (visium_print_operand_address): Likewise.
17088 * expmed.c (flip_storage_order): Likewise.
17089 * expmed.h (emit_cstore): Likewise.
17090 (flip_storage_order): Likewise.
17091 * genrecog.c (validate_pattern): Likewise.
17092 * hsa-gen.c (gen_hsa_addr): Likewise.
17093 * internal-fn.c (expand_arith_overflow): Likewise.
17094 * ira-color.c (allocno_copy_cost_saving): Likewise.
17095 * lra-assigns.c (find_hard_regno_for_1): Likewise.
17096 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
17097 (process_invariant_for_inheritance): Likewise.
17098 * lra-eliminations.c (move_plus_up): Likewise.
17099 * omp-low.c (lower_oacc_reductions): Likewise.
17100 * simplify-rtx.c (simplify_subreg): Likewise.
17101 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
17102 (TARGET_CHKP_BOUND_MODE): Likewise..
17103 * targhooks.c (default_chkp_bound_mode): Likewise.
17104 (default_setup_incoming_vararg_bounds): Likewise.
17105 * targhooks.h (default_chkp_bound_mode): Likewise.
17106 (default_setup_incoming_vararg_bounds): Likewise.
17107 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
17108 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
17109 (have_whole_vector_shift): Likewise.
17110 * tree-vect-stmts.c (vectorizable_load): Likewise.
17111 * doc/tm.texi: Regenerate.
17113 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17115 Graceful degrade if Binutils PR21472 is not available.
17118 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
17119 .rodata in flash test fails.
17120 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
17121 * confgure: Regenerate.
17122 * config.in: Regenerate.
17123 * config/avr/avr.c (avr_asm_named_section)
17124 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
17125 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
17126 (avr_asm_init_sections): Same.
17128 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17130 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
17131 (fma<VH:mode>4_intrinsic): Likewise.
17132 (*fmsub<VCVTF:mode>4): Likewise.
17133 (*fmsub<VH:mode>4_intrinsic): Likewise.
17135 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17138 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
17139 Don't depend on "optimize > 0".
17140 (out_movhi_r_mr, out_movqi_mr_r): Same.
17141 (out_movhi_mr_r, out_movqi_r_mr): Same.
17142 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
17143 io_address_operand on "optimize > 0".
17145 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17147 * tree-loop-distribution.c: Add general explanantion on the pass.
17148 (generate_loops_for_partition): Mark distributed loop.
17149 (pg_add_dependence_edges): New parameter. Handle alias data
17150 dependence specially and record it in the parameter if asked.
17151 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
17152 (init_partition_graph_vertices, add_partition_graph_edge): New.
17153 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
17154 (free_partition_graph_vdata, build_partition_graph): New.
17155 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
17156 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
17157 (data_ref_segment_size, latch_dominated_by_data_ref): New.
17158 (compute_alias_check_pairs, version_loop_by_alias_check): New.
17159 (version_for_distribution_p, finalize_partitions): New.
17160 (distribute_loop): Handle alias data dependence specially. Factor
17161 out loop fusion code as functions and call these functions.
17163 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17165 * tree-loop-distribution.c (classify_partition): New parameter and
17166 better handle reduction statement.
17167 (rdg_build_partitions): Revise comment.
17168 (distribute_loop): Compute statements in all partitions and pass it
17169 to classify_partition.
17171 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17173 * tree-loop-distribution.c (enum partition_type): New.
17174 (struct partition): New field type.
17175 (partition_merge_into): Add parameter. Update partition type.
17176 (data_dep_in_cycle_p, update_type_for_merge): New functions.
17177 (build_rdg_partition_for_vertex): Compute partition type.
17178 (rdg_build_partitions): Dump partition type.
17179 (distribute_loop): Update calls to partition_merge_into.
17181 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17183 * tree-loop-distribution.c (struct ddr_hasher): New.
17184 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
17186 (classify_partition): Call get_data_dependence.
17187 (pg_add_dependence_edges): Ditto.
17188 (distribute_loop): Release data dependence hash table.
17190 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17192 * tree-loop-distribution.c (ref_base_address): Delete.
17193 (similar_memory_accesses): Rename ...
17194 (share_memory_accesses): ... to this. Check if partitions access
17195 the same memory reference.
17196 (distribute_loop): Call share_memory_accesses.
17198 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17200 * tree-loop-distribution.c (struct partition): New field recording
17201 its data reference.
17202 (partition_alloc, partition_free): Init and release data refs.
17203 (partition_merge_into): Merge data refs.
17204 (build_rdg_partition_for_vertex): Collect data refs for partition.
17205 (pg_add_dependence_edges): Change parameters from vector to bitmap.
17207 (distribute_loop): Remve data refs from vertice data of partition
17210 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17212 * tree-loop-distribution.c (params.h): Include header file.
17213 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
17214 (datarefs_vec): New global var.
17215 (create_rdg_vertices): Use datarefs_vec directly.
17216 (free_rdg): Don't free data references.
17217 (build_rdg): Update use. Don't free data references.
17218 (distribute_loop): Compute global variable for data references.
17219 Bail out if there are too many data references.
17221 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17223 * tree-loop-distribution.c (loop_nest): New global var.
17224 (build_rdg): Use loop directly, rather than loop nest.
17225 (pg_add_dependence_edges): Remove loop nest parameter. Use global
17227 (distribute_loop): Compute global variable loop nest. Update use.
17229 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17231 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
17232 (partition_merge_into): New parameter. Dump reason for fusion.
17233 (distribute_loop): Update use of partition_merge_into.
17235 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17237 * tree-loop-distribution.c (bb_top_order_index): New.
17238 (bb_top_order_index_size, bb_top_order_cmp): New.
17239 (stmts_from_loop): Use topological order.
17240 (pass_loop_distribution::execute): Compute and release topological
17241 order for basic blocks.
17243 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17245 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
17248 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17250 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
17251 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
17252 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
17253 * internal-fn.def (LOOP_DIST_ALIAS): New.
17254 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
17255 (fold_loop_internal_call): ... this.
17256 (vect_loop_dist_alias_call): New function.
17257 (set_uid_loop_bbs): Call fold_loop_internal_call.
17258 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
17261 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
17264 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
17265 Require dead FLAGS_REG at the beginning of a peephole.
17267 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
17270 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
17271 arguments in the call to __builtin_ia32_sbb_u32.
17272 (_subborrow_u64): Swap _X and _Y arguments in the call to
17273 __builtin_ia32_sbb_u64.
17275 2017-07-04 Jakub Jelinek <jakub@redhat.com>
17278 * tree-vrp.c (compare_assert_loc): Turn into a function template
17279 with stable template parameter. Only test if a->e is NULL,
17280 !a->e == !b->e has been verified already. Use e == NULL or
17281 e != NULL instead of e or ! e tests. If stable is true, don't use
17282 iterative_hash_expr, on the other side allow a or b or both NULL
17283 and sort the NULLs last.
17284 (process_assert_insertions): Sort using compare_assert_loc<false>
17285 instead of compare_assert_loc, later sort using
17286 compare_assert_loc<true> before calling process_assert_insertions_for
17287 in a loop. Use break instead of continue once seen NULL pointer.
17289 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
17291 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17292 Cortex-R7 and Cortex-R8 processors.
17294 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17296 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
17297 uninitialized while src is not.
17299 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
17301 * common/config/arm/arm-common.c: Adjust include path for
17303 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
17304 (arm-cpu.h): Create in build directory. Adjust dependency rules.
17305 (arm-cpu-data.h): Likewise.
17306 (arm-cpu-cdata.h): Likewise.
17307 * config/arm/arm-cpu.h: Delete.
17308 * config/arm/arm-cpu-cdata.h: Delete.
17309 * config/arm/arm-cpu-data.h: Delete.
17311 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
17313 * config/arm/arm-cpus.in (cortex-a55): New.
17314 (cortex-a75): Likewise.
17315 (cortex-a75.cortex-a55): Likewise.
17316 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
17318 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
17319 * config/arm/arm-cpu-cdata.h: Regenerate.
17320 * config/arm/arm-cpu-data.h: Regenerate.
17321 * config/arm/arm-cpu.h: Regenerate.
17322 * config/arm/arm-tables.opt: Regenerate.
17323 * config/arm/arm-tune.md: Regenerate.
17325 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17327 * haifa-sched.c (sched_create_recovery_edges): Update profile.
17329 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17331 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
17334 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
17336 PR tree-optimization/81292
17337 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
17338 full_string_p, also call adjust_related_strinfos if the adjustment
17339 is simple, otherwise invalidate related strinfos.
17341 2017-07-04 Martin Liska <mliska@suse.cz>
17344 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
17345 newly created variable as DECL_IGNORED_P.
17347 2017-07-04 Martin Liska <mliska@suse.cz>
17350 * ipa-inline.c (inline_small_functions):
17351 Use xstrdup_for_dump.
17353 2017-07-04 Tom de Vries <tom@codesourcery.com>
17355 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
17357 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
17360 * config/darwin.c (darwin_function_switched_text_sections):
17363 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
17365 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
17367 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17369 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
17371 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17373 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
17374 min_profitable_iters, and th as inclusive lower bounds.
17375 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
17376 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
17377 for min_profitable_iters and min_profitable_estimate.
17378 (vect_transform_loop): Treat th as an inclusive lower bound.
17379 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
17381 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
17384 * config/darwin.c (darwin_function_switched_text_sections):
17385 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
17386 in two pieces, and suppress the use of buf.
17388 2017-07-03 Nathan Sidwell <nathan@acm.org>
17390 * hash-table.h (hash_table_mod1): Fix indentation.
17392 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17394 PR middle-end/81290
17395 * predict.c (force_edge_cold): Be more careful about propagation
17397 * profile-count.h (profile_probability::guessed,
17398 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
17400 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
17402 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
17404 * doc/invoke.texi (rcpc architecture extension): Document it.
17406 2017-07-03 Richard Biener <rguenther@suse.de>
17408 PR tree-optimization/60510
17409 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
17410 the scalar reduction PHI and use it.
17411 (vectorizable_reduction): Properly guard the single_defuse_cycle
17412 path for non-SLP reduction chains where we cannot use it.
17413 Rework reduc_def/index and vector type deduction. Rework
17414 vector operand gathering during reduction op code-gen.
17415 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
17416 chains dissolve the chain and leave it to non-SLP reduction
17419 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17421 * tree-data-ref.h (dr_alignment): Declare.
17422 * tree-data-ref.c (dr_alignment): New function.
17423 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
17424 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
17426 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
17428 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17430 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
17431 and base_misalignment fields.
17432 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
17433 * tree-data-ref.c: Include builtins.h.
17434 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
17435 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
17436 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
17437 * tree-vect-data-refs.c: Include tree-cfg.h.
17438 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
17439 fields instead of calculating an alignment here.
17440 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
17441 innermost_loop_behavior fields.
17443 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17445 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
17447 (DR_STEP_ALIGNMENT): New macro.
17448 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
17449 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
17450 (create_data_ref): Print it.
17451 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
17452 to tell whether the step preserves vector (mis)alignment.
17453 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17454 Move the check for an integer step and generalise to all INTEGER_CST.
17455 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
17456 Print the outer step alignment.
17458 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17460 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
17461 with offset_alignment.
17462 (DR_ALIGNED_TO): Delete.
17463 (DR_OFFSET_ALIGNMENT): New macro.
17464 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
17465 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
17466 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
17467 (create_data_ref): Likewise.
17468 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17469 (vect_analyze_data_refs): Likewise.
17470 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
17471 creating dummy innermost behavior.
17473 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17475 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
17476 with a "innermost_loop_behavior *" and refeence tree.
17477 * tree-data-ref.c (dr_analyze_innermost): Likewise.
17478 (create_data_ref): Update call accordingly.
17479 * tree-predcom.c (find_looparound_phi): Likewise.
17481 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17483 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
17484 fields with dr_wrt_vec_loop.
17485 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
17486 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
17487 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
17488 (vect_dr_behavior): New function.
17489 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17490 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
17491 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
17492 track whether the step preserves the misalignment.
17493 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17494 Use vect_dr_behavior.
17495 (vect_setup_realignment): Update call accordingly.
17496 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
17497 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
17498 call to vect_create_addr_base_for_vector_ref.
17499 (vect_create_cond_for_align_checks): Likewise.
17500 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
17501 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
17502 (vect_recog_mask_conversion_pattern): Likewise.
17503 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
17504 (new_stmt_vec_info): Remove redundant zeroing.
17506 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17508 * common/config/arm/arm-common.c (arm_be8_option): New function.
17509 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
17510 (ISA_ARMv6): Add isa_bit_be8.
17511 * config/arm/arm.h (arm_be8_option): Add prototype.
17512 (BE8_SPEC_FUNCTION): New define.
17513 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
17514 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
17515 (mlittle-endian): Similarly.
17516 (mbe8, mbe32): New options.
17517 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
17518 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
17520 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17522 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
17524 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17526 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
17527 (cleanup_tree_cfg_bb): Use it.
17528 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
17530 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
17532 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17535 * loop-doloop.c (add_test): Update profile.
17537 2017-07-03 Martin Liska <mliska@suse.cz>
17540 * sanopt.c (rewrite_usage_of_param): New function.
17541 (sanitize_rewrite_addressable_params): Likewise.
17542 (pass_sanopt::execute): Call rewrite_usage_of_param.
17544 2017-07-03 Richard Biener <rguenther@suse.de>
17546 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
17547 back to using VIEW_CONVERT_EXPR.
17549 2017-07-03 Martin Liska <mliska@suse.cz>
17552 * doc/extend.texi: Document when a resolver function is
17553 generated for target_clones.
17555 2017-07-03 Martin Liska <mliska@suse.cz>
17557 * asan.c (asan_emit_stack_protection): Unpoison just red zones
17558 and shadow memory of auto variables which are subject of
17559 use-after-scope sanitization.
17560 (asan_expand_mark_ifn): Add do set only when is_poison.
17562 2016-07-03 Richard Biener <rguenther@suse.de>
17564 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
17566 (vect_force_simple_reduction): Record reduction def -> phi mapping.
17567 (vectorizable_reduction): Perform reduction PHI creation when
17568 visiting a reduction PHI and adjust and simplify code generation
17569 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
17570 (vect_transform_loop): Visit reduction PHIs.
17571 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
17572 defs into the SLP tree.
17573 (vect_build_slp_tree): Reduction defs terminate the recursion.
17574 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
17576 (vect_get_vec_defs_for_stmt_copy): Export.
17577 (vect_get_vec_defs): Likewise.
17578 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
17580 (vect_get_vec_defs_for_stmt_copy): Declare.
17581 (vect_get_vec_defs): Likewise.
17583 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17585 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
17586 parameter with a "loop" parameter and use it instead of the
17587 loop containing DR_STMT. Don't check simple_iv when doing
17588 BB analysis. Describe the two analysis modes in the comment.
17590 2017-07-03 Tom de Vries <tom@codesourcery.com>
17592 PR tree-optimization/69468
17593 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
17594 (find_same_succ_bb): Handle ignore_edge_flags.
17596 2017-07-03 Tom de Vries <tom@codesourcery.com>
17598 PR tree-optimization/81192
17599 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
17601 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
17603 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
17605 2017-07-03 Tom de Vries <tom@codesourcery.com>
17607 PR tree-optimization/81192
17608 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
17609 BB_SAME_SUCC (bb) == NULL.
17611 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17613 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
17616 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17618 * dumpfile.c: Include profile-count.h
17619 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
17621 (insert_cond_bb): Update profile.
17622 * tree-cfg.h (insert_cond_bb): Update prototype.
17623 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
17624 * tree-dump.c: Do not include tree-cfg.
17626 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17628 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
17630 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17632 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
17635 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17637 * tree-complex.c (expand_complex_div_wide): update profile.
17639 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17640 Alan Hayward <alan.hayward@arm.com>
17641 David Sherwood <david.sherwood@arm.com>
17643 * Makefile.in (MACHMODE_H): Remove insn-modes.h
17644 (CORETYPES_H): New define.
17645 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
17646 (insn-modes-inline.h, s-modes-inline-h): New rules.
17647 (generated_files): Add insn-modes-inline.h.
17648 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
17649 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
17650 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
17651 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
17652 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
17653 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
17654 (build/gencodes.o, build/genconditions.o): Likewise.
17655 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
17656 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
17657 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
17658 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
17659 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
17660 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
17661 * coretypes.h: Include everything up to real.h for generators.
17662 Include insn-modes.h first. Include wide-int-print.h after
17663 wide-int.h. Include insn-modes-inline.h and then machmode.h.
17664 * machmode.h: Don't include insn-modes.h here.
17665 * function-tests.c: Remove includes of signop.h, machmode.h,
17666 double-int.h and wide-int.h.
17668 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
17670 * optc-save-gen.awk: Likewise.
17671 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
17672 * godump.c: Remove include of wide-int-print.h.
17673 * pretty-print.h: Likewise.
17674 * wide-int-print.cc: Likewise.
17675 * wide-int.cc: Likewise.
17676 * hash-map-tests.c: Remove include of signop.h.
17677 * hash-set-tests.c: Likewise.
17678 * rtl-tests.c: Likewise.
17679 * mkconfig.sh: Remove include of machmode.h.
17680 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
17682 (emit_insn_modes_inline_h): ...this new function. Emit the code
17683 into an insn-modes-inline.h header file, adding appropriate
17684 include guards and end comments.
17685 (emit_insn_modes_c_header): Remove include of machmode.h.
17686 (emit_min_insn_modes_c_header): Include coretypes.h rather than
17688 (main): Handle -i flag and call emit_insn_modes_inline_h when
17691 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17693 * tree-ssa-strlen.c (strinfo): Rename the length field to
17694 nonzero_chars. Add a full_string_p field.
17695 (compare_nonzero_chars, zero_length_string_p): New functions.
17696 (get_addr_stridx): Add an offset_out parameter.
17697 Use compare_nonzero_chars.
17698 (get_stridx): Update accordingly. Use compare_nonzero_chars.
17699 (new_strinfo): Update after above changes to strinfo.
17700 (set_endptr_and_length): Set full_string_p.
17701 (get_string_length): Update after above changes to strinfo.
17702 (unshare_strinfo): Update call to new_strinfo.
17703 (maybe_invalidate): Likewise.
17704 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
17705 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
17706 as a uhwi instead of an shwi. Update after above changes to
17707 strinfo and new_strinfo.
17708 (zero_length_string): Assert that chainsi contains full strings.
17709 Use zero_length_string_p. Update call to new_strinfo.
17710 (adjust_related_strinfos): Update after above changes to strinfo.
17711 Copy full_string_p from origsi.
17712 (adjust_last_stmt): Use zero_length_string_p.
17713 (handle_builtin_strlen): Update after above changes to strinfo and
17714 new_strinfo. Install the lhs as the string length if the previous
17715 entry didn't describe a full string.
17716 (handle_builtin_strchr): Update after above changes to strinfo
17718 (handle_builtin_strcpy): Likewise.
17719 (handle_builtin_strcat): Likewise.
17720 (handle_builtin_malloc): Likewise.
17721 (handle_pointer_plus): Likewise.
17722 (handle_builtin_memcpy): Likewise. Track nonzero characters
17723 that aren't necessarily followed by a nul terminator.
17724 (handle_char_store): Likewise.
17726 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17728 PR tree-optimization/80769
17729 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
17730 for malloc and calloc. Document the new invariant that all related
17731 strinfos have delayed lengths or none do.
17732 (verify_related_strinfos): Move earlier in file.
17733 (set_endptr_and_length): New function, split out from...
17734 (get_string_length): ...here. Also set the lengths of related
17736 (zero_length_string): Assert that chainsi has known (rather than
17738 (adjust_related_strinfos): Likewise.
17740 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17742 PR tree-optimization/81136
17743 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
17744 assert that two references with the same misalignment have the same
17745 compile-time misalignment if those compile-time misalignments
17748 2017-07-01 Andi Kleen <ak@linux.intel.com>
17750 * print-tree.c (print_node): Print all attributes.
17752 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17754 * cfg.c (scale_bbs_frequencies): New function.
17755 * cfg.h (scale_bbs_frequencies): Declare it.
17756 * cfgloopanal.c (single_likely_exit): Cleanup.
17757 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
17759 (scale_loop_profile): Likewise.
17760 (loop_version): Likewise.
17761 (create_empty_loop_on_edge): Update.
17762 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
17763 scale_loop_frequencies, scale_loop_profile, loopify,
17764 loop_version): Update prototypes.
17765 * modulo-sched.c (sms_schedule): Update.
17766 * predict.c (unlikely_executed_edge_p): Also check probability.
17767 (probably_never_executed_edge_p): Fix typo.
17768 * tree-if-conv.c (version_loop_for_if_conversion): Update.
17769 * tree-parloops.c (gen_parallel_loop): Update.
17770 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
17771 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
17772 * tree-ssa-loop-split.c (split_loop): Update.
17773 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
17774 * tree-vect-loop-manip.c (vect_do_peeling): Update.
17775 (vect_loop_versioning): Update.
17776 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
17778 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17780 * trans-mem.c (split_bb_make_tm_edge): Update profile.
17782 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17784 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
17785 to keep profile consistent.
17787 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17789 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
17790 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
17791 * profile-count.h (max_safe_multiplier): Make unsigned.
17792 (profile_count::guessed_zero): New.
17794 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17796 * bb-reorder.c (fix_up_crossing_landing_pad,
17797 fix_crossing_conditional_branches): Use make_single_succ_edge
17798 to keep profile consistent.
17800 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17802 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
17805 2017-07-01 Jakub Jelinek <jakub@redhat.com>
17808 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
17809 the right scopes, make sure cond_jump isn't preserved between multiple
17810 iterations. Search for fallthru edge whenever there are 3+ edges and
17811 use find_fallthru_edge for it.
17813 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17815 Patch by Alexander Monakov <amonakov@ispras.ru>
17816 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
17817 probabilities consistently.
17819 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17821 * pa.c (pa_expand_compare_and_swap_loop): Update call of
17822 emit_cmp_and_jump_insns.
17824 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17827 * tree-inline.c (expand_call_inline): Combine profile statuses.
17829 2017-06-30 Andrew Pinski <apinski@cavium.com>
17831 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
17832 fold_stmt returned true.
17834 2017-06-30 Nathan Sidwell <nathan@acm.org>
17836 * ggc.h (empty_string): Delete.
17837 * cfgexpand.c (expand_asm_stmt): Use plain "".
17838 * optabs.c (expand_asm_memory_barrier): Likewise.
17839 * stringpool.c (empty_string): Delete.
17840 (digit_vector, digit_string): Delete.
17841 (ggc_alloc_string): Use plain "", don't optimize single digit
17842 strings. Use ggc_alloc_atomic.
17844 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
17846 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
17847 comparison set and one other set, use the cost of the non-comparison
17850 2017-06-30 Nathan Sidwell <nathan@acm.org>
17852 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
17855 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
17857 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
17858 loops. Remove now unneeded calls to gimple_switch_set_label() that
17859 just set removed labels to NULL_TREE.
17861 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
17863 * tree-ssanames.c (set_range_info_raw): Abstract from ...
17864 (set_range_info): ...here. Only call set_range_info_raw if domain
17866 (set_nonzero_bits): Call set_range_info_raw.
17867 * tree-ssanames.h (set_range_info_raw): New.
17869 2017-06-30 Jakub Jelinek <jakub@redhat.com>
17872 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
17873 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
17874 of nonimmediate_operand and <store_mask_constraint> instead of m
17875 for the input operand. For V8FI iterator, always split if input
17876 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
17877 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
17878 <store_mask_predicate> instead of register_operand and
17879 <store_mask_constraint> instead of v for the input operand. Make
17880 sure both operands aren't MEMs for if not <mask_applied>.
17882 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
17884 * lto-wrapper.c (copy_file) Close both file descriptors before
17887 2017-06-30 Martin Liska <mliska@suse.cz>
17890 * multiple_target.c (create_dispatcher_calls): Make ifunc
17891 also for function that don't have calls or are not referenced.
17893 2017-06-30 Richard Biener <rguenther@suse.de>
17895 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
17896 analyze the first scalar stmt. Move vector type computation
17897 for the BB case here from ...
17898 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
17899 live operation processing in the SLP case properly.
17901 2017-06-30 Richard Biener <rguenther@suse.de>
17903 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
17905 2017-06-30 Martin Liska <mliska@suse.cz>
17908 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
17909 before BUILT_IN_UNWIND_RESUME when ASAN is used.
17911 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
17913 * doc/invoke.texi (AArch64): Add missing options and remove redundant
17916 2017-06-30 Richard Biener <rguenther@suse.de>
17918 PR tree-optimization/81249
17919 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
17920 condition reduction result to original scalar type.
17922 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17924 * profile-count.h (enum profile_quality): Fix typos and whitespace
17927 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17929 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
17930 type for branch probabilities.
17932 2017-06-29 Julian Brown <julian@codesourcery.com>
17933 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17935 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
17936 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
17937 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
17938 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
17940 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17942 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
17943 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
17944 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
17945 CC usage from generic code to here.
17946 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
17947 CC usage into the target macros.
17949 2017-06-29 Maya Rashish <coypu@sdf.org>
17951 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
17954 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17956 * arm/arm-builtins.c: Include profile-count.h
17957 * except.c (sjlj_emit_function_enter): Use
17958 profile_probability::unlikely.
17960 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17962 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
17963 and tocrel_offset be pointer args rather than implicitly using
17965 (legitimate_constant_pool_address_p, rs6000_emit_move,
17966 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
17967 tocrel_offset and use in toc_relative_expr_p call.
17968 (print_operand, print_operand_address): Use static tocrel_base_oac
17969 and tocrel_offset_oac.
17970 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
17973 2017-06-29 Maya Rashish <coypu@sdf.org>
17975 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
17977 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
17979 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
17980 objects, take into account only the alignment of 'op0' and 'mode1' if
17983 2017-06-29 Steve Ellcey <sellcey@cavium.com>
17985 * ccmp.c (ccmp_tree_comparison_p): New function.
17986 (ccmp_candidate_p): Update to use above function.
17987 (get_compare_parts): New function.
17988 (expand_ccmp_next): Update to use new functions.
17989 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
17991 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
17992 take mode as argument.
17993 * ccmp.h (expand_ccmp_expr): Add mode as argument.
17994 * expr.c (expand_expr_real_1): Pass mode as argument.
17996 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
17998 * combine.c (combine_instructions): Print insns to dump_file, together
18001 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18003 * asan.c (asan_emit_stack_protection): Update.
18004 (create_cond_insert_point): Update.
18005 * auto-profile.c (afdo_propagate_circuit): Update.
18006 * basic-block.h (struct edge_def): Turn probability to
18007 profile_probability.
18008 (EDGE_FREQUENCY): Update.
18009 * bb-reorder.c (find_traces_1_round): Update.
18010 (better_edge_p): Update.
18011 (sanitize_hot_paths): Update.
18012 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
18013 (make_single_succ_edge): Update.
18014 (check_bb_profile): Update.
18015 (dump_edge_info): Update.
18016 (update_bb_profile_for_threading): Update.
18017 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
18018 probabilitycount to 0.
18019 * cfgbuild.c (compute_outgoing_frequencies): Update.
18020 * cfgcleanup.c (try_forward_edges): Update.
18021 (outgoing_edges_match): Update.
18022 (try_crossjump_to_edge): Update.
18023 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
18024 (expand_gimple_tailcall): Update.
18025 (construct_init_block): Use make_single_succ_edge.
18026 (construct_exit_block): Use make_single_succ_edge.
18027 * cfghooks.c (verify_flow_info): Update.
18028 (redirect_edge_succ_nodup): Update.
18029 (split_edge): Update.
18030 (account_profile_record): Update.
18031 * cfgloopanal.c (single_likely_exit): Update.
18032 * cfgloopmanip.c (scale_loop_profile): Update.
18033 (set_zero_probability): Remove.
18034 (duplicate_loop_to_header_edge): Update.
18035 * cfgloopmanip.h (loop_version): Update prototype.
18036 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
18037 (force_nonfallthru_and_redirect): Update.
18038 (update_br_prob_note): Update.
18039 (rtl_verify_edges): Update.
18040 (purge_dead_edges): Update.
18041 (rtl_lv_add_condition_to_bb): Update.
18042 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
18043 * cgraphunit.c (init_lowered_empty_function): Update.
18044 (cgraph_node::expand_thunk): Update.
18045 * cilk-common.c: Include profile-count.h
18046 * dojump.c (inv): Remove.
18047 (jumpifnot): Update.
18048 (jumpifnot_1): Update.
18049 (do_jump_1): Update.
18051 (do_jump_by_parts_greater_rtx): Update.
18052 (do_compare_rtx_and_jump): Update.
18053 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
18054 do_jump_1. do_compare_rtx_and_jump): Update prototype.
18055 * dwarf2cfi.c: Include profile-count.h
18056 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
18057 (sjlj_emit_dispatch_table): Likewise.
18058 * explow.c: Include profile-count.h
18059 * expmed.c (emit_store_flag_force): Update.
18060 (do_cmp_and_jump): Update.
18061 * expr.c (compare_by_pieces_d::generate): Update.
18062 (compare_by_pieces_d::finish_mode): Update.
18063 (emit_block_move_via_loop): Update.
18064 (store_expr_with_bounds): Update.
18065 (store_constructor): Update.
18066 (expand_expr_real_2): Update.
18067 (expand_expr_real_1): Update.
18068 * expr.h (try_casesi, try_tablejump): Update prototypes.
18069 * gimple-pretty-print.c (dump_probability): Update.
18070 (dump_profile): New.
18071 (dump_gimple_label): Update.
18072 (dump_gimple_bb_header): Update.
18073 * graph.c (draw_cfg_node_succ_edges): Update.
18074 * hsa-gen.c (convert_switch_statements): Update.
18075 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
18076 (find_if_case_1): Update.
18077 (find_if_case_2): Update.
18078 * internal-fn.c (expand_arith_overflow_result_store): Update.
18079 (expand_addsub_overflow): Update.
18080 (expand_neg_overflow): Update.
18081 (expand_mul_overflow): Update.
18082 (expand_vector_ubsan_overflow): Update.
18083 * ipa-cp.c (good_cloning_opportunity_p): Update.
18084 * ipa-split.c (split_function): Use make_single_succ_edge.
18085 * ipa-utils.c (ipa_merge_profiles): Update.
18086 * loop-doloop.c (add_test): Update.
18087 (doloop_modify): Update.
18088 * loop-unroll.c (compare_and_jump_seq): Update.
18089 (unroll_loop_runtime_iterations): Update.
18090 * lra-constraints.c (lra_inheritance): Update.
18091 * lto-streamer-in.c (input_cfg): Update.
18092 * lto-streamer-out.c (output_cfg): Update.
18093 * mcf.c (adjust_cfg_counts): Update.
18094 * modulo-sched.c (sms_schedule): Update.
18095 * omp-expand.c (expand_omp_for_init_counts): Update.
18096 (extract_omp_for_update_vars): Update.
18097 (expand_omp_ordered_sink): Update.
18098 (expand_omp_for_ordered_loops): Update.
18099 (expand_omp_for_generic): Update.
18100 (expand_omp_for_static_nochunk): Update.
18101 (expand_omp_for_static_chunk): Update.
18102 (expand_cilk_for): Update.
18103 (expand_omp_simd): Update.
18104 (expand_omp_taskloop_for_outer): Update.
18105 (expand_omp_taskloop_for_inner): Update.
18106 * omp-simd-clone.c (simd_clone_adjust): Update.
18107 * optabs.c (expand_doubleword_shift): Update.
18108 (expand_abs): Update.
18109 (emit_cmp_and_jump_insn_1): Update.
18110 (expand_compare_and_swap_loop): Update.
18111 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
18112 * predict.c (predictable_edge_p): Update.
18113 (edge_probability_reliable_p): Update.
18114 (set_even_probabilities): Update.
18115 (combine_predictions_for_insn): Update.
18116 (combine_predictions_for_bb): Update.
18117 (propagate_freq): Update.
18118 (estimate_bb_frequencies): Update.
18119 (force_edge_cold): Update.
18120 * profile-count.c (profile_count::dump): Add missing space into dump.
18121 (profile_count::debug): Add newline.
18122 (profile_count::differs_from_p): Explicitly convert to unsigned.
18123 (profile_count::stream_in): Update.
18124 (profile_probability::dump): New member function.
18125 (profile_probability::debug): New member function.
18126 (profile_probability::differs_from_p): New member function.
18127 (profile_probability::differs_lot_from_p): New member function.
18128 (profile_probability::stream_in): New member function.
18129 (profile_probability::stream_out): New member function.
18130 * profile-count.h (profile_count_quality): Rename to ...
18131 (profile_quality): ... this one.
18132 (profile_probability): New.
18133 (profile_count): Update.
18134 * profile.c (compute_branch_probabilities): Update.
18135 * recog.c (peep2_attempt): Update.
18136 * sched-ebb.c (schedule_ebbs): Update.
18137 * sched-rgn.c (find_single_block_region): Update.
18138 (compute_dom_prob_ps): Update.
18139 (schedule_region): Update.
18140 * sel-sched-ir.c (compute_succs_info): Update.
18141 * stmt.c (struct case_node): Update.
18142 (do_jump_if_equal): Update.
18143 (get_outgoing_edge_probs): Update.
18144 (conditional_probability): Update.
18145 (emit_case_dispatch_table): Update.
18146 (expand_case): Update.
18147 (expand_sjlj_dispatch_table): Update.
18148 (emit_case_nodes): Update.
18149 * targhooks.c: Update.
18150 * tracer.c (better_p): Update.
18151 (find_best_successor): Update.
18152 * trans-mem.c (expand_transaction): Update.
18153 * tree-call-cdce.c: Update.
18154 * tree-cfg.c (gimple_split_edge): Upate.
18155 (move_sese_region_to_fn): Upate.
18156 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
18157 * tree-eh.c (lower_resx): Upate.
18158 (cleanup_empty_eh_move_lp): Upate.
18159 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18160 * tree-inline.c (copy_edges_for_bb): Update.
18161 (copy_cfg_body): Update.
18162 * tree-parloops.c (gen_parallel_loop): Update.
18163 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
18164 (gimple_gen_time_profiler): Update.
18165 * tree-ssa-dce.c (remove_dead_stmt): Update.
18166 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
18167 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
18168 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
18169 (unloop_loops): Update.
18170 (try_peel_loop): Update.
18171 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18172 * tree-ssa-loop-split.c (connect_loops): Update.
18173 (split_loop): Update.
18174 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
18175 (hoist_guard): Update.
18176 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
18177 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
18178 (value_replacement): Update.
18179 * tree-ssa-reassoc.c (branch_fixup): Update.
18180 * tree-ssa-tail-merge.c (replace_block_by): Update.
18181 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
18182 (create_edge_and_update_destination_phis): Update.
18183 (compute_path_counts): Update.
18184 (recompute_probabilities): Update.
18185 (update_joiner_offpath_counts): Update.
18186 (freqs_to_counts_path): Update.
18187 (duplicate_thread_path): Update.
18188 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
18189 (struct switch_conv_info): Update.
18190 (gen_inbound_check): Update.
18191 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
18192 (vect_do_peeling): Update.
18193 (vect_loop_versioning): Update.
18194 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18195 (optimize_mask_stores): Update.
18196 * ubsan.c (ubsan_expand_null_ifn): Update.
18197 * value-prof.c (gimple_divmod_fixed_value): Update.
18198 (gimple_divmod_fixed_value_transform): Update.
18199 (gimple_mod_pow2): Update.
18200 (gimple_mod_pow2_value_transform): Update.
18201 (gimple_mod_subtract): Update.
18202 (gimple_mod_subtract_transform): Update.
18203 (gimple_ic): Update.
18204 (gimple_stringop_fixed_value): Update.
18205 (gimple_stringops_transform): Update.
18206 * value-prof.h: Update.
18208 2017-06-29 Carl Love <cel@us.ibm.com>
18210 * config/rs6000/rs6000-c.c: Add support for built-in functions
18211 vector signed int vec_signed (vector float);
18212 vector signed long long vec_signed (vector double);
18213 vector signed int vec_signed2 (vector double, vector double);
18214 vector signed int vec_signede (vector double);
18215 vector signed int vec_signedo (vector double);
18216 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
18217 instruction generator.
18218 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18219 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
18220 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
18222 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
18223 vunsignede_v2df): Add define_expands.
18224 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
18225 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
18226 VEC_UNSIGNEDO): Add definitions.
18227 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18228 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
18229 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
18230 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
18231 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
18232 * config/rs6000/altivec.h (vec_signed, vec_signed2,
18233 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
18234 vec_unsignede, vec_unsignedo): Add builtin defines.
18235 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
18237 * doc/extend.texi: Update the built-in documentation file for the
18238 new built-in functions.
18240 2017-06-29 Richard Biener <rguenther@suse.de>
18242 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
18243 reduction chains to LOOP_VINFO_REDUCTIONS.
18244 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
18245 SLP reductions after processing reduction chains.
18247 2017-06-29 Nathan Sidwell <nathan@acm.org>
18249 * builtins.c (fold_builtin_FUNCTION): Use
18250 lang_hooks.decl_printable_name.
18252 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
18254 PR middle-end/81194
18255 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
18256 with only one label.
18257 * stmt.c (expand_case): Assert NCASES is greater than one.
18259 2017-06-29 Richard Biener <rguenther@suse.de>
18261 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
18263 (group_case_labels): Likewise.
18264 (find_taken_edge): Push sanity checking on val to workers...
18265 (find_taken_edge_cond_expr): ... here
18266 (find_taken_edge_switch_expr): ... and here, handle cases
18267 with just a default label.
18268 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
18269 (group_case_labels): Likewise.
18270 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
18271 group_case_labels does anything cleanup the CFG again.
18273 2017-06-29 Bin Cheng <bin.cheng@arm.com>
18275 PR tree-optimization/81196
18276 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
18277 exit condition comparing two IVs.
18279 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
18281 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
18282 profile to the dummy entry at the end of the list of architectures.
18283 * config/arm/arm-cpu-cdata.h: Regenerated.
18285 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18286 Michael Collison <michael.collison@arm.com>
18289 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
18291 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
18292 (*aarch64_reg_<mode>3_minus_mask): New pattern.
18293 (*aarch64_<optab>_reg_di3_mask2): New pattern.
18294 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
18295 of shift when the shift amount is masked with constant equal to
18296 the size of the mode.
18297 * config/aarch64/predicates.md (subreg_lowpart_operator): New
18300 2017-06-29 Martin Liska <mliska@suse.cz>
18302 * config/i386/i386.opt: Change range from [1,5] to [0,5].
18304 2017-06-29 Yury Gribov <tetra2005@gmail.com>
18307 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
18310 (edge_growth_cache_entry::edge_growth_cache_entry): New
18312 (reset_edge_growth_cache): Update to use constructor.
18314 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18316 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
18317 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
18318 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
18320 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
18322 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
18323 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
18325 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
18327 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
18328 (*-linux-uclibc*): Add t-uclibc tmake_file.
18329 * config/t-musl: New.
18330 * config/t-uclibc: New.
18332 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
18334 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
18336 (gen_comm_data): Emit architectural setting of arch_prof.
18337 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
18339 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
18340 (armv8-m.base, armv8-m.main): Likewise.
18341 * arm-protos.h (arm_build_target): Add profile field.
18342 (arch_option): Likewise.
18343 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
18345 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
18346 arm_active_target.profile.
18348 2017-06-28 Richard Biener <rguenther@suse.de>
18350 PR middle-end/81227
18351 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
18352 TYPE_OVERFLOW_WRAPS.
18353 * match.pd (negate_expr_p): Likewise.
18354 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
18355 fold_build2, not fold_binary.
18357 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
18359 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
18360 Convert memory address to Pmode.
18361 (aarch64_print_operand): Assert MEM operands are always Pmode.
18363 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
18366 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
18367 Remove redundant if.
18368 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
18369 * config/arm/aarch-common-protos.h
18370 (aarch_forward_to_shift_is_not_shifted_re): Remove.
18371 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
18373 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
18376 * multiple_target.c (create_dispatcher_calls): Set the default
18377 clone to be static, not public.
18379 2017-06-28 Richard Biener <rguenther@suse.de>
18381 * tree-vect-loop.c (vectorizable_reduction): Move special
18382 cond reduction IV var creation ...
18383 (vect_create_epilog_for_reduction): ... here. Remove induction_index
18384 parameter. Use STMT_VINFO_VECTYPE.
18385 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
18388 2017-06-28 Martin Liska <mliska@suse.cz>
18391 * ipa-visibility.c (non_local_p): Handle visibility.
18393 2017-06-28 Martin Liska <mliska@suse.cz>
18396 * common.opt: Add IntegerRange to various options.
18397 * opt-functions.awk (integer_range_info): New function.
18398 * optc-gen.awk: Add integer_range_info to cl_options struct.
18399 * opts-common.c (decode_cmdline_option): Handle
18400 CL_ERR_INT_RANGE_ARG.
18401 (cmdline_handle_error): Likewise.
18402 * opts.c (print_filtered_help): Show valid interval in
18403 when --help is provided.
18404 * opts.h (struct cl_option): Add range_min and range_max fields.
18405 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
18407 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
18409 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
18410 (x * C EQ/NE y * C): New transformation.
18412 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
18414 * genmultilib (combination_space): Accept '+' in option names.
18416 2017-06-28 Martin Liska <mliska@suse.cz>
18419 * asan.c (instrument_derefs): Bail out inner references
18420 that are hard register variables.
18422 2017-06-28 Jakub Jelinek <jakub@redhat.com>
18425 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
18426 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
18428 2017-06-28 Richard Biener <rguenther@suse.de>
18430 * tree-vectorizer.h (vect_get_vec_defs): Remove.
18431 (vect_get_slp_defs): Adjust.
18432 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
18434 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
18436 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
18437 get_initial_defs_for_reduction instead of vect_get_vec_defs.
18438 (vectorizable_reduction): Adjust.
18439 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
18441 (vect_get_slp_defs): Likewise.
18442 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
18443 (vectorizable_bswap): Adjust.
18444 (vectorizable_call): Likewise.
18445 (vectorizable_conversion): Likewise.
18446 (vectorizable_assignment): Likewise.
18447 (vectorizable_shift): Likewise.
18448 (vectorizable_operation): Likewise.
18449 (vectorizable_store): Likewise.
18450 (vectorizable_condition): Likewise.
18451 (vectorizable_comparison): Likewise.
18453 2017-06-28 Michael Collison <michael.collison@arm.com>
18456 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
18458 (arm_gen_movmemqi): Removed unused variable 'i'.
18459 Convert 'for' loop into 'while' loop.
18460 (arm_expand_prologue): Remove last unnecessary set of insn.
18461 (thumb_pop): Remove unused variable 'pushed_words'.
18462 (thumb_exit): Remove last unnecessary set of regs_to_pop.
18464 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18466 * config/s390/predicates.md: Use s390_rel_address_ok_p.
18467 * config/s390/s390-protos.h: Add prototype of
18468 s390_rel_address_ok_p.
18469 * config/s390/s390.c (s390_got_symbol): New function.
18470 (s390_rel_address_ok_p): New function.
18471 (legitimize_pic_address): Use s390_rel_address_ok_p.
18472 (s390_load_got): Use s390_got_symbol.
18473 (s390_option_override): Issue error if
18474 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
18475 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
18477 * config/s390/s390.opt: New option mpic-data-is-text-relative.
18479 2017-06-27 Andrew Pinski <apinski@cavium.com>
18481 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
18482 (X * copysign (1.0, X)): New pattern.
18483 (X * copysign (1.0, -X)): New pattern.
18484 (copysign (-1.0, CST)): New pattern.
18486 2017-06-27 Joseph Myers <joseph@codesourcery.com>
18488 * genmultilib (combination_space): Remove variable.
18489 Validate reuse rules against regular expression for any sequence
18490 of multilib options in any order.
18492 2017-06-27 Michael Collison <michael.collison@arm.com>
18494 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
18495 call aarch64_split_simd_combine.
18496 * (aarch64_combine_internal<mode>): Delete pattern.
18497 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
18498 Allow register and subreg operands.
18500 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18502 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
18503 specific need, just fallback on defaults.
18504 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
18506 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18507 Olivier Hainque <hainque@adacore.com>
18509 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
18511 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
18512 targets. Pick a default if no particular attempt applied.
18513 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
18516 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18518 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
18519 (x86_64-wrs-vxworks7): Likewise.
18521 2017-06-27 Marek Polacek <polacek@redhat.com>
18524 * ubsan.c (instrument_null): Check get_base_address's result for null.
18526 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18528 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
18530 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18532 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
18533 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
18534 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
18535 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
18536 New function types.
18537 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
18538 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
18539 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
18540 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
18541 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
18542 BUILT_IN_FEUPDATEENV): New builtins.
18543 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
18544 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
18545 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
18546 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
18548 (builtin_structptr_types): Adjust size.
18549 * tree.c (builtin_structptr_types): Add four entries.
18551 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18552 Olivier Hainque <hainque@adacore.com>
18554 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
18555 (TLS_SYM): New local macro, forcing reference to __tls__ on
18556 link command lines for VxWorks 7 RTPs, triggering initialization
18558 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
18559 OS features TLS support, true for RTPs on VxWorks 7.
18560 * config/vxworks.c (vxworks_override_options): Setup emutls
18563 2017-06-27 Jakub Jelinek <jakub@redhat.com>
18565 * predict.c (test_prediction_value_range): Use -1U instead of -1
18566 to avoid narrowing conversion warning.
18567 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
18568 to avoid narrowing conversion warning.
18569 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
18571 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
18573 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18575 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
18576 64bit configurations.
18577 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
18578 (SIZE_TYPE): Likewise.
18579 * config/vxworks.c (vxworks_emutls_var_fields): Use
18580 long_unsigned_type_node instead of unsigned_type_node as the offset
18581 field type, which is "pointer" mode in emutls.c.
18583 2017-06-27 Jakub Jelinek <jakub@redhat.com>
18586 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
18588 PR middle-end/81207
18589 * gimple-fold.c (replace_call_with_call_and_fold): Handle
18590 gimple_vuse copying separately from gimple_vdef copying.
18592 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18594 * value-prof.c (free_hist): Remove call to memset and the enclosing if
18597 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
18598 Olivier Hainque <hainque@adacore.com>
18600 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
18601 for all vxworks7 targets.
18602 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
18603 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
18604 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
18605 variations for VX6/VX7 and 32/64bits later on in ...
18606 (VXWORKS_LIB_SPEC): Leverage new macros.
18607 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
18608 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
18610 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
18612 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
18613 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
18615 2017-06-26 Carl Love <cel@us.ibm.com>
18617 * config/rs6000/rs6000-c.c: Add support for built-in functions
18618 vector bool char vec_reve (vector bool char);
18619 vector signed char vec_reve (vector signed char);
18620 vector unsigned char vec_reve (vector unsigned char);
18621 vector bool int vec_reve (vector bool int);
18622 vector signed int vec_reve (vector signed int);
18623 vector unsigned int vec_reve (vector unsigned int);
18624 vector bool long long vec_reve (vector bool long long);
18625 vector signed long long vec_reve (vector signed long long);
18626 vector unsigned long long vec_reve (vector unsigned long long);
18627 vector bool short vec_reve (vector bool short);
18628 vector signed short vec_reve (vector signed short);
18629 vector double vec_reve (vector double);
18630 vector float vec_reve (vector float);
18631 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
18632 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
18633 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
18634 (altivec_vreve): New pattern.
18635 * config/rs6000/altivec.h (vec_reve): New define.
18636 * doc/extend.texi (vec_rev): Update the built-in documentation file
18637 for the new built-in functions.
18639 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18641 PR tree-optimization/71815
18642 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
18644 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
18646 (slsr_process_phi): Likewise.
18647 (replace_uncond_cands_and_profitable_phis): Don't replace a
18648 multiply candidate with a stride of 1 (copy or cast).
18649 (phi_incr_cost): Call uses_consumed_by_stmt rather than
18651 (lowest_cost_path): Likewise.
18652 (total_savings): Likewise.
18654 2017-06-26 Richard Biener <rguenther@suse.de>
18657 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
18658 Use def_builtin_pure for all gather builtins.
18660 2017-06-26 Richard Biener <rguenther@suse.de>
18662 PR tree-optimization/81203
18663 * tree-tailcall.c (find_tail_calls): Do not move stmts into
18664 non-dominating BBs.
18666 2017-06-26 Marek Polacek <polacek@redhat.com>
18669 * doc/invoke.texi: Document -Wmultistatement-macros.
18671 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
18673 * doc/sourcebuild.texi (ARM-specific attributes): Document new
18674 arm_neon_ok_no_float_abi effective target.
18676 2017-06-26 Richard Biener <rguenther@suse.de>
18678 PR tree-optimization/80928
18679 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
18680 (copy_bbs): Set BB_DUPLICATED flag early.
18681 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
18683 (execute_on_shrinking_pred): Likewise.
18684 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
18685 BB_DUPLICATED blocks.
18686 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
18687 iterate over all PHIs considering removal of *gsi.
18689 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
18691 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
18694 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18696 * config/rs6000/rs6000-string.c: (expand_block_clear,
18697 do_load_for_compare, select_block_compare_mode,
18698 compute_current_alignment, expand_block_compare,
18699 expand_strncmp_align_check, expand_strn_compare,
18700 expand_block_move, rs6000_output_load_multiple)
18701 Move functions related to string/block move/compare
18702 to a separate file.
18703 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
18704 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
18705 for this function which is now used in two files.
18706 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
18707 * config.gcc: Add rs6000-string.o to extra_objs for
18708 targets powerpc*-*-* and rs6000*-*-*.
18710 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
18713 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
18714 32-bit, since indexed is not valid for DImode.
18715 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
18716 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
18717 (define_peephole2 for Altivec d-form load): Add 32-bit support.
18718 (define_peephole2 for Altivec d-form store): Likewise.
18721 * multiple_target.c (create_dispatcher_calls): Only create the
18722 dispatcher call if the function is the default clone of a
18723 versioned function.
18725 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
18727 PR middle-end/80902
18728 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
18729 a call, force the call to not be a tail call.
18731 2017-06-23 Jeff Law <law@redhat.com>
18733 * doc/contrib.texi: Add entry for Steven Pemberton's work on
18736 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
18738 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
18739 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
18740 handling for early expansion of vector shifts (sl,sr,sra,rl).
18741 (builtin_function_type): Add vector shift right instructions
18742 to the unsigned argument list.
18744 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
18746 rtl-optimizatoin/79286
18747 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
18748 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
18749 trap. PIC register plus a const unspec without offset can never trap.
18751 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
18753 * tree.h (builtin_structptr_type): New type.
18754 (builtin_structptr_types): Declare new array.
18755 * tree.c (builtin_structptr_types): New array.
18756 (free_lang_data, build_common_tree_nodes): Use it.
18758 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
18761 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
18764 2017-06-22 Matt Turner <mattst88@gmail.com>
18766 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
18767 Lake models to skylake case. Assume skylake for unknown
18768 models with clflushopt.
18770 2017-06-22 Jeff Law <law@redhat.com>
18772 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
18773 frame sizes that do not satisfy aarch64_uimm12_shift.
18775 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
18777 * profile-count.h (apply_probability,
18778 apply_scale, probability_in): Fix checks for zero.
18780 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18782 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
18783 * doc/cppdiropts.texi (-I @var{dir}): Document it.
18785 2016-06-22 Richard Biener <rguenther@suse.de>
18787 * tree-vect-loop.c (vect_model_reduction_cost): Handle
18788 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
18789 REDUC_MAX_EXPR support.
18790 (vectorizable_reduction): Likewise.
18791 (vect_create_epilog_for_reduction): Likewise.
18793 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
18795 * match.pd (A / (1 << B) -> A >> B): New.
18796 * generic-match-head.c: Include optabs-tree.h.
18797 * gimple-match-head.c: Likewise.
18798 * optabs-tree.h (target_supports_op_p): New.
18799 * optabs-tree.c (target_supports_op_p): New.
18801 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18803 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
18804 $gcc_cv_ld --help output.
18805 (gcc_cv_ld_demangle): Likewise.
18806 (gcc_cv_ld_eh_frame_hdr): Likewise.
18807 (gcc_cv_ld_pie): Likewise.
18808 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
18809 (gcc_cv_ld_buildid): Likewise.
18810 (gcc_cv_ld_sysroot): Likewise.
18811 (ld_bndplt_support): Likewise.
18812 (ld_pushpopstate_support): Likewise.
18813 * configure: Regenerate.
18814 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
18816 2017-06-21 Jakub Jelinek <jakub@redhat.com>
18819 * config/i386/sse.md (round<mode>2): Renumber match_dup and
18820 operands indexes to avoid gap between operands and match_dups.
18822 2017-06-21 Andrew Pinski <apinski@cavium.com>
18824 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
18825 Increment Arith_shift and Arith_shift_reg by 1.
18826 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
18828 * config/aarch64/aarch64.c (thunderx_tunings): Enable
18829 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
18830 (aarch64_strip_extend): Add new argument and test for it.
18831 (aarch64_cheap_mult_shift_p): New function.
18832 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
18833 add a cost if it is true.
18834 Update calls to aarch64_strip_extend.
18835 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
18837 2017-06-21 Andrew Pinski <apinski@cavium.com>
18839 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
18841 (thunderxt88): Likewise.
18842 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
18843 (thunderx_prefetch_tune): New variable.
18844 (thunderx2t99_prefetch_tune): Update for the correct values.
18845 (thunderxt88_tunings): New variable.
18846 (thunderx_tunings): Use thunderx_prefetch_tune instead of
18847 generic_prefetch_tune.
18848 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
18850 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18852 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
18853 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
18854 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
18855 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
18856 (aarch64_atomic_cas<mode>, GPI): Likewise.
18858 2017-06-21 Martin Liska <mliska@suse.cz>
18860 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
18861 statements on cold and hot labels.
18862 * predict.c (tree_estimate_probability_bb): Remove the
18863 prediction from this place.
18865 2017-06-21 Martin Liska <mliska@suse.cz>
18867 PR tree-optimization/79489
18868 * gimplify.c (maybe_add_early_return_predict_stmt): New
18870 (gimplify_return_expr): Call the function.
18871 * predict.c (tree_estimate_probability_bb): Remove handling
18873 * predict.def: Update comment about early return predictor.
18874 * gimple-predict.h (is_gimple_predict): New function.
18875 * predict.def: Change default value of early return to 66.
18876 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
18878 * passes.def: Put pass_strip_predict_hints to the beginning of
18881 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
18883 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
18884 FUNCTION_DECL declarations.
18885 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
18887 (dwaf2out_decl): Likewise.
18888 * godump.c (go_early_global_decl): Skip call to the real debug hook
18889 for FUNCTION_DECL declarations.
18890 * passes.c (rest_of_decl_compilation): Skip call to the
18891 early_global_decl debug hook for FUNCTION_DECL declarations, unless
18892 -fdump-go-spec is passed.
18894 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
18896 * config/i386/i386.c (struct builtin_isa): New field pure_p.
18897 Reorder for compactness.
18898 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
18899 (def_builtin_pure, def_builtin_pure2): New functions.
18900 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
18902 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
18904 * match.pd (nop_convert): New predicate.
18905 ((A +- CST1) +- CST2): Allow some NOP conversions.
18907 2017-06-21 Jakub Jelinek <jakub@redhat.com>
18910 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
18911 with ctors/dtors if GOVD_SHARED is set.
18913 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
18915 * config/aarch64/aarch64.md (movti_aarch64):
18916 Emit mov rather than orr.
18917 (movtf_aarch64): Likewise.
18918 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
18919 Emit mov rather than orr.
18921 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
18923 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
18924 Swap alternatives, make integer dup more expensive.
18926 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
18928 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
18929 Return true for non-tls symbols.
18931 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
18933 * config/aarch64/aarch64-cores.def (cortex-a55): New.
18934 (cortex-a75): Likewise.
18935 (cortex-a75.cortex-a55): Likewise.
18936 * config/aarch64/aarch64-tune.md: Regenerate.
18937 * doc/invoke.texi (-mtune): Document new values for -mtune.
18939 2017-06-21 Tom de Vries <tom@codesourcery.com>
18941 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
18942 stack_size feature.
18943 (Effective-Target Keywords, Other attributes): Suggest using
18944 dg-add-options stack_size feature to get stack limit in stack_size
18945 effective target documentation.
18947 2017-06-21 Julian Brown <julian@codesourcery.com>
18948 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18950 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
18951 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
18952 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
18954 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
18955 attribute type list for neon_multiply.
18956 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
18957 attribute type list for neon_multiply.
18958 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
18959 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
18960 attribute type list for neon_multiply.
18961 * config/arm/types.md (crypto_pmull): Add.
18962 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
18963 attribute type list.
18965 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
18967 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
18970 2017-06-20 Jakub Jelinek <jakub@redhat.com>
18972 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
18973 to make sure not to dereference a NULL cost_classes_ptr pointer.
18975 2017-06-20 Carl Love <cel@us.ibm.com>
18977 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18978 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
18979 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
18980 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
18981 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
18982 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
18983 VMULOSW): New enum "unspec" values.
18984 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
18985 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
18986 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
18987 altivec_vmulosw): New patterns.
18988 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
18989 VMULOSW): Add definitions.
18991 2017-06-20 Julia Koval <julia.koval@intel.com>
18993 * config/i386/i386.c: Fix rounding expand for new pattern.
18994 * config/i386/subst.md: Fix pattern (parallel -> unspec).
18996 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
18998 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
18999 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
19001 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19003 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
19006 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19008 * config/aarch64/aarch64-cores.def: Rearrange to sort by
19009 architecture, then by implementer ID.
19010 * config/aarch64/aarch64-tune.md: Regenerate.
19012 2017-06-20 Richard Biener <rguenther@suse.de>
19014 PR middle-end/81097
19015 * fold-const.c (split_tree): Fold to type before negating.
19017 2017-06-20 David Malcolm <dmalcolm@redhat.com>
19019 * diagnostic-show-locus.c
19020 (selftest::test_fixit_deletion_affecting_newline): New function.
19021 (selftest::diagnostic_show_locus_c_tests): Call it.
19023 2017-06-20 Andreas Schwab <schwab@suse.de>
19026 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
19029 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19031 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
19032 __ARM_FEATURE_COPROC according to support.
19034 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19036 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
19037 Rewritten to avoid overflow for > 32-bit pointers.
19040 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
19041 by removing enum keyword.
19042 (ubsan_type_descriptor): Likewise. Formatting fix.
19045 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
19046 splitter): Require TARGET_SSE2 in the condition.
19048 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
19051 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
19052 for doing vector set of SFmode on ISA 3.0.
19053 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
19054 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
19056 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
19057 SFmode value into a V4SF variable that was extracted from another
19058 V4SF variable without converting the element to double precision
19059 and back to single precision vector format.
19060 (vsx_insert_extract_v4sf_p9_2): Likewise.
19062 2017-06-19 Jakub Jelinek <jakub@redhat.com>
19064 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
19065 in UWHI to avoid undefined overflow.
19068 * ubsan.h (enum ubsan_encode_value_phase): New.
19069 (ubsan_encode_value): Change second argument to
19070 enum ubsan_encode_value_phase with default value of
19071 UBSAN_ENCODE_VALUE_GENERIC.
19072 * ubsan.c (ubsan_encode_value): Change second argument to
19073 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
19074 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
19075 create_tmp_var_raw instead of create_tmp_var and use a
19077 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
19078 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
19079 ubsan_encode_value callers.
19082 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
19083 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
19084 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
19086 2017-06-19 Richard Biener <rguenther@suse.de>
19088 PR middle-end/81118
19089 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
19090 estimates if we changed anything.
19092 2017-06-19 Richard Biener <rguenther@suse.de>
19094 PR tree-optimization/80887
19095 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
19096 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
19097 simplified lookups, then reset mprts_hook.
19098 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
19100 (try_to_simplify): Likewise.
19102 2017-06-19 Martin Liska <mliska@suse.cz>
19105 * gimplify.c (gimplify_switch_expr):
19106 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
19108 2017-06-19 Martin Liska <mliska@suse.cz>
19110 * doc/install.texi: Document that PGO runs in 4 stages.
19112 2017-06-19 Martin Liska <mliska@suse.cz>
19115 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
19116 to dispatcher function name.
19117 * multiple_target.c (replace_function_decl): New function.
19118 (create_dispatcher_calls): Redirect both edges and references.
19120 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
19122 * profile-count.c (profile_count::dump): Dump quality.
19123 (profile_count::differs_from_p): Update for unsigned val.
19124 * profile-count.h (profile_count_quality): New enum.
19125 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
19127 2017-06-19 Richard Biener <rguenther@suse.de>
19129 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
19130 struct function as arg.
19131 (estimate_numbers_of_iterations): Export overload with loop arg.
19132 (free_numbers_of_iterations_estimates_loop): Use an overload of
19133 free_numbers_of_iterations_estimates instead.
19134 * tree-cfg.c (remove_bb): Adjust.
19135 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
19136 * tree-parloops.c (gen_parallel_loop): Likewise.
19137 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
19139 (tree_unroll_loops_completely): Likewise.
19140 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
19141 Use an overload instead and export.
19142 (estimated_loop_iterations): Adjust.
19143 (max_loop_iterations): Likewise.
19144 (likely_max_loop_iterations): Likewise.
19145 (estimate_numbers_of_iterations): Take struct function as arg
19147 (loop_exits_before_overflow): Adjust.
19148 (free_numbers_of_iterations_estimates_loop): Use an overload.
19149 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
19150 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
19152 2017-06-19 Richard Biener <rguenther@suse.de>
19155 * ipa-prop.c (find_constructor_constant_at_offset): Handle
19156 RANGE_EXPR conservatively.
19158 2017-06-16 Carl Love <cel@us.ibm.com>
19160 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19161 definitions for vec_float, vec_float2, vec_floato,
19162 vec_floate built-ins.
19163 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
19164 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
19166 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
19167 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
19168 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
19169 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
19170 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
19171 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
19172 vec_floato): Add builtin defines.
19173 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
19174 Update the built-in documentation file for the new built-in
19177 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19179 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
19180 (mthumb): Mark as the negative of -marm.
19182 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19184 * doc/invoke.texi (ARM Options, -mcpu): Document supported
19186 (ARM Options, -mtune): Document that this accepts the same
19187 extension options as -mcpu.
19188 (ARM Options, -mfpu): Document addition of -mfpu=auto.
19190 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19192 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
19193 permitted extensions.
19195 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19197 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
19198 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
19199 (armv8-m.main): Add option +nodsp.
19200 * config/arm/arm-cpu-cdata.h: Regenerated.
19202 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19204 * config/arm/t-fuchsia: New file.
19205 * config.gcc (arm*-*-fuchsia*): Use it.
19207 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19209 * config/arm/t-symbian: Rewrite for new option infrastructure.
19211 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19213 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
19214 (MULTILIB_REQUIRED): Likewise.
19216 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19218 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
19219 (MULTILIB_RESUE): Likewise.
19220 (MULTILIB_MATCHES): Likewise.
19221 (MULTLIB_REQUIRED): Likewise.
19223 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19225 * config/arm/t-rtems: Rewrite for new option framework.
19227 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19229 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
19230 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
19231 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
19232 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
19233 * config/arm/t-multilib: ... here.
19234 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
19235 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
19236 armv7-a and armv8*-a when A-profile libraries have not been built.
19237 * config/arm/t-rmprofile: Rewrite.
19239 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19241 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
19242 with a backslash. Remove the backslash after substituting unescaped
19244 * doc/fragments.texi (MULTILIB_REUSE): Document it.
19246 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19248 * config.gcc: (arm*-*-*): When building a-profile libraries, force
19249 the driver to pass through the default setting of -mfloat-abi.
19250 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
19252 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
19253 (all_feat_combs): New rule.
19254 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
19256 * config/arm/t-aprofile: Rewrite.
19258 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19260 * config/arm/arm.h (FPUTYPE_AUTO): Define.
19261 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
19262 fpu is not specified by the user/command-line.
19263 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
19264 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
19265 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
19266 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
19267 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
19268 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
19270 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19272 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
19273 * config/arm/t-arm-elf: Rewritten.
19275 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19277 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
19278 have some floating-point instructions.
19279 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
19280 (TARGET_MAYBE_HARD_FLOAT): New macro.
19281 * config/arm/arm-builtins.c (arm_init_builtins): Use
19282 TARGET_MAYBE_HARD_FLOAT.
19283 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
19285 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19287 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
19288 (configargs.h): Include it.
19289 (arm_print_hint_for_fpu_option): New function.
19290 (arm_parse_fpu_option): New function.
19291 (candidate_extension): New class.
19292 (arm_canon_for_multilib): New function.
19293 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
19294 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
19295 (ARCH_CANONICAL_SPECS): New macro.
19296 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
19298 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19300 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
19301 are set after handling multilib fragments. Set target_cpu_default2
19304 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19306 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
19308 (arm*-*-*): Set target_cpu_default2 to a quoted string.
19309 * config/arm/parsecpu.awk (check_cpu): Validate any extension
19311 (check_arch): Likewise.
19312 * config/arm/arm.c (arm_configure_build_target): Handle
19313 TARGET_CPU_DEFAULT being a string constant. Scan any feature
19314 options in the default.
19316 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19318 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
19319 when an option is an alias of another.
19320 * config/arm/parsecpu.awk (optalias): New parser token.
19321 (gen_comm_data): Mark non-alias options as such. Emit entries
19322 for extension aliases.
19323 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
19324 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
19325 (armv6kz, armv6zk, armv6t2): Likewise.
19326 (armv7): Make vfpv3-d16 an alias.
19327 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
19329 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
19330 Sort in canonical order.
19331 (armv8-a): Sort in canonical order.
19332 (armv8.1-a, armv8.2-a): Likewise.
19333 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
19335 (cortex-a9): Sort in canonical order.
19336 * config/arm/arm.c (selftests.h): Include it.
19337 (arm_test_cpu_arch_data): New function.
19338 (arm_run_self_tests): New function.
19339 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
19340 (targetm): Move declaration to the end of the file.
19341 * arm-cpu-cdata.h: Regenerated.
19343 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19345 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
19346 call to target_mode_check describing the type of option passed.
19347 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
19348 (arm_target_thumb_only): Use arm_parse_arch_option_name or
19349 arm_parse_cpu_option_name to match parameters against list of
19351 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
19352 arm_arch_core_flags data structure.
19353 * config/arm/arm-cpu_cdata.h: Regenerated.
19355 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19357 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
19359 (arm_print_hint_for_cpu_option): Likewise.
19360 (arm_print_hint_for_arch_option): Likewise.
19361 (arm_parse_cpu_option_name): Likewise.
19362 (arm_parse_arch_option_name): Likewise.
19363 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
19364 of entries in the all_fpus list.
19365 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
19366 (arm_parse_cpu_option_name): Declare.
19367 (arm_parse_arch_option_name): Declare.
19368 (arm_parse_option_features): Declare.
19369 (arm_intialize_isa): Declare.
19370 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
19372 (gen_comm_data): ... here. Make definitions non-static.
19373 * config/arm/arm-cpu-data.h: Regenerated.
19374 * config/arm/arm-cpu-cdata.h: Regenerated.
19376 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19378 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
19379 (cpu_arch_extension): New structure.
19380 (cpu_arch_option, arch_option, cpu_option): New structures.
19381 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
19382 architecture types.
19383 (gen_data): Generate new format data tables.
19384 * config/arm/arm.c (cpu_tune): New structure.
19385 (cpu_option, processors): Delete.
19386 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
19387 (arm_print_hint_for_cpu_option): ... this and ...
19388 (arm_print_hint_for_arch_option): ... this.
19389 (arm_parse_arch_cpu_name): Delete. Replace with ...
19390 (arm_parse_cpu_option_name): ... this and ...
19391 (arm_parse_arch_option_name): ... this.
19392 (arm_unrecognized_feature): Change type of target parameter to
19394 (arm_parse_arch_cpu_features): Delete. Replace with ...
19395 (arm_parse_option_features): ... this.
19396 (arm_configure_build_target): Rework to use new configuration data
19398 (arm_print_tune_info): Rework for new configuration data tables.
19399 * config/arm/arm-cpu-data.h: Regenerated.
19400 * config/arm/arm-cpu.h: Regenerated.
19402 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19404 * Makefile.in (OBJS): Move sbitmap.o from here ...
19405 (OBJS-libcommon): ... to here.
19407 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19409 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
19410 (ISA_ALL_CRYPTO): New macro.
19411 (ISA_ALL_SIMD): New macro
19412 (ISA_ALL_FP): New macro.
19413 * config/arm/arm.c (fpu_bitlist): Update initializer.
19414 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
19416 (arm9e): Add fpu. Add option for nofp
19417 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
19418 (arm926ej-s, arm1026ej-s): Likewise.
19419 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
19420 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
19421 neon-fp16, neon-vfpv4, nofp and nosimd.
19422 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
19423 (cortex-a8): Add fpu. Add option for nofp.
19424 (cortex-a9): Add fpu. Add options for nosimd and nofp.
19425 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
19426 (cortex-r4f): Add fpu.
19427 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
19428 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
19430 (cortex-r8): Likewise.
19431 (cortex-m4): Add fpu. Add option for nofp.
19432 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
19433 (cortex-a17.cortex-a7): Likewise.
19434 (cortex-a32): Add fpu. Add options for crypto and nofp.
19435 (cortex-a35, cortex-a53): Likewise.
19436 (cortex-a57): Add fpu. Add option for crypto.
19437 (cortex-a72, cortex-a73): Likewise.
19438 (exynos-m1): Likewise.
19439 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
19440 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
19441 (cortex-m33): Add fpu. Add option for nofp.
19442 * config/arm/arm-cpu-cdata.h: Regenerated
19443 * config/arm/arm-cpu-data.h: Regenerated.
19445 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19447 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
19448 (armv5te, armv5tej): Likewise.
19449 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
19450 (armv7): Add options fp and vfpv3-d16.
19451 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
19452 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
19454 (armv7ve): Likewise.
19455 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
19456 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
19457 (armv8-a): Add nocrypto option.
19458 (armv8.1-a, armv8.2-a): Likewise.
19459 (armv8-m.main): add options fp, fp.dp and nofp.
19461 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19463 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
19465 (armv8-a+crc): Delete.
19466 (armv8.1-a): Add options simd, crypto and nofp.
19467 (armv8.2-a): Add options fp16, simd, crypto and nofp.
19468 (armv8.2-a+fp16): Delete.
19469 (armv8-m.main): Add option dsp.
19470 (armv8-m.main+dsp): Delete.
19471 (cortex-a8): Add fpu. Add nofp option.
19472 (cortex-a9): Add fpu. Add nofp and nosimd options.
19473 * config/arm/parsecpu.awk (gen_data): Generate option tables and
19474 link to main cpu and architecture data structures.
19475 (gen_comm_data): Only put isa attributes from the main architecture
19477 (option): New statement for architecture and CPU entries.
19478 * arm.c (struct cpu_option): New structure.
19479 (struct processors): Add entry for options.
19480 (arm_unrecognized_feature): New function.
19481 (arm_parse_arch_cpu_name): Ignore any characters after the first
19483 (arm_parse_arch_cpu_feature): New function.
19484 (arm_configure_build_target): Separate out any CPU and architecture
19485 features and parse separately. Don't error out if -mfpu=auto is
19486 used with only an architecture string.
19487 (arm_print_asm_arch_directives): New function.
19488 (arm_file_start): Call it.
19489 * config/arm/arm-cpu-cdata.h: Regenerated.
19490 * config/arm/arm-cpu-data.h: Likewise.
19491 * config/arm/arm-tables.opt: Likewise.
19493 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19495 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
19496 assembler when it is not -mfpu=auto.
19498 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19500 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
19501 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
19502 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
19503 (ASM_CPU_SPEC): Rewrite.
19504 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
19505 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
19506 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
19507 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
19508 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
19509 copied string is NUL-terminated. Also strip any characters prefixed
19511 (arm_rewrite_selected_arch): New function.
19512 (arm_rewrite_march): New function.
19514 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19516 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
19517 (x_arm_cpu_string, x_arm_tune_string): Likewise.
19518 (march, mcpu, mtune): Convert to string-based options.
19519 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
19520 (arm_parse_arch_cpu_name): New function.
19521 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
19522 identify selected architecture or CPU.
19523 (arm_option_save): New function.
19524 (TARGET_OPTION_SAVE): Redefine.
19525 (arm_option_restore): Restore string options.
19526 (arm_option_print): Print string options.
19528 2017-06-16 Martin Sebor <msebor@redhat.com>
19530 PR tree-optimization/80933
19531 PR tree-optimization/80934
19532 * builtins.c (fold_builtin_3): Do not handle bcmp here.
19533 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
19534 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
19535 (gimple_fold_builtin): Call them.
19537 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19539 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
19540 as unlikely; update profile.
19542 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19544 * predict.c (force_edge_cold): Handle declaring edges impossible
19547 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19549 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
19551 (try_unroll_loop_completely): Fix reporting.
19553 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19555 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
19557 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
19560 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
19561 if given a non-constant argument for an intrinsic which requires a
19564 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19566 * profile.c (compare_freqs): New function.
19567 (branch_prob): Sort edge list.
19568 (find_spanning_tree): Assume that the list is priority sorted.
19570 2017-06-16 Richard Biener <rguenther@suse.de>
19572 PR tree-optimization/81090
19573 * passes.def (pass_record_bounds): Remove.
19574 * tree-pass.h (make_pass_record_bounds): Likewise.
19575 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
19576 make_pass_record_bounds): Likewise.
19577 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
19578 not free niter estimates at the beginning but at the end.
19579 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
19581 2017-06-16 Richard Biener <rguenther@suse.de>
19583 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
19584 initializer to workaround ICE in host GCC 4.8.
19586 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19588 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
19590 (clone_inlined_nodes): Update.
19592 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19594 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
19595 prefetch settings, and enable prefetching by default at -O3.
19597 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19599 * config/aarch64/aarch64.c (aarch64_override_options_internal):
19600 Set flag_prefetch_loop_arrays according to tuning data.
19602 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19604 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
19605 New tune structure.
19606 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
19607 [Unrelated to main purpose of the patch] Place the pointer field last
19608 to enable type checking errors when tune structure are wrongly merged.
19609 * config/aarch64/aarch64.c (generic_prefetch_tune,)
19610 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
19611 (thunderx2t99_prefetch_tune): New tune constants.
19612 (tune_params *_tunings): Update all tunings (no functional change).
19613 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
19614 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
19615 from tunings structures.
19617 2017-06-16 Jakub Jelinek <jakub@redhat.com>
19620 * ubsan.c (instrument_null): Add T argument, use it instead
19621 of computing it based on IS_LHS.
19622 (instrument_object_size): Likewise.
19623 (pass_ubsan::execute): Adjust instrument_null and
19624 instrument_object_size callers to pass gimple_get_lhs or
19625 gimple_assign_rhs1 result to it. Use instrument_null instead of
19626 calling get_base_address and instrument_mem_ref. Handle
19627 aggregate call arguments for object-size sanitization.
19629 2017-06-16 Yury Gribov <tetra2005@gmail.com>
19631 PR tree-optimization/81089
19632 * tree-vrp.c (is_masked_range_test): Validate operands of
19635 2017-06-15 Martin Sebor <msebor@redhat.com>
19638 * dumpfile.c (dump_register): Avoid calling memset to initialize
19639 a class with a default ctor.
19640 * gcc.c (struct compiler): Remove const qualification.
19641 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
19642 * hash-table.h: Ditto.
19643 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
19645 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
19646 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
19648 * params.h (struct param_info): Make struct members non-const.
19649 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
19650 with default initialization.
19651 * vec.h (vec_copy_construct, vec_default_construct): New helper
19653 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
19654 with vec_copy_construct.
19655 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
19656 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
19657 * doc/invoke.texi (-Wclass-memaccess): Document.
19659 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19661 * emit-rtl.h (is_leaf): Update comment about local
19662 register allocator.
19664 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19667 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
19668 for a variable to have a section before checking if the section has a
19670 Set section to.persistent if persistent attribute is set.
19671 Warn if .persistent attribute is used on an automatic variable.
19673 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
19675 PR rtl-optimization/80474
19676 * reorg.c (update_block): Do not ignore instructions in a delay slot.
19678 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
19680 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
19683 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
19685 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
19686 (casesi): Emit bounds checking as RTL.
19687 (casesi_internal_mips16_<mode>): Remove bounds checking.
19689 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
19691 * config/xtensa/xtensa.c (xtensa_option_override): Append
19692 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
19693 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
19694 xtensa_doloop_hooks): Define unconditionally.
19695 (xtensa_reorg_loops): Only call reorg_loops in the presence of
19697 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
19698 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
19699 for it in xtensa_option_override.
19700 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
19701 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
19703 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
19705 * doc/cppopts.texi: Document '-' special value to -MF.
19707 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
19709 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
19710 (cortex_a53_fconst): Likewise.
19711 (cortex_a53_fpmul): Likewise.
19712 (cortex_a53_f_load_64): Likewise.
19713 (cortex_a53_f_load_many): Likewise.
19714 (cortex_a53_advsimd_alu): Likewise.
19715 (cortex_a53_advsimd_alu_q): Likewise.
19716 (cortex_a53_advsimd_mul): Likewise.
19717 (cortex_a53_advsimd_mul_q): Likewise.
19718 (fpmac bypass): Add new bypass for fpmac-fpmac case.
19719 Add missing fmul, r2f_cvt and fconst cases.
19721 2017-06-14 Richard Biener <rguenther@suse.de>
19723 PR middle-end/81088
19724 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
19726 (fold_binary_loc): When associating do not treat pre-existing
19727 TREE_OVERFLOW on literal constants as a reason to allow
19728 TREE_OVERFLOW on associated literal constants.
19730 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19732 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
19733 (MASK_FEATURES): New macro.
19734 * config/sparc/sparc.c (sparc_option_override): Remove the special
19735 handling of -mfpu and generalize it to all MASK_FEATURES switches.
19737 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19739 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
19740 a division of 0 if non-call exceptions are enabled.
19742 2017-06-14 Andrew Pinski <apinski@cavium.com>
19743 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19746 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
19747 Improve vector initialization code gen for only variable case.
19749 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19751 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
19753 2017-06-14 Richard Biener <rguenther@suse.de>
19755 PR tree-optimization/81083
19756 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
19759 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19761 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
19762 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
19763 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
19764 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
19765 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
19766 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
19768 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19770 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
19771 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
19773 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19775 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
19777 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19779 * config/rs6000/t-rtems: Don't handle SPE.
19781 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19783 * config/rs6000/t-linux: Don't handle SPE.
19785 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19787 * config/rs6000/eabispe.h: Delete file.
19789 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19791 * config/rs6000/t-spe: Delete file.
19793 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19795 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
19796 (rs6000_legitimate_offset_address_p): Return false for anything in
19797 V2SImode or V2SFmode.
19799 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19801 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
19802 except V2SF and V2SI. Rearrange the vector modes, and add comments.
19803 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
19805 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
19806 (rs6000_legitimate_offset_address_p): Ditto.
19807 (rs6000_emit_move): Ditto.
19808 (rs6000_init_builtins): Remove V4HI_type_node.
19810 2017-06-13 Martin Liska <mliska@suse.cz>
19813 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
19814 (gate_asan): Likewise.
19815 * asan.h (asan_no_sanitize_address_p): Remove the function.
19816 (sanitize_flags_p): New function.
19817 * builtins.def: Fix coding style.
19818 * common.opt: Use renamed enum value.
19819 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
19820 * doc/extend.texi: Document no_sanitize attribute.
19821 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
19822 to SANITIZE_UNDEFINED_NONDEFAULT.
19823 * gcc.c (sanitize_spec_function): Use the renamed enum value.
19824 * gimple-fold.c (optimize_atomic_compare_exchange_p):
19825 Use sanitize_flags_p.
19826 * gimplify.c (gimplify_function_tree): Likewise.
19827 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
19828 * opts.c (parse_no_sanitize_attribute): New function.
19829 (common_handle_option): Use renamed enum value.
19830 * opts.h (parse_no_sanitize_attribute): Declare.
19831 * tree.c (sanitize_flags_p): New function.
19832 * tree.h: Declared here.
19833 * tsan.c: Use sanitize_flags_p.
19834 * ubsan.c (ubsan_expand_null_ifn): Likewise.
19835 (instrument_mem_ref): Likewise.
19836 (instrument_bool_enum_load): Likewise.
19837 (do_ubsan_in_current_function): Remove the function.
19838 (pass_ubsan::execute): Use sanitize_flags_p.
19839 * ubsan.h: Remove do_ubsan_in_current_function
19840 * tree-cfg.c (print_no_sanitize_attr_value): New function.
19841 (dump_function_to_file): Use it here.
19843 2017-06-13 Martin Jambor <mjambor@suse.cz>
19845 PR tree-optimization/80803
19846 PR tree-optimization/81063
19847 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
19848 (propagate_subaccesses_across_link): Enqueue subtree whenever
19849 necessary instead of relying on the caller.
19851 2017-06-13 Martin Jambor <mjambor@suse.cz>
19853 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
19854 that have a first_link.
19855 (sort_and_splice_var_accesses): Do not check first_link before
19857 (subtree_mark_written_and_enqueue): Likewise.
19858 (propagate_all_subaccesses): Likewise and do not stop at first
19859 parent with a first_link.
19861 2017-06-13 Martin Jambor <mjambor@suse.cz>
19863 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
19866 2017-06-13 Yury Gribov <tetra2005@gmail.com>
19868 * match.pd: New pattern.
19870 2017-06-13 Yury Gribov <tetra2005@gmail.com>
19872 * tree-vrp.c (is_masked_range_test): New function.
19873 (register_edge_assert_for): Determine ranges for
19876 2017-06-13 Yury Gribov <tetra2005@gmail.com>
19878 PR tree-optimization/67328
19879 * fold-const.c (maskable_range_p): New function.
19880 (build_range_check): Generate bittests if possible.
19882 2017-06-13 Martin Liska <mliska@suse.cz>
19884 * gimple-pretty-print.c (dump_probability): Add new argument.
19885 (dump_edge_probability): Dump both probability and count.
19886 (dump_gimple_label): Likewise.
19887 (dump_gimple_bb_header): Likewise.
19889 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
19892 * config/avr/avr-devices.c: Fix indentation.
19893 * config/avr/gen-avr-mmcu-specs.c: Dito.
19895 2017-06-13 Richard Biener <rguenther@suse.de>
19897 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
19898 instead get vector type from stmt_info.
19899 (vectorizable_reduction): Adjust. Remove dead code.
19901 2017-06-13 Richard Biener <rguenther@suse.de>
19903 PR middle-end/81065
19904 * fold-const.c (extract_muldiv_1): Remove bogus distribution
19905 case of C * (x * C2 + C3).
19906 (fold_addr_of_array_ref_difference): Properly fold index difference.
19908 2017-06-12 David S. Miller <davem@davemloft.net>
19911 * config/sparc/sparc.md (return expander): Emit frame blockage if
19912 function uses alloca.
19914 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
19916 * combine.c (make_field_assignment): Check len rather than the mode
19917 precision when calling force_to_mode.
19919 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
19921 Support multilibs and devices that see flash in RAM address range.
19924 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
19925 (avr_mcu_t) <flash_pm_offset>: New field.
19926 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
19927 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
19928 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
19929 (AVR_TINY_PM_OFFSET): Remove macro.
19930 * config/avr/avr.opt (-mshort-calls): New option.
19931 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
19932 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
19933 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
19934 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
19935 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
19936 instead of avr_arch->have_jmp_call.
19937 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
19938 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
19939 avr_arch->flash_pm_offset to define.
19940 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
19941 new field flash_pm_offset. Add entry for avrxmega3.
19942 (avr_texinfo): Add entry for avrxmega3.
19943 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
19944 attiny212, attiny214,
19945 attiny412, attiny414, attiny416, attiny417,
19946 attiny814, attiny816, attiny817,
19947 attiny1614, attiny1616, attiny1617,
19948 attiny3214, attiny3216, attiny3217.
19949 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
19950 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
19951 (avr_print_operand_address) [AVR_TINY]: Same.
19952 (avr_asm_init_sections) <readonly_data_section>: Only patch
19953 callback if avr_arch->flash_pm_offset = 0.
19954 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
19955 for rodata if avr_arch->flash_pm_offset != 0.
19956 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
19957 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
19958 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
19959 (m_options): Append opt_rcall.
19960 (m_dirnames): Append dir_rcall.
19961 * config/avr/t-multilib: Regenerate.
19963 * configure.ac [target=avr]: Check whether avrxmega3 default
19964 linker description file works as needed.
19965 * configure: Regenerate.
19966 * doc/avr-mmcu.texi: Regenerate.
19967 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
19968 <__AVR_ARCH__>: Document avrxmega3 and 103.
19969 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
19970 <__AVR_SHORT_CALLS__>: Document it.
19971 <__AVR_PM_BASE_ADDRESS__>: Document it.
19972 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
19973 (AVR Variable Attributes) <progmem>: Document this is
19974 not needed for avrxmega3.
19975 (AVR Named Address Spaces) <__flash>: Dito.
19977 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
19979 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
19981 2017-06-12 Doug Rupp <rupp@adacore.com>
19983 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
19984 Append vxworks-stdint.h to the tm_file list.
19985 * config/vxworks-stdint.h: New file.
19987 2017-06-12 Martin Liska <mliska@suse.cz>
19989 PR tree-optimization/81041
19990 * tree-profile.c (gimple_gen_ic_func_profiler):
19991 Create an extra BB in profile-generate
19992 (gimple_gen_time_profiler): Likewise.
19994 2017-06-12 Jakub Jelinek <jakub@redhat.com>
19996 PR tree-optimization/81003
19997 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
19998 (update_range_test): Use it instead of force_gimple_operand_gsi.
20000 2017-06-12 Richard Biener <rguenther@suse.de>
20002 PR tree-optimization/81053
20003 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
20004 with backedge value not defined in loop. Simplify def stmt
20007 2017-06-11 Tom de Vries <tom@codesourcery.com>
20010 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
20012 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
20013 nvptx_cannot_force_const_mem.
20015 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20017 * opts.c (finish_options): Move test for flag_split_stack after
20018 it has been initialized.
20020 2017-06-11 Jason Merrill <jason@redhat.com>
20022 * tree.h (id_equal): New.
20023 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
20024 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
20025 instead of strcmp of IDENTIFIER_POINTER.
20027 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20029 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
20030 (mark_all_inlined_calls_cdtor): Fix formating.
20031 (inline_transform): Rescale profile before inlining.
20033 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20035 * cgraph.h (cgraph_edge::clone): Update prototype.
20036 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
20037 (cgraph_node::create_clone): Update.
20038 (cgraph_node::create_version_clone): Update.
20039 * tree-inline.c (copy_bb): Update.
20040 (expand_call_inline): Update.
20042 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20044 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
20045 factored out from ...
20046 (rs6000_emit_prologue): ... here.
20048 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20050 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
20051 factored out from ...
20052 (rs6000_emit_prologue): ... here.
20054 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20056 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
20058 (handle_missing_profiles): Fix computation of tp_first_run.
20059 (counts_to_freqs): Do not touch freqs when count is 0.
20061 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20063 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
20066 2017-06-10 Tom de Vries <tom@codesourcery.com>
20068 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
20069 attributes): Document signal effective target.
20071 2017-06-10 Tom de Vries <tom@codesourcery.com>
20073 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
20074 Document effective target stack_size.
20076 2017-06-09 David Malcolm <dmalcolm@redhat.com>
20078 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
20079 to the edit_context if they can be auto-applied.
20081 2017-06-9 Ian Lance Taylor <iant@golang.org>
20083 * opts.c (finish_options): If -fsplit-stack, disable implicit
20084 -forder-blocks-and-partition.
20085 * doc/invoke.texi (Optimize Options): Document that when using
20086 -fsplit-stack -forder-blocks-and-partition is not implicitly
20089 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20091 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
20092 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
20093 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
20094 * builtins.def (abort, trap, unreachable): Declare cold.
20095 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
20096 * tree-core.h (ECF_COLD): New.
20097 * tree.c (set_call_expr_flags): Handle ECF_COLD.
20098 (build_common_builtin_nodes): Mark unreachable and abort as cold.
20100 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20102 * predict.c (unlikely_executed_stmt_p): Cleanup.
20104 2017-06-09 Richard Biener <rguenther@suse.de>
20106 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
20107 model if the ref is always written to.
20109 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20111 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
20113 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20115 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
20118 2017-06-09 Tom de Vries <tom@codesourcery.com>
20121 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
20122 "target cannot support label values" when encountering LABEL_REF.
20124 2017-06-09 Martin Liska <mliska@suse.cz>
20126 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
20127 (gimple_gen_ic_func_profiler): Emit direct comparison
20128 of __gcov_indirect_call_callee with NULL.
20129 (gimple_gen_time_profiler): Change probability from
20130 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
20132 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20134 * profile.c (edge_gcov_counts): Turn to pointer.
20135 (compute_branch_probabilities, compute_branch_probabilities): Update.
20136 (branch_prob): Do not clear edge_gcov_count.
20137 * profile.h (edge_gcov_counts): Turn to pointer.
20138 (edge_gcov_count): Update.
20140 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20142 * gimple.h (gimple_check_failed): Mark cold.
20144 2017-06-09 Richard Biener <rguenther@suse.de>
20146 PR tree-optimization/66623
20147 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
20148 refactor check_reduction into two parts, properly computing
20149 whether we have to check reduction validity for outer loop
20152 2017-06-09 Richard Biener <rguenther@suse.de>
20154 PR tree-optimization/79483
20155 * graphite-scop-detection.c (order): New global.
20156 (get_order): Compute bb to order mapping that satisfies code
20157 generation constraints.
20158 (cmp_pbbs): New helper.
20159 (build_scops): Start domwalk at entry block, sort generated
20162 2017-06-09 Richard Biener <rguenther@suse.de>
20164 PR middle-end/81007
20165 * ipa-polymorphic-call.c
20166 (ipa_polymorphic_call_context::restrict_to_inner_class):
20167 Skip FIELD_DECLs with error_mark_node type.
20168 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
20171 2017-06-09 Martin Liska <mliska@suse.cz>
20173 * predict.c (struct branch_predictor): New struct.
20174 (test_prediction_value_range): New test.
20175 (predict_c_tests): New function.
20176 * selftest-run-tests.c (selftest::run_tests): Run the function.
20177 * selftest.h: Declare new tests.
20179 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
20182 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
20183 gen_add3_insn did not fail.
20184 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
20185 r0, construct that number in a temporary reg and add that reg to r0.
20186 If asked to put the result in r0 as well, fail.
20188 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
20190 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
20191 for early expansion of vec_eqv.
20193 2017-06-08 Jakub Jelinek <jakub@redhat.com>
20195 PR middle-end/81005
20196 * ubsan.c (instrument_null): Avoid pointless code temporary.
20197 (pass_ubsan::execute): Instrument aggregate arguments of calls.
20199 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
20203 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
20206 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
20209 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20211 * predict.c (unlikely_executed_edge_p): Move ahead.
20212 (probably_never_executed_edge_p): Use it.
20214 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20216 PR middle-end/79988
20217 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
20218 gimple_call_builtin_p call.
20220 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20222 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
20223 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
20224 rtl_check_failed_type2, rtl_check_failed_code1,
20225 rtl_check_failed_code2, rtl_check_failed_code_mode,
20226 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
20227 rtvec_check_failed_bounds, rtl_check_failed_flag,
20228 _fatal_insn_not_found, _fatal_insn): Likewise.
20229 * tree.h (tree_contains_struct_check_failed,
20230 tree_check_failed, tree_not_check_failed,
20231 tree_class_check_failed, tree_range_check_failed,
20232 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
20233 tree_vec_elt_check_failed, phi_node_elt_check_failed,
20234 tree_operand_check_failed, omp_clause_check_failed,
20235 omp_clause_operand_check_failed, omp_clause_range_check_failed):
20238 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20240 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
20241 flag_branch_probabilities.
20242 * ipa-inline.c (edge_badness): Likewise.
20243 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
20244 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
20245 * predict.c (maybe_hot_frequency_p): Likewise.
20246 (probably_never_executed): Likewise.
20247 * sched-ebb.c (schedule_ebbs): Likewise.
20248 * sched-rgn.c (find_single_block_region): Likewise.
20249 * tracer.c (tail_duplicate): Likewise.
20251 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20253 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
20254 longer requires x_flag_profile_use.
20256 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20258 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
20259 instead of flag_reorder_blocks_and_partition.
20260 * dbxout.c (dbxout_function_end): Likewise.
20261 * dwarf2out.c (gen_subprogram_die): Likewise.
20262 * haifa-sched.c (sched_create_recovery_edges): Likewise.
20263 * hw-doloop.c (reorg_loops): Likewise.
20264 * varasm.c (assemble_start_function,
20265 assemble_end_function): Likewise.
20266 (decide_function_section): Do not check for
20267 flag_reorder_blocks_and_partition.
20269 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20271 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20273 (chkp_get_hard_register_fake_addr_expr): Ditto.
20274 (chkp_build_addr_expr): Add check for hard reg case.
20275 (chkp_parse_array_and_component_ref): Ditto.
20276 (chkp_find_bounds_1): Ditto.
20277 (chkp_process_stmt): Don't generate bounds store for
20280 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20282 * predict.c (maybe_hot_bb_p): Do not check profile status.
20283 (maybe_hot_edge_p): Likewise.
20284 (probably_never_executed): Check for zero counts even if profile
20286 (unlikely_executed_edge_p): New function.
20287 (unlikely_executed_stmt_p): New function.
20288 (unlikely_executed_bb_p): New function.
20289 (set_even_probabilities): Use unlikely predicates.
20290 (combine_predictions_for_bb): Likewise.
20291 (predict_paths_for_bb): Likewise.
20292 (predict_paths_leading_to_edge): Likewise.
20293 (determine_unlikely_bbs): New function.
20294 (estimate_bb_frequencies): Use it.
20295 (compute_function_frequency): Use zero counts even if profile is
20297 * profile-count.h: Fix typo.
20299 2017-08-08 Julia Koval <julia.koval@intel.com>
20301 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
20302 _mm512_mask_cvtsepi16_storeu_epi8,
20303 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
20304 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
20305 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
20306 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
20307 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
20308 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
20309 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
20310 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
20311 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
20312 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
20313 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
20314 __builtin_ia32_pmovuswb256mem_mask,
20315 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
20316 __builtin_ia32_pmovwb512mem_mask): New builtins.
20318 2017-08-08 Julia Koval <julia.koval@intel.com>
20320 PR target/73350,80862
20321 * config/i386/subst.md (round): Fix round pattern.
20322 * config/i386/i386.c (ix86_erase_embedded_rounding):
20323 Fix erasing rounding for the fixed pattern.
20325 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20327 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
20329 2017-06-08 Martin Liska <mliska@suse.cz>
20331 PR gcov-profile/80911
20332 * gcov.c (block_info::block_info): New constructor.
20334 2017-06-07 Carl Love <cel@us.ibm.com>
20336 * config/rs6000/rs6000-c: The return type of the following
20337 built-in functions was implemented as int not long long. Fix sign
20338 of return value for the unsigned version of vec_mulo and vec_mule.
20339 vector unsigned long long vec_bperm (vector unsigned long long,
20340 vector unsigned char)
20341 vector signed long long vec_mule (vector signed int,
20343 vector unsigned long long vec_mule (vector unsigned int,
20344 vector unsigned int)
20345 vector signed long long vec_mulo (vector signed int,
20347 vector unsigned long long vec_mulo (vector unsigned int,
20348 vector unsigned int)
20349 * doc/extend.texi: Fix the documentation for the built-in
20352 2017-06-07 Carl Love <cel@us.ibm.com>
20355 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
20358 2017-06-07 Carl Love <cel@us.ibm.com>
20360 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
20361 support, Generate doublehv for signed int/float for BE case only.
20363 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
20365 * doc/invoke.texi (mcx16): Rewrite.
20367 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20369 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
20370 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
20371 *mov<mode>_softfloat, and an anonymous splitter): Use
20372 nonimmediate_operand instead of rs6000_nonimmediate_operand.
20374 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20376 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
20378 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
20379 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
20380 (rs6000_debug_reg_global): Adjust.
20381 (rs6000_init_hard_regno_mode_ok): Adjust.
20382 (rs6000_dbx_register_number): Adjust.
20383 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
20384 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
20385 Remove SPE_ACC and SPEFSCR.
20386 (REG_ALLOC_ORDER): Ditto.
20387 (FRAME_POINTER_REGNUM): Change to 111.
20388 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
20389 (REG_CLASS_NAMES): Ditto.
20390 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
20391 (REGISTER_NAMES): Ditto.
20392 (ADDITIONAL_REG_NAMES): Ditto.
20393 (rs6000_reg_names): Ditto.
20394 * config/rs6000/rs6000.md: Renumber some register number
20397 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20399 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
20401 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
20402 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
20404 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
20405 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
20406 Delete the SPE high registers.
20407 (REG_ALLOC_ORDER): Ditto.
20408 (enum reg_class): Remove SPE_HIGH_REGS.
20409 (REG_CLASS_NAMES): Ditto.
20410 (REG_CLASS_CONTENTS): Delete the SPE high registers.
20411 (REGISTER_NAMES): Ditto.
20412 (rs6000_reg_names): Ditto.
20413 * doc/tm.texi.in: Remove SPE as example.
20414 * doc/tm.texi: Regenerate.
20416 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20418 * config/rs6000/8540.md (ppc8540_brinc): Delete.
20419 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
20420 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
20421 * config/rs6000/rs6000.md (type): Remove "brinc".
20423 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20425 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
20426 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
20427 * config/rs6000/linuxspe.h: Delete file.
20428 * config/rs6000/rs6000.md: Don't include spe.md.
20429 * config/rs6000/spe.h: Delete file.
20430 * config/rs6000/spe.md: Delete file.
20431 * config/rs6000/t-rs6000: Remove spe.md.
20433 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20435 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
20436 (reg_or_none500mem_operand): Delete.
20437 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
20438 instead of reg_or_none500mem_operand.
20440 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20442 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
20443 handling of SPE flags.
20444 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
20446 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20448 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
20450 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
20451 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
20452 paired_divv2sf3): Similar.
20453 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
20454 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
20455 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
20457 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
20458 Rename the paired_* instruction patterns.
20459 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
20461 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
20462 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
20463 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
20464 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
20465 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
20466 PAIRED_VECTOR_MODE.
20467 (struct machine_function): Delete field spe_insn_chain_scanned_p.
20468 (spe_func_has_64bit_regs_p): Delete.
20469 (spe_expand_predicate_builtin): Delete.
20470 (spe_expand_evsel_builtin): Delete.
20471 (TARGET_DWARF_REGISTER_SPAN): Do not define.
20472 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
20473 (invalid_e500_subreg): Delete.
20474 (rs6000_legitimize_address): Always force_reg op2 as well, for
20475 paired single memory accesses.
20476 (rs6000_member_type_forces_blk): Delete.
20477 (rs6000_spe_function_arg): Delete.
20478 (rs6000_expand_unop_builtin): Delete SPE handling.
20479 (rs6000_expand_binop_builtin): Ditto.
20480 (spe_expand_stv_builtin): Delete.
20481 (bdesc_2arg_spe): Delete.
20482 (spe_expand_builtin): Delete.
20483 (spe_expand_predicate_builtin): Delete.
20484 (spe_expand_evsel_builtin): Delete.
20485 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
20486 (spe_init_builtins): Delete.
20487 (spe_func_has_64bit_regs_p): Delete.
20488 (savres_routine_name): Delete "info" parameter. Adjust callers.
20489 (rs6000_emit_stack_reset): Ditto.
20490 (rs6000_dwarf_register_span): Delete.
20491 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
20492 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
20493 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
20495 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
20497 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
20498 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
20499 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
20500 mulv2sf3, divv2sf3): Delete expanders.
20502 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20504 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
20506 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20508 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
20509 * config/rs6000/rs6000.c: Ditto.
20511 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20513 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
20514 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
20515 comparison_operator.
20517 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20519 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
20520 * config/rs6000/rs6000.opt: Ditto.
20521 * config/rs6000/t-rtems: Ditto.
20523 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20525 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
20526 TARGET_E500_SINGLE by 0, simplify.
20527 * config/rs6000/rs6000.c: Ditto.
20528 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
20529 (spe_build_register_parallel): Delete.
20530 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
20531 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
20532 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
20533 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
20534 (E500_CONVERT): Delete.
20535 * config/rs6000/spe.md: Remove many patterns and all define_constants.
20537 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20539 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
20540 * config/rs6000/dfp.md: Ditto.
20541 (negdd2, *negdd2_fpr): Merge.
20542 (absdd2, *absdd2_fpr): Merge.
20543 (negtd2, *negtd2_fpr): Merge.
20544 (abstd2, *abstd2_fpr): Merge.
20545 * config/rs6000/e500.h: Delete file.
20546 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
20547 TARGET_FPRS by 1 and simplify.
20548 * config/rs6000/rs6000-c.c: Ditto.
20549 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
20550 TARGET_DF_SPE by 0.
20551 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
20553 * config/rs6000/rs6000.md: Ditto.
20554 (floatdidf2, *floatdidf2_fpr): Merge.
20555 (move_from_CR_gt_bit): Delete.
20556 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
20557 (E500_CR_IOR_COMPARE): Delete.
20558 (All patterns that require !TARGET_FPRS): Delete.
20559 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
20561 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20563 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
20565 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20567 * graphds.c (add_edge): Intitialize edge's attached data.
20568 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
20569 pointer parameter. Call pointed function on each edge during
20570 graph traversing. Skip traversing the edge when the function
20572 (graphds_dfs, graphds_scc): Ditto.
20573 (for_each_edge): New parameter. Pass the new parameter to callback
20575 * graphds.h (skip_edge_callback): New function pointer type.
20576 (graphds_dfs, graphds_scc): New function pointer parameter.
20577 (graphds_edge_callback, for_each_edge): New parameter.
20579 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20581 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
20582 out code checking if runtime alias check is possible to below ...
20583 Call the new function.
20584 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
20585 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
20587 2017-06-07 Marek Polacek <polacek@redhat.com>
20590 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
20591 TYPE_OVERFLOW_WRAPS check.
20593 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20595 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
20596 if versioning is required.
20597 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
20598 peeling with the check for versioning.
20600 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20602 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
20603 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
20604 Set true to new parameter if new ssa variable is defined.
20605 (vect_gen_vector_loop_niters): Refactor. Set range information
20606 for the new vector loop bound variable.
20607 (vect_do_peeling): Ditto.
20609 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20611 * tree-affine.c (ssa.h): Include header file.
20612 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
20613 has wrapping overflow behavior.
20615 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20617 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
20619 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20621 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
20622 (tree_to_aff_combination): ... here.
20624 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20626 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
20627 reg_pressure model function.
20628 (ivopts_global_cost_for_size): Delete.
20629 (determine_set_costs, iv_ca_recount_cost): Call new model function
20630 ivopts_estimate_reg_pressure.
20632 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20634 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
20635 expensive than udiv. Remove floating point cases from mod.
20637 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20639 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
20640 Increase idiv cost.
20642 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20644 * config/aarch64/aarch64.md
20645 (copysignsf3): Fix mask generation.
20647 2017-06-07 Jakub Jelinek <jakub@redhat.com>
20649 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
20651 (class dump_manager): Add register_dumps method.
20652 * dumpfile.c: Include langhooks.h.
20653 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
20654 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
20655 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
20656 (dump_manager::dump_register): Start with 512 entries instead of 32.
20657 (dump_manager::register_dumps): New method.
20658 * toplev.c (general_init): Instead of invoking register_dumps
20659 langhook, invoke register_dumps method on the dump manager.
20660 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
20663 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
20665 * doc/md.texi: Clarify the restrictions on a define_insn condition.
20666 Say that # requires an associated define_split to exist, and that
20667 the define_split must be suitable for use after register allocation.
20669 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20671 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
20672 (compute_outgoing_frequencies): Also initialize zero counts.
20673 (find_many_sub_basic_blocks): Do not produce uninitialized profile
20674 around loops; preserve more of profile when nothing changes.
20676 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
20678 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
20680 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
20681 * config/arm/arm-cpu-cdata.h: Regenerate.
20682 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
20683 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
20684 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
20685 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
20686 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
20688 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
20690 * config/arm/t-rmprofile: Likewise.
20691 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
20693 2017-06-06 David S. Miller <davem@davemloft.net>
20696 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
20697 blockage if function uses alloca.
20699 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20701 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
20702 New "uid" fields to hold pretty-print IDs of group and ref.
20703 Memory references are now identified as <group_id>:<ref_id>
20704 instead of using [random] addresses.
20705 (dump_mem_details): Simplify, no functional change.
20706 (dump_mem_ref): Simplify and make output more concise.
20707 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
20708 (find_or_create_group): Initialize group uid.
20709 (record_ref): Initialize ref uid. Improve debug output.
20710 (prune_group_by_reuse, should_issue_prefetch_p,)
20711 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
20712 (mark_nontemporal_store, determine_loop_nest_reuse):
20713 Improve debug output.
20715 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20717 * dbgcnt.def (prefetch): New debug counter.
20718 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
20719 (schedule_prefetches): Stop issueing prefetches if debug counter
20722 2017-06-06 Tom de Vries <tom@codesourcery.com>
20724 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
20725 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
20727 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20729 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
20730 Use aarch64_reg_or_zero predicate for operand 4.
20731 (aarch64_compare_and_swap<mode> define_insn_and_split):
20732 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
20733 (aarch64_store_exclusive<mode>): Likewise for operand 2.
20735 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
20737 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
20738 (arm_compute_save_core_reg_mask): This.
20739 (thumb1_compute_save_reg_mask): Rename into ...
20740 (thumb1_compute_save_core_reg_mask): This.
20741 (arm_compute_save_reg0_reg12_mask): Adapt comment.
20742 (arm_compute_frame_layout): Likewise.
20744 2017-06-06 Richard Biener <rguenther@suse.de>
20746 PR tree-optimization/80974
20747 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
20748 keep or clear leaders SSA info.
20750 2017-06-06 Tom de Vries <tom@codesourcery.com>
20752 * config/nvptx/nvptx.c (split_mode_p): New function.
20753 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
20755 2017-06-06 Tom de Vries <tom@codesourcery.com>
20757 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
20759 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20762 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
20765 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20767 * shrink-wrap.c (handle_simple_exit): Update profile.
20768 (try_shrink_wrapping): Upate profile.
20770 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20772 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
20773 (tree_guess_outgoing_edge_probabilities): New.
20774 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
20775 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
20777 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20779 * ipa-split.c (split_function): Initialize return bb profile.
20781 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20783 * profile.c (compute_branch_probabilities): Also initialize
20784 EXIT_BLOCK profile.
20786 2017-06-06 Richard Biener <rguenther@suse.de>
20788 PR tree-optimization/80928
20789 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
20790 (vect_analyze_loop_operations): Properly guard analysis for
20792 (vect_transform_loop): Likewise.
20793 (vect_analyze_loop_2): Also reset SLP type on PHIs.
20794 (vect_model_induction_cost): Do not cost for pure SLP.
20795 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
20796 of induction in inner loop vectorization.
20797 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
20798 (vect_get_and_check_slp_defs): Handle vect_induction_def.
20799 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
20801 (vect_analyze_slp_cost_1): Cost induction.
20802 (vect_detect_hybrid_slp_stmts): Handle PHIs.
20803 (vect_get_slp_vect_defs): Likewise.
20804 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
20805 (vect_transform_stmt): Handle SLP reductions.
20806 * tree-vectorizer.h (vectorizable_induction): Adjust.
20808 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
20810 * config/rs6000/rs6000.c (make_resolver_func): Update
20811 init_lowered_empty_function call.
20813 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
20815 * doc/invoke.texi: Document the -fprofile-abs-path option.
20816 * common.opt (fprofile-abs-path): New option.
20817 * gcov-io.h (gcov_write_filename): Declare.
20818 * gcov-io.c (gcov_write_filename): New function.
20819 * coverage.c (coverage_begin_function): Use gcov_write_filename.
20820 * profile.c (output_location): Likewise.
20822 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20824 * shring-wrap.c: Revert accidental commit.
20826 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
20828 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
20830 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20832 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
20834 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
20835 profile in callgraph edge.
20836 * profile-count.h (apply_probability): If THIS is 0, then result is 0
20837 (apply_scale): Likewise.
20838 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
20839 Also scale profile when inlining function with zero profile.
20840 (initialize_cfun): Update exit block profile even when it is zero.
20841 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
20842 when profile is read.
20844 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
20846 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
20847 (CLONE_*): New constants to define the processors we can generate
20848 code for with the target_clone attribute.
20849 (rs6000_clone_map): New array to identify which clone processors
20850 the current program is running on.
20851 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
20852 target_clone attribute.
20853 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
20854 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
20855 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
20856 (cpu_expand_builtin): Add support for target_clone attribute.
20857 (rs6000_valid_attribute_p): Allow "default" attribute.
20858 (get_decl_name): New debug function to simplify printing the
20859 current function name in debugging statements.
20860 (rs6000_clone_priority): New functions to support the target_clone
20861 attribute, and be able to generate code to switch between ISA 2.05
20862 through ISA 3.0 (power6 through power9).
20863 (rs6000_compare_version_priority): Likewise.
20864 (rs6000_get_function_versions_dispatcher): Likewise.
20865 (make_resolver_func): Likewise.
20866 (add_condition_to_bb): Likewise.
20867 (dispatch_function_versions): Likewise.
20868 (rs6000_generate_version_dispatcher_body): Likewise.
20869 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
20870 (fusion_gpr_load_p): Fix a spacing issue.
20871 * doc/extend.texi (Common Function Attributes): Document that the
20872 PowerPC supports the target_clone attribute.
20874 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
20876 * config/arm/arm.h: explain F symbol found in description of ARM
20877 register allocation in its legend.
20879 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20881 * config/mips/frame-header-opt.c: Include profile-count.h.
20882 * config/riscv/riscv.c: Include profile-count.h
20884 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20886 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
20888 (sm_set_flag_if_changed): Add bbs field.
20889 (execute_sm_if_changed_flag_set): Pass BBS.
20890 (execute_sm): Update.
20892 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20894 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
20897 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20899 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
20900 (peephole2): New peephole2 to emit the above.
20901 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
20903 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20905 * config/aarch64/aarch64.c (define_peephole2 above
20906 *sub_<shift>_<mode>): New peephole.
20908 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
20910 * config/i386/i386.c (make_resolver_func): Update.
20911 * Makefile.in: Add profile-count.h and profile-count.o
20912 * auto-profile.c (afdo_indirect_call): Update to new API.
20913 (afdo_set_bb_count): Update.
20914 (afdo_propagate_edge): Update.
20915 (afdo_propagate_circuit): Update.
20916 (afdo_calculate_branch_prob): Update.
20917 (afdo_annotate_cfg): Update.
20918 * basic-block.h: Include profile-count.h
20919 (struct edge_def): Turn count to profile_count.
20920 (struct basic_block_def): Likewie.
20921 (REG_BR_PROB_BASE): Move to profile-count.h
20922 (RDIV): Move to profile-count.h
20923 * bb-reorder.c (max_entry_count): Turn to profile_count.
20924 (find_traces): Update.
20925 (rotate_loop):Update.
20926 (connect_traces):Update.
20927 (sanitize_hot_paths):Update.
20928 * bt-load.c (migrate_btr_defs): Update.
20929 * cfg.c (RDIV): Remove.
20930 (init_flow): Use alloc_block.
20931 (alloc_block): Uninitialize count.
20932 (unchecked_make_edge): Uninitialize count.
20933 (check_bb_profile): Update.
20934 (dump_edge_info): Update.
20935 (dump_bb_info): Update.
20936 (update_bb_profile_for_threading): Update.
20937 (scale_bbs_frequencies_int): Update.
20938 (scale_bbs_frequencies_gcov_type): Update.
20939 (scale_bbs_frequencies_profile_count): New.
20940 * cfg.h (update_bb_profile_for_threading): Update.
20941 (scale_bbs_frequencies_profile_count): Declare.
20942 * cfgbuild.c (compute_outgoing_frequencies): Update.
20943 (find_many_sub_basic_blocks): Update.
20944 * cfgcleanup.c (try_forward_edges): Update.
20945 (try_crossjump_to_edge): Update.
20946 * cfgexpand.c (expand_gimple_tailcall): Update.
20947 (construct_exit_block): Update.
20948 * cfghooks.c (verify_flow_info): Update.
20949 (dump_bb_for_graph): Update.
20950 (split_edge): Update.
20951 (make_forwarder_block): Update.
20952 (duplicate_block): Update.
20953 (account_profile_record): Update.
20954 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
20955 (get_estimated_loop_iterations): Update.
20956 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
20957 (single_likely_exit): Update.
20958 * cfgloopmanip.c (scale_loop_profile): Update.
20960 (set_zero_probability): Update.
20961 (lv_adjust_loop_entry_edge): Update.
20962 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
20963 (purge_dead_edges): Update.
20964 (rtl_account_profile_record): Update.
20965 * cgraph.c (cgraph_node::create): Uninitialize count.
20966 (symbol_table::create_edge): Uninitialize count.
20967 (cgraph_update_edges_for_call_stmt_node): Update.
20968 (cgraph_edge::dump_edge_flags): Update.
20969 (cgraph_node::dump): Update.
20970 (cgraph_edge::maybe_hot_p): Update.
20971 * cgraph.h: Include profile-count.h
20972 (create_clone), create_edge, create_indirect_edge): Update.
20973 (cgraph_node): Turn count to profile_count.
20974 (cgraph_edge0: Likewise.
20975 (make_speculative, clone): Update.
20976 (create_edge): Update.
20977 (init_lowered_empty_function): Update.
20978 * cgraphclones.c (cgraph_edge::clone): Update.
20979 (duplicate_thunk_for_node): Update.
20980 (cgraph_node::create_clone): Update.
20981 * cgraphunit.c (cgraph_node::analyze): Update.
20982 (cgraph_node::expand_thunk): Update.
20983 * final.c (dump_basic_block_info): Update.
20984 * gimple-streamer-in.c (input_bb): Update.
20985 * gimple-streamer-out.c (output_bb): Update.
20986 * graphite.c (print_global_statistics): Update.
20987 (print_graphite_scop_statistics): Update.
20988 * hsa-brig.c: Include basic-block.h.
20989 * hsa-dump.c: Include basic-block.h.
20990 * hsa-gen.c (T sum_slice): Update.
20991 (convert_switch_statements):Update.
20992 * hsa-regalloc.c: Include basic-block.h.
20993 * ipa-chkp.c (chkp_produce_thunks): Update.
20994 * ipa-cp.c (struct caller_statistics): Update.
20995 (init_caller_stats): Update.
20996 (gather_caller_stats): Update.
20997 (ipcp_cloning_candidate_p): Update.
20998 (good_cloning_opportunity_p): Update.
20999 (get_info_about_necessary_edges): Update.
21000 (dump_profile_updates): Update.
21001 (update_profiling_info): Update.
21002 (update_specialized_profile): Update.
21003 (perhaps_add_new_callers): Update.
21004 (decide_about_value): Update.
21005 (ipa_cp_c_finalize): Update.
21006 * ipa-devirt.c (struct odr_type_warn_count): Update.
21007 (struct decl_warn_count): Update.
21008 (struct final_warning_record): Update.
21009 (possible_polymorphic_call_targets): Update.
21010 (ipa_devirt): Update.
21011 * ipa-fnsummary.c (redirect_to_unreachable): Update.
21012 * ipa-icf.c (sem_function::merge): Update.
21013 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
21014 * ipa-inline.c (compute_uninlined_call_time): Update.
21015 (compute_inlined_call_time): Update.
21016 (want_inline_small_function_p): Update.
21017 (want_inline_self_recursive_call_p): Update.
21018 (edge_badness): Update.
21019 (lookup_recursive_calls): Update.
21020 (recursive_inlining): Update.
21021 (inline_small_functions): Update.
21022 (dump_overall_stats): Update.
21023 (dump_inline_stats): Update.
21024 * ipa-profile.c (ipa_profile_generate_summary): Update.
21025 (ipa_propagate_frequency): Update.
21026 (ipa_profile): Update.
21027 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
21028 * ipa-utils.c (ipa_merge_profiles): Update.
21029 * loop-doloop.c (doloop_modify): Update.
21030 * loop-unroll.c (report_unroll): Update.
21031 (unroll_loop_runtime_iterations): Update.
21032 * lto-cgraph.c (lto_output_edge): Update.
21033 (lto_output_node): Update.
21034 (input_node): Update.
21035 (input_edge): Update.
21036 (merge_profile_summaries): Update.
21037 * lto-streamer-in.c (input_cfg): Update.
21038 * lto-streamer-out.c (output_cfg): Update.
21039 * mcf.c (create_fixup_graph): Update.
21040 (adjust_cfg_counts): Update.
21041 (sum_edge_counts): Update.
21042 * modulo-sched.c (sms_schedule): Update.
21043 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
21044 * predict.c (maybe_hot_count_p): Update.
21045 (probably_never_executed): Update.
21046 (dump_prediction): Update.
21047 (combine_predictions_for_bb): Update.
21048 (propagate_freq): Update.
21049 (handle_missing_profiles): Update.
21050 (counts_to_freqs): Update.
21051 (rebuild_frequencies): Update.
21052 (force_edge_cold): Update.
21053 * predict.h: Include profile-count.h
21054 (maybe_hot_count_p, counts_to_freqs): UPdate.
21055 * print-rtl-function.c: Do not include cfg.h
21056 * print-rtl.c: Include basic-block.h
21057 * profile-count.c: New file.
21058 * profile-count.h: New file.
21059 * profile.c (is_edge_inconsistent): Update.
21060 (correct_negative_edge_counts): Update.
21061 (is_inconsistent): Update.
21062 (set_bb_counts): Update.
21063 (read_profile_edge_counts): Update.
21064 (compute_frequency_overlap): Update.
21065 (compute_branch_probabilities): Update; Initialize and deinitialize
21067 (branch_prob): Update.
21068 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
21069 (edge_gcov_count): New.
21070 (bb_gcov_count): New.
21071 * shrink-wrap.c (try_shrink_wrapping): Update.
21072 * tracer.c (better_p): Update.
21073 * trans-mem.c (expand_transaction): Update.
21074 (ipa_tm_insert_irr_call): Update.
21075 (ipa_tm_insert_gettmclone_call): Update.
21076 * tree-call-cdce.c: Update.
21077 * tree-cfg.c (gimple_duplicate_sese_region): Update.
21078 (gimple_duplicate_sese_tail): Update.
21079 (gimple_account_profile_record): Update.
21080 (execute_fixup_cfg): Update.
21081 * tree-inline.c (copy_bb): Update.
21082 (copy_edges_for_bb): Update.
21083 (initialize_cfun): Update.
21084 (freqs_to_counts): Update.
21085 (copy_cfg_body): Update.
21086 (expand_call_inline): Update.
21087 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
21088 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
21089 (try_unroll_loop_completely): Update.
21090 (try_peel_loop): Update.
21091 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
21092 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
21093 * tree-ssa-loop-split.c (connect_loops): Update.
21094 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
21095 * tree-ssa-reassoc.c (branch_fixup): Update.
21096 * tree-ssa-tail-merge.c (replace_block_by): Update.
21097 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
21098 (compute_path_counts): Update.
21099 (update_profile): Update.
21100 (recompute_probabilities): Update.
21101 (update_joiner_offpath_counts): Update.
21102 (estimated_freqs_path): Update.
21103 (freqs_to_counts_path): Update.
21104 (clear_counts_path): Update.
21105 (ssa_fix_duplicate_block_edges): Update.
21106 (duplicate_thread_path): Update.
21107 * tree-switch-conversion.c (case_bit_test_cmp): Update.
21108 (struct switch_conv_info): Update.
21109 * tree-tailcall.c (decrease_profile): Update.
21110 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
21111 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
21112 * value-prof.c (check_counter): Update.
21113 (gimple_divmod_fixed_value): Update.
21114 (gimple_mod_pow2): Update.
21115 (gimple_mod_subtract): Update.
21116 (gimple_ic_transform): Update.
21117 (gimple_stringop_fixed_value): Update.
21118 * value-prof.h (gimple_ic): Update.
21120 2017-06-02 Carl Love <cel@us.ibm.com>
21122 * config/rs6000/rs6000-c: Add support for built-in functions
21123 vector double vec_doublee (vector signed int);
21124 vector double vec_doublee (vector unsigned int);
21125 vector double vec_doublee (vector float);
21126 vector double vec_doubleh (vector signed int);
21127 vector double vec_doubleh (vector unsigned int);
21128 vector double vec_doubleh (vector float);
21129 vector double vec_doublel (vector signed int);
21130 vector double vec_doublel (vector unsigned int);
21131 vector double vec_doublel (vector float);
21132 vector double vec_doubleo (vector signed int);
21133 vector double vec_doubleo (vector unsigned int);
21134 vector double vec_doubleo (vector float);.
21135 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
21136 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
21138 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
21139 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
21140 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
21142 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
21143 vec_doublel, vec_doubleh.
21144 * doc/extend.texi: Update the built-in documentation file for the
21145 new built-in functions.
21147 2017-06-02 David Malcolm <dmalcolm@redhat.com>
21150 * ipa-inline-analysis.c (free_growth_caches): Set
21151 edge_removal_hook_holder to NULL after removing it.
21153 2017-06-02 Sudakshina Das <sudi.das@arm.com>
21155 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
21156 comparision with zero.
21158 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21159 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21160 for early expansion of vec_min and vec_max builtins.
21161 (builtin_function_type): Add min/max unsigned variants to those
21162 identified as having unsigned arguments.
21164 2017-06-02 Olivier Hainque <hainque@adacore.com>
21166 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
21168 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21170 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
21171 Use VALL_F16 iterator rather than VALL.
21173 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21175 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
21176 Emit CBNZ inside loop when doing a strong exchange and comparing
21177 against zero. Generate the CC flags after the loop.
21179 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
21181 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
21182 (dl_section_ref): New.
21183 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
21184 On AIX, append an expression to subtract the size of the
21185 section length to dl_section_ref.
21187 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21189 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21190 for early expansion of vector absolute builtins.
21192 2017-06-02 Richard Biener <rguenther@suse.de>
21194 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
21195 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
21197 2017-06-02 Richard Biener <rguenther@suse.de>
21199 PR tree-optimization/80948
21200 * tree-tailcall.c (find_tail_calls): Track stmts to move in
21201 stmt order as well.
21203 2017-06-02 Richard Biener <rguenther@suse.de>
21205 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
21207 * tree-vect-stmts.c (process_use): Do not mark backedge defs
21208 for inductions as relevant.
21210 2017-06-02 Richard Biener <rguenther@suse.de>
21212 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
21213 (vectorizable_induction): ... this. Remove dead code.
21215 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
21217 * builtins. (expand_builtin_alloca): Remove second parameter and
21218 infer its value from the first parameter instead.
21219 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
21221 2017-06-02 Jakub Jelinek <jakub@redhat.com>
21223 PR rtl-optimization/80903
21224 * loop-doloop.c (add_test): Unshare sequence.
21226 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
21228 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
21230 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
21232 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
21234 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
21235 xlogue_layout::get_instance, logue_layout::xlogue_layout,
21236 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
21237 (xlogue_layout::get_stub_rtx): Make static.
21238 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
21239 (xlogue_layout::compute_stub_managed_regs): Rename to...
21240 (xlogue_layout::count_stub_managed_regs): ...this.
21241 (xlogue_layout::is_stub_managed_reg): New function.
21242 (xlogue_layout::m_stub_names): Rename to...
21243 (xlogue_layout::s_stub_names): ...this, make static.
21244 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
21245 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
21246 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
21247 xlogue_layout::s_stub_names): Instantiate statics.
21248 (stub_managed_regs): Remove.
21249 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
21250 (disable_call_ms2sysv_xlogues): Rename to...
21251 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
21252 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
21254 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
21255 change after reload_completed.
21256 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
21258 (ix86_expand_prologue): Likewise.
21259 (ix86_expand_epilogue): Likewise.
21260 (ix86_expand_split_stack_prologue): Likewise.
21261 (ix86_compute_frame_layout): Remove frame parameter ...
21262 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
21263 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
21265 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
21266 (ix86_frame): Move from here ...
21267 * config/i386/i386.h (ix86_frame): ... to here.
21268 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
21269 complete ix86_frame data structure instead. Remove some_ld_name.
21271 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
21273 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
21274 symbols that hold a DECL_VALUE_EXPR.
21276 2017-06-01 Martin Jambor <mjambor@suse.cz>
21278 PR tree-optimization/80898
21279 * tree-sra.c (process_subtree_disqualification): Removed.
21280 (disqualify_candidate): Do not acll
21281 process_subtree_disqualification.
21282 (subtree_mark_written_and_enqueue): New function.
21283 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
21284 RHS has been disqualified and re-queue LHS if necessary. Apart
21285 from that, ignore disqualified RHS.
21287 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21289 * config/s390/s390.c (s390_emit_epilogue): Disable early return
21290 address fetch for z10 or later.
21292 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21294 * config/arc/arc.md (tst_movb): Add guard when splitting.
21296 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21298 * config/arc/arc.c (arc_can_eliminate): Test against
21299 arc_frame_pointer_needed.
21301 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21303 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
21304 to prevent store reordering.
21305 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
21306 (type): Add block type.
21307 (stack_tie): Define special instruction to be used in
21310 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21312 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
21313 constraint. It is not valid for the pattern.
21314 (noncommutative_binary_comparison): Likewise.
21316 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21318 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
21321 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21323 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
21324 be used by the reg-alloc.
21326 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21328 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
21329 reg-alloc when having mul64 or mul32x16 instructions.
21330 (mulsidi3): Likewise.
21331 (umulsidi3): Likewise.
21332 (mulsi32x16): New pattern.
21333 (mulsi64): Likewise.
21334 (mulsidi64): Likewise.
21335 (umulsidi64): Likewise.
21336 (MUL32x16_REG): Define.
21337 (mul64_600): Use MUL32x16_REG.
21338 (mac64_600): Likewise.
21339 (umul64_600): Likewise.
21340 (umac64_600): Likewise.
21342 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21344 * config/arc/arc.md (mulsi3_700): Make it commutative.
21346 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
21348 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
21350 (*sign_extendsidi2_insn): Likewise for movstosw.
21352 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
21354 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
21355 the type of the input discriminant value. Convert the
21356 discriminant value of signedness vary.
21358 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
21360 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
21361 Add to -Wall section.
21363 2017-06-01 Richard Biener <rguenther@suse.de>
21365 PR middle-end/66313
21366 * fold-const.c (fold_plusminus_mult_expr): If the factored
21367 factor may be zero use a wrapping type for the inner operation.
21368 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
21369 and handle moved defs.
21370 (process_assignment): Properly guard the unary op case. Return a
21371 tri-state indicating that moving the stmt before the call may allow
21372 to continue. Pass through to_move.
21373 (find_tail_calls): Handle moving unrelated defs before
21376 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
21379 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
21380 splitter result in the canonical way.
21382 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
21384 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
21385 also for 32bit target. Update insn attributes.
21386 (zero-extendsidi2 splitter): Allow all registers for operand 1.
21388 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
21390 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
21391 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
21392 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
21393 (_mm_maskz_min_ss): New intrinsics.
21395 2017-05-31 Martin Liska <mliska@suse.cz>
21397 * tree-vect-loop.c (vect_create_epilog_for_reduction):
21398 Change comment style to one we normally use.
21399 (vectorizable_reduction): Likewise.
21400 (vectorizable_induction): Likewise.
21401 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
21402 (vectorizable_call): Likewise.
21403 (vectorizable_simd_clone_call): Likewise.
21404 (vectorizable_conversion): Likewise.
21405 (vectorizable_assignment): Likewise.
21406 (vectorizable_shift): Likewise.
21407 (vectorizable_operation): Likewise.
21408 (vectorizable_store): Likewise.
21409 (vectorizable_load): Likewise.
21410 * tree-vectorizer.h: Likewise.
21412 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
21414 * passes.c (emergency_dump_function): New.
21415 * tree-pass.h (emergency_dump_function): Declare.
21416 * plugin.c (plugins_internal_error_function): Remove.
21417 * plugin.h (plugins_internal_error_function): Remove declaration.
21418 * toplev.c (internal_error_function): New static function. Use it...
21419 (general_init): ...here.
21421 2017-05-31 Graham Markall <graham.markall@embecosm.com>
21423 * config/arc/arc.c (arc_print_operand): Handle constant operands.
21424 (arc_rtx_costs): Add costs for new patterns.
21425 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
21426 * config/arc/predicates.md: Add _1_2_3_operand predicate.
21428 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21430 * tree-ssa-strlen.c (get_next_strinfo): New function.
21431 (get_stridx_plus_constant): Use it.
21432 (zero_length_string): Likewise.
21433 (adjust_related_strinfos): Likewise.
21434 (adjust_last_stmt): Likewise.
21436 2017-05-31 Richard Biener <rguenther@suse.de>
21439 * config/i386/i386.c (ix86_expand_builtin): Remove assert
21440 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
21442 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21444 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
21445 loop_vinfo argument and use of dependence distance vectors.
21446 Check instead whether the two references differ only in their
21447 initial value and assume that they have the same alignment if the
21448 difference is a multiple of the vector alignment.
21449 (vect_analyze_data_refs_alignment): Update call accordingly.
21451 2017-05-31 Martin Liska <mliska@suse.cz>
21454 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
21456 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21458 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
21459 (create_intersect_range_checks): Move from ...
21460 * tree-data-ref.c (create_intersect_range_checks_index)
21461 (create_intersect_range_checks): ... to here.
21462 (create_runtime_alias_checks): New function factored from ...
21463 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
21464 here. Call above function.
21465 * tree-data-ref.h (create_runtime_alias_checks): New function.
21467 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21469 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
21470 segment length for dr_b and compute it in wide_int.
21472 2017-05-31 Richard Biener <rguenther@suse.de>
21474 PR tree-optimization/80906
21475 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
21476 and pass through iv_map.
21477 (copy_bb_and_scalar_dependences): Adjust.
21478 (translate_pending_phi_nodes): Likewise.
21479 (copy_loop_close_phi_args): Handle code-generating IVs instead
21482 2017-05-30 David Malcolm <dmalcolm@redhat.com>
21484 * diagnostic-color.c (color_dict): Add "type-diff".
21485 (parse_gcc_colors): Update comment.
21486 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
21487 -fdiagnostics-show-template-tree and -fno-elide-type.
21488 (GCC_COLORS): Add type-diff to example.
21490 (-fdiagnostics-show-template-tree): New.
21491 (-fno-elide-type): New.
21492 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
21493 the pp_format_decoder callback. Call any m_format_postprocessor's
21495 (pretty_printer::pretty_printer): Initialize
21496 m_format_postprocessor.
21497 (pretty_printer::~pretty_printer): Delete any
21498 m_format_postprocessor.
21499 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
21500 (class format_postprocessor): New class.
21501 (struct pretty_printer::format_decoder): Document the new parameters.
21502 (struct pretty_printer::m_format_postprocessor): New field.
21503 * tree-diagnostic.c (default_tree_printer): Update for new
21504 bool and const char ** params.
21505 * tree-diagnostic.h (default_tree_printer): Likewise.
21507 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
21509 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
21510 (lwa_operand): Delete rs6000_gen_cell_microcode test.
21511 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
21512 rs6000_gen_cell_microcode code.
21513 (rs6000_final_prescan_insn): Delete.
21514 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
21515 "warn-cell-microcode" entries.
21516 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
21517 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
21518 throughout. Change cc_reg_not_micro_cr0_operand to
21519 cc_reg_not_cr0_operand throughout.
21520 (*extendhi<mode>2_noload): Delete.
21521 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
21522 (mwarn-cell-microcode): Delete.
21523 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
21524 -mgen-cell-microcode and -mwarn-cell-microcode.
21526 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
21529 * config/i386/constraints.md (Yd): New constraint.
21531 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
21532 and (?Yd, r) alternatives. Update insn attributes.
21533 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
21534 and (?*Yd, r) alternatives. Update insn attributes.
21535 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
21537 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
21539 * gimplify.c (gimplify_modify_expr): Don't create a
21540 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
21543 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
21545 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
21547 2017-05-30 Richard Biener <rguenther@suse.de>
21549 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
21550 and reduc_def fields.
21551 (STMT_VINFO_REDUC_TYPE): New define.
21552 (STMT_VINFO_REDUC_DEF): Likewise.
21553 (vect_force_simple_reduction): Adjust prototype.
21554 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
21555 (vect_is_simple_reduction): Remove check_reduction argument.
21556 (vect_force_simple_reduction): Adjust and set
21557 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21558 (vectorizable_reduction): Do not re-do reduction analysis
21559 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21560 * tree-parloops.c (gather_scalar_reductions): Adjust.
21562 2017-05-30 Richard Biener <rguenther@suse.de>
21564 PR middle-end/80901
21565 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
21568 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
21570 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
21571 Introduce unknown_misalignment parameter and remove vf.
21572 (vect_peeling_hash_get_lowest_cost):
21573 Pass unknown_misalignment parameter.
21574 (vect_enhance_data_refs_alignment):
21575 Fix unsupportable data ref treatment.
21577 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21579 * tree-vect-data-refs.c (vect_get_data_access_cost):
21580 Workaround for SLP handling.
21581 (vect_enhance_data_refs_alignment):
21582 Compute costs for doing no peeling at all, compare to the best
21583 peeling costs so far and avoid peeling if cheaper.
21585 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21587 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
21588 Return peeling info and set costs to zero for unlimited cost
21590 (vect_enhance_data_refs_alignment): Also inspect all datarefs
21591 with unknown misalignment. Compute and costs for unknown
21592 misalignment, compare them to the costs for known misalignment
21593 and choose the cheapest for peeling.
21595 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21597 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
21598 (vect_get_peeling_costs_all_drs): Create function.
21599 (vect_peeling_hash_get_lowest_cost):
21600 Use vect_get_peeling_costs_all_drs.
21601 (vect_peeling_supportable): Create function.
21602 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
21604 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21606 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
21607 DR_HAS_NEGATIVE_STEP.
21608 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
21609 (vect_enhance_data_refs_alignment): Use.
21610 (vect_duplicate_ssa_name_ptr_info): Use.
21611 * tree-vectorizer.h (dr_misalignment): Use.
21612 (known_alignment_for_access_p): Use.
21614 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21617 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
21619 (has_section_name): New function.
21621 2017-05-30 Martin Liska <mliska@suse.cz>
21624 * auto-profile.c (get_function_decl_from_block): Fix
21627 2017-05-30 Richard Biener <rguenther@suse.de>
21629 PR middle-end/80876
21630 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
21632 2017-05-30 Martin Liska <mliska@suse.cz>
21634 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
21635 * dumpfile.h (struct dump_file_info): Remove ctors.
21637 2017-05-30 Martin Liska <mliska@suse.cz>
21639 * predict.def: Fix GNU coding style.
21641 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
21643 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
21644 Mark 'to' argument with ATTRIBUTE_UNUSED.
21646 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
21648 * config/xtensa/xtensa.c (xtensa_emit_call): Use
21649 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
21650 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
21653 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
21655 * doc/install.texi (Options specification): Restore entry of
21656 --enable-sjlj-exceptions.
21658 2017-05-27 Michael Eager <eager@eagercon.com>
21661 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
21663 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
21665 * config/microblaze/microblaze.h
21666 (FIXED_REGISTERS): Update in macro.
21667 (CALL_USED_REGISTERS): Update in macro.
21669 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
21671 * doc/install.texi: Add links to macOS binary distributions.
21673 2017-05-27 Jakub Jelinek <jakub@redhat.com>
21677 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21679 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
21681 2017-05-26 Martin Liska <mliska@suse.cz>
21683 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
21685 2017-05-26 Martin Liska <mliska@suse.cz>
21687 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
21688 always leading ';; '.
21689 (dump_bb_info): Likewise.
21690 (brief_dump_cfg): Likewise.
21691 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
21692 * dumpfile.c: Remove usage of TDF_VERBOSE.
21693 * dumpfile.h (enum dump_kind): Likewise.
21694 (dump_gimple_bb_header): Do not use TDF_COMMENT.
21695 * print-tree.c (debug_verbose): Remove.
21696 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
21697 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
21698 * tree-diagnostic.c (default_tree_printer): Replace
21699 TDF_DIAGNOSTIC with TDF_SLIM.
21701 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21703 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
21704 in parameter loop, rather than loop_vinfo.
21705 (create_intersect_range_checks): Ditto.
21706 (vect_create_cond_for_alias_checks): Update call to above functions.
21708 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21710 PR tree-optimization/80815
21711 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
21712 for merging runtime alias checks. Handle negative DR_STEPs.
21714 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21716 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
21718 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
21719 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
21720 out code pruning runtime alias checks.
21721 * tree-data-ref.c (prune_runtime_alias_test_list): New function
21722 factored out from above.
21723 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
21725 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
21727 (prune_runtime_alias_test_list): New decalaration.
21729 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21731 * tree-vect-data-refs.c (compare_tree): Rename and move ...
21732 * tree-data-ref.c (data_ref_compare_tree): ... to here.
21733 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
21734 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
21735 (operator==, comp_dr_with_seg_len_pair): Ditto.
21736 (vect_prune_runtime_alias_test_list): Ditto.
21738 2017-05-26 Martin Liska <mliska@suse.cz>
21741 * params.def: Bound partial-inlining-entry-probability param.
21743 2017-05-26 Marek Polacek <polacek@redhat.com>
21746 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
21749 2017-05-26 Richard Biener <rguenther@suse.de>
21751 PR tree-optimization/80842
21752 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
21755 2017-05-26 Richard Biener <rguenther@suse.de>
21757 PR tree-optimization/80844
21758 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
21760 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
21762 * doc/md.texi (Machine Constraints): Update x86 family
21763 machine constraints section to match 'config/i386/constraints.md'.
21765 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
21767 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
21769 2017-05-25 Nathan Sidwell <nathan@acm.org>
21771 * doc/invoke.texi (--enable-languages): Update documentation.
21773 2017-05-25 Martin Liska <mliska@suse.cz>
21775 * dumpfile.c: Add TDF_FOLDING.
21776 * dumpfile.h (enum dump_kind): Likewise.
21777 * genmatch.c (dt_simplify::gen_1): Use it.
21779 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21781 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
21783 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21785 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
21786 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
21788 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21790 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
21791 * match.pd (X == C): Rewrite it here.
21792 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
21793 with_certain_nonzero_bits2): New predicates.
21794 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
21796 2017-05-24 Nathan Sidwell <nathan@acm.org>
21798 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
21801 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
21804 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
21806 * config/powerpcspe: New port. Files are copied from the rs6000
21807 port, with "rs6000" in filenames replaced by "powerpcspe".
21809 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
21811 PR rtl-optimization/80754
21812 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
21814 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
21816 * config/sparc/sparc.md (length): Return the correct value for -mflat
21817 sibcalls to match output_sibcall.
21819 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
21823 * config/rs6000/rs6000.c (struct machine_function): Add new field
21825 (rs6000_get_separate_components): Init that field, use it.
21826 (rs6000_components_for_bb): Use the field.
21828 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
21830 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
21832 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
21834 PR middle-end/80823
21835 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
21837 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21840 * config/s390/s390.c (s390_check_qrst_address): Check incoming
21841 address against address_operand predicate.
21842 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
21844 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
21846 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
21847 parameters passed indirectly.
21849 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
21851 * config/i386/i386.md (*movdi_internal): Remove SSE4
21852 alternative 18 (?r, *v). Update insn attributes.
21853 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
21854 Update insn attributes.
21855 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
21856 Update insn attributes.
21857 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
21858 alternative 1 (r, v). Remove isa attribute.
21859 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
21860 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
21861 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
21863 2017-05-23 Tom de Vries <tom@codesourcery.com>
21865 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
21868 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21870 * cgraphunit.c (symbol_table::process_new_functions): Update.
21871 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
21872 (inline_generate_summary): Rename to ...
21873 (ipa_fn_summary_generate): ... this one.
21874 (inline_read_summary): Rename to ...
21875 (ipa_fn_summary_read): ... this one.
21876 (inline_write_summary): Rename to ...
21877 (ipa_fn_summary_write): ... this one.
21878 (inline_free_summary): Rename to ...
21879 (ipa_free_fn_summary): ... this one.
21880 (pass_data_local_fn_summary, pass_local_fn_summary,
21881 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
21882 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
21883 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
21884 make_pass_ipa_fn_summary): New.
21885 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
21886 inline_write_summary, inline_free_summary): Remove.
21887 (ipa_free_fn_summary) : New.
21888 * ipa-inline.c (ipa_inline): Update.
21889 (pass_ipa_inline): Do not generate summaries.
21890 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
21892 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
21893 and add pass_ipa_fn_summary.
21894 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
21896 (make_pass_inline_parameters): Remove.
21898 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
21900 * omp-low.c (struct omp_context): Remove "default_kind" member.
21903 * omp-offload.c (execute_oacc_device_lower): Remove the
21904 parallelism dimensions function attributes for unparallelized
21905 OpenACC kernels constructs.
21907 2017-05-23 Martin Liska <mliska@suse.cz>
21909 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
21911 (cgraph_edge::make_speculative): Likewise.
21912 (cgraph_edge::resolve_speculation): Likewise.
21913 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
21914 (cgraph_node::dump): Likewise.
21915 * cgraph.h: Likewise.
21916 * cgraphunit.c (analyze_functions): Likewise.
21917 (symbol_table::compile): Likewise.
21918 * ipa-cp.c (print_all_lattices): Likewise.
21919 (determine_versionability): Likewise.
21920 (initialize_node_lattices): Likewise.
21921 (ipcp_verify_propagated_values): Likewise.
21922 (estimate_local_effects): Likewise.
21923 (update_profiling_info): Likewise.
21924 (create_specialized_node): Likewise.
21925 (perhaps_add_new_callers): Likewise.
21926 (decide_about_value): Likewise.
21927 (decide_whether_version_node): Likewise.
21928 (identify_dead_nodes): Likewise.
21929 (ipcp_store_bits_results): Likewise.
21930 * ipa-devirt.c (dump_targets): Likewise.
21931 (ipa_devirt): Likewise.
21932 * ipa-icf.c (sem_item::dump): Likewise.
21933 (sem_function::equals): Likewise.
21934 (sem_variable::equals): Likewise.
21935 (sem_item_optimizer::read_section): Likewise.
21936 (sem_item_optimizer::execute): Likewise.
21937 (congruence_class::dump): Likewise.
21938 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
21939 (dump_inline_summary): Likewise.
21940 (estimate_node_size_and_time): Likewise.
21941 (inline_analyze_function): Likewise.
21942 * ipa-inline-transform.c (inline_call): Likewise.
21943 * ipa-inline.c (report_inline_failed_reason): Likewise.
21944 (want_early_inline_function_p): Likewise.
21945 (edge_badness): Likewise.
21946 (update_edge_key): Likewise.
21947 (inline_small_functions): Likewise.
21948 * ipa-profile.c (ipa_profile): Likewise.
21949 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
21950 (ipa_make_edge_direct_to_target): Likewise.
21951 (remove_described_reference): Likewise.
21952 (ipa_impossible_devirt_target): Likewise.
21953 (propagate_controlled_uses): Likewise.
21954 (ipa_print_node_params): Likewise.
21955 (ipcp_transform_function): Likewise.
21956 * ipa-pure-const.c (pure_const_read_summary): Likewise.
21957 (propagate_pure_const): Likewise.
21958 * ipa-reference.c (generate_summary): Likewise.
21959 (read_write_all_from_decl): Likewise.
21960 (propagate): Likewise.
21961 (ipa_reference_read_optimization_summary): Likewise.
21962 * ipa-utils.c (ipa_merge_profiles): Likewise.
21963 * ipa.c (walk_polymorphic_call_targets): Likewise.
21964 (symbol_table::remove_unreachable_nodes): Likewise.
21965 (ipa_single_use): Likewise.
21966 * passes.c (execute_todo): Likewise.
21967 * predict.c (drop_profile): Likewise.
21968 * symtab.c (symtab_node::get_dump_name): New function.
21969 (symtab_node::dump_name): Likewise.
21970 (symtab_node::dump_asm_name): Likewise.
21971 (symtab_node::dump_references): Likewise.
21972 (symtab_node::dump_referring): Likewise.
21973 (symtab_node::dump_base): Likewise.
21974 (symtab_node::debug_symtab): Likewise.
21975 * tree-sra.c (convert_callers_for_node): Likewise.
21976 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
21977 * value-prof.c (init_node_map): Likewise.
21979 2017-05-23 Martin Liska <mliska@suse.cz>
21981 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
21982 and symtab_node::debug_symtab to symbol_table::debug.
21983 * cgraphunit.c (analyze_functions): Use the renamed function.
21984 (symbol_table::compile): Likewise.
21985 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
21986 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
21987 * passes.c (execute_todo): Likewise.
21988 * symtab.c (symbol_table::dump): New function.
21989 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
21991 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21993 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
21994 that nonconst implies exec.
21996 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21998 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
21999 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22000 (inline_edge_summary_vec): Turn into ...
22001 (ipa_call_summaries): ... this one.
22002 (redirect_to_unreachable, edge_set_predicate,
22003 evaluate_properties_for_edge, inline_summary_alloc,
22004 reset_ipa_call_summary, reset_inline_summary,
22005 inline_summary_t::duplicate): Update.
22006 (inline_edge_duplication_hook): Turn to ...
22007 (ipa_call_summary_t::duplicate): ... this one.
22008 (inline_edge_removal_hook): Turn to ...
22009 (ipa_call_summary_t::remove): ... this one.
22010 (dump_inline_edge_summary): Turn to ...
22011 (dump_ipa_call_summary): ... this one.
22012 (estimate_function_body_sizes): Update.
22013 (inline_update_callee_summaries): Update.
22014 (remap_edge_change_prob): Update.
22015 (remap_edge_summaries): Update.
22016 (inline_merge_summary): Update.
22017 (do_estimate_edge_time): Update.
22018 (inline_generate_summary): Update.
22019 (inline_read_section): Update.
22020 (inline_read_summary): Update.
22021 (inline_free_summary): Update.
22022 * ipa-inline.c (can_inline_edge_p): Update.
22023 (compute_inlined_call_time): Update.
22024 (want_inline_small_function_p): Update.
22025 (edge_badness): Update.
22026 (early_inliner): Update.
22027 * ipa-inline.h (inline_edge_summary): Turn to ...
22028 (ipa_call_summary): ... this one.
22029 (ipa_call_summary_t): New class.
22030 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22031 (ipa_call_summaries): New.
22032 (inline_edge_summary): Remove.
22033 (estimate_edge_growth): Update.
22034 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22035 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22036 * ipa-split.c (execute_split_functions): Update.
22037 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22039 2017-05-23 Tom de Vries <tom@codesourcery.com>
22041 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22042 attributes): Document rdrand effective target.
22044 2017-05-23 Tom de Vries <tom@codesourcery.com>
22046 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22047 attributes): Sort alphabetically.
22049 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
22051 * config/avr/genmultilib.awk: Use gsub instead of gensub.
22053 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
22056 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
22057 V2DF/V2DI splat into two separate patterns, one that handles
22058 registers, and the other that only handles memory. Drop support
22059 for splatting from a GPR on ISA 2.07 and then splitting the
22060 splat into direct move and splat.
22061 (vsx_splat_<mode>_reg): Likewise.
22062 (vsx_splat_<mode>_mem): Likewise.
22064 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
22066 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
22068 2017-05-22 Jakub Jelinek <jakub@redhat.com>
22070 PR middle-end/80809
22071 * omp-low.c (finish_taskreg_remap): New function.
22072 (finish_taskreg_scan): If unit size of ctx->record_type
22073 is non-constant, unshare the size expression and replace
22074 decls in it with possible outer var refs.
22076 PR middle-end/80809
22077 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
22078 GOVD_SHARED rather than GOVD_PRIVATE with it.
22079 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
22080 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
22082 PR middle-end/80853
22083 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
22084 as last argument to build_outer_var_ref for pointer bases of array
22085 section reductions.
22087 2017-05-19 Martin Sebor <msebor@redhat.com>
22089 * print-tree.c (print_node): Print DECL_READ_P flag.
22091 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22093 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
22094 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
22095 * cgraph.c: Likewise.
22096 * cgraphunit.c: Likewise.
22097 * gengtype.c: Likewise.
22098 * ipa-cp.c: Likewise.
22099 * ipa-devirt.c: Likewise.
22100 * ipa-icf.c: Likewise.
22101 * ipa-predicate.c: Likewise.
22102 * ipa-profile.c: Likewise.
22103 * ipa-prop.c: Likewise.
22104 * ipa-split.c: Likewise.
22106 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
22107 edge_predicate_pool, dump_inline_hints,
22108 inline_summary::account_size_time, redirect_to_unreachable,
22109 edge_set_predicate, set_hint_predicate,
22110 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
22111 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
22112 inline_summary_t::remove, remap_hint_predicate_after_duplication,
22113 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
22114 ipa_call_summary_t::remove, initialize_growth_caches,
22115 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
22116 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
22117 mark_modified, unmodified_parm_1, unmodified_parm,
22118 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
22119 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
22120 compute_bb_predicates, will_be_nonconstant_expr_predicate,
22121 will_be_nonconstant_predicate, record_modified_bb_info,
22122 get_minimal_bb, record_modified, param_change_prob,
22123 phi_result_unknown_predicate, predicate_for_phi_result,
22124 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
22125 estimate_function_body_sizes, compute_inline_parameters,
22126 compute_inline_parameters_for_curren, pass_data_inline_parameters,
22127 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
22128 inline_update_callee_summaries, remap_edge_change_prob,
22129 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
22130 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
22131 inline_analyze_function, inline_summary_t::insert,
22132 inline_generate_summary, read_ipa_call_summary, inline_read_section,
22133 inline_read_summary, write_ipa_call_summary, inline_write_summary,
22134 inline_free_summary): Move to ipa-fnsummary.h
22135 (predicate_t): Remove.
22136 * ipa-fnsummary.c: New file.
22137 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
22138 (enum inline_hints_vals, inline_hints, agg_position_info,
22139 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
22140 inline_summaries, ipa_call_summary, ipa_call_summary_t,
22141 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
22142 dump_inline_summary, dump_inline_hints, inline_generate_summary,
22143 inline_read_summary, inline_write_summary, inline_free_summary,
22144 inline_analyze_function, initialize_inline_failed,
22145 inline_merge_summary, inline_update_overall_summary,
22146 compute_inline_parameters): Move to ipa-fnsummary.h
22147 * ipa-fnsummary.h: New file.
22148 * ipa-inline-transform.h: Include ipa-inline.h.
22149 * ipa-inline.c: LIkewise.
22151 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22153 * ipa-inline.c (edge_badness): Use inlined_time instead of
22154 inline_summaries->get.
22156 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22158 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
22160 2017-05-22 Nathan Sidwell <nathan@acm.org>
22162 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
22163 (fdump-lang): Document 'raw' option.
22164 * dumpfile.h (TDI_tu): Delete.
22165 * dumpfile.c (dump_files): Remove translation-unit.
22166 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
22168 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
22170 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
22171 command option from $(AWK) call.
22172 * config/avr/genmultilib.awk: Simplify and rewrite so that it
22173 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
22174 [FORMAT]: Remove handling of variable.
22175 * config/avr/t-multilib: Regenerate.
22177 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22179 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
22181 (dump_inline_summary): Do not print self_time.
22182 (estimate_function_body_sizes): Do not set self_time.
22183 (compute_inline_parameters): Likewise.
22184 (inline_read_section, inline_write_summary): Do not stream self_time.
22185 * ipa-inline.h (inline_summary): Drop self_time.
22187 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22189 * ipa-inline-analysis.c (account_size_time): Rename to ...
22190 (inline_summary::account_size_time): ... this one.
22191 (reset_ipa_call_summary): Turn to ...
22192 (ipa_call_summary::reset): ... this one.
22193 (reset_inline_summary): Turn to ...
22194 (inline_summary::reset): ... this one.
22195 (inline_summary_t::remove): Update.
22196 (inline_summary_t::duplicate): Update.
22197 (ipa_call_summary_t::remove): Update.
22198 (dump_inline_summary): Update.
22199 (estimate_function_body_sizes): Update.
22200 (compute_inline_parameters): Update.
22201 (estimate_node_size_and_time): Update.
22202 (inline_merge_summary): Update.
22203 (inline_update_overall_summary): Update.
22204 (inline_read_section): Update.
22205 (inline_write_summary): Update.
22206 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
22207 add account_size_time and reset member functions.
22208 (ipa_call_summary): Add reset function.
22209 * ipa-predicate.h (predicate::operator &): Constify.
22211 2017-05-22 Richard Biener <rguenther@suse.de>
22213 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
22215 2017-05-19 Jason Merrill <jason@redhat.com>
22217 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
22219 2017-05-19 Marek Polacek <polacek@redhat.com>
22222 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
22223 TYPE_OVERFLOW_WRAPS checks.
22225 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
22227 * tree-core.h (enum omp_clause_default_kind): Add
22228 "OMP_CLAUSE_DEFAULT_PRESENT".
22229 * tree-pretty-print.c (dump_omp_clause): Handle it.
22230 * gimplify.c (enum gimplify_omp_var_data): Add
22231 "GOVD_MAP_FORCE_PRESENT".
22232 (gimplify_adjust_omp_clauses_1): Map it to
22233 "GOMP_MAP_FORCE_PRESENT".
22234 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
22236 * gimplify.c (oacc_default_clause): Clarify.
22238 2017-05-19 Nathan Sidwell <nathan@acm.org>
22240 LANG_HOOK_REGISTER_DUMPS
22241 * toplev.c (general_init): Call register dump lang hook.
22242 * doc/invoke.texi: Document -fdump-lang option family.
22243 * dumpfile.c (dump_files): Remove class dump here.
22244 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
22245 * dumpfile.h (tree_dump_index): Remove TDI_class.
22246 * langhooks-def.h (lhd_register_dumps): Declare.
22247 (LANG_HOOKS_REGISTER_DUMPS): Define.
22248 (LANG_HOOKS_INITIALIZER): Add it.
22249 * langhooks.c (lhd_register_dumps): Define.
22250 * langhooks.h (struct lang_hooks): Add register_dumps.
22252 2017-05-19 Nathan Sidwell <nathan@acm.org>
22254 * context.h (context::set_passes): New.
22255 * context.c (context::context): Do not create pass manager.
22256 * toplev.c (general_init): Create pass manager here.
22258 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
22260 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
22261 use this splitter if two add or or instructions would also work for
22262 the constant we want to generate.
22264 2017-05-19 Richard Biener <rguenther@suse.de>
22267 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
22268 predicate evaluation.
22270 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
22272 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
22274 * ipa-inline.c (want_inline_small_function_p): Do not cast to
22277 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
22279 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22280 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22281 (inline_edge_summary_vec): Turn into ...
22282 (ipa_call_summaries): ... this one.
22283 (redirect_to_unreachable, edge_set_predicate,
22284 evaluate_properties_for_edge, inline_summary_alloc,
22285 reset_ipa_call_summary, reset_inline_summary,
22286 inline_summary_t::duplicate): Update.
22287 (inline_edge_duplication_hook): Turn to ...
22288 (ipa_call_summary_t::duplicate): ... this one.
22289 (inline_edge_removal_hook): Turn to ...
22290 (ipa_call_summary_t::remove): ... this one.
22291 (dump_inline_edge_summary): Turn to ...
22292 (dump_ipa_call_summary): ... this one.
22293 (estimate_function_body_sizes): Update.
22294 (inline_update_callee_summaries): Update.
22295 (remap_edge_change_prob): Update.
22296 (remap_edge_summaries): Update.
22297 (inline_merge_summary): Update.
22298 (do_estimate_edge_time): Update.
22299 (inline_generate_summary): Update.
22300 (inline_read_section): Update.
22301 (inline_read_summary): Update.
22302 (inline_free_summary): Update.
22303 * ipa-inline.c (can_inline_edge_p): Update.
22304 (compute_inlined_call_time): Update.
22305 (want_inline_small_function_p): Update.
22306 (edge_badness): Update.
22307 (early_inliner): Update.
22308 * ipa-inline.h (inline_edge_summary): Turn to ...
22309 (ipa_call_summary): ... this one.
22310 (ipa_call_summary_t): New class.
22311 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22312 (ipa_call_summaries): New.
22313 (inline_edge_summary): Remove.
22314 (estimate_edge_growth): Update.
22315 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22316 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22317 * ipa-split.c (execute_split_functions): Update.
22318 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22320 2017-05-19 Richard Biener <rguenther@suse.de>
22322 PR middle-end/80764
22323 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
22325 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
22327 * config/rs6000/rs6000.c (struct machine_function): Add field
22328 fpr_is_wrapped_separately.
22329 (rs6000_get_separate_components): Use 64 components. Handle the
22330 new FPR components.
22331 (rs6000_components_for_bb): Handle the FPR components.
22332 (rs6000_emit_prologue_components): Handle the FPR components.
22333 (rs6000_emit_epilogue_components): Handle the FPR components.
22334 (rs6000_set_handled_components): Handle the FPR components.
22335 (rs6000_emit_prologue): Don't output prologue code for those FPRs
22336 that are already separately shrink-wrapped.
22337 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
22338 that are already separately shrink-wrapped.
22340 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
22343 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
22346 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
22347 (define_peephole2 for Altivec d-form load): Add peepholes to catch
22348 cases where the register allocator uses a move and an offsettable
22349 memory operation to/from a FPR register on ISA 2.06/2.07.
22350 (define_peephole2 for Altivec d-form store): Likewise.
22352 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
22355 * config/i386/mmx.md (*mov<mode>_internal): Enable
22356 alternatives 11, 12, 13 and 14 also for 32bit targets.
22357 Remove alternatives 15, 16, 17 and 18.
22358 * config/i386/sse.md (vec_concatv2di): Change
22359 alternative (!x, *y) to (x, ?!*Yn).
22361 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
22363 * dumpfile.h (enum dump_kind): Remove stray comma.
22365 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
22367 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
22368 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
22369 predicate::num_conditions
22370 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
22371 (CHANGED): turn into predicate::changed.
22372 (agg_position_info): Move to ipa-predicate.h
22373 (add_condition, predicate::add_clause, predicate::operator &=,
22374 predicate::or_with, predicate::evaluate, predicate::probability,
22375 dump_condition, dump_clause, predicate::dump,
22376 predicate::remap_after_duplication, predicate::remap_after_inlining,
22377 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
22378 (evaluate_conditions_for_known_args): Update.
22379 (set_cond_stmt_execution_predicate): Update.
22380 * ipa-inline.h: Include ipa-predicate.h
22381 (condition, inline_param_summary, conditions, agg_position_info,
22382 predicate): Move to ipa-predicate.h
22383 * ipa-predicate.c: New file.
22384 * ipa-predicate.h: New file.
22386 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
22388 * final.c (leaf_function_p): Check we are not in a sequence.
22390 2017-05-18 Martin Liska <mliska@suse.cz>
22392 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
22393 * dumpfile.c (dump_register): Use new enum dump_kind.
22394 (get_dump_file_name): Likewise.
22395 (dump_enable_all): Likewise.
22396 (dump_switch_p_1): Likewise.
22397 (enable_rtl_dump_file): Remove usage of TDF_RTL.
22398 * dumpfile.h (enum dump_kind): New enum type.
22399 (struct dump_file_info): Create constructor and
22400 format fields and comments.
22401 * passes.c (pass_manager::register_one_dump_file):
22403 * statistics.c (statistics_early_init): Likewise.
22404 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
22405 TDF_TREE with TDF_SLIM.
22406 (gather_memory_references_ref): Likewise.
22408 2017-05-18 Martin Liska <mliska@suse.cz>
22410 * vec.h (struct vnull): Use it.
22412 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
22414 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
22415 (true_predicate, false_predicate, true_predicate_p,
22416 false_predicate_p): Remove.
22417 (single_cond_predicate, not_inlined_predicate): Turn to member function
22419 (add_condition): Update.
22420 (add_clause): Turn to...
22421 (predicate::add_clause): ... this one; update; allow passing NULL
22423 (and_predicates): Turn to ...
22424 (predicate::operator &=): ... this one.
22425 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
22426 (or_predicates): Turn to ...
22427 (predicate::or_with): ... this one.
22428 (evaluate_predicate): Turn to ...
22429 (predicate::evaluate): ... this one.
22430 (predicate_probability): Turn to ...
22431 (predicate::probability): ... this one.
22432 (dump_condition): Update.
22433 (dump_predicate): Turn to ...
22434 (predicate::dump): ... this one.
22435 (account_size_time): Update.
22436 (edge_set_predicate): Update.
22437 (set_hint_predicate): UPdate.
22438 (evaluate_conditions_for_known_args): Update.
22439 (evaluate_properties_for_edge): Update.
22440 (remap_predicate_after_duplication): Turn to...
22441 (predicate::remap_after_duplication): ... this one.
22442 (remap_hint_predicate_after_duplication): Update.
22443 (inline_summary_t::duplicate): UPdate.
22444 (dump_inline_edge_summary): Update.
22445 (dump_inline_summary): Update.
22446 (set_cond_stmt_execution_predicate): Update.
22447 (set_switch_stmt_execution_predicate): Update.
22448 (compute_bb_predicates): Update.
22449 (will_be_nonconstant_expr_predicate): Update.
22450 (will_be_nonconstant_predicate): Update.
22451 (phi_result_unknown_predicate): Update.
22452 (predicate_for_phi_result): Update.
22453 (array_index_predicate): Update.
22454 (estimate_function_body_sizes): Update.
22455 (estimate_node_size_and_time): Update.
22456 (estimate_ipcp_clone_size_and_time): Update.
22457 (remap_predicate): Rename to ...
22458 (predicate::remap_after_inlining): ... this one.
22459 (remap_hint_predicate): Update.
22460 (inline_merge_summary): Update.
22461 (inline_update_overall_summary): Update.
22462 (estimate_size_after_inlining): Update.
22463 (read_predicate): Rename to ...
22464 (predicate::stream_in): ... this one.
22465 (read_inline_edge_summary): Update.
22466 (write_predicate): Rename to ...
22467 (predicate::stream_out): ... this one.
22468 (write_inline_edge_summary): Update.
22469 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
22470 (clause_t): Turn to uint32_t
22471 (predicate): Turn to class; implement constructor and operators
22473 (size_time_entry): Update.
22474 (inline_summary): Update.
22475 (inline_edge_summary): Update.
22477 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
22479 * fold-const.c (fold_binary_loc): Move transformation...
22480 * match.pd (C - X CMP X): ... here.
22482 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
22484 * config/sparc/sparc.c (sparc_option_override): Set function
22485 alignment for -mcpu=niagara7 to 64 to match the I$ line.
22486 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
22488 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
22490 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
22492 2017-05-18 Marek Polacek <polacek@redhat.com>
22495 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
22496 (pass_ubsan::execute): Call gimple_assign_single_p instead of
22497 gimple_assign_load_p.
22499 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
22501 PR middle-end/80692
22502 * real.c (do_compare): Give decimal_do_compare preference over
22503 comparing just the signs.
22505 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
22507 * doc/md.texi (Canonicalization of Instructions): Describe the
22508 canonical form of instructions that inherently set a condition
22511 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
22513 PR middle-end/80775
22514 * tree-cfg.c: Move deletion of unreachable case statements to after
22515 the merging of consecutive case labels.
22517 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
22519 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
22520 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
22521 restoring of callee-saved registers.
22523 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
22525 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
22526 * config/visium/visium.c (single_set_and_flags): Likewise.
22527 * config/visium/visium.md (Substitutions): Likewise.
22529 2017-05-17 Martin Liska <mliska@suse.cz>
22531 * cfg.c: Introduce dump_flags_t type and
22532 use it instead of int type.
22534 * cfghooks.c: Likewise.
22535 * cfghooks.h (struct cfg_hooks): Likewise.
22536 * cfgrtl.c: Likewise.
22537 * cfgrtl.h: Likewise.
22538 * cgraph.c (cgraph_node::get_body): Likewise.
22539 * coretypes.h: Likewise.
22540 * domwalk.c: Likewise.
22541 * domwalk.h: Likewise.
22542 * dumpfile.c (struct dump_option_value_info): Likewise.
22543 (dump_enable_all): Likewise.
22544 (dump_switch_p_1): Likewise.
22545 (opt_info_switch_p): Likewise.
22546 * dumpfile.h (enum tree_dump_index): Likewise.
22547 (struct dump_file_info): Likewise.
22548 * genemit.c: Likewise.
22549 * generic-match-head.c: Likewise.
22550 * gengtype.c (open_base_files): Likewise.
22551 * gimple-pretty-print.c: Likewise.
22552 * gimple-pretty-print.h: Likewise.
22553 * graph.c (print_graph_cfg): Likewise.
22554 * graphite-scop-detection.c (dot_all_sese): Likewise.
22555 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
22556 * loop-unroll.c (report_unroll): Likewise.
22557 * passes.c (pass_manager::register_one_dump_file): Likewise.
22558 * print-tree.c: Likewise.
22559 * statistics.c: Likewise.
22560 * tree-cfg.c: Likewise.
22561 * tree-cfg.h: Likewise.
22562 * tree-dfa.c: Likewise.
22563 * tree-dfa.h: Likewise.
22564 * tree-dump.c (dump_function): Likewise.
22565 * tree-dump.h (struct dump_info): Likewise.
22566 * tree-pretty-print.c: Likewise.
22567 * tree-pretty-print.h: Likewise.
22568 * tree-ssa-live.c: Likewise.
22569 * tree-ssa-live.h: Likewise.
22570 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
22571 * tree-vect-loop.c: Likewise.
22572 * tree-vect-slp.c: Likewise.
22574 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
22575 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22577 PR tree-optimization/80457
22578 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
22579 of all arguments to a statement as scalar_to_vec operations.
22580 (vectorizable_call): Adjust call to vect_model_simple_cost for
22582 (vectorizable_conversion): Likewise.
22583 (vectorizable_assignment): Likewise.
22584 (vectorizable_shift): Likewise.
22585 (vectorizable_operation): Likewise.
22586 (vectorizable_comparison): Likewise.
22587 (vect_is_simple_cond): Record the def types for operands.
22588 (vectorizable_condition): Likewise, call vect_model_simple_cost.
22589 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
22590 for statement argument count.
22592 2017-05-16 Carl Love <cel@us.ibm.com>
22594 * config/rs6000/rs6000-c: Add support for built-in functions
22595 vector unsigned long long vec_bperm (vector unsigned long long,
22596 vector unsigned char)
22597 vector signed long long vec_mule (vector signed int,
22599 vector unsigned long long vec_mule (vector unsigned int,
22600 vector unsigned int)
22601 vector signed long long vec_mulo (vector signed int,
22603 vector unsigned long long vec_mulo (vector unsigned int,
22604 vector unsigned int)
22605 vector signed char vec_sldw (vector signed char,
22606 vector signed char,
22608 vector unsigned char vec_sldw (vector unsigned char,
22609 vector unsigned char,
22611 vector signed short vec_sldw (vector signed short,
22612 vector signed short,
22614 vector unsigned short vec_sldw (vector unsigned short,
22615 vector unsigned short,
22617 vector signed int vec_sldw (vector signed int,
22620 vector unsigned int vec_sldw (vector unsigned int,
22621 vector unsigned int,
22623 vector signed long long vec_sldw (vector signed long long,
22624 vector signed long long,
22626 vector unsigned long long vec_sldw (vector unsigned long long,
22627 vector unsigned long long,
22629 * config/rs6000/rs6000-c: Add support for built-in functions
22630 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
22631 * config/rs6000/altivec.h: Add defintion for vec_sldw.
22632 * doc/extend.texi: Update the built-in documentation for the
22633 new built-in functions.
22635 2017-05-16 Marek Polacek <polacek@redhat.com>
22639 * tree.c (save_expr): Don't fold the expression.
22641 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
22643 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
22644 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
22645 and (?*y,m). Update insn attributes.
22647 2017-05-16 Martin Liska <mliska@suse.cz>
22649 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
22650 flags argument of print_gimple_stmt, print_gimple_expr,
22651 print_generic_stmt and print_generic_expr.
22652 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
22653 * coretypes.h: Likewise.
22654 * except.c (dump_eh_tree): Likewise.
22655 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
22656 * gimple-pretty-print.h: Likewise.
22657 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
22658 (backprop::push_to_worklist): Likewise.
22659 (backprop::pop_from_worklist): Likewise.
22660 (backprop::process_use): Likewise.
22661 (backprop::intersect_uses): Likewise.
22662 (note_replacement): Likewise.
22663 * gimple-ssa-store-merging.c
22664 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
22665 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
22666 (pass_store_merging::execute): Likewise.
22667 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
22668 (ssa_base_cand_dump_callback): Likewise.
22669 (dump_incr_vec): Likewise.
22670 (replace_refs): Likewise.
22671 (replace_mult_candidate): Likewise.
22672 (create_add_on_incoming_edge): Likewise.
22673 (create_phi_basis): Likewise.
22674 (insert_initializers): Likewise.
22675 (all_phi_incrs_profitable): Likewise.
22676 (introduce_cast_before_cand): Likewise.
22677 (replace_one_candidate): Likewise.
22678 * gimplify.c (gimplify_expr): Likewise.
22679 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
22680 (set_rename): Likewise.
22681 (rename_uses): Likewise.
22682 (copy_loop_phi_nodes): Likewise.
22683 (add_close_phis_to_merge_points): Likewise.
22684 (copy_loop_close_phi_args): Likewise.
22685 (copy_cond_phi_args): Likewise.
22686 (graphite_copy_stmts_from_block): Likewise.
22687 (translate_pending_phi_nodes): Likewise.
22688 * graphite-poly.c (print_pdr): Likewise.
22689 (dump_gbb_cases): Likewise.
22690 (dump_gbb_conditions): Likewise.
22691 (print_scop_params): Likewise.
22692 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
22693 (build_cross_bb_scalars_use): Likewise.
22694 (gather_bbs::before_dom_children): Likewise.
22695 * hsa-dump.c (dump_hsa_immed): Likewise.
22696 * ipa-cp.c (print_ipcp_constant_value): Likewise.
22697 (get_replacement_map): Likewise.
22698 * ipa-inline-analysis.c (dump_condition): Likewise.
22699 (estimate_function_body_sizes): Likewise.
22700 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
22701 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
22702 * ipa-prop.c (ipa_dump_param): Likewise.
22703 (ipa_print_node_jump_functions_for_edge): Likewise.
22704 (ipa_modify_call_arguments): Likewise.
22705 (ipa_modify_expr): Likewise.
22706 (ipa_dump_param_adjustments): Likewise.
22707 (ipa_dump_agg_replacement_values): Likewise.
22708 (ipcp_modif_dom_walker::before_dom_children): Likewise.
22709 * ipa-pure-const.c (check_stmt): Likewise.
22710 (pass_nothrow::execute): Likewise.
22711 * ipa-split.c (execute_split_functions): Likewise.
22712 * omp-offload.c (dump_oacc_loop_part): Likewise.
22713 (dump_oacc_loop): Likewise.
22714 * trans-mem.c (tm_log_emit): Likewise.
22715 (tm_memopt_accumulate_memops): Likewise.
22716 (dump_tm_memopt_set): Likewise.
22717 (dump_tm_memopt_transform): Likewise.
22718 * tree-cfg.c (gimple_verify_flow_info): Likewise.
22719 (print_loop): Likewise.
22720 * tree-chkp-opt.c (chkp_print_addr): Likewise.
22721 (chkp_gather_checks_info): Likewise.
22722 (chkp_get_check_result): Likewise.
22723 (chkp_remove_check_if_pass): Likewise.
22724 (chkp_use_outer_bounds_if_possible): Likewise.
22725 (chkp_reduce_bounds_lifetime): Likewise.
22726 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
22727 (chkp_mark_completed_bounds): Likewise.
22728 (chkp_register_incomplete_bounds): Likewise.
22729 (chkp_mark_invalid_bounds): Likewise.
22730 (chkp_maybe_copy_and_register_bounds): Likewise.
22731 (chkp_build_returned_bound): Likewise.
22732 (chkp_get_bound_for_parm): Likewise.
22733 (chkp_build_bndldx): Likewise.
22734 (chkp_get_bounds_by_definition): Likewise.
22735 (chkp_generate_extern_var_bounds): Likewise.
22736 (chkp_get_bounds_for_decl_addr): Likewise.
22737 * tree-chrec.c (chrec_apply): Likewise.
22738 * tree-data-ref.c (dump_data_reference): Likewise.
22739 (dump_subscript): Likewise.
22740 (dump_data_dependence_relation): Likewise.
22741 (analyze_overlapping_iterations): Likewise.
22742 * tree-inline.c (expand_call_inline): Likewise.
22743 (tree_function_versioning): Likewise.
22744 * tree-into-ssa.c (dump_defs_stack): Likewise.
22745 (dump_currdefs): Likewise.
22746 (dump_names_replaced_by): Likewise.
22747 (dump_update_ssa): Likewise.
22748 (update_ssa): Likewise.
22749 * tree-object-size.c (pass_object_sizes::execute): Likewise.
22750 * tree-parloops.c (build_new_reduction): Likewise.
22751 (try_create_reduction_list): Likewise.
22752 (ref_conflicts_with_region): Likewise.
22753 (oacc_entry_exit_ok_1): Likewise.
22754 (oacc_entry_exit_single_gang): Likewise.
22755 * tree-pretty-print.h: Likewise.
22756 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
22757 (get_scalar_evolution): Likewise.
22758 (add_to_evolution): Likewise.
22759 (get_loop_exit_condition): Likewise.
22760 (analyze_evolution_in_loop): Likewise.
22761 (analyze_initial_condition): Likewise.
22762 (analyze_scalar_evolution): Likewise.
22763 (instantiate_scev): Likewise.
22764 (number_of_latch_executions): Likewise.
22765 (gather_chrec_stats): Likewise.
22766 (final_value_replacement_loop): Likewise.
22767 (scev_const_prop): Likewise.
22768 * tree-sra.c (dump_access): Likewise.
22769 (disqualify_candidate): Likewise.
22770 (create_access): Likewise.
22771 (reject): Likewise.
22772 (maybe_add_sra_candidate): Likewise.
22773 (create_access_replacement): Likewise.
22774 (analyze_access_subtree): Likewise.
22775 (analyze_all_variable_accesses): Likewise.
22776 (sra_modify_assign): Likewise.
22777 (initialize_constant_pool_replacements): Likewise.
22778 (find_param_candidates): Likewise.
22779 (decide_one_param_reduction): Likewise.
22780 (replace_removed_params_ssa_names): Likewise.
22781 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
22782 * tree-ssa-copy.c (dump_copy_of): Likewise.
22783 (copy_prop_visit_cond_stmt): Likewise.
22784 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
22785 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
22786 (record_equivalences_from_stmt): Likewise.
22787 * tree-ssa-dse.c (compute_trims): Likewise.
22788 (delete_dead_call): Likewise.
22789 (delete_dead_assignment): Likewise.
22790 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
22791 (forward_propagate_into_cond): Likewise.
22792 (pass_forwprop::execute): Likewise.
22793 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
22794 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
22796 (move_computations_worker): Likewise.
22797 (execute_sm): Likewise.
22798 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
22799 (remove_exits_and_undefined_stmts): Likewise.
22800 (remove_redundant_iv_tests): Likewise.
22801 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
22802 (adjust_iv_update_pos): Likewise.
22803 * tree-ssa-math-opts.c (bswap_replace): Likewise.
22804 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
22805 (value_replacement): Likewise.
22806 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
22807 * tree-ssa-pre.c (print_pre_expr): Likewise.
22808 (get_representative_for): Likewise.
22809 (create_expression_by_pieces): Likewise.
22810 (insert_into_preds_of_block): Likewise.
22811 (eliminate_insert): Likewise.
22812 (eliminate_dom_walker::before_dom_children): Likewise.
22813 (eliminate): Likewise.
22814 (remove_dead_inserted_code): Likewise.
22815 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
22816 * tree-ssa-reassoc.c (get_rank): Likewise.
22817 (eliminate_duplicate_pair): Likewise.
22818 (eliminate_plus_minus_pair): Likewise.
22819 (eliminate_not_pairs): Likewise.
22820 (undistribute_ops_list): Likewise.
22821 (eliminate_redundant_comparison): Likewise.
22822 (update_range_test): Likewise.
22823 (optimize_range_tests_var_bound): Likewise.
22824 (optimize_vec_cond_expr): Likewise.
22825 (rewrite_expr_tree): Likewise.
22826 (rewrite_expr_tree_parallel): Likewise.
22827 (linearize_expr): Likewise.
22828 (break_up_subtract): Likewise.
22829 (linearize_expr_tree): Likewise.
22830 (attempt_builtin_powi): Likewise.
22831 (attempt_builtin_copysign): Likewise.
22832 (transform_stmt_to_copy): Likewise.
22833 (transform_stmt_to_multiply): Likewise.
22834 (dump_ops_vector): Likewise.
22835 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
22836 (print_scc): Likewise.
22837 (set_ssa_val_to): Likewise.
22838 (visit_reference_op_store): Likewise.
22839 (visit_use): Likewise.
22840 (sccvn_dom_walker::before_dom_children): Likewise.
22841 (run_scc_vn): Likewise.
22842 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
22844 (expr_hash_elt::print): Likewise.
22845 (const_and_copies::pop_to_marker): Likewise.
22846 (const_and_copies::record_const_or_copy_raw): Likewise.
22847 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
22848 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
22849 (dump_predicates): Likewise.
22850 (find_uninit_use): Likewise.
22851 (warn_uninitialized_phi): Likewise.
22852 (pass_late_warn_uninitialized::execute): Likewise.
22853 * tree-ssa.c (verify_vssa): Likewise.
22854 (verify_ssa): Likewise.
22855 (maybe_optimize_var): Likewise.
22856 * tree-vrp.c (dump_value_range): Likewise.
22857 (dump_all_value_ranges): Likewise.
22858 (dump_asserts_for): Likewise.
22859 (register_edge_assert_for_2): Likewise.
22860 (vrp_visit_cond_stmt): Likewise.
22861 (vrp_visit_switch_stmt): Likewise.
22862 (vrp_visit_stmt): Likewise.
22863 (vrp_visit_phi_node): Likewise.
22864 (simplify_cond_using_ranges_1): Likewise.
22865 (fold_predicate_in): Likewise.
22866 (evrp_dom_walker::before_dom_children): Likewise.
22867 (evrp_dom_walker::push_value_range): Likewise.
22868 (evrp_dom_walker::pop_value_range): Likewise.
22869 (execute_early_vrp): Likewise.
22871 2017-05-16 Richard Biener <rguenther@suse.de>
22873 * dwarf2out.c (loc_list_from_tree_1): Do not create
22874 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
22876 2017-05-16 Richard Biener <rguenther@suse.de>
22878 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
22880 (note_variable_value_in_expr): If we resolved the decl ref
22881 do not push to the stack.
22883 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
22885 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
22886 operations in fast-math mode.
22887 (vaddq_f16): Likewise.
22888 (vmul_f16): Likewise.
22889 (vmulq_f16): Likewise.
22890 (vsub_f16): Likewise.
22891 (vsubq_f16): Likewise.
22892 * config/arm/neon.md (add<mode>3): New.
22894 (fma:<VH:mode>3): New. Also remove outdated comment.
22897 2017-05-16 Martin Liska <mliska@suse.cz>
22901 * ipa-devirt.c (warn_types_mismatch): Fix typo.
22902 (odr_types_equivalent_p): Likewise.
22904 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
22906 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
22908 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
22911 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
22912 non-interunit SSE move alternatives with '?'.
22913 (zero-extendsidi peephole2): New peephole to skip intermediate
22914 general register in SSE zero-extend sequence.
22916 2017-05-15 Jeff Law <law@redhat.com>
22918 * reorg.c (relax_delay_slots): Create a new variable to hold
22919 the temporary target rather than clobbering TARGET_LABEL.
22921 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
22922 missing argument to extract_bit_field call.
22923 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
22925 2017-05-15 Martin Liska <mliska@suse.cz>
22928 * gcc.c (process_command): Do not allow empty argument of -o option.
22930 2017-05-15 Renlin Li <renlin.li@arm.com>
22932 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
22933 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
22934 * config/aarch64/constraints.md (Usf): Add long call check.
22935 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
22936 (call_value): Likewise.
22937 (sibcall): Likewise.
22938 (sibcall_value): Likewise.
22940 (call_value_insn): New.
22941 (sibcall_insn): Update rtx pattern.
22942 (sibcall_value_insn): Likewise.
22943 (call_internal): Remove.
22944 (call_value_internal): Likewise.
22945 (sibcall_internal): Likewise.
22946 (sibcall_value_internal): Likewise.
22947 (call_reg): Likewise.
22948 (call_symbol): Likewise.
22949 (call_value_reg): Likewise.
22950 (call_value_symbol): Likewise.
22952 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
22955 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
22957 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
22959 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
22960 compatible with CCGOCmode and with CCZmode.
22962 2017-05-14 Martin Sebor <msebor@redhat.com>
22964 PR middle-end/77671
22965 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
22966 (gimple_fold_builtin_snprintf): Same.
22967 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
22968 (gimple_fold_builtin_snprintf): Same.
22969 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
22970 of character types.
22971 (is_call_safe): New function.
22972 (try_substitute_return_value): Call it.
22973 (try_simplify_call): New function.
22974 (pass_sprintf_length::handle_gimple_call): Call it.
22976 2017-05-14 Martin Sebor <msebor@redhat.com>
22978 PR middle-end/80669
22979 * builtins.c (expand_builtin_stpncpy): Simplify.
22981 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
22983 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
22984 * config/i386/i386.h
22985 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
22986 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
22987 (struct machine_function): Add new members call_ms2sysv,
22988 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
22989 (struct machine_frame_state): New fields sp_realigned and
22990 sp_realigned_offset.
22991 * config/i386/i386.c
22992 (enum xlogue_stub): New enum.
22993 (enum xlogue_stub_sets): New enum.
22994 (class xlogue_layout): New class.
22995 (struct ix86_frame): New fields stack_realign_allocate_offset,
22996 stack_realign_offset and outlined_save_offset. Modify comments to
22997 detail stack layout when using out-of-line stubs.
22998 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
22999 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
23000 -mcall-ms2sysv-xlogues.
23001 (stub_managed_regs): New static variable.
23002 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
23003 registers managed by out-of-line stub.
23004 (disable_call_ms2sysv_xlogues): New function.
23005 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
23006 m->call_ms2sysv when appropriate and compute frame layout for
23008 (sp_valid_at, fp_valid_at): New inline functions.
23009 (choose_basereg): New function.
23010 (choose_baseaddr): Add align parameter, use choose_basereg and modify
23012 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
23013 Use align parameter of choose_baseaddr to generated aligned SSE movs
23015 (pro_epilogue_adjust_stack): Modify to track
23016 machine_frame_state::sp_realigned.
23017 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
23018 (ix86_nsaved_sseregs): Likewise.
23019 (ix86_emit_save_regs): Likewise.
23020 (ix86_emit_save_regs_using_mov): Likewise.
23021 (ix86_emit_save_sse_regs_using_mov): Likewise.
23022 (get_scratch_register_on_entry): Likewise.
23023 (gen_frame_set): New function.
23024 (gen_frame_load): Likewise.
23025 (gen_frame_store): Likewise.
23026 (emit_outlined_ms2sysv_save): Likewise.
23027 (emit_outlined_ms2sysv_restore): Likewise.
23028 (ix86_expand_prologue): Modify stack re-alignment code and call
23029 emit_outlined_ms2sysv_save when appropriate.
23030 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
23031 parameter rtx_insn *insn, which allows the function to be used to only
23032 generate the notes.
23033 (ix86_expand_epilogue): Modify validity checks of frame and stack
23034 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
23035 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
23036 * config/i386/predicates.md
23037 (save_multiple): New predicate.
23038 (restore_multiple): Likewise.
23039 * config/i386/sse.md
23040 (save_multiple<mode>): New pattern.
23041 (save_multiple_realign<mode>): Likewise.
23042 (restore_multiple<mode>): Likewise.
23043 (restore_multiple_and_return<mode>): Likewise.
23044 (restore_multiple_leave_return<mode>): Likewise.
23045 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
23047 2017-05-14 Julia Koval <julia.koval@intel.com>
23049 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
23050 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
23051 (__builtin_ia32_xsetbv): New builtins.
23052 * config/i386/i386.c (ix86_expand_special_args_builtin):
23054 (ix86_expand_builtin): Special expand for new intrinsics.
23055 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
23056 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
23057 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
23059 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23061 * cfganal.c (inverted_post_order_compute): Change argument type
23063 * cfganal.h (inverted_post_order_compute): Adjust prototype.
23064 * df-core.c (rest_of_handle_df_initialize): Adjust.
23065 (rest_of_handle_df_finish): Likewise.
23066 (df_analyze_1): Likewise.
23067 (df_analyze): Likewise.
23068 (loop_inverted_post_order_compute): Change argument to be a vec *.
23069 (df_analyze_loop): Adjust.
23070 (df_get_n_blocks): Likewise.
23071 (df_get_postorder): Likewise.
23072 * df.h (struct df_d): Change field to be a vec.
23073 * lcm.c (compute_laterin): Adjust.
23074 (compute_available): Likewise.
23075 * lra-lives.c (lra_create_live_ranges_1): Likewise.
23076 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
23077 * tree-ssa-pre.c (compute_antic): Likewise.
23079 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23081 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
23082 (depth_first_search::depth_first_search): Change structure init
23083 function to this constructor.
23084 (depth_first_search::add_bb): Rename function to this member.
23085 (depth_first_search::execute): Likewise.
23086 (flow_dfs_compute_reverse_finish): Adjust.
23088 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23090 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
23091 (longest_simple_path): Likewise.
23092 * shrink-wrap.c (spread_components): Likewise.
23093 (disqualify_problematic_components): Likewise.
23094 (emit_common_heads_for_components): Likewise.
23095 (emit_common_tails_for_components): Likewise.
23096 (insert_prologue_epilogue_for_components): Likewise.
23098 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23100 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
23103 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23105 * df-core.c (df_set_blocks): Start using auto_bitmap.
23106 (df_compact_blocks): Likewise.
23107 * df-problems.c (df_rd_confluence_n): Likewise.
23108 * df-scan.c (df_insn_rescan_all): Likewise.
23109 (df_process_deferred_rescans): Likewise.
23110 (df_update_entry_block_defs): Likewise.
23111 (df_update_exit_block_uses): Likewise.
23112 (df_entry_block_bitmap_verify): Likewise.
23113 (df_exit_block_bitmap_verify): Likewise.
23114 (df_scan_verify): Likewise.
23115 * lra-constraints.c (lra_constraints): Likewise.
23116 (undo_optional_reloads): Likewise.
23117 (lra_undo_inheritance): Likewise.
23118 * lra-remat.c (calculate_gen_cands): Likewise.
23119 (do_remat): Likewise.
23120 * lra-spills.c (assign_spill_hard_regs): Likewise.
23121 (spill_pseudos): Likewise.
23122 * tree-ssa-pre.c (bitmap_set_and): Likewise.
23123 (bitmap_set_subtract_values): Likewise.
23125 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23127 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
23128 management with auto_bitmap.
23129 (fix_inter_tick): Likewise.
23130 (fix_recovery_deps): Likewise.
23131 * ira.c (add_store_equivs): Likewise.
23132 (find_moveable_pseudos): Likewise.
23133 (split_live_ranges_for_shrink_wrap): Likewise.
23134 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
23135 (rtx_reuse_manager::seen_def_p): Likewise.
23136 (rtx_reuse_manager::set_seen_def): Likewise.
23137 * print-rtl.h (class rtx_reuse_manager): Likewise.
23139 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23141 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
23143 (migrate_btr_def): Likewise.
23144 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
23145 * df-core.c (loop_post_order_compute): Likewise.
23146 (loop_inverted_post_order_compute): Likewise.
23147 * hsa-common.h: Likewise.
23148 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
23149 * init-regs.c (initialize_uninitialized_regs): Likewise.
23150 * ipa-inline.c (resolve_noninline_speculation): Likewise.
23151 (inline_small_functions): Likewise.
23152 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
23153 * ira.c (combine_and_move_insns): Likewise.
23154 (build_insn_chain): Likewise.
23155 * loop-invariant.c (find_invariants): Likewise.
23156 * lower-subreg.c (propagate_pseudo_copies): Likewise.
23157 * predict.c (tree_predict_by_opcode): Likewise.
23158 (predict_paths_leading_to): Likewise.
23159 (predict_paths_leading_to_edge): Likewise.
23160 (estimate_loops_at_level): Likewise.
23161 (estimate_loops): Likewise.
23162 * shrink-wrap.c (try_shrink_wrapping): Likewise.
23163 (spread_components): Likewise.
23164 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
23165 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
23166 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
23167 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
23168 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
23169 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
23170 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
23171 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
23172 (mark_threaded_blocks): Likewise.
23173 (thread_through_all_blocks): Likewise.
23174 * tree-ssa.c (verify_ssa): Likewise.
23175 (execute_update_addresses_taken): Likewise.
23176 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
23178 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23180 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
23182 (post_order_compute): Likewise.
23183 (inverted_post_order_compute): Likewise.
23184 (pre_and_rev_post_order_compute_fn): Likewise.
23186 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23188 * genrecog.c (int_set::int_set): Explicitly construct our
23189 auto_vec base class.
23190 * vec.h (auto_vec::auto_vec): New constructor.
23192 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23194 * bitmap.h (class auto_bitmap): New constructor taking
23195 bitmap_obstack * argument.
23197 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23199 * bitmap.h (class auto_bitmap): Change type of m_bits to
23200 bitmap_head, and adjust ctor / dtor and member operators.
23202 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
23204 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
23205 when returned register mode doesn't match original mode.
23207 2017-05-12 Jeff Law <law@redhat.com>
23208 Jakub Jelinek <jakub@redhat.com>
23210 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
23211 we look for cc setter after the compare-elim changes.
23212 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
23213 within the vector to match what compare-elim now expects.
23214 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
23215 (xorsi3_flags, one_cmplsi2_flags): Likewise.
23217 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
23218 after the compare-elim changes.
23219 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
23220 the vector to match what compare-elim now expects.
23221 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
23222 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
23223 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
23224 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
23225 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
23227 * config/visium/visium.c (single_set_and_flags): Fix where
23228 we look for cc setter after the compare-elim changes.
23229 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
23230 with the vector to match what compare-elim now expects.
23231 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
23232 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
23233 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
23234 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
23235 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
23236 (neg<mode>2_insn_set_overflow): Likewise.
23238 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
23240 PR middle-end/79794
23241 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
23242 maybe_expand_insn call, set ops[0].target. If still set after call,
23243 set alt_rtl. Add extra arg to recursive calls.
23244 (extract_bit_field): Add alt_rtl argument. Pass to
23246 * expmed.h (extract_bit_field): Fix prototype.
23247 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
23248 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
23249 to extract_bit_field_calls.
23250 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
23251 Pass alt_rtl to extract_bit_field calls.
23252 * calls.c (store_unaligned_arguments_into_psuedos)
23253 load_register_parameters): Pass extra NULL to extract_bit_field calls.
23254 * optabs.c (maybe_legitimize_operand): Clear op->target when call
23256 * optabs.h (struct expand_operand): Add target bitfield.
23258 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
23260 * compare-elim.c (try_eliminate_compare): Canonicalize
23261 operation with embedded compare to
23262 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
23263 (set (reg) (operation)].
23265 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
23267 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
23270 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
23271 cost of adding a carry flag for ADC instruction.
23272 [case MINUS]: Ignore the cost of subtracting a carry flag
23273 for SBB instruction.
23275 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
23277 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
23279 * config/rs6000/bmiintrin.h: New file.
23280 * config/rs6000/bmi2intrin.h: New file.
23281 * config/rs6000/x86intrin.h: New file.
23283 2017-05-12 Jeff Law <law@redhat.com>
23285 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
23288 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
23290 PR middle-end/80707
23291 * tree-cfg.c: Remove cfg edges of unreachable case statements.
23293 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23295 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
23296 early expansion of vector divide builtins.
23297 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
23298 builtins identified as having unsigned arguments.
23300 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23302 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
23303 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
23304 expansion of vector logical operations (and, andc, or, xor,
23307 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23309 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
23310 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
23312 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23314 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
23315 early GIMPLE expansion of vector multiplies.
23317 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23319 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
23320 TARGET_HAVE_MOVT conditional.
23321 (movt splitter): Likewise.
23323 2017-05-12 Richard Biener <rguenther@suse.de>
23325 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
23326 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
23327 Fold all stmts not inplace.
23329 2017-05-12 Richard Biener <rguenther@suse.de>
23331 PR tree-optimization/80713
23332 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
23333 inserted_exprs bit for not removed stmts.
23335 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
23337 PR middle-end/69921
23338 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
23339 parallelized" attribute for parallelized OpenACC kernels.
23340 * omp-offload.c (execute_oacc_device_lower): Use it.
23342 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
23343 Set "oacc kernels" attribute.
23344 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
23345 parameter. Adjust all users.
23346 (oacc_fn_attrib_kernels_p): Remove function.
23347 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
23348 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
23349 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
23350 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
23351 assert "oacc kernels" attribute is set.
23353 2017-05-11 Carl Love <cel@us.ibm.com>
23355 * config/rs6000/rs6000-c: Add support for built-in functions
23356 vector unsigned char vec_popcnt (vector signed char)
23357 vector unsigned char vec_popcnt (vector unsigned char)
23358 vector unsigned short vec_popcnt (vector signed short)
23359 vector unsigned short vec_popcnt (vector unsigned short)
23360 vector unsigned int vec_popcnt (vector signed int)
23361 vector unsigned int vec_popcnt (vector unsigned int)
23362 vector unsigned long long vec_popcnt (vector signed long long)
23363 vector unsigned long long vec_popcnt (vector unsigned long long)
23364 vector signed long long vec_slo (vector signed long long,
23365 vector signed char)
23366 vector signed long long vec_slo (vector signed long long,
23367 vector unsigned char)
23368 vector unsigned long long vec_slo (vector unsigned long long,
23369 vector signed char)
23370 vector unsigned long long vec_slo (vector unsigned long long,
23371 vector unsigned char)
23372 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
23373 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
23374 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
23375 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
23376 * doc/extend.texi: Update the built-in documentation file for the
23377 new built-in functions.
23379 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
23381 * attribs.h (sorted_attr_string): Move machine independent
23382 functions for target clone support from the i386 port to common
23383 code. Rename ix86_function_versions to common_function_versions.
23384 Rename make_name to make_unique_name.
23385 (common_function_versions): Likewise.
23386 (make_unique_name): Likewise.
23387 (make_dispatcher_decl): Likewise.
23388 (is_function_default_version): Likewise.
23389 * attribs.c (attr_strcmp): Likewise.
23390 (sorted_attr_string): Likewise.
23391 (common_function_versions): Likewise.
23392 (make_unique_name): Likewise.
23393 (make_dispatcher_decl): Likewise.
23394 (is_function_default_version): Likewise.
23395 * config/i386/i386.c (attr_strcmp): Likewise.
23396 (sorted_attr_string): Likewise.
23397 (ix86_function_versions): Likewise.
23398 (make_name): Likewise.
23399 (make_dispatcher_decl): Likewise.
23400 (is_function_default_version): Likewise.
23401 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
23403 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23406 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23407 Account for direct move costs for vec_construct of integer
23410 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
23413 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
23414 (UNSPEC_STX_ATOMIC): Ditto.
23415 (loaddi_via_sse): New insn.
23416 (storedi_via_sse): Ditto.
23417 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
23418 Update corresponding peephole2 patterns.
23419 (atomic_storedi_fpu): Ditto.
23421 2017-05-11 Julia Koval <julia.koval@intel.com>
23423 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
23424 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
23426 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
23427 (__builtin_ia32_rsqrt14ss_mask): New builtins.
23428 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
23430 2017-05-11 Nathan Sidwell <nathan@acm.org>
23432 * graphite-poly.c: Include dumpfile.h.
23434 * dumpfle.h (dump_function): Declare here ...
23435 * tree-dump.h (dump_function): ... not here.
23436 * dumpfile.c: #include tree-cfg.h.
23437 (dump_function): Move here from ...
23438 * tree-dump.c (dump_function): ... here.
23439 * gimplify.c: #include splay-tree.h, not tree-dump.h.
23440 * graphite-poly.c: Don't include tree-dump.h.
23441 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
23442 * print-tree.c: Likewise.
23443 * stor-layout.c: Likewise.
23444 * tree-nested.c: Likewise.
23446 * dumpfile.c (dump_start): Use TDF_FLAGS.
23447 (dump_enable_all): Fix TDF_KIND check thinko.
23449 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
23451 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23452 array entries to represent two legal parameterizations of the
23453 overloaded __builtin_cmpb function, as represented by the
23454 P6_OV_BUILTIN_CMPB constant.
23455 (altivec_resolve_overloaded_builtin): Add special case handling
23456 for the __builtin_cmpb function, as represented by the
23457 P6_OV_BUILTIN_CMPB constant.
23458 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
23459 (BU_P6_64BIT_2): New macro.
23460 (BU_P6_OVERLOAD_2): New macro
23461 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
23462 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
23463 (CMPB): Add overload support to represent both 32-bit and 64-bit
23464 compare-bytes function.
23465 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
23466 support for TARGET_CMPB.
23467 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
23468 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23469 documentation of the __builtin_cmpb overloaded built-in function.
23471 2017-05-11 Richard Biener <rguenther@suse.de>
23473 PR tree-optimization/80705
23474 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
23475 bases are not vectorizable.
23477 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23479 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
23480 when counting register pressure.
23482 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23484 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
23485 (iv_ca_more_deps): Renamed to ...
23486 (iv_ca_compare_deps): ... this.
23487 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
23489 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23491 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
23493 (determine_group_iv_costs): ... here.
23494 (find_inv_vars_cb): Record inv var if it's not recorded before.
23496 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23498 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
23499 (get_shiftadd_cost): Ditto.
23501 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23503 * tree-ssa-address.c: Include header file.
23504 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
23506 (add_to_parts): Refactor.
23507 (addr_to_parts): New parameter. Update use of move_hint_to_base.
23508 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
23511 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23513 PR tree-optimization/53090
23514 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
23516 (extract_cond_operands): Detect condition with IV on both sides
23517 and return COMP_IV_EXPR_2.
23518 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
23519 (rewrite_use_compare): Simplify by removing call to function
23520 extract_cond_operands.
23522 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23524 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
23525 (extract_cond_operands): Detect condition comparing against non-
23526 invariant bound and return appropriate enum value.
23527 (find_interesting_uses_cond): Update use of extract_cond_operands.
23528 Handle its return value accordingly.
23529 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
23531 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23533 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
23534 nonlinear iv_use computation in loop invariant sensitive way.
23536 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23538 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
23539 (find_iv_candidates): Call relate_compare_use_with_all_cands.
23541 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23543 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
23544 (dump_cand): Support iv_cand.inv_exprs.
23545 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
23547 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
23550 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23552 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
23554 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
23555 as local function. Include necessary header files.
23556 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
23558 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23560 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
23562 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23564 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
23565 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
23566 RSHIFT_EXPR and BIT_NOT_EXPR.
23568 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23570 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
23571 (adjust_setup_cost): New parameter supporting round up adjustment.
23572 (struct address_cost_data): Delete.
23573 (force_expr_to_var_cost): Don't bound cost with spill_cost.
23574 (split_address_cost, ptr_difference_cost): Delete.
23575 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
23576 (struct ainc_cost_data): New struct.
23577 (get_address_cost_ainc): New function.
23578 (get_address_cost, get_computation_cost): Reimplement.
23579 (determine_group_iv_cost_address): Record inv_expr for all uses of
23581 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
23582 (iv_ca_has_deps): Reimplemented to ...
23583 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
23585 (iv_ca_extend): Call iv_ca_more_deps.
23587 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23589 * tree-ssa-address.c (struct mem_address): Move to header file.
23590 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
23591 * tree-ssa-address.h (struct mem_address): Move from C file.
23592 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
23594 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23596 * tree-affine.h (aff_combination_type): New interface.
23597 (aff_combination_zero_p): Remove static.
23598 (aff_combination_const_p): New interface.
23599 (aff_combination_singleton_var_p): New interfaces.
23601 2017-05-11 Richard Biener <rguenther@suse.de>
23603 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
23604 Skip unreachable blocks and destinations.
23605 (eliminate): Move stmt removal and fixup ...
23606 (fini_eliminate): ... here. Skip inserted exprs.
23607 (pass_pre::execute): Move fini_pre after fini_eliminate.
23608 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
23609 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
23610 PRE to get rid of dead code that has invalid SSA form and
23611 split critical edges again.
23613 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23615 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
23617 2017-05-11 Richard Biener <rguenther@suse.de>
23619 * passes.c (execute_function_todo): Verify loops if they are
23620 said to be up-to-date.
23621 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
23622 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
23624 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
23627 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
23628 handle calling assemble_external ourself.
23631 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
23632 modes with zero size. Enhance comment.
23634 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23636 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
23637 built-ins for vec_xl and vec_xst with short and char pointer
23640 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
23642 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
23643 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
23644 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
23645 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
23646 (_mm_maskz_min_round_ss): New intrinsics.
23647 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
23648 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
23649 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
23650 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
23651 (__builtin_ia32_minss_mask_round): New builtins.
23652 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23653 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23654 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
23656 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
23657 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
23659 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
23662 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
23664 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
23665 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
23666 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
23667 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
23668 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
23669 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
23670 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
23671 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23672 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
23673 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
23674 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
23675 (__builtin_ia32_mulss_mask_round): New builtins.
23676 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23677 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23678 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
23680 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
23681 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
23683 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
23686 2017-05-10 Julia Koval <julia.koval@intel.com>
23688 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
23689 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
23690 (_mm256_setr_m128i): New intrinsics.
23692 2017-05-10 Julia Koval <julia.koval@intel.com>
23694 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
23695 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
23696 (_mm_maskz_rcp14_ss): New intrinsics.
23697 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
23698 (__builtin_ia32_rcp14ss_mask): New builtins.
23699 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
23701 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
23703 PR tree-optimization/51513
23704 * tree-cfg.c (gimple_seq_unreachable_p): New function.
23705 (assert_unreachable_fallthru_edge_p): Use it.
23706 (group_case_labels_stmt): Likewise.
23707 * tree-cfg.h: Prototype it.
23708 * stmt.c: Include cfghooks.h and tree-cfg.h.
23709 (emit_case_dispatch_table) <gap_label>: New local variable.
23710 Use it to fill dispatch table gaps.
23711 Test for default_label before updating probabilities.
23712 (expand_case) <default_label>: Remove unneeded initialization.
23713 Test for unreachable default case statement and remove its edge.
23714 Set default_label accordingly.
23715 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
23717 2017-05-10 Carl Love <cel@us.ibm.com>
23719 * config/rs6000/rs6000-c: Add support for built-in functions
23720 vector signed char vec_neg (vector signed char)
23721 vector signed short int vec_neg (vector short int)
23722 vector signed int vec_neg (vector signed int)
23723 vector signed long long vec_neg (vector signed long long)
23724 vector float vec_neg (vector float)
23725 vector double vec_neg (vector double)
23726 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
23728 * config/rs6000/altivec.h: Add define for vec_neg
23729 * doc/extend.texi: Update the built-in documentation for the
23730 new built-in functions.
23732 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23734 PR tree-optimization/77644
23735 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
23737 2017-05-10 Nathan Sidwell <nathan@acm.org>
23739 * dumpfile.h (TDI_lang_all): New.
23740 (TDF_KIND): New. Renumber others
23741 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
23743 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
23745 (get_dump_file_name): Adjust suffix generation.
23746 (dump_enable_all): Use TDF_KIND.
23747 * doc/invoke.texi (-fdump-lang-all): Document.
23749 * dumpfile.h: Tabify.
23751 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
23754 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
23755 Move member access before delete.
23757 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
23759 * tree-inline.c (expand_call_inline): Split block at stmt
23762 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
23765 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
23766 are now unused after splitting mov{sf,sd}_hardfloat.
23767 (f32_lr2): Likewise.
23768 (f32_lm): Likewise.
23769 (f32_lm2): Likewise.
23770 (f32_li): Likewise.
23771 (f32_li2): Likewise.
23772 (f32_lv): Likewise.
23773 (f32_sr): Likewise.
23774 (f32_sr2): Likewise.
23775 (f32_sm): Likewise.
23776 (f32_sm2): Likewise.
23777 (f32_si): Likewise.
23778 (f32_si2): Likewise.
23779 (f32_sv): Likewise.
23780 (f32_dm): Likewise.
23781 (f32_vsx): Likewise.
23782 (f32_av): Likewise.
23783 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
23784 For movsf, order stores so the VSX stores occur before the GPR
23785 store which encourages the register allocator to use a traditional
23786 FPR instead of a GPR. For movsd, order the stores so that the GPR
23787 store comes before the VSX stores to allow the power6 to work.
23788 This is due to the power6 not having a 32-bit integer store
23789 instruction from a FPR.
23790 (movsf_hardfloat): Likewise.
23791 (movsd_hardfloat): Likewise.
23793 2017-05-09 Martin Sebor <msebor@redhat.com>
23795 PR translation/80280
23796 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
23799 PR translation/80280
23800 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
23801 data member added in r247778.
23802 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
23804 2017-05-09 Nathan Sidwell <nathan@acm.org>
23806 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
23808 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
23811 2017-05-09 Marek Polacek <polacek@redhat.com>
23813 * doc/invoke.texi: Fix typo.
23815 2017-05-09 Richard Biener <rguenther@suse.de>
23817 * tree-vrp.c (vrp_val_is_max): Adjust comment.
23818 (vrp_val_is_min): Likewise.
23819 (set_value_range_to_value): Likewise.
23820 (set_value_range_to_nonnegative): Likewise.
23821 (gimple_assign_nonzero_p): Likewise.
23822 (gimple_stmt_nonzero_p): Likewise.
23823 (vrp_int_const_binop): Likewise. Remove unreachable case.
23824 (adjust_range_with_scev): Adjust comments.
23825 (compare_range_with_value): Likewise.
23826 (extract_range_from_phi_node): Likewise.
23827 (test_for_singularity): Likewise.
23829 2017-05-09 Richard Biener <rguenther@suse.de>
23831 * tree-vrp.c (get_single_symbol): Add assert that we don't
23832 get overflowed constants as invariant part.
23833 (compare_values_warnv): Add comment before the TREE_NO_WARNING
23834 checks. Use wi::cmp instead of recursing for integer constants.
23835 (compare_values): Just ignore whether we assumed undefined
23836 overflow instead of failing the compare.
23837 (extract_range_for_var_from_comparison_expr): Add comment before the
23838 TREE_NO_WARNING sets.
23839 (test_for_singularity): Likewise.
23840 (extract_range_from_comparison): Do not disable optimization
23841 when we assumed undefined overflow.
23842 (extract_range_basic): Remove init of unused var.
23844 2017-05-09 Richard Biener <rguenther@suse.de>
23846 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
23847 (extract_range_from_multiplicative_op_1): Adjust.
23848 (extract_range_from_binary_expr_1): Use int_const_binop.
23850 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
23853 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
23854 rs6000_store_data_bypass_p in seven define_bypass directives and
23855 in several comments.
23856 * config/rs6000/rs6000-protos.h: Add prototype for
23857 rs6000_store_data_bypass_p function.
23858 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
23859 function implements slightly different (rs6000-specific) semantics
23860 than store_data_bypass_p, returning false rather than aborting
23861 with assertion error when arguments do not satisfy the
23862 requirements of store data bypass.
23863 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
23864 rs6000_store_data_bypass_p.
23866 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
23868 * config/xtensa/xtensa-protos.h
23869 (xtensa_initial_elimination_offset): New declaration.
23870 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
23871 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
23872 macro definition, add case for FRAME_POINTER_REGNUM when
23873 FRAME_GROWS_DOWNWARD.
23874 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
23875 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
23876 xtensa_initial_elimination_offset.
23878 2017-05-08 Nathan Sidwell <nathan@acm.org>
23880 * doc/invoke.texi: Alphabetize -fdump options.
23882 2017-05-08 Martin Sebor <msebor@redhat.com>
23884 PR translation/80280
23885 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
23887 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
23889 * target.def (compute_frame_layout): New optional target hook.
23890 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
23891 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
23892 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
23894 * reload1.c (verify_initial_elim_offsets): Likewise.
23895 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
23896 (use_simple_return_p): Call arm_compute_frame_layout if needed.
23897 (arm_get_frame_offsets): Split up into this ...
23898 (arm_compute_frame_layout): ... and this function.
23900 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
23902 * config/aarch64/constraints.md (Usa): New constraint.
23903 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
23905 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
23907 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
23908 with_multilib_list after it has been checked.
23910 2017-05-08 Richard Biener <rguenther@suse.de>
23912 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
23913 (bitmap_set_subtract_values): Likewise.
23915 2017-05-08 Richard Biener <rguenther@suse.de>
23917 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
23918 (gimple_assign_nonzero): ... this and remove strict_overflow_p
23920 (gimple_stmt_nonzero_warnv_p): Rename to ...
23921 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
23923 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
23924 (extract_range_basic): Adjust, do not disable propagation on
23925 strict overflow sensitive simplification.
23926 (vrp_visit_cond_stmt): Likewise.
23928 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
23930 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
23931 body size unconditionally.
23933 2017-05-07 Jeff Law <law@redhat.com>
23936 2017-05-06 Jeff Law <law@redhat.com>
23937 PR tree-optimization/78496
23938 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
23941 PR tree-optimization/78496
23942 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
23943 (simplify_stmt_using_ranges): Call it.
23944 (vrp_dom_walker::before_dom_children): Extract equivalences
23945 from an ASSERT_EXPR with an equality comparison against a
23948 2017-05-06 Jeff Law <law@redhat.com>
23950 PR tree-optimization/78496
23951 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
23954 PR tree-optimization/78496
23955 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
23956 (simplify_stmt_using_ranges): Call it.
23957 (vrp_dom_walker::before_dom_children): Extract equivalences
23958 from an ASSERT_EXPR with an equality comparison against a
23961 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
23963 * lra-constraints.c (lra_copy_reg_equiv): New function.
23964 (split_reg): Use it to copy equivalence information from the
23965 original register to the spill register.
23967 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
23969 PR rtl-optimization/75964
23970 * simplify-rtx.c (simplify_const_relational_operation): Remove
23971 invalid handling of comparisons of integer ABS.
23973 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
23975 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
23976 initialize to zero.
23977 (init_regs): Remove declaration.
23978 (function_arg_advance_32): Initialize error_p as boolean variable.
23980 2017-05-05 Nathan Sidwell <nathan@acm.org>
23982 * store-motion.c (remove_reachable_equiv_notes): Reformat long
23983 lines. Use for (;;).
23985 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23987 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
23988 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
23989 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
23990 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
23991 VF=2 that require versioning.
23993 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23995 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
23998 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24000 * diagnostic.h (diagnostic_override_option_index): Convert from
24001 macro to inline function.
24003 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24005 * diagnostic.c (last_module_changed_p): New function.
24006 (set_last_module): New function.
24007 (diagnostic_report_current_module): Convert macro usage to
24008 the above functions.
24009 * diagnostic.h (diagnostic_context::last_module): Strengthen
24010 from const line_map * to const line_map_ordinary *.
24011 (diagnostic_last_module_changed): Delete macro.
24012 (diagnostic_set_last_module): Delete macro.
24014 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24016 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
24017 with diagnostic_report_diagnostic.
24018 (diagnostic_n_impl_richloc): Likewise.
24019 * diagnostic.h (report_diagnostic): Delete macro.
24020 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
24021 with diagnostic_report_diagnostic.
24022 * substring-locations.c (format_warning_va): Likewise.
24024 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24026 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
24027 save/restor of format_spec. Move option-printing code to...
24028 (print_option_information): ...this new function, and
24029 reimplement by simply printing to the pretty_printer,
24030 rather than appending to the format string.
24032 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24034 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
24035 handling logic into...
24036 (update_effective_level_from_pragmas): ...this new function.
24038 2017-05-04 Andrew Waterman <andrew@sifive.com>
24040 * config/riscv/riscv.opt (mstrict-align): New option.
24041 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
24042 (SLOW_UNALIGNED_ACCESS): Define.
24043 (riscv_slow_unaligned_access): Declare.
24044 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
24046 (riscv_slow_unaligned_access): New variable.
24047 (rocket_tune_info): Set slow_unaligned_access to true.
24048 (optimize_size_tune_info): Set slow_unaligned_access to false.
24049 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
24050 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
24051 (riscv_option_override): Set riscv_slow_unaligned_access.
24052 * doc/invoke.texi: Add -mstrict-align to RISC-V.
24054 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
24056 * config/riscv/riscv.md: Unify indentation.
24058 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
24063 * config/rs6000/rs6000.md (u code attribute): Add FIX and
24065 (extendsi<mode>2): Add support for doing sign extension via
24066 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
24067 don't have ISA 3.0 instructions.
24068 (extendsi<mode>2 splitter): Likewise.
24069 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
24070 generate the normal insns since SImode can now go in vector
24071 registers. Disallow the special UNSPECs needed for previous
24072 machines to hide SImode being used. Add new insns
24073 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
24074 (fix_trunc<mode>si2_stfiwx): Likewise.
24075 (fix_trunc<mode>si2_internal): Likewise.
24076 (fixuns_trunc<mode>si2): Likewise.
24077 (fixuns_trunc<mode>si2_stfiwx): Likewise.
24078 (fctiw<u>z_<mode>_smallint): Likewise.
24079 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
24080 of floating point to 32-bit integer from doing a direct move to
24081 the GPR registers to do a store.
24082 (fctiwz_<mode>): Break long line.
24084 2017-05-05 Bin Cheng <bin.cheng@arm.com>
24086 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
24087 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
24088 (addr_list, addr_offset_valid_p): New.
24089 (split_address_groups): Check offset validity with above function.
24090 (gt-tree-ssa-loop-ivopts.h): Include header file.
24092 2017-05-05 Nathan Sidwell <nathan@acm.org>
24094 * config.gcc (arm*-*-*): Add missing 'fi'.
24096 2017-05-05 Steve Ellcey <sellcey@cavium.com>
24098 * doc/invoke.texi (-fopt-info): Explicitly say order of options
24099 included in -fopt-info does not matter.
24100 * doc/optinfo.texi (-fopt-info): Fix description of default
24101 behavour. Explicitly say order of options included in -fopt-info
24104 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
24106 * config.gcc: Allow combinations of aprofile and rmprofile values for
24107 --with-multilib-list.
24108 * config/arm/t-multilib: New file.
24109 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
24110 variables. Remove setting of ISA and floating-point ABI in
24111 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
24112 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
24113 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
24114 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
24116 * config/arm/t-rmprofile: Likewise except for the matches changes.
24117 * doc/install.texi (--with-multilib-list): Document the combination of
24118 aprofile and rmprofile values and warn about pitfalls in doing that.
24120 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24122 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
24123 (movdi_aarch64): Likewise.
24125 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24127 PR tree-optimization/80632
24128 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
24130 (build_arrays): Initialize it for virtual phis.
24131 (fix_phi_nodes): Use it for virtual phis.
24133 PR tree-optimization/80558
24134 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
24135 [x, y] op z into [x op, y op z] for op & or | if conditions
24138 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
24139 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24142 * config/arm/arm.md (use_literal_pool): Remove.
24143 (64-bit immediate split): No longer takes cost into consideration
24144 if arm_disable_literal_pool is enabled.
24145 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
24146 used when arm_disable_literal_pool is enabled.
24147 (arm_max_const_double_inline_cost): Remove use of
24148 arm_disable_literal_pool.
24149 (push_minipool_fix): Add assert.
24150 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
24151 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
24152 (no_literal_pool_sf_immediate): New.
24154 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24156 PR tree-optimization/80613
24157 * tree-ssa-dce.c (propagate_necessity): Remove cases for
24158 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
24160 2017-05-05 Richard Biener <rguenther@suse.de>
24162 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
24164 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
24166 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
24167 of this flag from insn conditions due to removal from r247495.
24169 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24171 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
24173 (arm_early_store_addr_dep_ptr): Likewise.
24174 * config/arm/aarch-common-protos.h
24175 (arm_early_load_addr_dep_ptr): Add prototype.
24176 (arm_early_store_addr_dep_ptr): Likewise.
24177 * config/arm/cortex-a53.md: Add new bypasses.
24179 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24181 * tree.c (next_type_uid): Change type to unsigned.
24182 (type_hash_canon): Decrement back next_type_uid if
24183 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
24184 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
24187 2017-05-04 Martin Sebor <msebor@redhat.com>
24189 * builtins.c: Fix a trivial typo in a comment.
24191 PR middle-end/79234
24192 * builtins.c (check_sizes): Adjust to handle reading past the end.
24193 Avoid printing excessive upper bound of ranges. Use %E to print
24194 tree nodes instead of converting them to %wu.
24195 (expand_builtin_memchr): New function.
24196 (compute_dest_size): Rename...
24197 (compute_objsize): ...to this.
24198 (expand_builtin_memcpy): Adjust.
24199 (expand_builtin_mempcpy): Adjust.
24200 (expand_builtin_strcat): Adjust.
24201 (expand_builtin_strcpy): Adjust.
24202 (check_strncat_sizes): Adjust.
24203 (expand_builtin_strncat): Adjust.
24204 (expand_builtin_strncpy): Adjust and simplify.
24205 (expand_builtin_memset): Adjust.
24206 (expand_builtin_bzero): Adjust.
24207 (expand_builtin_memcmp): Adjust.
24208 (expand_builtin): Handle memcmp.
24209 (maybe_emit_chk_warning): Check strncat just once.
24211 2017-05-04 Martin Sebor <msebor@redhat.com>
24213 PR preprocessor/79214
24214 PR middle-end/79222
24215 PR middle-end/79223
24216 * builtins.c (check_sizes): Add inlining context and issue
24217 warnings even when -Wno-system-headers is set.
24218 (check_strncat_sizes): Same.
24219 (expand_builtin_strncat): Same.
24220 (expand_builtin_memmove): New function.
24221 (expand_builtin_stpncpy): Same.
24222 (expand_builtin): Handle memmove and stpncpy.
24224 2017-05-04 Bin Cheng <bin.cheng@arm.com>
24226 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
24227 which is not used any more.
24229 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24231 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
24233 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24235 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
24236 (cortexa53_tunings): Likewise.
24237 (cortexa57_tunings): Likewise.
24238 (cortexa72_tunings): Likewise.
24239 (cortexa73_tunings): Likewise.
24241 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24243 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
24244 Set loop alignment to 8.
24246 2017-05-04 Martin Sebor <msebor@redhat.com>
24248 PR translation/80280
24249 * builtins.c (expand_builtin_object_size): Add missing quoting to
24250 %D and like directives.
24251 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
24252 (hsa_type_for_tree_type): Same.
24253 (verify_function_arguments): Same.
24254 * symtab.c (symbol_table::change_decl_assembler_name): Same.
24255 * varasm.c (get_section): Same.
24258 2017-05-04 Martin Sebor <msebor@redhat.com>
24260 PR translation/80280
24261 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
24263 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24265 * config/aarch64/aarch64.c (generic_addrcost_table):
24266 Change HI/TI mode setting.
24268 2017-05-04 Martin Jambor <mjambor@suse.cz>
24270 PR tree-optimization/80622
24271 * tree-sra.c (comes_initialized_p): New function.
24272 (build_accesses_from_assign): Only set write lazily when
24273 comes_initialized_p is false.
24274 (analyze_access_subtree): Use comes_initialized_p.
24275 (propagate_subaccesses_across_link): Assert !comes_initialized_p
24276 instead of testing for PARM_DECL.
24278 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24280 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
24281 constraint on operand 0 to allow more general addressing modes.
24282 Adjust output template.
24283 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
24285 * config/aarch64/aarch64-protos.h
24286 (aarch64_address_valid_for_prefetch_p): Declare prototype.
24287 * config/aarch64/constraints.md (Dp): New address constraint.
24288 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
24291 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
24293 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
24294 update use of estimate_ipcp_clone_size_and_time.
24295 (estimate_local_effects): Update use of
24296 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
24297 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
24298 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
24299 Return nonspecialized time.
24301 2017-05-04 Richard Biener <rguenther@suse.de>
24303 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
24304 for the last VUSE which def dominates the PHI. Directly call
24306 (get_continuation_for_phi_1): Remove.
24308 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
24310 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
24311 to explain the use of truncating division. Cap the number of
24312 iterations to the maximum given by nb_iterations_upper_bound,
24315 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24317 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
24318 * configure: Regenerate.
24319 * config.in: Regenerate.
24320 * config/i386/driver-mingw32.c: new file.
24321 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
24322 * config.host: Link driver-mingw32.o on MinGW host.
24323 * doc/install.texi: Document new --enable-mingw-wildcard configure
24326 2017-05-04 Marek Polacek <polacek@redhat.com>
24328 PR tree-optimization/80612
24329 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
24331 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24332 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
24334 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
24335 (movt splitter): Likewise.
24336 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
24337 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
24338 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
24339 block for Thumb-1 with MOVT.
24340 (thumb2_legitimate_address_p): Move code block ...
24341 (can_avoid_literal_pool_for_label_p): ... into this new function.
24342 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
24344 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
24345 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
24346 "M-profile targets with the MOVT instruction".
24348 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24350 * config/arm/arm-builtins.c (arm_init_builtins): Rename
24351 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
24352 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
24354 2017-05-04 Martin Liska <mliska@suse.cz>
24356 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
24357 variable cond_code.
24359 2017-05-04 Richard Biener <rguenther@suse.de>
24361 * tree.c (array_at_struct_end_p): Handle arrays at struct
24362 end with flexarrays more conservatively. Refactor and treat
24363 arrays of arrays or aggregates more strict. Fix
24364 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
24365 * tree.c (array_at_struct_end_p): Adjust prototype.
24366 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
24367 * gimple-fold.c (get_range_strlen): Likewise.
24368 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
24370 2017-05-04 Richard Biener <rguenther@suse.de>
24372 PR tree-optimization/31130
24373 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
24375 (supports_overflow_infinity): Likewise.
24376 (is_negative_overflow_infinity): Likewise.
24377 (is_positive_overflow_infinity): Likewise.
24378 (is_overflow_infinity): Likewise.
24379 (stmt_overflow_infinity): Likewise.
24380 (overflow_infinity_range_p): Likewise.
24381 (usable_range_p): Remove as always returning true.
24382 (make_overflow_infinity): Remove.
24383 (negative_overflow_infinity): Likewise.
24384 (positive_overflow_infinity): Likewise.
24385 (avoid_overflow_infinity): Likewise.
24386 (set_value_range): Adjust accordingly.
24387 (set_value_range_to_nonnegative): Likewise, remove now unused
24388 overflow_infinity arg.
24389 (vrp_operand_equal_p): Adjust.
24390 (update_value_range): Likewise.
24391 (range_int_cst_singleton_p): Likewise.
24392 (operand_less_p): Likewise.
24393 (compare_values_warnv): Likewise.
24394 (extract_range_for_var_from_comparison_expr): Likewise.
24395 (vrp_int_const_binop): Likewise.
24396 (zero_nonzero_bits_from_vr): Likewise.
24397 (extract_range_from_multiplicative_op_1): Likewise.
24398 (extract_range_from_binary_expr_1): Likewise.
24399 (extract_range_from_unary_expr): Likewise.
24400 (extract_range_from_comparison): Likewise.
24401 (extract_range_basic): Likewise.
24402 (adjust_range_with_scev): Likewise.
24403 (compare_ranges): Likewise.
24404 (compare_range_with_value): Likewise.
24405 (dump_value_range): Likewise.
24406 (test_for_singularity): Likewise, remove strict_overflow_p parameter
24408 (simplify_cond_using_ranges): Adjust.
24410 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
24412 * brig-builtins.def: Added a builtin for class_f64.
24413 * builtin-types.def: Added a builtin type needed by class_f64.
24415 2017-05-03 Jason Merrill <jason@redhat.com>
24417 * timevar.def: Add TV_CONSTEXPR.
24419 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24421 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
24423 2017-05-03 Martin Jambor <mjambor@suse.cz>
24425 * ipa-prop.c (ipa_update_after_lto_read): Removed.
24426 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
24427 * ipa-cp.c (ipcp_propagate_stage): Do not call
24428 ipa_update_after_lto_read.
24429 * ipa-inline.c (ipa_inline): Likewise.
24431 2017-05-03 Martin Jambor <mjambor@suse.cz>
24433 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
24434 tag. Added a default constructor and a destructor.
24435 (ipa_edge_args_sum_t): New class;
24436 (ipa_edge_args_sum): Declare.
24437 (ipa_edge_args_vector): Remove declaration.
24438 (IPA_EDGE_REF): Use ipa_edge_args_sum.
24439 (ipa_free_edge_args_substructures): Remove declaration.
24440 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
24441 (ipa_edge_args_info_available_for_edge_p): Likewise.
24442 * ipa-prop.c (ipa_edge_args_vector): Removed.
24443 (edge_removal_hook_holder): Likewise.
24444 (edge_duplication_hook_holder): Likewise.
24445 (ipa_edge_args_sum): New variable.
24446 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
24447 ipa_edge_args_vector.
24448 (ipa_free_edge_args_substructures): Likewise.
24449 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
24450 ipa_edge_args_vector.
24451 (ipa_edge_removal_hook): Turned into method
24452 ipa_edge_args_sum_t::remove.
24453 (ipa_edge_duplication_hook): Turned into method
24454 ipa_edge_args_sum_t::duplicate.
24455 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
24456 registering edge hooks.
24457 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
24458 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
24459 ipa_edge_args_sum instead of ipa_edge_args_vector.
24460 * ipa-profile.c (ipa_profile): Likewise.
24462 2017-05-03 Martin Jambor <mjambor@suse.cz>
24464 * symbol-summary.h (function_summary): New method exists.
24465 (function_summary::symtab_removal): Deallocate through release.
24466 (call_summary): New class.
24467 (gt_ggc_mx): New overload.
24468 (gt_pch_nx): Likewise.
24469 (gt_pch_nx): Likewise.
24471 2017-05-03 Jeff Law <law@redhat.com>
24473 PR tree-optimization/78496
24474 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
24475 from simplify_cond_using_ranges. Split off code to walk
24476 backwards through casts into ...
24477 (simplify_cond_using_ranges_2): New function.
24478 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
24479 (execute_vrp): After identifying jump threads, call
24480 simplify_cond_using_ranges_2.
24482 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
24485 * ipa-inline.h (inline_summary): Add ctor.
24486 (create_ggc): Do not use ggc_cleared_alloc.
24488 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
24489 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24491 * gcc.c (handle_braces): Support escaping in switch matching
24493 * doc/invoke.texi (Spec Files): Document it.
24494 Remove superfluous @code markup in items.
24496 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24498 * diagnostic-show-locus.c (struct column_range): New struct.
24499 (get_affected_columns): New function.
24500 (get_printed_columns): New function.
24501 (struct correction): New struct.
24502 (correction::ensure_capacity): New function.
24503 (correction::ensure_terminated): New function.
24504 (struct line_corrections): New struct.
24505 (line_corrections::~line_corrections): New dtor.
24506 (line_corrections::add_hint): New function.
24507 (layout::print_trailing_fixits): Reimplement in terms of the new
24509 (selftest::test_overlapped_fixit_printing): New function.
24510 (selftest::diagnostic_show_locus_c_tests): Call it.
24512 2017-05-03 Nathan Sidwell <nathan@acm.org>
24514 Canonicalize canonical type hashing
24515 * tree.h (type_hash_canon_hash): Declare.
24516 * tree.c (type_hash_list, attribute_hash_list): Move into
24517 type_hash_canon_hash.
24518 (build_type_attribute_qual_variant): Break out hash code calc into
24519 type_hash_canon_hash.
24520 (type_hash_canon_hash): New. Generic type hash computation.
24521 (build_range_type_1, build_array_type_1, build_function_type,
24522 build_method_type_directly, build_offset_type, build_complex_type,
24523 make_vector_type): Call it.
24525 2017-05-03 Richard Biener <rguenther@suse.de>
24527 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
24528 When all DRs have unknown misaligned do not always peel
24529 when there is a store but apply the same costing model as if
24530 there were only loads.
24532 2017-05-03 Richard Biener <rguenther@suse.de>
24535 PR tree-optimization/80492
24536 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
24537 compare_base_decls returning dont-know properly.
24539 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
24541 * config/arm/iterators.md (CCSI): New mode iterator.
24542 (arch): New mode attribute.
24543 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
24544 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
24545 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
24546 code iterator for success result mode.
24547 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
24548 the corresponding new insn generators.
24550 2017-05-03 Bin Cheng <bin.cheng@arm.com>
24553 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24554 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24556 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
24558 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
24559 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
24560 (DDR_A): Wrap DDR argument in brackets.
24561 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
24562 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
24563 (DDR_REVERSED_P): Likewise.
24565 2017-05-03 Jakub Jelinek <jakub@redhat.com>
24567 PR tree-optimization/79472
24568 * tree-switch-conversion.c (struct switch_conv_info): Add
24569 contiguous_range and default_case_nonstandard fields.
24570 (collect_switch_conv_info): Compute contiguous_range and
24571 default_case_nonstandard fields, don't clear final_bb if
24572 contiguous_range and only the default case doesn't have the required
24574 (check_all_empty_except_final): Set default_case_nonstandard instead
24575 of failing if contiguous_range and the default case doesn't have empty
24577 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
24578 and only the default case doesn't have the required constants. Skip
24580 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
24581 if default_case_nonstandard.
24582 (build_constructors): Build constant 1 just once. Assert that default
24583 values aren't inserted in between cases if contiguous_range. Skip
24585 (build_arrays): Skip virtual phis.
24586 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
24587 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
24588 Handle virtual phis.
24589 (gen_inbound_check): Handle default_case_nonstandard case.
24590 (process_switch): Adjust check_final_bb caller. Call
24591 gather_default_values with the first non-default case instead of
24592 default case if default_case_nonstandard.
24594 2017-05-02 Nathan Sidwell <nathan@acm.org>
24596 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
24597 check. Fix formatting.
24599 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
24601 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
24602 errors when comparing specialized and unspecialized times.
24604 2017-05-02 David Malcolm <dmalcolm@redhat.com>
24606 * diagnostic-show-locus.c
24607 (layout::should_print_annotation_line_p): Make private.
24608 (layout::print_annotation_line): Make private.
24609 (layout::annotation_line_showed_range_p): Make private.
24610 (layout::show_ruler): Make private.
24611 (layout::print_source_line): Make private. Pass in line and
24612 line_width, rather than calling location_get_source_line. Drop
24614 (layout::print_leading_fixits): New method.
24615 (layout::print_any_fixits): Rename to...
24616 (layout::print_trailing_fixits): ...this, and make private.
24617 Don't print newline fixits.
24618 (diagnostic_show_locus): Move logic for printing one row into...
24619 (layout::print_line): ...this new function. Move the
24620 location_get_source_line call and error-handling from
24621 print_source_line to here. Call print_leading_fixits, and rename
24622 print_any_fixits to print_trailing_fixits.
24623 (selftest::test_fixit_insert_containing_newline): Update now that
24624 newlines are partially supported.
24625 (selftest::test_fixit_insert_containing_newline_2): New test.
24626 (selftest::test_fixit_replace_containing_newline): Update comments.
24627 (selftest::diagnostic_show_locus_c_tests): Call the new test.
24628 * edit-context.c (class added_line): New class.
24629 (class edited_line): Describe newline handling in comment.
24630 (edited_line::actually_edited_p): New method.
24631 (edited_line::print_content): Delete redundant decl.
24632 (edited_line::m_predecessors): New field.
24633 (edited_file::print_content): Call edited_line::print_content.
24634 (edited_file::print_diff): Update to support newlines.
24635 (edited_file::print_diff_hunk): Likewise.
24636 (edited_file::print_run_of_changed_lines): New function.
24637 (edited_file::print_diff_line): Convert to...
24638 (print_diff_line): ...this.
24639 (edited_file::get_effective_line_count): New function.
24640 (edited_line::edited_line): Initialize new field m_predecessors.
24641 (edited_line::~edited_line): Clean up m_predecessors.
24642 (edited_line::apply_fixit): Handle newlines.
24643 (edited_line::get_effective_line_count): New function.
24644 (edited_line::print_content): New function.
24645 (edited_line::print_diff_lines): New function.
24646 (selftest::test_applying_fixits_insert_containing_newline): New
24648 (selftest::test_applying_fixits_replace_containing_newline): New
24650 (selftest::insert_line): New function.
24651 (selftest::test_applying_fixits_multiple_lines): Add example of
24653 (selftest::edit_context_c_tests): Call the new tests.
24655 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24657 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
24658 parameter cand. Update dump information.
24659 (get_computation_cost): Update uses.
24661 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24663 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
24664 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
24665 (get_computation_at, rewrite_use_address): Update use of
24666 get_computation_aff.
24668 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24670 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
24671 (get_computation): Delete.
24672 (get_computation_cost): Implement like get_computation_cost_at.
24673 Use get_computation_at.
24674 (get_computation_cost_at): Delete.
24675 (rewrite_use_nonlinear_expr): Use get_computation_at.
24676 (rewrite_use_compare, remove_unused_ivs): Ditto.
24678 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24680 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
24682 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24684 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
24685 (ivopts_global_cost_for_size): Rename parameter and update uses.
24686 (iv_ca_recount_cost): Update uses.
24687 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
24688 candidates seperately in n_invs and n_cands.
24689 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
24691 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24693 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
24694 (find_inv_vars_cb): New.
24695 (find_depends): Renamed to ...
24696 (find_inv_vars): ... this.
24697 (add_candidate_1, force_var_cost): Call find_inv_vars.
24698 (split_address_cost, determine_group_iv_cost_cond): Ditto.
24700 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24702 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
24703 inv_vars. Add inv_exprs.
24704 (struct iv_cand): Rename depends_on to inv_vars.
24705 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
24706 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
24707 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
24708 (dump_cand): Dump inv_vars.
24709 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
24710 (record_invariant, find_depends, add_candidate_1): Ditto.
24711 (set_group_iv_cost, force_var_cost): Ditto.
24712 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
24713 (get_computation_cost_at, get_computation_cost): Ditto.
24714 (determine_group_iv_cost_generic): Ditto.
24715 (determine_group_iv_cost_address): Ditto.
24716 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
24717 (determine_group_iv_costs): Ditto.
24718 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
24719 (iv_ca_set_remove_invariants): Renamed to ...
24720 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
24721 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
24722 (iv_ca_set_add_invariants): Renamed to ...
24723 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
24724 (iv_ca_set_cp): Use iv_ca_set_add_invs.
24725 (iv_ca_has_deps): Support inv_vars and inv_exprs.
24726 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
24727 (create_new_ivs): Remove useless dump.
24729 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24731 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
24733 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
24734 (iv_ca_set_no_cp, create_new_iv): Ditto.
24736 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24738 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24740 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
24742 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
24743 function tree_check2.
24745 2017-05-02 Martin Liska <mliska@suse.cz>
24747 * doc/gcov.texi: Add missing preposition.
24748 * gcov.c (function_info::function_info): Properly fill up
24749 all member variables.
24751 2017-05-02 Tamar Christina <tamar.christina@arm.com>
24753 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
24755 2017-05-02 Tamar Christina <tamar.christina@arm.com>
24757 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
24759 2017-05-02 Martin Liska <mliska@suse.cz>
24762 * lto-streamer-in.c (lto_read_tree_1): Remove
24763 LTO_STREAMER_DEBUG.
24764 * lto-streamer.c (struct tree_hash_entry): Likewise.
24765 (struct tree_entry_hasher): Likewise.
24766 (tree_entry_hasher::hash): Likewise.
24767 (tree_entry_hasher::equal): Likewise.
24768 (lto_streamer_init): Likewise.
24769 (lto_orig_address_map): Likewise.
24770 (lto_orig_address_get): Likewise.
24771 (lto_orig_address_remove): Likewise.
24772 * lto-streamer.h: Likewise.
24773 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
24774 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
24776 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
24778 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
24779 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
24780 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
24781 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
24782 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
24783 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
24784 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
24785 (mm_maskz_sub_ss): New intrinsics.
24786 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24787 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
24788 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
24789 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
24790 (__builtin_ia32_subss_mask_round): New builtins.
24791 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24792 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24793 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
24795 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
24796 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
24798 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
24801 2017-05-02 Martin Jambor <mjambor@suse.cz>
24803 PR tree-optimization/78687
24804 * tree-sra.c (access): New field parent.
24805 (process_subtree_disqualification): New function.
24806 (disqualify_candidate): Call it.
24807 (build_accesses_from_assign): Reset write flag if creating an
24809 (build_access_subtree): Fill in parent field and also prpagate
24810 down grp_write flag.
24811 (create_artificial_child_access): New parameter set_grp_write, set
24812 grp_write to its value.
24813 (propagate_subaccesses_across_link): Also propagate grp_write flag
24815 (propagate_all_subaccesses): Push the closest parent back to work
24816 queue if add_access_to_work_queue returned true.
24818 2017-05-02 Richard Biener <rguenther@suse.de>
24820 * common.opt (fstrict-overflow): Alias negative to fwrapv.
24821 * doc/invoke.texi (fstrict-overflow): Remove all traces of
24822 -fstrict-overflow documentation.
24823 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
24824 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
24825 flag_strict_overflow.
24826 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
24827 * lto-opts.c (lto_write_options): Do not stream it.
24828 * lto-wrapper.c (merge_and_complain): Do not handle it.
24829 * opts.c (default_options_table): Do not set -fstrict-overflow.
24830 (finish_options): Likewise do not clear it when sanitizing.
24831 * simplify-rtx.c (simplify_const_relational_operation): Do not
24832 test flag_strict_overflow.
24834 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
24836 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
24837 using enabled attribute.
24838 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
24839 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
24840 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
24841 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
24842 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
24843 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
24844 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
24845 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
24846 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
24847 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
24849 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
24851 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
24853 2017-05-02 Richard Biener <rguenther@suse.de>
24855 PR tree-optimization/80591
24857 2017-04-10 Richard Biener <rguenther@suse.de>
24859 * tree-ssa-structalias.c (find_func_aliases): Properly handle
24862 2017-05-02 Richard Biener <rguenther@suse.de>
24864 PR tree-optimization/80549
24865 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
24866 (cleanup_tree_cfg_noloop): Create forwarders to known loop
24867 headers if they do not have a preheader.
24869 2017-05-02 Martin Liska <mliska@suse.cz>
24872 * common.opt: Fix typo.
24873 * doc/invoke.texi: Likewise.
24875 2017-05-01 Jan Beulich <jbeulich@suse.com>
24877 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
24878 swapping, add (x,x,m,x,n) alternative.
24880 2017-05-01 Nathan Sidwell <nathan@acm.org>
24882 * calls.c (combine_pending_stack_adjustment_and_call): Remove
24883 unnecessary unadjusted_alignment check.
24885 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
24888 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
24890 * gimplify.c (gimplify_cilk_detach): New function.
24891 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
24892 * tree-core.h: Document EXPR_CILK_SPAWN.
24893 * tree.h (EXPR_CILK_SPAWN): Define.
24895 2017-05-01 David Malcolm <dmalcolm@redhat.com>
24897 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
24898 to use new fixit_hint representation, using the "replace" logic.
24899 (get_line_span_for_fixit_hint): Likewise.
24900 (layout::print_any_fixits): Likewise.
24901 (selftest::test_one_liner_many_fixits): Rename to...
24902 (selftest::test_one_liner_many_fixits_1): ...this, and update
24903 comment and expected output to reflect that the multiple fix-it
24904 hints are now consolidated into one insertion.
24905 (selftest::test_one_liner_many_fixits_2): New test.
24906 (selftest::test_diagnostic_show_locus_one_liner): Update for
24908 (selftest::test_fixit_consolidation): Update for fix-it API
24910 * diagnostic.c (print_parseable_fixits): Likewise.
24911 * edit-context.c (edited_line::m_line_events): Convert from
24912 auto_vec <line_event *> to auto_vec <line_event>.
24913 (class line_event): Convert from abstract base class to a concrete
24914 class, taking over the role of replace_event.
24915 (class insert_event): Delete.
24916 (class replace_event): Rename to class line_event. Convert to
24918 (edit_context::add_fixits): Reimplement.
24919 (edit_context::apply_insert): Delete.
24920 (edit_context::apply_replace): Rename to...
24921 (edit_context::apply_fixit): ...this. Convert to half-open range.
24922 (edited_file::apply_insert): Delete.
24923 (edited_file::apply_replace): Rename to...
24924 (edited_file::apply_fixit): ...this.
24925 (edited_line::~edited_line): Drop deletion of events.
24926 (edited_line::apply_insert): Delete.
24927 (edited_line::apply_replace): Rename to...
24928 (edited_line::apply_fixit): ...this. Convert to half-open range.
24929 Update for change to type of m_line_events.
24930 * edit-context.h (edit_context::apply_insert): Delete.
24931 (edit_context::apply_replace): Rename to...
24932 (edit_context::apply_fixit): ...this.
24934 2017-05-01 Martin Sebor <msebor@redhat.com>
24936 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
24939 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
24942 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
24943 __get_cpuid_max returns 0.
24944 (__get_cpuid_count): Ditto.
24946 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
24948 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
24949 replacement expression is another instance of one of its arguments.
24951 2017-05-01 Jakub Jelinek <jakub@redhat.com>
24954 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
24955 check for stack push/pop autoinc.
24956 * config/i386/i386.c (ix86_agi_dependent): Return false
24957 if the only reason why modified_in_p returned true is that
24958 addr is SP based and set_insn is a push or pop.
24960 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
24962 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
24965 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
24968 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
24969 (account_size_time): Use two predicates - exec_pred and
24971 (evaluate_conditions_for_known_args): Compute both clause and
24973 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
24974 (inline_summary_t::duplicate): Update.
24975 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
24977 (compute_inline_parameters): Likewise.
24978 (estimate_edge_size_and_time): Update caluclation of time.
24979 (estimate_node_size_and_time): Compute both time and nonspecialized
24981 (estimate_ipcp_clone_size_and_time): Update.
24982 (inline_merge_summary): Update.
24983 (do_estimate_edge_time): Update.
24984 (do_estimate_edge_size): Update.
24985 (do_estimate_edge_hints): Update.
24986 (inline_read_section, inline_write_summary): Stream both new predicates.
24987 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
24989 (compute_inlined_call_time): Cleanup.
24990 (big_speedup_p): Update.
24991 (edge_badness): Update.
24992 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
24993 (size_time_entry): Replace predicate by exec_predicate and
24994 nonconst_predicate.
24995 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
24996 (estimate_edge_time): Return also nonspec_time.
24997 (reset_edge_growth_cache): Update.
24999 2017-04-29 Jakub Jelinek <jakub@redhat.com>
25001 PR rtl-optimization/80491
25002 * ifcvt.c (noce_process_if_block): When looking for x setter
25003 with missing else_bb, don't check only the insn right before
25004 cond_earliest, but look for the last insn that x is modified in
25005 within the same bb.
25007 PR rtl-optimization/80491
25008 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
25010 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
25012 PR tree-optimization/80487
25013 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
25015 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25017 PR tree-optimization/79697
25018 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
25019 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
25020 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
25022 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
25023 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
25025 2017-04-28 Martin Sebor <msebor@redhat.com>
25027 PR tree-optimization/80523
25028 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
25029 (init_target_to_host_charmap, target_to_host, target_strtol10): New
25031 (maybe_warn, format_directive, parse_directive): Use new functions.
25032 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
25034 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
25036 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
25038 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
25040 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
25041 target_header_dir): Set correctly.
25042 * configure: Regenerated.
25043 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25044 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25045 instead of SYSTEM_HEADER_DIR.
25047 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
25049 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
25050 (estimate_local_effects): Likewise.
25051 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
25052 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
25053 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
25054 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
25055 do_estimate_edge_time, estimate_edge_time): Likewise.
25056 * ipa-inline-analysis.c (estimate_node_size_and_time,
25057 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
25058 (estimate_time_after_inlining): Remove.
25060 2017-04-28 Martin Liska <mliska@suse.cz>
25062 * doc/gcov.texi: Enhance documentation of gcov.
25064 2017-04-28 Martin Liska <mliska@suse.cz>
25066 * doc/gcov.texi: Sort options in alphabetic order.
25067 * doc/gcov-dump.texi: Likewise.
25068 * doc/gcov-tool.texi: Likewise.
25069 * gcov.c (print_usage): Likewise.
25070 * gcov-dump.c (print_usage): Likewise.
25071 * gcov-tool.c (print_merge_usage_message): Likewise.
25072 (print_rewrite_usage_message): Likewise.
25073 (print_overlap_usage_message): Likewise.
25075 2017-04-28 Martin Liska <mliska@suse.cz>
25077 PR gcov-profile/53915
25078 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
25080 2017-04-28 Martin Liska <mliska@suse.cz>
25082 PR gcov-profile/79891
25083 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
25084 is marked by compiler as living on a line.
25085 (get_cycles_count): Remove usage of the union.
25086 (output_intermediate_file): Likewise.
25087 (find_source): Fix GNU coding style.
25088 (accumulate_line_counts): Remove old non-all block mode.
25089 (output_lines): Remove usage of the union.
25090 * profile.c (output_location): Include all BBs, even if
25091 belonging to a same line (and file) as a previous BB.
25093 2017-04-28 Martin Liska <mliska@suse.cz>
25095 * gcov.c (process_args): Handle new argument 'w'.
25096 (read_graph_file): Assign ID to BBs.
25097 (output_branch_count): Display BB # if verbose flag is set.
25098 (output_lines): Likewise for arcs.
25099 (print_usage): Add '--verbose' option help.
25100 * doc/gcov.texi: Document --verbose (-w) option.
25102 2017-04-28 Martin Liska <mliska@suse.cz>
25104 * gcov.c (struct block_location_info): New struct.
25105 (process_file): Fill up the new structure.
25106 (read_graph_file): Replace usage of encoding by the newly added
25108 (add_line_counts): Likewise.
25109 (accumulate_line_counts): Remove usage of the union.
25110 (function_info::function_info): New function.
25111 (function_info::~function_info): Likewise.
25112 (process_file): Call delete instead of release_function.
25113 (release_function): Release the function.
25114 (release_structures): Call delete instead of release_function.
25115 (solve_flow_graph): Replace usage of num_blocks.
25116 (find_exception_blocks): Likewise.
25117 (output_lines): Fix GNU coding style.
25119 2017-04-28 Martin Liska <mliska@suse.cz>
25122 * coverage.c (coverage_remove_note_file): New function.
25123 * coverage.h: Declare the function.
25124 * toplev.c (finalize): Clean if an error has been seen.
25126 2017-04-28 Martin Liska <mliska@suse.cz>
25128 PR gcov-profile/80031
25129 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
25130 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
25131 * gcov.c (read_graph_file): Read just number of blocks.
25132 * profile.c (branch_prob): Do not stream 0 flags per a basic
25135 2017-04-28 Martin Liska <mliska@suse.cz>
25137 * gcov-dump.c (tag_*): Add new argument to declarations.
25138 (dump_gcov_file): Likewise.
25139 (tag_blocks): Add and use new argument depth.
25140 (tag_arcs): Likewise.
25141 (tag_lines): Likewise.
25142 (tag_counters): Likewise.
25143 (tag_summary): Likewise.
25144 (dump_working_sets): Use depth to do a proper indentation.
25146 2017-04-28 Jakub Jelinek <jakub@redhat.com>
25149 * cgraph.h (symtab_node::debug_symtab): No longer inline.
25150 * symtab.c (symtab_node::debug_symtab): Move definition here.
25152 2017-04-28 Richard Biener <rguenther@suse.de>
25154 * lto-streamer.h (LTO_major_version): Bump to 7.
25156 2017-04-28 Richard Biener <rguenther@suse.de>
25158 * tree-vrp.c (assert_info): New struct.
25159 (add_assert_info): New helper.
25160 (register_edge_assert_for_2): Refactor to add asserts to a vector
25162 (register_edge_assert_for_1): Likewise.
25163 (register_edge_assert_for): Likewise.
25164 (finish_register_edge_assert_for): New helper actually registering
25165 asserts where live on edge.
25166 (find_conditional_asserts): Adjust.
25167 (find_switch_asserts): Likewise.
25168 (evrp_dom_walker::try_find_new_range): Generalize.
25169 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
25171 2017-04-27 Marek Polacek <polacek@redhat.com>
25174 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
25175 arg10 and arg11 to itype.
25177 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
25179 * doc/extend.texi (Object Size Checking): Improve grammar.
25181 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
25184 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
25185 that the logic for permitting reciprocal estimates matches that
25188 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25191 * tree.c (type_cache_hasher::equal): Only compare
25192 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
25193 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
25194 non-aggregate element types.
25195 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
25196 about the flag on ARRAY_TYPEs in the comment, formatting fix.
25198 2017-04-27 Richard Biener <rguenther@suse.de>
25200 PR middle-end/80533
25201 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
25202 stripping ARRAY_REFs from MEM_EXPR make sure we're not
25203 keeping a reference to a trailing array.
25205 2017-04-27 Richard Biener <rguenther@suse.de>
25207 PR middle-end/80539
25208 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
25209 being in loop-closed SSA form conservatively.
25210 (chrec_fold_multiply_poly_poly): Likewise.
25212 2017-04-27 Tamar Christina <tamar.christina@arm.com>
25214 PR middle-end/79665
25215 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
25216 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
25218 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25221 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
25222 (aarch64_function_arg_alignment): Return unsigned int again, but still
25223 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
25224 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
25225 Don't emit -Wpsabi note.
25226 (aarch64_function_arg_boundary): Likewise.
25227 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
25230 2017-04-26 Nathan Sidwell <nathan@acm.org>
25232 * tree.h (crc32_unsigned_n): Declare.
25233 (crc32_unsigned, crc32_unsigned): Make inline.
25234 * tree.c (crc32_unsigned_bits): Replace with ...
25235 (crc32_unsigned_n): ... this.
25236 (crc32_unsigned, crc32_byte): Remove.
25237 (crc32_string): Remove unnecessary braces.
25239 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25241 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
25242 * ipa-inline-analysis.c (MAX_TIME): Remove.
25243 (account_size_time): Use sreal for time.
25244 (dump_inline_summary): Update.
25245 (estimate_function_body_sizes): Update.
25246 (estimate_edge_size_and_time): Update.
25247 (estimate_calls_size_and_time): Update.
25248 (estimate_node_size_and_time): Update.
25249 (inline_merge_summary): Update.
25250 (inline_update_overall_summary): Update.
25251 (estimate_time_after_inlining): Update.
25252 (inline_read_section): Update.
25253 (inline_write_summary): Update.
25254 * ipa-inline.c (compute_uninlined_call_time): Update.
25255 (compute_inlined_call_time): Update.
25256 (recursive_inlining): Update.
25257 (inline_small_functions): Update.
25258 (dump_overall_stats): Update.
25259 * ipa-inline.h: Include sreal.h.
25260 (size_time_entry): Turn time to sreal.
25261 (inline_summary): Turn self_time nad time to sreal.
25263 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25265 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
25267 (sreal::stream_out, sreal::stream_in): New.
25268 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
25270 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25272 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
25275 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
25278 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
25280 (dimode_scalar_chain::compute_convert_gain): Ditto.
25281 (dimode_scalar_chain::make_vector_copies): Ditto.
25282 (dimode_scalar_chain::convert_reg): Ditto.
25283 (dimode_scalar_chain::convert_insn): Ditto.
25284 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
25285 (VI248_AVX512BW_1): New mode iterator.
25286 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
25287 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
25290 2017-04-25 Martin Sebor <msebor@redhat.com>
25292 PR tree-optimization/80497
25293 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
25294 constants are representable in HOST_WIDE_INT.
25295 (parse_directive): Ditto.
25297 2017-04-25 Martin Sebor <msebor@redhat.com>
25300 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
25301 (new_zero_array): Adjust signature.
25302 (dom_info::dom_init): Used unsigned rather that size_t.
25303 (dom_info::dom_info): Same.
25305 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25306 Jakub Jelinek <jakub@redhat.com>
25309 * config/arm/arm.c: Include gimple.h.
25310 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
25311 returns negative, increment ncrn only if it returned positive.
25312 (arm_needs_doubleword_align): Return int instead of bool,
25313 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
25314 members, but if there is any such non-FIELD_DECL
25315 > PARM_BOUNDARY aligned decl, return -1 instead of false.
25316 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
25317 returns negative, increment nregs only if it returned positive.
25318 (arm_setup_incoming_varargs): Likewise.
25319 (arm_function_arg_boundary): Emit -Wpsabi note if
25320 arm_needs_doubleword_align returns negative, return
25321 DOUBLEWORD_ALIGNMENT only if it returned positive.
25323 2017-04-25 Marek Polacek <polacek@redhat.com>
25326 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
25327 first argument to type.
25329 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
25332 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
25333 type checks to test for compatibility instead of equality.
25335 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25336 Jakub Jelinek <jakub@redhat.com>
25339 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
25341 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
25342 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
25343 the alignment computation, but return their maximum in warn_alignment.
25344 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
25345 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
25347 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
25348 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
25351 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25353 * config/arc/simdext.md (dmpyh): Fix typo.
25355 2017-04-25 Richard Biener <rguenther@suse.de>
25357 PR tree-optimization/80492
25358 * alias.c (compare_base_decls): Handle registers with asm
25359 specification conservatively.
25360 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
25361 compare_base_decls returning dont-know properly.
25363 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25365 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
25366 (legitimate_offset_address_p): New function.
25367 (arc_legitimate_address_p): Use above function.
25369 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25371 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
25373 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25375 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
25376 ACCH registers whenever they are available.
25378 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25380 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
25381 double regs fix when not used.
25383 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25385 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
25387 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
25388 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
25390 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25392 * config/arc/arc.c (arc_output_addsi): Check for h-register class
25393 when emitting short ADD instructions.
25395 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25397 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
25399 (cmpsi_cc_c_insn): Likewise.
25400 (cbranchsi4_scratch): Compute proper instruction length using
25401 compact_hreg_operand.
25402 * config/arc/predicates.md (compact_hreg_operand): New predicate.
25404 2017-04-25 Richard Biener <rguenther@suse.de>
25406 PR middle-end/80509
25407 * passes.c (pass_manager::pass_manager): Initialize
25408 m_name_to_pass_map.
25410 2017-04-25 Richard Biener <rguenther@suse.de>
25412 PR tree-optimization/79201
25413 * tree-ssa-sink.c (statement_sink_location): Handle calls.
25415 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25418 * config/s390/vector.md: Split MEM->GPR vector moves for
25419 non-s_operand addresses.
25421 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25424 * config/s390/predicates.md (reload_const_wide_int_operand): New
25426 * config/s390/s390.md ("movti"): Remove d/P alternative.
25427 ("movti_bigconst"): New pattern definition.
25429 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25432 * s390-protos.h (s390_expand_cs_hqi): Removed.
25433 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
25434 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
25435 modes as well as CCZ1mode and CCZmode.
25436 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
25437 signature of s390_emit_compare_and_swap.
25438 (s390_expand_cs_hqi): Likewise, make static.
25439 (s390_expand_cs_tdsi): Generate an explicit compare before trying
25440 compare-and-swap, in some cases.
25441 (s390_expand_cs): Wrapper function.
25442 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
25444 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
25445 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
25446 patterns for small and large integers. Forbid symref memory operands.
25447 Move expander to s390.c. Require cc register.
25448 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
25449 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
25450 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
25451 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
25452 symref memory operands. Remove CC mode and call s390_match_ccmode
25454 ("atomic_exchange<mode>"): Allow and implement all integer modes.
25456 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25458 * config/s390/s390.md (define_peephole2): New peephole to help
25459 combining the load-and-test pattern with volatile memory.
25461 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25463 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
25464 with CCZmode for TARGET_Z196.
25466 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25468 PR rtl-optimization/80501
25469 * combine.c (make_compound_operation_int): Set subreg_code to SET
25470 even for AND with mask of the sign bit of mode.
25472 PR rtl-optimization/80500
25473 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
25474 sum's initial value.
25476 2017-04-25 Julian Brown <julian@codesourcery.com>
25477 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25479 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
25481 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
25483 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
25485 2017-04-25 Julian Brown <julian@codesourcery.com>
25486 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25488 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
25489 (thunderx2t99_sha): New Reservation.
25491 2017-04-25 Julian Brown <julian@codesourcery.com>
25492 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25494 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
25495 type for 1-element load.
25497 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
25499 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
25501 2017-04-24 Martin Jambor <mjambor@suse.cz>
25503 PR tree-optimization/80293
25504 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
25505 char arrays not totally scalarizable if it is false.
25506 (analyze_all_variable_accesses): Pass correct value in the new
25507 parameter. Add a statistics counter.
25509 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
25511 PR middle-end/79931
25512 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
25514 2017-04-24 Richard Biener <rguenther@suse.de>
25516 PR tree-optimization/80494
25517 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
25518 out for complex types.
25520 2017-04-24 Richard Biener <rguenther@suse.de>
25522 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
25523 * tree-ssa-sccvn.c (print_scc): Print SCC size.
25524 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
25525 (DFS): Adjust and never fail.
25526 (sccvn_dom_walker::fail): Remove.
25527 (sccvn_dom_walker::before_dom_children): Adjust.
25528 (run_scc_vn): Likewise and never fail.
25529 * tree-ssa-pre.c (pass_pre::execute): Adjust.
25530 (pass_fre::execute): Likewise.
25532 2017-04-24 Richard Biener <rguenther@suse.de>
25534 PR tree-optimization/79725
25535 * tree-ssa-sink.c (statement_sink_location): Return whether
25536 failure reason was zero uses. Move that check later.
25537 (sink_code_in_bb): Deal with zero uses by removing the stmt
25540 2017-04-24 Richard Biener <rguenther@suse.de>
25543 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
25544 pointer-based references.
25546 2017-04-24 Richard Biener <rguenther@suse.de>
25549 * pass_manager.h (pass_manager::operator new): Remove.
25550 (pass_manager::operator delete): Likewise.
25551 * passes.c (pass_manager::operator new): Remove.
25552 (pass_manager::operator delete): Likewise.
25553 (pass_manager::pass_manager): Zero individual pass members.
25555 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
25558 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
25559 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
25560 Check "XEXP (src, 1)" operand here.
25561 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
25562 Check "XEXP (src, 1)" operand here.
25563 (dimode_scalar_chain::make_vector_copies): Detect count register
25564 of a shift instruction. Zero extend count register from QImode
25565 to DImode to satisfy vector shift pattern count operand predicate.
25566 Substitute vector shift count operand with a DImode copy.
25567 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
25570 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
25572 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
25573 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
25574 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
25575 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
25576 (UNSPEC_NOREX_MEM): Remove definition.
25578 2017-04-21 Richard Biener <rguenther@suse.de>
25580 PR tree-optimization/79547
25581 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
25582 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
25583 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
25584 without any constraints.
25586 2017-04-21 Richard Biener <rguenther@suse.de>
25588 PR tree-optimization/78847
25589 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
25591 2017-04-21 Richard Biener <rguenther@suse.de>
25593 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
25594 (build_distinct_type_copy): Likewise.
25595 (build_variant_type_copy): Likewise.
25596 * tree.c (build_qualified_type): Pass down mem-stat info.
25597 (build_distinct_type_copy): Likewise.
25598 (build_variant_type_copy): Likewise.
25600 2017-04-21 Richard Biener <rguenther@suse.de>
25602 PR tree-optimization/80237
25603 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
25605 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
25606 for a simplified result.
25608 2016-04-21 Richard Biener <rguenther@suse.de>
25610 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
25611 sth as strict as a simple_iv but a chrec without symbols and an
25612 operand defined in the loop we are peeling (and not some subloop).
25613 (propagate_constants_for_unrolling): Propagate all constants.
25615 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25618 * config/i386/i386.c (print_reg): Remove assert for disalowed
25619 regno values, call output_operand_lossage instead.
25621 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25624 * config/i386/constraints.md (Yc): New register constraint.
25625 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
25626 Use Yc constraint for alternative 2 of operand 0. Remove
25627 preferred_for_speed attribute.
25629 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
25631 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
25632 lastprivate clauses in SIMT case.
25634 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
25636 * doc/invoke.texi (-Wextra-semi): Document new warning option.
25638 2017-04-20 Richard Biener <rguenther@suse.de>
25640 PR tree-optimization/57796
25641 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
25642 as N scalar stores.
25643 (vect_model_load_cost): Cost gathers as N scalar loads.
25645 2017-04-20 Richard Biener <rguenther@suse.de>
25647 * ggc-page.c (ggc_allocated_p): Rename to ...
25648 (safe_lookup_page_table_entry): ... this and return the lookup
25650 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
25652 2017-04-20 Richard Biener <rguenther@suse.de>
25654 PR tree-optimization/80453
25655 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
25656 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
25657 from the conditions.
25658 (vn_phi_eq): Pass them down.
25659 (vn_phi_lookup): Record them.
25660 (vn_phi_insert): Likewise.
25662 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
25664 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
25665 uninitialized variable warning to avoid buffer overrun.
25667 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
25670 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
25671 is suppressed for '{ 0 }' in C.
25673 2017-04-20 Jakub Jelinek <jakub@redhat.com>
25675 * BASE-VER: Set to 8.0.0.
25677 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
25679 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
25680 priority .init_array and .fini_array section with SECTION_NOTYPE
25683 2017-04-20 Jakub Jelinek <jakub@redhat.com>
25685 PR middle-end/80423
25686 * tree.h (build_array_type): Add typeless_storage default argument.
25687 * tree.c (type_cache_hasher::equal): Also compare
25688 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
25689 (build_array_type): Add typeless_storage argument, set
25690 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
25692 (build_nonshared_array_type): Adjust build_array_type_1 caller.
25693 (build_array_type): Likewise. Add typeless_storage argument.
25695 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
25696 Jakub Jelinek <jakub@redhat.com>
25698 PR tree-optimization/80426
25699 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
25700 operation on symbolic operands, also compute the overflow for the
25701 invariant part when the operation degenerates into a negation.
25703 2017-04-19 Jakub Jelinek <jakub@redhat.com>
25706 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
25707 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
25710 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
25712 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
25715 * config/avr/avr.c (tree.h): Include it.
25716 (cgraph.h): Include it.
25717 (avr_encode_section_info): Don't warn for uninitialized progmem
25718 variable if it's just an alias.
25720 2017-04-19 Richard Biener <rguenther@suse.de>
25723 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
25724 when needed by AutoPGO.
25726 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
25729 * doc/lto.texi: Remove an extra 'that'.
25731 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
25733 PR rtl-optimization/80429
25734 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
25735 are only used in debug insns.
25737 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
25738 Vladimir Makarov <vmakarov@redhat.com>
25740 * config/sparc/predicates.md (input_operand): Add comment. Return
25741 true for any memory operand when LRA is in progress.
25742 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
25744 2017-04-18 Jeff Law <law@redhat.com>
25747 * mips.md ({return,simple_return}_internal): Do not overwrite
25750 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25752 PR tree-optimization/80443
25753 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
25754 instead of adding 1, subtract -1 and similarly instead of subtracting
25757 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
25759 PR rtl-optimization/80357
25760 * haifa-sched.c (tmp_bitmap): New variable.
25761 (model_recompute): Handle duplicate use records.
25762 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
25763 (free_global_sched_pressure_data): Free it.
25765 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
25768 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
25769 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25770 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25771 instead of SYSTEM_HEADER_DIR.
25773 2017-04-18 Jeff Law <law@redhat.com>
25775 PR middle-end/80422
25776 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
25777 predecessors after walking up the insn chain.
25779 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25782 * dwarf2out.c (modified_type_die): Try harder not to emit internal
25783 sizetype type into debug info.
25785 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
25788 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
25789 unneeded test for TARGET_UPPER_REGS_SF.
25790 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
25792 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25795 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
25796 instead of gsi_after_labels.
25798 2017-04-18 Jeff Law <law@redhat.com>
25800 * regcprop.c (maybe_mode_change): Avoid creating copies of the
25804 2017-04-13 Jeff Law <law@redhat.com>
25805 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
25806 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
25808 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
25811 * config/avr/avr.c (intl.h): Include it.
25812 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
25814 2017-04-18 Martin Liska <mliska@suse.cz>
25816 PR gcov-profile/78783
25817 * gcov-tool.c (gcov_output_files): Validate that destination
25818 file is either removed by the tool or by a user.
25820 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
25821 Guy Benyei <guybe@mellanox.com>
25823 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
25824 block, and do not negate it, the stored id is already negative.
25826 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
25828 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
25830 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
25833 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
25834 masks of options that should be turned off if the VSX vector
25835 options are turned off.
25836 (OTHER_P8_VECTOR_MASKS): Likewise.
25837 (OTHER_VSX_VECTOR_MASKS): Likewise.
25838 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
25839 rs6000_disable_incompatible_switches to validate no type switches
25841 (rs6000_incompatible_switch): New function to disallow turning on
25842 other vector options if -mno-vsx, -mno-power8-vector, or
25843 -mno-power9-vector are specified.
25845 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
25847 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
25849 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
25851 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
25852 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
25853 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
25854 (ARG_POINTER_CFA_OFFSET): Likewise.
25856 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
25858 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
25859 conditions to take advantage of various optimizations.
25861 2017-04-13 Jeff Law <law@redhat.com>
25863 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
25864 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
25865 (zero_extendsidi2_dext): Likewise.
25867 2017-04-13 Jakub Jelinek <jakub@redhat.com>
25870 * fold-const.c (fold_ternary_loc): Revert
25871 use op0 instead of fold_convert_loc (loc, type, arg0) part of
25874 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
25876 PR rtl-optimization/80343
25877 * lra-remat.c (update_scratch_ops): Assign original hard reg to
25878 new scratch pseudo.
25880 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
25883 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
25884 to ubsan_encode_value.
25886 2017-04-13 Jeff Law <law@redhat.com>
25888 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
25889 appearing in DEBUG_INSNs.
25891 2017-04-13 Martin Liska <mliska@suse.cz>
25893 PR gcov-profile/80413
25894 * gcov-io.c (gcov_write_string): Copy to buffer just when
25895 allocated size is greater than zero.
25897 2017-04-13 Jakub Jelinek <jakub@redhat.com>
25900 * dwarf2out.c (decls_for_scope): Ignore declarations of
25901 current_function_decl in BLOCK_NONLOCALIZED_VARS.
25903 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
25906 * ipa-visibility.c (non_local_p): Fix typos.
25907 (localize_node): When localizing symbol in same comdat group,
25908 dissolve the group only when we know external symbols are going
25910 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
25912 2017-04-12 Jakub Jelinek <jakub@redhat.com>
25914 PR tree-optimization/79390
25915 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
25916 order does not result in usable sequence, retry with reversed operand
25922 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
25923 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
25924 op0 instead of fold_convert_loc (loc, type, arg0).
25926 2017-04-12 Jeff Law <law@redhat.com>
25928 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
25929 has a delay slot in the generated code.
25931 * config/cris/cris.md (cris_preferred_reload_class): Return
25932 GENNONACR_REGS rather than GENERAL_REGS.
25934 2017-04-12 Jakub Jelinek <jakub@redhat.com>
25937 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
25938 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
25939 signedness of the result type.
25941 2017-04-12 Richard Biener <rguenther@suse.de>
25942 Jeff Law <law@redhat.com>
25944 PR tree-optimization/80359
25945 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
25946 trim stores to TARGET_MEM_REFs.
25948 2017-04-12 Richard Biener <rguenther@suse.de>
25950 PR tree-optimization/79390
25951 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
25952 threading case even more.
25954 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
25957 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
25958 for quad_address_p for TImode, instead of just not indexed_address.
25960 2017-04-12 Richard Biener <rguenther@suse.de>
25961 Bernd Edlinger <bernd.edlinger@hotmail.de>
25963 PR middle-end/79671
25964 * alias.c (component_uses_parent_alias_set_from): Handle
25965 TYPE_TYPELESS_STORAGE.
25966 (get_alias_set): Likewise.
25967 * tree-core.h (tree_type_common): Add typeless_storage flag.
25968 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
25969 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
25970 for types containing members with TYPE_TYPELESS_STORAGE.
25971 (place_field): Likewise.
25972 (layout_type): Likewise for ARRAY_TYPE.
25973 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
25974 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
25975 TYPE_TYPELESS_STORAGE.
25976 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
25978 2017-04-12 Jakub Jelinek <jakub@redhat.com>
25981 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
25982 first argument to type.
25984 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25988 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
25989 CONST0_RTX (mode) rather than const0_rtx where appropriate.
25990 (rs6000_expand_binop_builtin): Likewise.
25991 (rs6000_expand_ternop_builtin): Likewise; also add missing
25992 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
25993 vshasigma built-ins.
25994 * doc/extend.texi: Document that vec_xxpermdi's third argument
25995 must be a constant.
25997 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
25999 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
26000 Use shift_const cost parameter when calculating gain of STV shifts.
26002 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
26004 PR rtl-optimization/70478
26005 * lra-constraints.c (process_alt_operands): Check memory for
26006 disfavoring memory insn operand.
26008 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26010 PR middle-end/80100
26011 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
26012 left shift in unsigned HOST_WIDE_INT type.
26014 PR rtl-optimization/80385
26015 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
26016 (not (neg X)) into (plus X -1) for complex or non-integral modes.
26019 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
26020 if they have any depend clauses.
26022 2017-04-11 Martin Liska <mliska@suse.cz>
26025 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
26026 * ipa-split.c (split_function): Create a local comdat symbol
26027 if caller is in a comdat group.
26029 2017-04-11 Martin Liska <mliska@suse.cz>
26032 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
26035 2017-04-11 Martin Sebor <msebor@redhat.com>
26037 PR middle-end/80364
26038 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
26039 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
26041 (directive::set_width, directive::set_precision, format_character):
26043 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
26046 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
26049 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
26050 conflict, set target->arch_name instead of target->cpu_name.
26052 2017-04-11 Richard Biener <rguenther@suse.de>
26054 PR tree-optimization/80374
26055 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
26056 build_zero_cst, remove fold_convertible_p check again.
26058 2017-04-11 Martin Liska <mliska@suse.cz>
26061 * ubsan.c (instrument_object_size): Do not instrument register
26064 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26067 * config/i386/i386-builtin-types.def
26068 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
26069 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
26070 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
26071 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
26072 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
26073 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
26074 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
26075 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
26076 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
26077 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
26078 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
26079 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
26080 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
26081 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
26082 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
26083 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
26084 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
26085 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
26086 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
26087 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
26088 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
26089 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
26090 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
26091 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
26092 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
26093 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
26094 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
26095 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
26096 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
26097 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
26098 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
26099 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
26100 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
26101 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
26102 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
26103 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
26104 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
26105 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
26106 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
26107 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
26108 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
26109 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
26110 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
26111 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
26112 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
26113 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
26115 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
26116 flag to second_arg_count, handle 4 argument function type _COUNT
26117 aliases, handle second_arg_count on second argument rather than last.
26119 2017-04-10 Jeff Law <law@redhat.com>
26121 PR tree-optimization/80374
26122 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
26123 record anything if we can not convert integer_zero_node to the
26126 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
26129 * config/rs6000/rs6000.c (rs6000_option_override_internal):
26130 Enhance special handling given to the TARGET_P9_MINMAX option in
26131 relation to certain other options.
26133 2017-04-10 Bin Cheng <bin.cheng@arm.com>
26135 PR tree-optimization/80153
26136 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
26137 remove POINTER_PLUS_EXPR's base part directly, rather than through
26140 2017-04-10 Richard Biener <rguenther@suse.de>
26141 Bin Cheng <bin.cheng@arm.com>
26143 PR tree-optimization/80153
26144 * tree-affine.c (aff_combination_to_tree): Get base pointer from
26145 the first element of pointer type aff_tree. Build result expr in
26147 (add_elt_to_tree): Convert to type unconditionally. Remove other
26148 fold_convert calls.
26149 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
26150 (rewrite_use_nonlinear_expr): Check invariant using iv information.
26152 2017-04-10 Richard Biener <rguenther@suse.de>
26154 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26157 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
26159 PR rtl-optimization/70478
26160 * lra-constraints.c (curr_small_class_check): New.
26161 (update_and_check_small_class_inputs): New.
26162 (process_alt_operands): Update curr_small_class_check. Disfavor
26163 alternative insn memory operands. Check available regs for small
26166 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26169 * config/mips/mips.opt (-mvirt): Update description.
26170 * doc/invoke.texi (-mvirt): Likewise.
26172 2017-04-10 Richard Biener <rguenther@suse.de>
26174 PR middle-end/80362
26175 * fold-const.c (fold_binary_loc): Look at unstripped ops when
26176 looking for NEGATE_EXPR in -A / -B to A / B folding.
26178 2017-04-10 Martin Liska <mliska@suse.cz>
26180 PR gcov-profile/80224
26181 * gcov.c (print_usage): Fix usage string.
26182 (get_gcov_intermediate_filename): Remove.
26183 (output_gcov_file): Use both for normal and intermediate format.
26184 (generate_results): Do not initialize special file for
26185 intermediate format.
26187 2017-04-10 Richard Biener <rguenther@suse.de>
26189 PR tree-optimization/80304
26190 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
26193 2017-04-10 Nathan Sidwell <nathan@acm.org>
26196 * config/rs6000/rs6000.c (rs6000_vector_type): New.
26197 (rs6000_init_builtins): Use it.
26199 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26201 * config/arm/arm.md (<mrc>): Add mode to SET source.
26202 (<mrrc>): Likewise.
26204 2017-04-10 Richard Biener <rguenther@suse.de>
26206 PR middle-end/80344
26207 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
26209 2017-04-10 Jakub Jelinek <jakub@redhat.com>
26212 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
26213 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
26214 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
26215 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
26216 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
26217 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
26218 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
26219 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
26220 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
26221 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
26222 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
26223 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
26224 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
26225 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
26226 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
26227 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
26228 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
26229 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
26230 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
26231 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
26232 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
26233 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
26234 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
26236 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
26238 PR rtl-optimization/70478
26239 * lra-constraints.c: Reverse the last patch.
26241 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
26243 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
26244 Add comment for WCHAR_T.
26246 2017-04-08 Martin Liska <mliska@suse.cz>
26249 2017-04-07 Martin Liska <mliska@suse.cz>
26252 * ipa-split.c (split_function): Add function part to a same comdat
26255 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26258 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
26260 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
26262 * rs6000/rs6000.c (vec_load_pendulum): Rename...
26263 (vec_pairing): ...to this.
26264 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
26265 (rs6000_sched_init): Adjust for name change.
26266 (struct rs6000_sched_context): Likewise.
26267 (rs6000_init_sched_context): Likewise.
26268 (rs6000_set_sched_context): Likewise.
26270 2017-04-07 Jakub Jelinek <jakub@redhat.com>
26276 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
26278 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
26279 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
26280 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
26282 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
26284 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
26286 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
26288 PR rtl-optimization/70703
26289 * ira-color.c (update_conflict_hard_regno_costs): Use
26290 int64_t instead of HOST_WIDE_INT.
26292 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
26294 PR rtl-optimization/70478
26295 * lra-constraints.c (process_alt_operands): Disfavor alternative
26296 insn memory operands.
26298 2017-04-07 Jeff Law <law@redhat.com>
26300 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
26301 CALL and NOTE_INSN_CALL_ARG_LOCATION.
26303 2017-04-07 Martin Liska <mliska@suse.cz>
26306 * config/aarch64/aarch64.c (aarch64_process_target_attr):
26307 Show error message instead of an ICE.
26309 2017-04-07 Martin Liska <mliska@suse.cz>
26312 * ipa-split.c (split_function): Add function part to a same comdat
26315 2017-04-07 Richard Biener <rguenther@suse.de>
26317 PR middle-end/80341
26318 * tree.c (get_unwidened): Also handle ! for_type case for
26320 * convert.c (do_narrow): Split out from ...
26321 (convert_to_integer_1): ... here. Do not pass final truncation
26322 type to get_unwidened for TRUNC_DIV_EXPR.
26324 2017-04-07 Richard Biener <rguenther@suse.de>
26326 * tree-affine.c (wide_int_ext_for_comb): Take type rather
26328 (aff_combination_const): Adjust.
26329 (aff_combination_scale): Likewise.
26330 (aff_combination_add_elt): Likewise.
26331 (aff_combination_add_cst): Likewise.
26332 (aff_combination_convert): Likewise.
26333 (add_elt_to_tree): Likewise. Remove unused argument.
26334 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
26336 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
26338 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
26340 * config/arm/arm.c (arm_default_short_enums): Use
26341 ARM_DEFAULT_SHORT_ENUMS.
26342 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
26344 2017-04-06 Jakub Jelinek <jakub@redhat.com>
26347 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
26348 members with redundant out-of-class redeclaration.
26350 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
26353 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
26354 * config/i386/i386.md (*zero_extendsidi2):
26355 Add (?*x,*x) and (?*v,*v) alternatives.
26357 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
26360 * config/i386/i386.c (ix86_expand_builtin)
26361 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
26362 mode from insn data. Convert operands to insn operand mode.
26363 Copy operands that don't satisfy insn predicate to a register.
26365 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
26367 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
26370 2017-04-06 Richard Biener <rguenther@suse.de>
26372 PR tree-optimization/80334
26373 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
26374 preserve alignment of accesses.
26376 2017-04-06 Richard Biener <rguenther@suse.de>
26378 PR tree-optimization/80262
26379 * tree-sra.c (build_ref_for_offset): Preserve address-space
26381 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
26382 Drop useless address-space information on MEM_REF offsets.
26384 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
26386 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
26388 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
26390 PR rtl-optimization/70703
26391 * ira-color.c (update_conflict_hard_regno_costs): Use
26392 HOST_WIDE_INT instead of long.
26394 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
26397 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
26398 not defined for x86_64 target. Add -mmmx target option when __SSE2__
26400 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
26401 for x86_64 target. Handle -m3dnowa option.
26403 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
26405 PR rtl-optimization/70703
26406 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
26407 (update_conflict_hard_regno_costs): Use long instead of unsigned
26408 arithmetic for cost calculation.
26410 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26411 Bernd Edlinger <bernd.edlinger@hotmail.de>
26414 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
26417 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
26420 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
26421 ptr_mode with Pmode throughout.
26422 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
26423 into probe_stack_range and use DImode.
26425 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
26428 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
26429 call_eh_return is true.
26431 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26433 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
26434 Initialize last_match_fntype_index.
26436 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26439 * tree-nvr.c: Include internal-fn.h.
26440 (pass_return_slot::execute): Ignore internal calls without
26443 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26444 Richard Biener <rguenther@suse.de>
26447 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
26448 captures used multiple times, except for the last use.
26449 * generic-match-head.c: Include gimplify.h.
26451 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26453 PR tree-optimization/79390
26454 * target.h (struct noce_if_info): Declare.
26455 * targhooks.h (default_noce_conversion_profitable_p): Declare.
26456 * target.def (noce_conversion_profitable_p): New target hook.
26457 * ifcvt.h (struct noce_if_info): New type, moved from ...
26458 * ifcvt.c (struct noce_if_info): ... here.
26459 (noce_conversion_profitable_p): Renamed to ...
26460 (default_noce_conversion_profitable_p): ... this. No longer
26462 (noce_try_store_flag_constants, noce_try_addcc,
26463 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
26464 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
26465 instead of noce_conversion_profitable_p.
26466 * config/i386/i386.c: Include ifcvt.h.
26467 (ix86_option_override_internal): Don't override
26468 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
26469 (ix86_noce_conversion_profitable_p): New function.
26470 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
26471 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
26472 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
26473 * doc/tm.texi: Regenerated.
26475 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26477 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
26480 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26483 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
26484 instructions for small multiply cores.
26486 2017-04-04 Jeff Law <law@redhat.com>
26488 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
26490 (mips_expand_vec_perm_const): Initialize elements in orig_perm
26491 that are not set by the loop over the elements.
26493 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26496 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
26497 int mode, convert_modes it to mode as unsigned, otherwise use
26498 lowpart_subreg to mode rather than SImode.
26499 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
26500 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
26501 Use DImode instead of SImode for the shift count operand.
26502 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
26505 2017-04-04 Richard Biener <rguenther@suse.de>
26507 PR middle-end/80281
26508 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
26509 arithmetic done for the negate or the plus. Simplify.
26510 (A - (-B) -> A + B): Likewise.
26511 * fold-const.c (split_tree): Make sure to not negate pointers.
26513 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
26515 PR rtl-optimization/60818
26516 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
26517 a compare of comparisons with the thing compared if this results
26518 in a different machine mode.
26520 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
26522 * alias.c (base_alias_check): Fix typo in comment.
26523 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
26524 * cgraphunit.c (symbol_table::compile): Likewise.
26525 * collect2.c (maybe_run_lto_and_relink): Likewise.
26526 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
26527 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
26528 * config/avr/avr.c (avr_map_op_t): Likewise.
26529 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
26530 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
26531 * config/epiphany/epiphany.md (movcc): Likewise.
26532 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
26533 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
26535 * config/mips/mips.c (mips_save_restore_reg): Likewise.
26536 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
26537 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
26538 * config/sh/sh.c (sh_rtx_costs): Likewise.
26539 * fold-const.c (fold_truth_andor): Likewise.
26540 * genautomata.c (collapse_flag): Likewise.
26541 * gengtype.h (struct type::u::s): Likewise.
26542 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
26543 * input.c (FORMAT_AMOUNT): Likewise.
26544 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
26545 (known_aggs_to_agg_replacement_list): Likewise.
26546 * ipa-inline-analysis.c: Likewise.
26547 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
26548 * ipa-polymorphic-call.c
26549 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
26550 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
26551 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
26553 * modulo-sched.c (apply_reg_moves): Likewise.
26554 * omp-expand.c (build_omp_regions_1): Likewise.
26555 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
26556 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
26557 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
26558 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
26559 * value-prof.c: Likewise.
26560 * var-tracking.c (val_reset): Likewise.
26562 2017-04-03 Richard Biener <rguenther@suse.de>
26564 PR tree-optimization/80275
26565 * fold-const.c (split_address_to_core_and_offset): Handle
26568 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
26570 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
26571 descriptors is at least equal to that of functions.
26573 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
26575 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
26577 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
26580 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
26581 (mov<IMOD4:mode>): New expander.
26582 (*mov<IMOD4:mode>_internal): New insn and split pattern.
26584 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
26586 PR rtl-optimization/79405
26587 * fwprop.c (propagations_left): New variable.
26588 (forward_propagate_into): Decrement it.
26589 (fwprop_init): Initialize it.
26590 (fw_prop): If the variable has reached zero, stop propagating.
26591 (fwprop_addr): Ditto.
26593 2017-03-31 Jakub Jelinek <jakub@redhat.com>
26596 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
26597 a FUNCTION_DECL, pass it as decl instead of origin to
26600 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
26602 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
26605 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
26608 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
26609 TARGET_VSX_SMALL_INTEGER.
26611 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26613 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
26614 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
26616 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26618 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
26619 extraction from odd-numbered MSA register.
26621 2017-03-31 Jakub Jelinek <jakub@redhat.com>
26623 PR middle-end/80173
26624 * expmed.c (store_bit_field_1): Don't attempt to create
26625 a word subreg out of hard registers wider than word if they
26626 have HARD_REGNO_NREGS of 1 for their mode.
26628 PR middle-end/80163
26629 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
26630 conversions to integer types wider than word and pointer.
26633 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
26634 (rtx_equal_for_cselib_p): Pass 0 to it.
26635 * cselib.c (cselib_hasher::equal): Likewise.
26636 (rtx_equal_for_cselib_1): Add depth argument. If depth
26637 is 128, don't look up VALUE locs and punt. Increment
26638 depth in recursive calls when walking VALUE locs.
26640 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
26642 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
26643 (make_gcov_file_name): Use the canonical path name for generating
26645 (read_line): Fix handling of files with ascii null bytes.
26647 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
26649 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
26650 to initialise a vector register instead
26651 of using a const_int.
26653 2017-03-30 Jakub Jelinek <jakub@redhat.com>
26655 PR translation/80189
26656 * gimplify.c (omp_default_clause): Use %qs instead of %s in
26657 diagnostic messages.
26659 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
26662 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
26663 (dfp_diex_<mode>): Update mode of operand 1.
26664 * doc/extend.texi (dxex, dxexq): Document change to return type.
26665 (diex, diexq): Document change to argument type.
26667 2017-03-30 Martin Jambor <mjambor@suse.cz>
26670 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
26671 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
26672 it reflects the signature changes performed at the callee side.
26673 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
26674 to cgraph_build_function_type_skip_args.
26675 (build_function_decl_skip_args): Adjust call to the above function.
26677 2017-03-30 Jakub Jelinek <jakub@redhat.com>
26680 * config/i386/sse.md
26681 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
26682 register as dest whenever it is a MEM not rtx_equal_p to the
26683 corresponding dup operand, and when forcing into reg move the
26684 reg into the memory afterwards.
26685 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
26686 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
26687 for the force_reg mode.
26688 (avx512vl_vextractf128<mode>): Use register as dest either
26689 always when a MEM, or when it is a MEM not rtx_equal_p to the
26690 corresponding dup operand, or even not when it is a CONST_VECTOR
26691 depending on the mode and lo vs. hi.
26692 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
26694 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
26695 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
26696 Likewise. Require that operands[2] is even.
26697 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
26698 Remove extraneous parens. Require that operands[2] is a multiple
26700 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
26701 operands[0] is a MEM if <mask_applied>, the predicates/constraints
26702 disallow memory then.
26704 2017-03-30 Richard Biener <rguenther@suse.de>
26706 PR tree-optimization/77498
26707 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
26708 to non-constants over backedges.
26710 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
26712 PR rtl-optimization/80233
26713 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
26714 as last_combined_insn. Do not test for BARRIER_P separately.
26716 2017-03-29 Andreas Schwab <schwab@suse.de>
26719 * calls.c (prepare_call_address): Convert funexp to Pmode before
26720 copying to temp reg.
26722 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26724 PR tree-optimization/80158
26725 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
26726 Handle possible future case of more than one alternate
26728 (replace_rhs_if_not_dup): Likewise.
26729 (replace_one_candidate): Likewise.
26731 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
26733 PR rtl-optimization/80193
26734 * ira.c (ira): Do not check allocation for LRA.
26736 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
26738 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
26739 (nvptx_output_simt_exit): Declare.
26740 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
26741 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
26742 (init_softstack_frame): Move initialization of crtl->is_leaf to...
26743 (nvptx_declare_function_name): ...here. Emit declaration of local
26744 memory space buffer for omp_simt_enter insn.
26745 (nvptx_output_unisimt_switch): New.
26746 (nvptx_output_softstack_switch): New.
26747 (nvptx_output_simt_enter): New.
26748 (nvptx_output_simt_exit): New.
26749 * config/nvptx/nvptx.h (struct machine_function): New fields
26750 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
26751 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
26752 (UNSPECV_SIMT_EXIT): Ditto.
26753 (omp_simt_enter_insn): New insn.
26754 (omp_simt_enter): New expansion.
26755 (omp_simt_exit): New insn.
26756 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
26758 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
26759 (expand_GOMP_SIMT_ENTER_ALLOC): New.
26760 (expand_GOMP_SIMT_EXIT): New.
26761 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
26762 (GOMP_SIMT_ENTER_ALLOC): Ditto.
26763 (GOMP_SIMT_EXIT): Ditto.
26764 * target-insns.def (omp_simt_enter): New insn.
26765 (omp_simt_exit): Ditto.
26766 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
26768 (lower_rec_simd_input_clauses): Implement SIMT privatization.
26769 (lower_rec_input_clauses): Likewise.
26770 (lower_lastprivate_clauses): Handle SIMT privatization.
26772 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
26773 (ompdevlow_adjust_simt_enter): New.
26774 (find_simtpriv_var_op): New.
26775 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
26776 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
26778 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
26779 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
26780 (copy_decl_for_dup_finish): Ditto.
26782 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
26784 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
26787 * config/i386/i386.c (ix86_option_override_internal): Always
26788 allow -mpreferred-stack-boundary=3 for 64-bit targets.
26790 2017-03-28 Bin Cheng <bin.cheng@arm.com>
26792 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
26794 2017-03-28 Bin Cheng <bin.cheng@arm.com>
26796 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
26797 mark new edge's irreducible flag accordign to it.
26798 (vect_do_peeling): Check loop preheader edge's irreducible flag
26799 and pass it to function slpeel_add_loop_guard.
26801 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
26803 PR tree-optimization/80218
26804 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
26805 Update block frequencies and counts.
26807 2017-03-28 Richard Biener <rguenther@suse.de>
26809 PR tree-optimization/78644
26810 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
26811 of a simplification result we may not use it at all.
26813 2017-03-28 Richard Biener <rguenther@suse.de>
26816 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
26817 without arguments, generate default definition of a SSA name.
26819 2017-03-28 Richard Biener <rguenther@suse.de>
26821 PR middle-end/80222
26822 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
26823 TYPE_REF_CAN_ALIAS_ALL references.
26824 * fold-const.c (fold_indirect_ref_1): Likewise.
26826 2017-03-28 Martin Liska <mliska@suse.cz>
26829 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
26830 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
26832 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
26833 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
26835 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
26836 (EXTRA_SPECS): Define.
26837 (SUBTARGET_EXTRA_SPECS): Likewise.
26838 (SUBTARGET_CPP_SPEC): Likewise.
26839 * config/arc/elf.h (EXTRA_SPECS): Renamed to
26840 SUBTARGET_EXTRA_SPECS.
26841 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
26843 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
26845 * config/arc/simdext.md (vst64_insn): Update pattern.
26846 (vld32wh_insn): Likewise.
26847 (vld32wl_insn): Likewise.
26848 (vld64_insn): Likewise.
26849 (vld32_insn): Likewise.
26851 2017-03-28 Marek Polacek <polacek@redhat.com>
26854 * fold-const.c (fold_comparison): Use protected_set_expr_location
26855 instead of SET_EXPR_LOCATION.
26857 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
26859 * tree.c (add_expr): Avoid name lookup warning.
26861 2017-03-27 Jeff Law <law@redhat.com>
26863 PR tree-optimization/80216
26864 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
26865 function name. Limit recursion depth.
26866 (record_temporary_equivalences): Corresponding changes.
26868 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
26870 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
26873 2017-03-27 Jakub Jelinek <jakub@redhat.com>
26876 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
26878 * cfgcleanup.c (reg_note_cfa_p): New array.
26879 (insns_have_identical_cfa_notes): New function.
26880 (old_insns_match_p): Don't cross-jump in between /f
26881 and non-/f instructions. If both i1 and i2 are frame related,
26882 verify all CFA notes, their order and content.
26884 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
26887 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
26888 HImode and SImode with zero extend to DImode to one insn.
26889 (bswap<mode>2_extenddi): Likewise.
26890 (bswapsi2_extenddi): Likewise.
26891 (bswaphi2_extendsi): Likewise.
26892 (bswaphi2): Combine bswap HImode and SImode into one insn.
26893 Separate memory insns from swapping register.
26894 (bswapsi2): Likewise.
26895 (bswap<mode>2): Likewise.
26896 (bswaphi2_internal): Delete, no longer used.
26897 (bswapsi2_internal): Likewise.
26898 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
26899 store, and gpr<-gpr swap insns.
26900 (bswap<mode>2_store): Likewise.
26901 (bswaphi2_reg): Register only splitter, combine with the splitter.
26902 (bswaphi2 splitter): Likewise.
26903 (bswapsi2_reg): Likewise.
26904 (bswapsi2 splitter): Likewise.
26905 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
26906 the insns into load, store, and register/register insns.
26907 (bswapdi2_ldbrx): Likewise.
26908 (bswapdi2_load): Likewise.
26909 (bswapdi2_store): Likewise.
26910 (bswapdi2_reg): Likewise.
26912 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
26914 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
26915 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
26917 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
26920 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
26922 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
26923 special handling for target option conflicts between dform
26924 options (-mpower9-dform, -mpower9-dform-vector,
26925 -mpower9-dform-scalar) and -mno-direct-move.
26927 2017-03-27 Richard Biener <rguenther@suse.de>
26929 PR tree-optimization/80181
26930 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
26932 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
26934 * config/arc/predicates.md (move_double_src_operand): Replace the
26935 call to move_double_src_operand with a call to address_operand.
26937 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
26939 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
26940 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
26941 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
26943 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
26945 * config/arc/predicates.md (long_immediate_loadstore_operand):
26946 Consider scaled addresses cases.
26948 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
26950 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
26951 restored when in interrupt.
26952 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
26953 doesn't have delay slot.
26955 2017-03-27 Richard Biener <rguenther@suse.de>
26958 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
26959 inlined thunk clones.
26961 2017-03-27 Jakub Jelinek <jakub@redhat.com>
26964 * asan.c (instrument_derefs): Copy over last operand from
26965 original COMPONENT_REF to the new COMPONENT_REF with
26966 DECL_BIT_FIELD_REPRESENTATIVE.
26967 * ubsan.c (instrument_object_size): Likewise.
26969 2017-03-27 Richard Biener <rguenther@suse.de>
26971 PR tree-optimization/80170
26972 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
26973 sure DR/SCEV didnt fold in constants we do not see when looking
26974 at the reference base alignment.
26976 2017-03-27 Richard Biener <rguenther@suse.de>
26978 PR middle-end/80171
26979 * gimple-fold.c (fold_ctor_reference): Properly guard against
26980 NULL return value from canonicalize_constructor_val.
26982 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
26985 * config/i386/i386.c (ix86_expand_builtin)
26986 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
26987 flags reg setting and flags reg using instructions.
26988 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
26989 clobbering instructions to zero extend op2.
26991 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
26993 * doc/install.texi (Configuration) <--with-aix-soname>:
26994 Update link to AIX ld.
26996 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
26998 PR rtl-optimization/80160
26999 PR rtl-optimization/80159
27000 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
27001 reg_alternate_class into account.
27003 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
27006 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
27007 to consider in curr_insn_transform.
27009 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27011 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
27012 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
27013 and emit_mode_inner.
27015 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27017 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
27018 argument to the overloaded builtin variants. Use the new flag to
27019 deprecate certain builtin variants.
27020 * config/s390/s390-builtin-types.def: Add new builtin types.
27021 * config/s390/s390-builtins.h: Support new flags field for
27022 overloaded builtins.
27023 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
27024 (s390_macro_to_expand): Enable vector float data type.
27025 (s390_cpu_cpp_builtins_internal): Indicate support of the new
27026 builtins by incrementing the __VEC__ version number.
27027 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
27029 (s390_resolve_overloaded_builtin): Emit error messages depending
27030 on the builtin flags.
27031 * config/s390/s390.c (s390_expand_builtin): Support additional
27032 flags argument. Change error message to match the messages
27033 emitted in s390-c.c.
27034 * config/s390/s390.md: New UNSPEC_* constants.
27035 (op_type): Add new instruction types.
27036 * config/s390/vecintrin.h: Add new builtins and test data class
27038 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
27039 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
27040 (VEC_INEXACT, VEC_NOINEXACT): New constants.
27041 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
27042 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
27043 ("vec_mergel<mode>"): V_HW -> VEC_HW.
27045 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
27046 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
27047 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
27048 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
27050 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
27051 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
27052 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
27053 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
27055 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
27056 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
27057 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
27058 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
27059 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
27060 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
27061 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
27063 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
27064 ("vec_scatter_element<V_HW_4:mode>_DI")
27065 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
27066 ("vec_fpint<mode>", "vflls")
27067 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
27068 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
27069 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
27070 ("*vec_cmphe<mode>_cc"): ... these.
27072 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
27073 mode constant instead of magic value.
27075 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27077 * config/s390/s390.c (s390_expand_vec_compare): Support other
27078 vector floating point modes than just V2DF.
27079 (s390_expand_vcond): Likewise.
27080 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
27081 (s390_cannot_change_mode_class): Prevent mode changes between TF
27082 and V1TF in vector registers.
27083 * config/s390/s390.md (DF, SF): New mode attributes.
27084 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
27085 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
27086 SFmode support for VRs.
27087 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
27089 (VFT, VF_HW): New mode iterators.
27090 (vw, sdx): New mode attributes.
27091 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
27092 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
27093 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
27094 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
27095 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
27096 also the new vector floating point modes. Renaming to ...
27098 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
27099 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
27100 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
27101 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
27102 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
27103 ("vec_unordered<mode>"): ... these.
27105 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
27106 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
27107 ("*vec_extendv2df"): New insn definitions.
27109 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27111 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
27112 ("mulditi3_2", "*muldi3_sign"): New patterns.
27113 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
27114 rename the pattern definition.
27116 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27118 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
27120 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
27122 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27124 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
27125 instruction if possible.
27126 * config/s390/vector.md (vec_halfnumelts): New mode
27128 ("*vec_vllezlf<mode>"): New pattern.
27130 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27132 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
27133 ("popcountv4si2", "popcountv2di2"): Rename to ...
27134 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
27135 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
27137 ("popcount<mode>2_vxe"): New pattern.
27139 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27141 * common/config/s390/s390-common.c (processor_flags_table): Add
27143 * config.gcc: Add arch12.
27144 * config/s390/driver-native.c (s390_host_detect_local_cpu):
27145 Default to arch12 for unknown CPU model numbers.
27146 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
27147 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
27148 PROCESSOR_max sanity check.
27149 * config/s390/s390-opts.h (enum processor_type): Add
27151 * config/s390/s390.c (processor_table): Add arch12.
27152 (s390_expand_builtin): Add check for B_VXE flag.
27153 (s390_issue_rate): Add PROCESSOR_ARCH12.
27154 (s390_get_sched_attrmask): Likewise.
27155 (s390_get_unit_mask): Likewise.
27156 (s390_sched_score): Enable z13 scheduling for arch12.
27157 (s390_sched_reorder): Likewise.
27158 (s390_sched_variable_issue): Likewise.
27159 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
27161 (s390_tune_attr): Use z13 scheduling also for arch12.
27162 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
27163 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
27164 (TARGET_VXE_P): New macros.
27165 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
27166 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
27167 * config/s390/s390.opt: Add arch12 as processor_type.
27169 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27171 * config/s390/s390.md
27172 ("fixuns_truncdddi2", "fixuns_trunctddi2")
27173 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
27174 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
27176 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
27177 Rename expanders to ...
27179 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
27180 ("fixuns_truncdddi2_emu"): ... these.
27182 ("fixuns_trunc<mode>si2_emu"): New expander.
27184 ("*fixuns_truncdfdi2_z13"): Rename to ...
27185 ("*fixuns_truncdfdi2_vx"): ... this.
27187 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27189 * config/s390/2964.md: Remove the single element vector compare
27190 instructions which are no longer used.
27191 * config/s390/s390.c (s390_select_ccmode): Remove handling of
27193 (s390_canonicalize_comparison): Remove handling of DFmode
27195 (s390_expand_vec_compare_scalar): Remove function.
27196 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
27197 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
27199 ("*cmp<mode>_ccs"): Add wfcdb instruction.
27201 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27203 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
27205 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
27206 will anyway by matched by mov<mode>_64dfp.
27208 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27210 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
27211 vlef/vstef. Add missing operand to vleif.
27213 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27215 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
27216 pair for all vector types with 64 bit elements.
27217 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
27218 * config/s390/vector.md (V_HW_64): ... here.
27219 (V_128_NOSINGLE): New mode iterator.
27220 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
27221 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
27222 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
27223 ("*vec_load_pairv2di"): Change to ...
27224 ("*vec_load_pair<mode>"): ... this one.
27226 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27228 * config/s390/constraints.md: Add comments.
27229 (jKK): Reject element sizes > 8 bytes.
27230 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
27232 * config/s390/s390.md: Add the s_operand checks formerly in
27233 s390_split_ok_p to various splitters where they are still
27235 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
27236 for 128 bit vectors. Plus two splitters.
27238 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27240 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
27243 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27246 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
27247 error if the boundary argument is not constant.
27249 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27251 PR rtl-optimization/80112
27252 * loop-doloop.c (doloop_condition_get): Don't check condition
27253 if cmp isn't SET with IF_THEN_ELSE src.
27255 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27257 PR tree-optimization/80158
27258 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
27259 replacing a candidate statement, also replace it for the
27260 candidate's alternate interpretation.
27261 (replace_rhs_if_not_dup): Likewise.
27262 (replace_one_candidate): Likewise.
27264 2017-03-24 Richard Biener <rguenther@suse.de>
27266 PR tree-optimization/80167
27267 * graphite-isl-ast-to-gimple.c
27268 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
27270 (translate_isl_ast_to_gimple::get_rename): Likewise.
27272 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
27274 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
27275 handling of certain combinations of target options, including the
27276 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
27277 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
27279 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27282 * config/arm/arm.md (*load_multiple): Add reload_completed to
27283 matching condition.
27285 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27286 Richard Biener <rguenth@suse.de>
27288 PR tree-optimization/79908
27289 PR tree-optimization/80136
27290 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27291 been cast away, gimplify_and_add suffices.
27293 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
27295 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
27297 2017-03-23 Richard Biener <rguenther@suse.de>
27299 PR tree-optimization/80032
27300 * gimplify.c (gimple_push_cleanup): Forced unconditional
27301 cleanups still have to go to the conditional_cleanups
27304 2017-03-22 Jakub Jelinek <jakub@redhat.com>
27306 PR tree-optimization/80072
27307 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
27309 (next_operand_entry_id): Change type to unsigned int.
27310 (sort_by_operand_rank): Make sure to return the right return value
27311 even if unsigned fields are bigger than INT_MAX.
27312 (struct oecount): Change cnt and id type to unsigned int.
27313 (oecount_hasher::equal): Formatting fix.
27314 (oecount_cmp): Make sure to return the right return value
27315 even if unsigned fields are bigger than INT_MAX.
27316 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
27319 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
27320 TREE_READONLY on result if writing it more than once.
27323 * doc/invoke.texi (-fsanitize=thread): Document that with
27324 -fnon-call-exceptions atomics are not able to throw
27328 * tsan.c: Include tree-eh.h.
27329 (instrument_builtin_call): Call maybe_clean_eh_stmt or
27330 maybe_clean_or_replace_eh_stmt where needed.
27331 (instrument_memory_accesses): Add cfg_changed argument.
27332 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
27333 if it returned true.
27334 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
27336 PR rtl-optimization/63191
27337 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
27338 wrapper function, moved the whole old content into ...
27339 (ix86_delegitimize_address_1): ... this. New inline function.
27340 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
27341 true as last argument instead of ix86_delegitimize_address.
27343 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
27345 * config/aarch64/aarch64.c (generic_branch_cost): Copy
27346 cortexa57_branch_cost.
27348 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
27350 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
27352 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27355 * doc/md.texi (Constraints): Document wA constraint.
27356 * config/rs6000/constraints.md (wA): New.
27357 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
27358 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
27359 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
27360 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
27362 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
27365 * gimplify.c (is_oacc_declared): New function.
27366 (oacc_default_clause): Use it to set default flags for acc declared
27367 variables inside parallel regions.
27368 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
27369 declared variables.
27370 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
27371 declare attribute to any decl as necessary.
27373 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
27376 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
27377 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
27378 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
27379 (arm_arch_lpae): This.
27380 * config/arm/arm.c (arm_arch7ve): Rename into ...
27381 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
27382 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
27385 2017-03-22 Martin Liska <mliska@suse.cz>
27388 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
27389 error message instead of an ICE.
27391 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27393 * doc/extend.texi (6.11 Additional Floating Types): Revise.
27395 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
27397 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
27399 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27402 2017-03-21 Martin Sebor <msebor@redhat.com>
27404 * doc/extend.texi: Use "cannot" instead of "can't."
27405 * doc/hostconfig.texi: Same.
27406 * doc/install.texi: Same.
27407 * doc/invoke.texi: Same.
27408 * doc/loop.texi: Same.
27409 * doc/md.texi: Same.
27410 * doc/objc.texi: Same.
27411 * doc/rtl.texi: Same.
27412 * doc/tm.texi: Same.
27413 * doc/tm.texi.in: Same.
27414 * doc/trouble.texi: Same.
27416 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
27419 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
27420 (collect_checksum_attributes): Set it.
27421 (die_checksum_ordered): Use it.
27423 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27425 PR tree-optimization/79908
27426 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
27427 change: For a VA_ARG whose LHS has been cast away, use
27428 force_gimple_operand to construct the side effects.
27430 2017-03-21 David Malcolm <dmalcolm@redhat.com>
27432 PR translation/80001
27433 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
27434 more amenable to translation.
27435 (oacc_loop_auto_partitions): Likewise.
27437 2017-03-21 Marek Polacek <polacek@redhat.com>
27438 Martin Sebor <msebor@redhat.com>
27440 PR tree-optimization/80109
27441 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
27442 on INTEGRAL_TYPE_P.
27444 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27445 Segher Boessenkool <segher@kernel.crashing.org>
27448 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
27449 check reg_used_between_p between insn and one of succ or succ2
27450 depending on if succ is artificial insn not inserted into insn
27453 2017-03-21 Martin Liska <mliska@suse.cz>
27455 PR gcov-profile/80081
27456 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
27457 * doc/gcc.texi: Include gcov-dump stuff.
27458 * doc/gcov-dump.texi: New file.
27460 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
27462 PR rtl-optimization/79150
27463 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
27464 conditional jump, if the jump is the last insn of the loop.
27466 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27467 Richard Biener <rguenth@suse.de>
27469 PR tree-optimization/79908
27470 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27471 been cast away, use force_gimple_operand to construct the side
27474 2017-03-21 Martin Liska <mliska@suse.cz>
27476 PR libfortran/79956
27477 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
27480 2017-03-21 Brad Spengler <spender@grsecurity.net>
27483 * plugin.c (htab_hash_plugin): New function.
27484 (add_new_plugin): Use it and adjust.
27485 (parse_plugin_arg_opt): Adjust.
27486 (init_one_plugin): Likewise.
27488 2017-03-21 Richard Biener <rguenther@suse.de>
27490 PR tree-optimization/80032
27491 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
27492 if set force the cleanup to happen unconditionally.
27493 (gimplify_target_expr): Push inserted clobbers with force_uncond
27494 to avoid them being removed by control-dependent DCE.
27496 2017-03-21 Richard Biener <rguenther@suse.de>
27498 PR tree-optimization/80122
27499 * tree-inline.c (copy_bb): Do not expans va-arg packs or
27500 va_arg_pack_len when the inlined call stmt requires pack
27502 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
27504 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27507 * tsan.c (instrument_builtin_call): If the memory model argument
27508 is not a constant, assume it is valid.
27511 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
27514 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27516 PR rtl-optimization/79910
27517 * combine.c (can_combine_p): Do not allow combining an I0 or I1
27518 if its dest is used by an insn before I2 (other than the combined
27519 insns themselves, which are properly handled already).
27521 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27524 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
27526 * combine.c (record_used_regs): New static function.
27527 (try_combine): Handle situations where there is an additional
27528 instruction between I2 and I3 which needs to have a LOG_LINK
27532 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27534 * combine.c (try_combine): Delete redundant i1 test. Call
27535 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27537 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27540 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
27541 alternatives 13/14.
27543 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27545 PR tree-optimization/80054
27546 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
27547 the optimization if a PHI or any of its arguments is not dominated
27548 by the candidate's basis. Use gphi* rather than gimple* as
27550 (replace_profitable_candidates): Clean up a gimple* variable that
27551 should be a gphi* variable.
27553 2017-03-20 Martin Sebor <msebor@redhat.com>
27556 * doc/extend.texi (attribute constructor): Document present limitation.
27558 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
27561 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
27562 __POWER9_VECTOR__ #ifdef control, change template definition to
27563 use Power9-specific built-in function.
27564 (vec_any_eq): Likewise.
27565 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
27566 to control outcomes from this test.
27567 (vector_ae_<mode>p): For VEC_F modes, likewise.
27569 2017-03-20 Ian Lance Taylor <iant@google.com>
27571 * config/i386/i386.c (ix86_function_regparm): Save an extra
27572 register for -fsplit-stack with DECL_STATIC_CHAIN.
27574 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
27577 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
27578 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
27580 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
27582 * config/riscv/riscv.c (riscv_print_operand): Use "fence
27584 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
27587 2017-03-20 Marek Polacek <polacek@redhat.com>
27590 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
27592 2017-03-20 Richard Biener <rguenther@suse.de>
27594 PR tree-optimization/80113
27595 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
27596 allocate extra SSA name for PHI def.
27597 (add_close_phis_to_outer_loops): Likewise.
27598 (add_close_phis_to_merge_points): Likewise.
27599 (copy_loop_close_phi_args): Likewise.
27600 (copy_cond_phi_nodes): Likewise.
27602 2017-03-20 Martin Liska <mliska@suse.cz>
27604 PR middle-end/79753
27605 * tree-chkp.c (chkp_build_returned_bound): Do not build
27606 returned bounds for a LHS that's not a BOUNDED_P type.
27608 2017-03-20 Martin Liska <mliska@suse.cz>
27612 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
27613 COMPLEX_CST and VECTOR_CST.
27615 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27618 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
27619 target operand. A new splitter adds the clobber statement in case
27620 the target operand is dead anyway.
27622 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
27624 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
27625 to age-old versions of binutils and glibc.
27627 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
27629 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
27631 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
27633 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
27635 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
27637 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
27638 requirement for binutils 2.13.
27640 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27642 * combine.c (try_combine): Delete redundant i1 test. Call
27643 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27645 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
27647 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
27648 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
27650 <riscv64-*-elf>: Re-arrange section
27651 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
27652 <riscv32-*-linux>: Likewise.
27653 <riscv64-*-elf>: Likewise
27654 <riscv64-*-linux>: Likewise.
27656 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
27659 * aarch64.opt(verbose-cost-dump): Fix typo.
27661 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
27664 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
27665 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
27667 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
27669 * reload.c (find_reloads): When reloading a nonoffsettable address,
27670 use RELOAD_OTHER for it and its address reloads.
27672 PR rtl-optimization/79910
27673 * combine.c (record_used_regs): New static function.
27674 (try_combine): Handle situations where there is an additional
27675 instruction between I2 and I3 which needs to have a LOG_LINK
27678 2017-03-17 Jeff Law <law@redhat.com>
27680 PR tree-optimization/71437
27681 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
27682 conditional in the hash table first.
27683 (vrp_dom_walker::before_dom_children): Extract condition from
27684 ASSERT_EXPR. Record condition, its inverion and any implied
27685 conditions as well.
27687 2017-03-17 Marek Polacek <polacek@redhat.com>
27688 Markus Trippelsdorf <markus@trippelsdorf.de>
27690 PR tree-optimization/80079
27691 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
27694 2017-03-17 Richard Biener <rguenther@suse.de>
27696 PR middle-end/80075
27697 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
27698 Properly verify the LHS before the RHS possibly claims to be
27700 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
27703 2017-03-17 Martin Jambor <mjambor@suse.cz>
27705 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
27706 (List of -O2 options): Likewise.
27707 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
27710 2017-03-17 Tom de Vries <tom@codesourcery.com>
27712 * gcov-dump.c (print_usage): Print bug_report_url.
27714 2017-03-17 Richard Biener <rguenther@suse.de>
27716 PR middle-end/80050
27717 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
27718 (parser::peek): Likewise.
27720 2017-03-17 Richard Biener <rguenther@suse.de>
27722 PR tree-optimization/80048
27723 * sese.c (free_sese_info): Properly release rename_map and
27724 copied_bb_map elements.
27726 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
27728 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
27729 Add linked-list forward and backlinks. Insert on
27730 construction, remove on destruction.
27731 (class pass_store_merging): Add m_stores_head field.
27732 (pass_store_merging::terminate_and_process_all_chains):
27733 Iterate over m_stores_head list.
27734 (pass_store_merging::terminate_all_aliasing_chains):
27736 (pass_store_merging::execute): Check for debug stmts first.
27737 Push new chains onto the m_stores_head stack.
27739 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
27742 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
27743 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
27744 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
27746 2017-03-16 Jeff Law <law@redhat.com>
27748 PR tree-optimization/71437
27749 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
27750 member function. Implementation moved into after_dom_children
27751 member function and into the threader's thread_outgoing_edges
27753 (dom_opt_dom_walker::after_dom_children): Simplify by moving
27754 some code into new thread_outgoing_edges.
27755 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
27756 definition. Simplify marker handling (do it here). Assume we always
27757 have the available expression and the const/copies tables.
27758 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
27760 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
27761 * tree-vrp.c (equiv_stack): No longer file scoped.
27762 (vrp_dom_walker): New class.
27763 (vrp_dom_walker::before_dom_children): New member function.
27764 (vrp_dom_walker::after_dom_children): Likewise.
27765 (identify_jump_threads): Setup domwalker. Use it rather than
27766 walking edges in a random order by hand. Simplify setup/finalization.
27767 (finalize_jump_threads): Remove.
27768 (vrp_finalize): Do not call identify_jump_threads here.
27769 (execute_vrp): Do it here instead and call thread_through_all_blocks
27772 PR tree-optimization/71437
27773 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
27775 (simplify_stmt_for_jump_threading): Add basic_block argument. All
27777 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
27778 (dom_opt_dom_walker::thread_across_edge): Remove
27779 handle_dominating_asserts argument. All callers changed.
27780 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
27781 changes. Remove calls to lhs_of_dominating_assert. Other
27782 uses of handle_dominating_asserts turn into unconditional code
27783 (simplify_control_stmt_condition_1): Likewise.
27784 (simplify_control_stmt_condition): Likewise.
27785 (thread_through_normal_block, thread_across_edge): Likewise.
27786 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
27787 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
27788 object if it is not an SSA_NAME.
27789 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
27790 before calling into the VRP specific simplifiers.
27791 (identify_jump_threads): Remove handle_dominating_asserts
27794 2017-03-16 Jakub Jelinek <jakub@redhat.com>
27797 * tree-diagnostic.c (default_tree_printer): No longer static.
27798 * tree-diagnostic.h (default_tree_printer): New prototype.
27800 2017-03-16 Tamar Christina <tamar.christina@arm.com>
27802 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
27803 Change ins into fmov.
27805 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27807 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
27808 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
27809 Use h_con constraint for operand 1.
27810 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
27811 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
27813 2017-03-15 Jeff Law <law@redhat.com>
27815 PR tree-optimization/71437
27816 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
27817 (record_temporary_equivalences): Use it.
27819 PR tree-optimization/71437
27820 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
27821 tree-ssa-scopedtables.
27822 (lookup_avail_expr, build_and_record_new_cond): Likewise.
27823 (record_conditions, record_cond, vuse_eq): Likewise.
27824 (record_edge_info): Adjust to API tweak of record_conditions.
27825 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
27826 (record_temporary_equivalences, optimize_stmt): Likewise.
27827 (eliminate_redundant_computations): Likewise.
27828 (record_equivalences_from_stmt): Likewise.
27829 * tree-ssa-scopedtables.c: Include options.h and params.h.
27830 (vuse_eq): New function, moved from tree-ssa-dom.c
27831 (build_and_record_new_cond): Likewise.
27832 (record_conditions): Likewise. Accept vector of conditions rather
27833 than edge_equivalence structure for first argument.
27834 for the first argument.
27835 (avail_exprs_stack::lookup_avail_expr): New member function, moved
27836 from tree-ssa-dom.c.
27837 (avail_exprs_stack::record_cond): Likewise.
27838 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
27839 from tree-ssa-dom.c.
27840 (avail_exprs_stack): Add new member functions lookup_avail_expr
27842 (record_conditions): Declare.
27844 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
27847 * lra-constraints.c (process_alt_operands): Increase reject for
27848 reloading an input/output operand.
27850 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
27853 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
27854 insns to convert from signed/unsigned char/short to IEEE 128-bit
27856 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
27858 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
27861 * config/i386/i386.c (ix86_vector_duplicate_value): Create
27862 subreg of inner mode for values already in registers.
27864 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
27866 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
27867 iteration reg is used after the loop.
27869 2017-03-14 Martin Sebor <msebor@redhat.com>
27871 PR tree-optimization/79800
27872 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
27873 precision in negative-positive range.
27874 (format_floating): Call non-const overload with adjusted precision.
27876 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
27879 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
27882 2017-03-14 Martin Sebor <msebor@redhat.com>
27884 PR middle-end/80020
27885 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
27886 * builtins.def (aligned_alloc): Use it.
27889 * Makefile.in (GTFILES): Add calls.c.
27890 * calls.c: Include "gt-calls.h".
27892 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
27894 PR rtl-optimization/79728
27895 * regs.h (struct target_regs): New field
27896 x_contains_allocatable_regs_of_mode.
27897 (contains_allocatable_regs_of_mode): New macro.
27898 * reginfo.c (init_reg_sets_1): Initialize it, and change
27899 contains_reg_of_mode so it includes global regs as well.
27900 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
27901 rather than contains_regs_of_mode.
27903 2017-03-14 Martin Liska <mliska@suse.cz>
27905 * doc/invoke.texi: Document options that can't be combined with
27906 -fcheck-pointer-bounds.
27908 2017-03-14 Martin Liska <mliska@suse.cz>
27910 PR middle-end/79831
27911 * doc/invoke.texi (-Wchkp): Document the option.
27913 2017-03-14 Martin Liska <mliska@suse.cz>
27915 * Makefile.in: Install gcov-dump.
27917 2017-03-14 Martin Liska <mliska@suse.cz>
27919 * multiple_target.c (expand_target_clones): Bail out for
27920 an invalid attribute.
27922 2017-03-14 Richard Biener <rguenther@suse.de>
27924 * alias.c (struct alias_set_entry): Pack properly.
27925 * cfgloop.h (struct loop): Likewise.
27926 * cse.c (struct set): Likewise.
27927 * ipa-utils.c (struct searchc_env): Likewise.
27928 * loop-invariant.c (struct invariant): Likewise.
27929 * lra-remat.c (struct cand): Likewise.
27930 * recog.c (struct change_t): Likewise.
27931 * rtl.h (struct address_info): Likewise.
27932 * symbol-summary.h (function_summary): Likewise.
27933 * tree-loop-distribution.c (struct partition): Likewise.
27934 * tree-object-size.c (struct object_size_info): Likewise.
27935 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
27936 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
27937 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
27938 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
27939 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
27940 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
27941 (struct _stmt_vec_info): Likewise.
27943 2017-03-14 Martin Liska <mliska@suse.cz>
27946 * multiple_target.c (create_dispatcher_calls): Check that
27947 a target can create a function dispatcher.
27949 2017-03-14 Martin Liska <mliska@suse.cz>
27952 * multiple_target.c (expand_target_clones): Drop local.local
27953 flag for default implementation.
27955 2017-03-14 Richard Biener <rguenther@suse.de>
27957 PR tree-optimization/80030
27958 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
27960 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
27962 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
27963 gcc_fallthrough() instead of __attribute__((fallthrough));
27965 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
27967 * doc/gcc.texi: Remove "up" link to (DIR).
27968 * doc/gccint.texi: Ditto.
27970 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
27972 * doc/install.texi (Specific) <avr>: Remove reference to
27975 2017-03-13 Jeff Law <law@redhat.com>
27977 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
27978 attribute rather than comments.
27980 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
27981 match_scratch operand is highest.
27983 2017-03-13 Martin Liska <mliska@suse.cz>
27985 PR middle-end/78339
27986 * ipa-pure-const.c (warn_function_noreturn): If the declarations
27987 is a CHKP clone, use original declaration.
27989 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27991 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
27992 (arc_conditional_register_usage): Use a different allocation order
27993 when optimizing for size.
27994 * common/config/arc/arc-common.c (arc_option_optimization_table):
27995 Section anchors default on when optimizing for size.
27997 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27999 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
28001 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28003 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
28004 * config/arc/arc.md (cpu_facility): Add cd variant.
28005 (*movqi_insn): Add code density variant.
28006 (*movhi_insn): Likewise.
28007 (*movqi_insn): Likewise.
28008 (*addsi3_mixed): Likewise.
28009 (subsi3_insn): Likewise.
28011 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28013 * config/arc/arc.md (movsi_cond_exec): Update constraint.
28015 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28017 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
28018 expressions with MINUS and UNARY ops.
28020 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28023 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
28025 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
28026 between vec_select and vector argument.
28027 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
28028 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
28029 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
28030 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
28031 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
28032 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
28034 2017-03-13 Richard Biener <rguenther@suse.de>
28037 * params.def (vect-max-peeling-for-alignment): Fix typo.
28039 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28041 * doc/install.texi (Specific) <mips-*-*>: Remove description of
28042 issue that only occurred with binutils below 2.18.
28044 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28046 * doc/install.texi (Specific) <cris-axis-elf>: No longer
28047 refer to binutils 2.11/2.12 minimum.
28049 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28051 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
28052 ftp.kernel.org and simplify binutils requirement.
28054 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
28056 * doc/invoke.texi (Warning Options): Fix spelling of link-time
28058 (Optimize Options): Ditto. Also remove redundancy.
28060 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28062 PR translation/79848
28063 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
28065 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
28066 to G_ to avoid double translation.
28068 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28070 PR translation/79923
28071 * auto-profile.c (get_combined_location): Convert leading
28072 character of diagnostics to lower case and remove trailing period.
28073 (read_profile): Likewise for various diagnostics.
28074 * config/arm/arm.c (arm_option_override): Remove trailing period
28075 from various diagnostics.
28076 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
28077 (msp430_expand_delay_cycles): Likewise.
28079 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28082 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
28083 full command-line argument, rather than just "str".
28084 (aarch64_validate_march): Likewise.
28085 (aarch64_validate_mtune): Likewise.
28087 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
28089 PR rtl-optimization/78911
28090 * lra-assigns.c (must_not_spill_p): New function.
28091 (spill_for): Use it.
28093 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28095 PR tree-optimization/79981
28096 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
28097 ATOMIC_COMPARE_EXCHANGE ifn result.
28098 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
28099 IFN_ATOMIC_COMPARE_EXCHANGE.
28101 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28104 * opts.c (parse_sanitizer_options): Add missing question mark to
28105 "did you mean" message.
28107 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28109 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
28111 (VMULEUH_UNS): Likewise.
28112 (VMULOUB_UNS): Likewise.
28113 (VMULOUH_UNS): Likewise.
28114 * config/rs6000/rs6000.c (builtin_function_type): Remove
28115 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
28117 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28120 * read-rtl-function.c (function_reader::read_rtx_operand): Update
28121 x with result of extra_parsing_for_operand_code_0.
28122 (function_reader::extra_parsing_for_operand_code_0): Convert
28123 return type from void to rtx, returning x. When reading
28124 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
28125 larger size containing struct block_symbol.
28127 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
28129 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
28130 -mfloat128-hardware without -m64.
28132 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
28135 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
28136 entries to the case statement that marks unsigned arguments to
28137 overloaded functions.
28139 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
28141 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
28142 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
28144 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
28147 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
28148 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
28150 2017-03-10 Martin Liska <mliska@suse.cz>
28154 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
28155 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
28158 2017-03-10 Olivier Hainque <hainque@adacore.com>
28160 * tree-switch-conversion (array_value_type): Start by resetting
28161 candidate type to it's main variant.
28163 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28165 PR rtl-optimization/79909
28166 * combine.c (try_combine): Use simplify_replace_rtx on individual
28167 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
28168 of the whole CALL_INSN_FUNCTION_USAGE.
28170 PR tree-optimization/79972
28171 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
28172 get_range_info on SSA_NAMEs. Formatting fixes.
28174 2017-03-10 Richard Biener <rguenther@suse.de>
28175 Jakub Jelinek <jakub@redhat.com>
28177 PR tree-optimization/77975
28178 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
28179 edge to be constant.
28180 (get_val_for): For constant x return it. Formatting fix.
28181 (loop_niter_by_eval): Avoid pointless looping if the next iteration
28182 would use the same bases as the current one.
28184 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28186 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
28187 instead of vec_select for V1TImode.
28188 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
28190 (VSX_LE_128): Add V1TI to this mode iterator.
28191 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
28192 (*vsx_le_perm_store_<mode>): Likewise.
28193 (pre-reload splitter for VSX stores): Likewise.
28194 (post-reload splitter for VSX stores): Likewise.
28195 (*vsx_xxpermdi2_le_<mode>): Likewise.
28196 (*vsx_lxvd2x2_le_<mode>): Likewise.
28197 (*vsx_stxvd2x2_le_<mode>): Likewise.
28199 2017-03-09 Michael Eager <eager@eagercon.com>
28201 Correct failures with --enable-checking=yes,rtl.
28203 * config/microblaze/microblaze.c (microblaze_expand_shift):
28204 Replace GET_CODE test with CONST_INT_P and INTVAL test with
28205 test for const0_rtx.
28206 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
28207 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
28209 2017-03-09 Richard Biener <rguenther@suse.de>
28211 PR tree-optimization/79977
28212 * graphite-scop-detection.c (scop_detection::merge_sese):
28213 Handle the case of extra exits to blocks dominating the entry.
28215 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
28217 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28220 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
28222 PR rtl-optimization/79949
28223 * lra-constraints.c (process_alt_operands): Check memory when
28224 trying to predict a cycle. Print about the overall increase.
28226 2017-03-09 Richard Biener <rguenther@suse.de>
28228 PR middle-end/79971
28229 * gimple-expr.c (useless_type_conversion_p): Preserve
28230 TYPE_SATURATING for fixed-point types.
28232 2017-03-09 Richard Biener <rguenther@suse.de>
28235 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
28236 alignment of BLKmode params.
28238 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28241 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
28242 (VALL_NO_V2Q): Likewise.
28244 * config/aarch64/aarch64-simd.md
28245 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
28247 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
28248 VALL_NO_V2Q mode iterator.
28249 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
28251 2017-03-09 Martin Liska <mliska@suse.cz>
28253 PR tree-optimization/79631
28254 * tree-chkp-opt.c (chkp_is_constant_addr): Call
28255 tree_int_cst_sign_bit just for INTEGER constants.
28257 2017-03-09 Martin Liska <mliska@suse.cz>
28261 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
28264 2017-03-09 Marek Polacek <polacek@redhat.com>
28267 * tree.c (inchash::add_expr): Handle TREE_VEC.
28269 2017-03-09 Martin Liska <mliska@suse.cz>
28272 (chkp_narrow_size_and_offset): New function.
28273 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
28274 (void chkp_parse_bit_field_ref): New function.
28275 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
28276 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
28278 2017-03-09 Martin Liska <mliska@suse.cz>
28281 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
28282 (chkp_find_bounds_1): Remove gcc_unreachable.
28284 2017-03-09 Jakub Jelinek <jakub@redhat.com>
28287 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
28288 BUILT_IN_SYNC*, determine the access type from the size suffix and
28289 always build a MEM_REF with that type. Handle forgotten
28290 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
28293 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
28294 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
28295 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
28296 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
28297 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
28298 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
28299 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
28300 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
28301 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
28302 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
28303 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
28304 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
28305 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
28306 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
28307 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
28308 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
28309 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
28310 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
28311 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
28312 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
28313 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
28314 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
28315 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
28316 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
28317 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
28318 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
28319 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
28320 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
28321 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
28322 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
28323 definitions outside of __OPTIMIZE__ guarded section.
28326 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
28327 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
28328 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
28329 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
28332 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28334 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
28335 ("vfenez<mode>"): Add missing constraints.
28337 2017-03-08 Martin Sebor <msebor@redhat.com>
28340 * config/nds32/nds32.c (nds32_option_override):
28341 Fix misspelled diagnostic.
28343 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28346 * gimplify.c (gimplify_omp_for): Replace index var in outer
28347 taskloop statement with an artificial variable and add
28348 OMP_CLAUSE_PRIVATE clause for it.
28350 2017-03-08 Richard Biener <rguenther@suse.de>
28352 PR tree-optimization/79955
28353 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
28354 for accesses that are completely outside of the variable.
28356 2017-03-08 Andrew Haley <aph@redhat.com>
28358 PR tree-optimization/79943
28359 * tree-ssa-loop-split.c (compute_new_first_bound): When
28360 calculating the new upper bound, (END-BEG) should be added, not
28363 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28365 * config/avr/avr.md (setmemhi): Make sure match_dup
28366 operand number comes before match_scratch.
28368 2017-03-08 Richard Biener <rguenther@suse.de>
28370 PR tree-optimization/79920
28371 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
28372 with ncopies == 1 to ...
28373 (vect_transform_slp_perm_load): ... here. Properly compute
28374 all element loads by iterating VF times over the group. Do
28375 not handle ncopies (computed in a broken way) in
28376 vect_create_mask_and_perm.
28378 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28381 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
28382 is a uniform vector, use uniform_vector_p return value instead of
28383 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
28385 2017-03-07 Marek Polacek <polacek@redhat.com>
28387 PR middle-end/79809
28388 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
28389 (alloca_call_type): Likewise.
28391 2017-03-07 Martin Liska <mliska@suse.cz>
28393 * gcov.c (process_args): Put comment to correct location.
28395 2017-03-07 Martin Liska <mliska@suse.cz>
28397 PR middle-end/68270
28398 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
28399 Use array_at_struct_end_p instead of DECL_CHAIN (field).
28400 (chkp_narrow_bounds_for_field): Likewise.
28401 (chkp_parse_array_and_component_ref): Pass one more argument to
28404 2017-03-07 Richard Biener <rguenther@suse.de>
28406 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
28409 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
28411 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
28412 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
28414 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28417 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
28418 to end of description.
28419 (PARAM_MAX_STORES_TO_MERGE): Likewise.
28421 2017-03-07 Jakub Jelinek <jakub@redhat.com>
28423 PR rtl-optimization/79901
28424 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
28426 (*avx512f_<code><mode>3<mask_name>): ... this.
28427 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
28428 iterator instead of VI8_AVX2_AVX512BW.
28430 PR rtl-optimization/79901
28431 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
28432 min/max expander, expand it using expand_vec_cond_expr.
28435 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
28438 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28441 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
28442 to void * for PCH reasons.
28443 * dwarf2out.c (output_loc_operands, output_die): Cast
28444 v.val_vec.array to unsigned char *.
28446 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
28449 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
28452 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
28454 PR rtl-optimization/79571
28455 * lra-constraints.c (process_alt_operands): Calculate static
28456 reject and subtract it from overall when only addresses will be
28459 2017-03-06 Julia Koval <julia.koval@intel.com>
28462 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
28463 incoming stack boundary to 128 for 64-bit targets.
28465 2017-03-06 Richard Biener <rguenther@suse.de>
28467 PR tree-optimization/79894
28468 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
28469 to NULL after folding it.
28471 2017-03-06 Richard Biener <rguenther@suse.de>
28473 PR tree-optimization/79824
28474 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
28475 check disabling peeling for gaps.
28477 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
28479 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28480 attributes): Document gettimeofday.
28482 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28484 * config/s390/s390.c (s390_option_override_internal): Set
28485 PARAM_MIN_VECT_LOOP_BOUND
28487 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28489 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
28490 * config/s390/s390.md: Likewise.
28492 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28495 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
28496 (<avx2_avx512>_perm<mode>): Rename to ...
28497 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
28499 (<avx512>_perm<mode>_mask): Rename to ...
28500 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
28502 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
28503 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
28504 instead of VI8F_256_512.
28505 (avx512f_perm<mode>): New define_expand.
28506 (avx512f_perm<mode>_mask): Likewise.
28507 (avx512f_perm<mode>_1<mask_name>): New define_insn.
28508 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
28510 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28512 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
28513 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
28515 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
28517 2017-03-06 Martin Liska <mliska@suse.cz>
28520 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
28521 when having a SSA NAME w/o VAR_DECL assigned to it.
28523 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28525 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
28526 msa_dpsub_<su>_d): Fix MODE for vec_select.
28528 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28530 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
28532 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
28534 2017-03-06 Richard Biener <rguenther@suse.de>
28536 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
28537 * plugin.c (register_plugin_info): Likewise.
28538 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
28540 2017-03-05 Jakub Jelinek <jakub@redhat.com>
28542 * config/i386/sse.md (sse_storehps, sse_storelps,
28543 avx_<castmode><avxsizesuffix>_<castmode>,
28544 avx512f_<castmode><avxsizesuffix>_<castmode>,
28545 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
28546 in condition that at least one operand is not a MEM.
28548 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28550 PR middle-end/79805
28551 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
28552 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
28554 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
28555 gimple_call_nothrow_p flag based on whether original builtin can throw.
28556 If it can, emit following stmts on the fallthrough edge.
28557 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
28558 don't create new bb if inserting just debug stmts on the edge, try to
28559 insert them on the fallthru bb or just reset debug stmts.
28561 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
28564 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
28565 restore recog_data (including the operand rtxes inside it) around
28566 the call to get_insn_template.
28568 2017-03-03 Martin Sebor <msebor@redhat.com>
28570 PR tree-optimization/79699
28571 * context.c (context::~context): Free MPFR caches to avoid
28572 a memory leak on program exit.
28574 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28576 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
28577 Use wide_int::ulow () instead of .elt (0).
28579 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
28581 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
28582 (*pushxf): Limit oF constraint to 32bit targets and add oC
28583 constraint for 64bit targets.
28584 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
28585 (*pushdf): Change rmF constraint to rmC.
28587 2017-03-03 Martin Liska <mliska@suse.cz>
28589 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
28590 Remove unused variable.
28592 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28595 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
28596 is a memory operand, increase num_memory.
28597 (ix86_expand_args_builtin): Likewise.
28599 2017-03-03 Jan Hubicka <jh@suse.cz>
28602 * ipa-devirt.c (maybe_record_node): Properly handle
28603 __cxa_pure_virtual visibility.
28605 2017-03-03 Martin Liska <mliska@suse.cz>
28607 PR tree-optimization/79803
28608 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
28610 (pass_loop_prefetch::execute): Disabled optimization if an
28611 assumption about L1 cache size is not met.
28613 2017-03-03 Martin Liska <mliska@suse.cz>
28615 PR rtl-optimization/79574
28616 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
28617 (hash_scan_set): Likewise.
28618 (dump_hash_table): Likewise.
28619 (hoist_code): Likewise.
28621 2017-03-03 Richard Biener <rguenther@suse.de>
28623 * fixed-value.c (fixed_from_string): Restore use of elt (1)
28624 in place of uhigh ().
28625 (fixed_convert_from_real): Likewise.
28627 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
28630 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
28632 2017-03-03 Richard Biener <rguenther@suse.de>
28634 PR middle-end/79818
28635 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
28636 TYPE_OVERFLOW_UNDEFINED check.
28638 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28640 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
28642 (vector_ae_<mode>_p): Likewise.
28643 (vector_nez_<mode>_p): Likewise.
28644 (vector_ne_v2di_p): Likewise.
28645 (vector_ae_v2di_p): Likewise.
28646 (vector_ne_<mode>_p): Likewise.
28647 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
28649 (vsx_tsqrt<mode>2_fe): Likewise.
28651 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
28654 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
28656 2017-03-02 Jakub Jelinek <jakub@redhat.com>
28658 PR rtl-optimization/79780
28659 * cprop.c (one_cprop_pass): When second and further conditional trap
28660 in a single basic block is turned into an unconditional trap, turn it
28661 into a deleted note to avoid RTL verification failures.
28663 2017-03-02 Richard Biener <rguenther@suse.de>
28665 * fold-const.c (const_binop): Use ulow () instead of elt (0).
28667 2017-03-02 Richard Biener <rguenther@suse.de>
28669 PR tree-optimization/79345
28671 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
28672 param and abort the walk, returning -1 if it is hit.
28673 (walk_aliased_vdefs): Take a limit param and pass it on.
28674 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
28675 defaulting to 0 and return a signed int.
28676 * tree-ssa-uninit.c (struct check_defs_data): New struct.
28677 (check_defs): New helper.
28678 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
28679 about uninitialized memory.
28680 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
28681 bogus uninitialized warning.
28682 (fixed_convert_from_real): Likewise.
28684 2017-03-02 Bin Cheng <bin.cheng@arm.com>
28686 PR tree-optimization/66768
28687 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
28688 iv_use if base object can't be determined.
28690 2017-03-02 Jakub Jelinek <jakub@redhat.com>
28692 PR tree-optimization/79345
28693 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
28694 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
28695 (get_pattern_stats): Initialize it.
28696 * genemit.c (gen_expand): Verify match_scratch numbers come after
28697 match_operand/match_dup numbers.
28698 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
28699 match_scratch numbers.
28700 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
28702 * config/s390/s390.md (trunctdsd2): Likewise.
28704 2017-03-02 Richard Biener <rguenther@suse.de>
28706 * wide-int.h (wide_int_storage::operator=): Implement in terms
28709 2017-03-02 Richard Biener <rguenther@suse.de>
28711 PR tree-optimization/79777
28712 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
28713 the to insert expression to sth existing.
28715 2017-03-01 Martin Sebor <msebor@redhat.com>
28717 PR middle-end/79692
28718 * gimple-ssa-sprintf.c
28719 (directive::known_width_and_precision): New function.
28720 (format_integer): Use it.
28721 (get_mpfr_format_length): Consider the full range of precision
28722 when computing %g output with the # flag. Set the likely byte
28723 count to 3 rather than 1 when precision is indeterminate.
28724 (format_floating): Correct the lower bound of precision.
28726 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28728 * doc/invoke.texi: Document default code model for 64-bit Linux.
28730 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28733 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
28734 udiv rather than div since input pattern is unsigned.
28736 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
28738 * config/i386/i386.c (print_reg): Warn for values of
28739 unsupported size in integer register.
28741 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
28744 * config/rs6000/predicates.md (current_file_function_operand): Do
28745 not allow self calls to be local if the function is replaceable.
28747 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
28750 * config/rs6000/altivec.h (vec_ctz and others): Change the
28751 preprocessor macro that controls conditional compilation from
28752 _ARCH_PWR9 to __POWER9_VECTOR__.
28753 (vec_all_ne): Change parameterization of __altivec_scalar_pred
28754 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
28755 control (instead of _ARCH_PWR9 control) so that template
28756 definition uses power9-specific function.
28757 (vec_any_eq): Likewise.
28758 (vec_all_ne): Change macro definition to use a power9-specific
28759 expansion under #ifdef __POWER9_VECTOR__ control (instead of
28760 _ARCH_PWR9 control).
28761 (vec_any_eq) Likewise.
28762 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
28763 expansion for CMPNEF to remove support for xvcmpnesp instruction.
28764 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
28765 support for xvcmpnedp instruction.
28766 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
28767 macro expansion so that Power9 implementation of vec_all_ne does
28768 not use the AltiVec predicate framework.
28769 (VCMPNEH_P): Likewise.
28770 (VCMPNEW_P): Likewise.
28771 (VCMPNED_P): Likewise.
28772 (VCMPNEFP_P): Likewise.
28773 (VCMPNEDP_P): Likewise.
28774 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
28775 implementation of vec_any_eq to not use AltiVec predicate
28777 (VCMPAEH_P): Likewise.
28778 (VCMPAEW_P): Likewise.
28779 (VCMPAED_P): Likewise.
28780 (VCMPAEFP_P): Likewise.
28781 (VCMPAEDP_P): Likewise.
28782 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
28783 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
28784 not use the AltiVec predicate framework.
28785 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
28786 of vec_any_eq to not use AltiVec predicate framework.
28787 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
28788 support for predefined __POWER9_VECTOR__ macro to indicate that
28789 Power9 instruction selection is enabled.
28790 (altivec_overloaded_builtins): Remove extraneous
28791 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
28792 function argument types RS6000_BTI_bool_V16QI and
28793 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
28794 entry for overloaded function argument types RS6000_BTI_bool_V4SI
28795 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
28796 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
28797 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
28798 Power9 for implementations of vec_cmpne. Change the signature for
28799 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
28800 (representing vec_all_ne) to remove the previously described first
28801 argument of type RS6000_BTI_INTSI, as this was an artifact of
28802 reliance on the AltiVec predicate framework, which is no longer
28803 used in the implementation of these functions. Add
28804 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
28805 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
28806 since, unlike the AltiVec predicate framework implementation, we
28807 do not share function descriptors between vec_alle and vec_anyeq.
28808 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
28809 set of modes that receive special treatment even when
28810 TARGET_P9_VECTOR is true. The special treatment emits code that
28811 does not depend on Power9 instructions.
28812 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
28813 define_expand to not rely on AltiVec predicate framework.
28814 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
28816 (vector_ne_v2di_p): Change this define_expand to not rely on
28817 AltiVec predicate framework.
28818 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
28820 (vector_ne_<mode>_p): Change this define_expand to not rely on
28821 AltiVec predicate framework.
28822 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
28824 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
28825 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
28826 define_insn pattern.
28827 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
28828 define_insn pattern because the xvcmpne<VSs>. instruction is not
28830 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
28831 instruction is not supported.
28833 2017-03-01 Jakub Jelinek <jakub@redhat.com>
28835 * config/nvptx/nvptx.c: Include intl.h.
28837 2017-03-01 Martin Jambor <mjambor@suse.cz>
28840 * ipa-prop.h (ipa_bits): Removed field known.
28841 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
28842 to pointers. Adjusted their comments to warn about their sharing.
28843 (ipcp_transformation_summary): Change bits to a vector of pointers.
28844 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
28845 (ipa_get_ipa_bits_for_value): Declare.
28846 * tree-vrp.h (value_range): Mark as GTY((for_user)).
28847 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
28848 (ipa_bits_hash_table): Likewise.
28849 (ipa_vr_ggc_hash_traits): Likewise.
28850 (ipa_vr_hash_table): Likewise.
28851 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
28852 being pointers and vr_known being removed.
28853 (ipa_set_jf_unknown): Likewise.
28854 (ipa_get_ipa_bits_for_value): New function.
28855 (ipa_set_jfunc_bits): Likewise.
28856 (ipa_get_value_range): New overloaded functions.
28857 (ipa_set_jfunc_vr): Likewise.
28858 (ipa_compute_jump_functions_for_edge): Use the above functions to
28859 construct bits and vr parts of jump functions.
28860 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
28861 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
28863 (ipcp_grow_transformations_if_necessary): Also allocate
28864 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
28866 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
28867 them. Fix too long lines.
28868 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
28869 vr_known being removed.
28870 (ipa_read_jump_function): Use new setter functions to construct bits
28871 and vr parts of jump functions or set them to NULL.
28872 (write_ipcp_transformation_info): Adjust for bits being pointers.
28873 (read_ipcp_transformation_info): Likewise.
28874 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
28876 Include gt-ipa-prop.h.
28877 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
28879 (ipcp_store_bits_results): Likewise.
28880 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
28881 Do not write to existing jump functions but use a temporary instead.
28883 2017-03-01 Jakub Jelinek <jakub@redhat.com>
28886 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
28887 attempt to use its first operand as BIT_FIELD_REF base.
28889 2017-03-01 Richard Biener <rguenther@suse.de>
28891 PR middle-end/79721
28892 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
28893 interpolating formula in wrapping arithmetic.
28894 (chrec_apply): Convert chrec_evaluate return value to wanted type.
28896 2017-03-01 Jakub Jelinek <jakub@redhat.com>
28898 PR tree-optimization/79734
28899 * tree-vect-generic.c (expand_vector_condition): Optimize
28900 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
28901 Handle VEC_COND_EXPR where comparison has different inner width from
28902 type's inner width.
28904 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
28906 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
28907 markup, and similar issues. Remove @opindex entries for things
28908 that aren't options. Add missing -mmpy-option entries.
28910 2017-02-28 Jakub Jelinek <jakub@redhat.com>
28912 PR tree-optimization/79737
28913 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
28914 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
28915 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
28916 instead of byte_size. Formatting fix.
28917 (shift_bytes_in_array_right): Formatting fix.
28919 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
28922 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
28923 condition on optimize for the leaf function test.
28925 2017-02-28 Martin Liska <mliska@suse.cz>
28928 * read-rtl-function.c (function_reader::handle_unknown_directive):
28929 Bail out when one uses -flto.
28931 2017-02-28 Martin Liska <mliska@suse.cz>
28933 * common.opt: Replace space with tabular for options of <number>
28935 * config/i386/i386.opt: Show <number> value for
28936 -mlarge-data-threshold.
28937 * opts.c (print_filtered_help): Do not display number in hexadecimal
28940 2017-02-28 Martin Liska <mliska@suse.cz>
28942 * common.opt: Fix --help=option -Q for options which are of
28945 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
28947 * config/i386/i386.c (print_reg): Error out for values
28948 of 8-bit size in invalid integer register.
28950 2017-02-28 Martin Sebor <msebor@redhat.com>
28952 PR tree-optimization/79691
28953 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
28955 2017-02-28 Jakub Jelinek <jakub@redhat.com>
28958 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
28959 gcc_unreachable with output_operand_lossage.
28961 2017-02-28 Richard Biener <rguenther@suse.de>
28963 PR tree-optimization/79740
28964 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
28966 (visit_nary_op): Insert the nary into the hashtable if we
28967 pattern-matched sth.
28968 * tree-ssa-pre.c (eliminate_insert): Robustify.
28970 2017-02-28 Richard Biener <rguenther@suse.de>
28972 PR middle-end/79731
28973 * fold-const.c (decode_field_reference): Reject out-of-bound
28976 2017-02-28 Jakub Jelinek <jakub@redhat.com>
28978 * config/i386/i386.c: Include intl.h.
28979 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
28980 instead of just cond ? "..." : "...".
28981 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
28982 * coverage.c (read_counts_file): Likewise.
28983 * omp-offload.c: Include intl.h.
28984 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
28985 of just cond ? "..." : "...".
28986 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
28987 of just cond ? "..." : "...".
28989 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
28992 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
28994 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
28995 'tune for' CPU name.
28996 * config/arm/arm-cpu-data.h: Regenerated.
28998 2017-02-28 Richard Biener <rguenther@suse.de>
29000 PR tree-optimization/79732
29001 * tree-inline.c (expand_call_inline): Do not shadow var.
29003 2017-02-28 Richard Biener <rguenther@suse.de>
29005 PR tree-optimization/79723
29006 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
29007 address-space properly.
29009 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
29011 * doc/optinfo.texi (Optimization groups): Fix option used for
29013 * doc/invoke.texi (-fopt-info): Document "omp".
29014 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
29015 (OPTGROUP_ALL): Add OPTGROUP_OMP.
29016 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
29017 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
29018 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
29020 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
29022 * dumpfile.c (optgroup_options): Instead of "openmp", associate
29023 OPTGROUP_OMP with "omp".
29025 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
29028 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
29029 for arithmetic shift of unsigned V2DI.
29031 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
29033 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
29034 arc/linux.h headers.
29035 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
29036 (LINK_SPEC): Likewise.
29037 (ARC_TLS_EXTRA_START_SPEC): Likewise.
29038 (EXTRA_SPECS): Likewise.
29039 (STARTFILE_SPEC): Likewise.
29040 (ENDFILE_SPEC): Likewise.
29041 (LIB_SPEC): Likewise.
29042 (TARGET_SDATA_DEFAULT): Likewise.
29043 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
29044 (MULTILIB_DEFAULTS): Likewise.
29045 (DWARF2_UNWIND_INFO): Likewise.
29046 * config/arc/big.h: New file.
29047 * config/arc/elf.h: Likewise.
29048 * config/arc/linux.h: Likewise.
29049 * config/arc/t-uClibc: Remove.
29051 2017-02-27 Bin Cheng <bin.cheng@arm.com>
29053 PR tree-optimization/77536
29054 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
29055 (tree_transform_and_unroll_loop): Use above function to compute the
29056 estimated niter of unrolled loop and use it when scaling profile.
29057 Also use count info rather than frequency if it's non-zero.
29058 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
29059 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
29060 (vect_transform_loop): Call above function.
29062 2017-02-27 Richard Biener <rguenther@suse.de>
29064 PR tree-optimization/45397
29065 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
29066 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
29067 (visit_nary_op): Add pattern matching for CSEing sign-changed
29068 or truncated operations with wider ones.
29070 2017-02-27 Richard Biener <rguenther@suse.de>
29072 PR tree-optimization/79690
29073 * tree-vect-stmts.c (vectorizable_store): Use vector type
29074 built from the DR with address-space.
29076 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
29078 * doc/invoke.texi (Optimize Options): Refine the description
29079 of asan-use-after-return.
29081 2017-02-25 Alan Modra <amodra@gmail.com>
29083 PR rtl-optimization/79584
29084 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
29085 base, not ad->base_term, the reg within base. Remove assertion
29086 that ad->base == ad->base_term. Replace gen_int_mode using
29087 bogus mode with const0_rtx.
29089 2017-02-25 Jakub Jelinek <jakub@redhat.com>
29091 PR middle-end/79396
29092 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
29093 FMA_EXPR like tcc_binary or tcc_unary.
29095 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
29098 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
29100 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
29101 (output_loc_operands): Handle DW_OP_call_ref and
29102 DW_OP_GNU_variable_value.
29103 (struct variable_value_struct): New type.
29104 (struct variable_value_hasher): Likewise.
29105 (variable_value_hash): New variable.
29106 (string_types): Remove.
29107 (copy_loc_descr): New function.
29108 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
29109 (prepend_loc_descr_to_each): New function.
29110 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
29111 instead of add_loc_descr_to_each if the first argument is single
29112 location list and the second has multiple.
29113 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
29114 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
29115 when looking for variable value which doesn't have other location info.
29116 (loc_list_from_tree): Formatting fix.
29117 (gen_array_type_die): Simplify DW_AT_string_length handling.
29118 (adjust_string_types): Remove.
29119 (gen_subprogram_die): Don't call adjust_string_types nor test/set
29120 string_types. Call resolve_variable_values.
29121 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
29122 (resolve_addr_in_expr): Likewise. Add A argument.
29123 (copy_deref_exprloc): Remove deref argument. Adjust for the
29124 original expression being DW_OP_GNU_variable_value with optionally
29125 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
29126 optionally after it.
29127 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
29128 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
29129 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
29130 (variable_value_hasher::hash, variable_value_hasher::equal): New
29132 (resolve_variable_value_in_expr, resolve_variable_value,
29133 resolve_variable_values, note_variable_value_in_expr,
29134 note_variable_value): New functions.
29135 (dwarf2out_early_finish): Call note_variable_value on all toplevel
29138 2017-02-24 Jakub Jelinek <jakub@redhat.com>
29141 * opts.h (handle_generated_option): Add GENERATED_P argument.
29142 * opts-common.c (handle_option): Adjust function comment.
29143 (handle_generated_option): Add GENERATED_P argument, pass it to
29145 (control_warning_option): Pass false to handle_generated_option
29147 * opts.c (maybe_default_option): Pass true to handle_generated_option
29149 * optc-gen.awk: Likewise.
29151 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29153 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
29154 a REG, look at the REG it is a SUBREG of.
29155 (splitter for cmpeqsi_t): Ditto.
29157 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29159 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
29160 the special USEs with the pattern of the insn, not the insn itself.
29162 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
29165 * doc/invoke.texi: Document -mload-store-pairs.
29167 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29168 Sandra Loosemore <sandra@codesourcery.com>
29170 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
29171 argument isn't a CONST_INT.
29172 (nios2_alternate_compare_const): Assert op is a CONST_INT.
29173 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
29174 (nios2_validate_compare): Bypass alternate compare logic if *op2
29175 is not a CONST_INT.
29176 (ldstwm_operation_p): Return false if first_base is not a REG or
29177 if first_offset is not a CONST_INT.
29179 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29181 * config/cris/cris.md: Use correct operand in a define_peephole2.
29183 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29185 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
29187 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29189 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
29190 this_insn if it is an INSN or JUMP_INSN.
29191 (force_offsettable): Look at base, not at addr.
29192 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
29193 on things that aren't necessarily CONST_INTs.
29195 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
29197 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
29198 -mfpmath=sse is the default also for x86-32 targets with SSE2
29199 instruction set when @option{-ffast-math} is enabled
29201 2017-02-24 Jeff Law <law@redhat.com>
29203 PR rtl-optimizatoin/79286
29204 * ira.c (update_equiv_regs): Drop may_trap_p exception to
29207 2017-02-24 Richard Biener <rguenther@suse.de>
29209 PR tree-optimization/79389
29210 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
29213 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
29215 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
29216 function comment to reflect reality.
29217 (loop_exits_before_overflow): Fix typo in function description.
29219 2017-02-24 Richard Biener <rguenther@suse.de>
29221 PR tree-optimization/79389
29222 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
29223 properly that a threading opportunity exists. Detect conditional
29224 copy/constant propagation opportunities.
29226 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
29228 * config/visium/visium.md (type): Add trap.
29229 (b): New mode attribute.
29230 (*btst): Rename into...
29231 (*btst<mode>): ...this and adjust.
29232 (*cbranchsi4_btst_insn): Rename into...
29233 (*cbranch<mode>4_btst_insn): ...this and adjust.
29234 (trap): New define_insn.
29236 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29238 PR tree-optimization/79389
29239 * ifcvt.c (struct noce_if_info): Add rev_cond field.
29240 (noce_reversed_cond_code): New function.
29241 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
29242 reversed_comparison_code. Formatting fix.
29243 (noce_try_store_flag): Test rev_cond != NULL in addition to
29244 reversed_comparison_code.
29245 (noce_try_store_flag_constants): Likewise.
29246 (noce_try_store_flag_mask): Likewise.
29247 (noce_try_addcc): Use rev_cond if non-NULL instead of
29248 reversed_comparison_code.
29249 (noce_try_cmove_arith): Likewise. Formatting fixes.
29250 (noce_try_minmax, noce_try_abs): Clear rev_cond.
29251 (noce_find_if_block): Initialize rev_cond.
29252 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
29253 instead of false as last argument never attempt to reverse it
29256 2017-02-23 Bin Cheng <bin.cheng@arm.com>
29258 PR tree-optimization/79663
29259 * tree-predcom.c (combine_chains): Process refs in reverse order
29260 only for ZERO length chains, and add explaining comment.
29262 2017-02-23 Jeff Law <law@redhat.com>
29264 PR tree-optimization/79578
29265 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
29266 in call to operand_equal_p.
29268 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
29271 * config/i386/cpuid.h: Fix another undefined behavior.
29273 2017-02-23 Richard Biener <rguenther@suse.de>
29275 PR tree-optimization/79683
29276 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
29277 vector types for data-refs.
29279 2017-02-23 Martin Liska <mliska@suse.cz>
29281 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
29283 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29285 PR middle-end/79665
29286 * internal-fn.c (get_range_pos_neg): Moved to ...
29287 * tree.c (get_range_pos_neg): ... here. No longer static.
29288 * tree.h (get_range_pos_neg): New prototype.
29289 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
29290 are known to be in between 0 and signed maximum inclusive, try to
29291 expand both unsigned and signed divmod and use the cheaper one from
29294 2017-02-22 Jeff Law <law@redhat.com>
29296 PR tree-optimization/79578
29297 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
29298 to compare base operands.
29300 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
29303 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
29304 gpc_reg_operand instead of fpr_reg_operand.
29306 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
29308 * config/mips/mips.c (mips_return_in_memory): Force FP
29309 vector types to be returned in memory for o32 ABI.
29311 2017-02-22 Jakub Jelinek <jakub@redhat.com>
29313 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
29314 instead of DW_TAG_member for static data member declarations and don't
29315 set no_linkage_name for static inline data members.
29316 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
29319 2017-02-22 Martin Liska <mliska@suse.cz>
29321 * doc/invoke.texi: Replace inequality signs with square brackets
29324 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
29327 * lra-constraints.c (simplify_operand_subreg): Handle
29328 WORD_REGISTER_OPERATIONS targets.
29330 2017-02-22 Jakub Jelinek <jakub@redhat.com>
29333 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
29334 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
29335 elimination by swapping fld*.
29337 2017-02-22 Richard Biener <rguenther@suse.de>
29339 PR tree-optimization/79673
29340 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
29341 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
29342 irrelevant address-space qualifiers and avoiding a
29343 ADDR_SPACE_CONVERT_EXPR from fold_convert.
29345 2017-02-22 Richard Biener <rguenther@suse.de>
29347 PR tree-optimization/79666
29348 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
29349 to not symbolically negate if that may introduce undefined
29352 2017-02-22 Martin Liska <mliska@suse.cz>
29355 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
29356 * data-streamer-out.c (streamer_write_gcov_count_stream):
29358 * value-prof.c (stream_out_histogram_value): Make assert more
29359 precise based on type of counter.
29361 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
29364 * config/i386/i386.md (standard_x87sse_constant_load splitter):
29365 Use nonimmediate_operand instead of memory_operand for operand 1.
29366 (float-extend standard_x87sse_constant_load splitter): Ditto.
29368 2017-02-21 Jeff Law <law@redhat.com>
29370 PR tree-optimization/79621
29371 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
29372 blocks with edges to themselves.
29374 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29377 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
29378 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
29379 Use gimple_call_builtin_p.
29382 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
29383 on temporarily removed DEBUG_INSNs.
29385 PR tree-optimization/79649
29386 * tree-loop-distribution.c (classify_partition): Give up on
29387 non-generic address space loads/stores.
29389 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
29391 * doc/loop.texi (Loop manipulation): Remove nonexistent
29392 tree_ssa_loop_version from the documentation.
29393 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
29395 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29398 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
29399 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
29400 * config/rs6000/rs6000.c: Include except.h.
29401 (rs6000_expand_split_stack_prologue): Call
29402 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
29404 2017-02-21 Martin Jambor <mjambor@suse.cz>
29407 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
29408 have been analyzed.
29410 2017-02-21 Martin Jambor <mjambor@suse.cz>
29412 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
29413 for backward compatibility only.
29414 * doc/invoke.texi (Option Summary): Remove all references to
29415 -fipa-cp-alignment.
29417 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
29421 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29423 * lra-constraints.c (curr_insn_transform): Handle
29424 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29426 2017-02-21 Martin Liska <mliska@suse.cz>
29428 * config/i386/i386.opt: Replace -masm-dialect with -masm.
29430 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
29432 PR translation/79638
29433 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
29435 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
29438 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
29439 (arm_function_ok_for_sibcall): Return false for an indirect call by
29440 descriptor if all the argument registers are used.
29441 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
29442 alignment of the function.
29444 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29446 PR tree-optimization/61441
29447 * simplify-rtx.c (simplify_const_unary_operation): For
29448 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
29449 the sNaN unmodified.
29451 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
29453 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29454 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29455 instead of SYSTEM_HEADER_DIR.
29457 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
29458 Martin Liška <mliska@suse.cz>
29460 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
29461 Fix typos and grammar, use active voice, and clarify.
29463 2017-02-20 Marek Polacek <polacek@redhat.com>
29465 PR middle-end/79537
29466 * gimplify.c (gimplify_expr): Handle unused *&&L;.
29469 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
29471 2017-02-20 Jakub Jelinek <jakub@redhat.com>
29474 * config/i386/i386.c (ix86_expand_builtin): Handle
29475 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
29476 ix86_builtins_isa[fcode].isa as a requirement of those
29477 flags and any other flag in the bitmask.
29478 (ix86_init_mmx_sse_builtins): Use 0 instead of
29479 ~OPTION_MASK_ISA_64BIT as mask.
29480 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
29481 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
29482 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
29483 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
29485 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29488 * lra-constraints.c (split_reg): Check requested split mode
29489 is supported by the register.
29491 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29493 * lra-constraints.c (simplify_operand_subreg): Remove early
29496 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29499 * lra-constraints.c (curr_insn_transform): Tighten condition
29500 for converting SUBREG reloads from OP_OUT to OP_INOUT.
29502 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29505 * lra-constraints.c (curr_insn_transform): Handle
29506 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29508 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
29511 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
29513 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
29515 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
29518 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
29521 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29523 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
29526 2017-02-18 Jakub Jelinek <jakub@redhat.com>
29528 * final.c (last_columnnum, override_columnnum): New variables.
29529 (final_start_function): Set last_columnnum, pass it to begin_prologue
29530 hook and pass 0 to dwarf2out_begin_prologue.
29531 (final_scan_insn): Update override_columnnum. Pass last_columnnum
29532 to source_line debug hook.
29533 (notice_source_line): Compute last_columnnum and for debug_column_info
29534 return true on column changes.
29535 * debug.h (struct gcc_debug_hooks): Add column argument to
29536 source_line and begin_prologue hooks.
29537 (debug_nothing_int_charstar_int_bool): Remove prototype.
29538 (debug_nothing_int_int_charstar,
29539 debug_nothing_int_int_charstar_int_bool): New prototypes.
29540 (dwarf2out_begin_prologue): Add column argument.
29541 * debug.c (do_nothing_debug_hooks): Adjust source_line and
29542 begin_prologue hooks.
29543 (debug_nothing_int_charstar_int_bool): Remove.
29544 (debug_nothing_int_int_charstar,
29545 debug_nothing_int_int_charstar_int_bool): New functions.
29546 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
29547 through to dwarf2out_source_line.
29548 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
29549 (dwarf2out_source_line): Add column argument, emit it if requested.
29550 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
29552 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29553 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29554 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
29555 through to dwarf2out_begin_prologue.
29556 (vmsdbgout_source_line): Add column argument, pass it through to
29557 dwarf2out_source_line.
29558 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
29559 dbxout_source_line caller.
29560 (dbxout_source_line): Add column argument.
29562 * common.opt (gno-column-info, gcolumn-info): New options.
29563 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
29564 (check_die): Also test for multiple DW_AT_decl_column attributes.
29565 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
29566 DW_AT_decl_column if requested.
29567 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
29569 (gen_variable_die): Likewise.
29570 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
29571 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
29574 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
29575 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
29576 (ix86_handle_option): Handle OPT_m3dnowa.
29577 * doc/invoke.texi (-m3dnowa): Document.
29578 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
29579 -m3dnowa instead of -m3dnow -march=athlon.
29582 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
29583 instead of gcc_assert for K, r and R code checks. Formatting fixes.
29585 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29588 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
29589 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
29590 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
29591 generator for vsx_xxpermdi_<mode>_be.
29592 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
29593 force big-endian semantics.
29594 (vsx_xxpermdi_<mode>_be): New define_expand with same
29595 implementation as previous version of vsx_xxpermdi_<mode>.
29597 2017-02-17 Jakub Jelinek <jakub@redhat.com>
29599 PR tree-optimization/79327
29600 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
29601 variable, its initialization and use.
29603 2017-02-17 Julia Koval <julia.koval@intel.com>
29605 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
29606 (OPTION_MASK_ISA_PKU_UNSET): New.
29607 (ix86_handle_option): Handle -mrdpid.
29608 * config/i386/cpuid.h (bit_RDPID): New.
29609 * config/i386/driver-i386.c (host_detect_local_cpu):
29610 Detect RDPID feature.
29611 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
29612 * config/i386/i386-c.c (ix86_target_macros_internal):
29614 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
29615 (ix86_valid_target_attribute_inner_p): Add "rdpid".
29616 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
29617 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
29618 * config/i386/i386.md (define_insn "rdpid"): New.
29619 * config/i386/i386.opt Add -mrdpid.
29620 * config/i386/immintrin.h (_rdpid_u32): New.
29622 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
29624 PR rtl-optimization/79541
29625 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
29626 instead of transforming it into USE.
29628 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
29630 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
29631 If HONOR_SNANS (SFmode) force the input to a register.
29632 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
29633 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
29634 an frsp or similar insn.
29636 2017-02-17 Martin Liska <mliska@suse.cz>
29638 PR rtl-optimization/79577
29639 * params.def (selsched-max-sched-times): Increase minimum to 1.
29641 2017-02-17 Martin Liska <mliska@suse.cz>
29643 PR rtl-optimization/79574
29644 * gcse.c (want_to_gcse_p): Prevent integer overflow.
29646 2017-02-17 Martin Liska <mliska@suse.cz>
29648 PR tree-optimization/79529
29649 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
29650 ssa_defined_default_def_p to handle cases which are implicitly
29652 * tree-ssa.c (ssa_defined_default_def_p): New function.
29653 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
29654 which are implicitly defined.
29655 * tree-ssa.h (ssa_defined_default_def_p): Declare.
29657 2017-02-17 Richard Biener <rguenther@suse.de>
29659 PR middle-end/79576
29660 * params.def (max-ssa-name-query-depth): Limit to 10.
29662 2017-02-17 Richard Biener <rguenther@suse.de>
29664 PR tree-optimization/79552
29665 * tree-ssa-structalias.c (visit_loadstore): Properly verify
29668 2017-02-17 Richard Biener <rguenther@suse.de>
29671 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
29673 2017-02-17 Marek Polacek <polacek@redhat.com>
29675 PR middle-end/79536
29676 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
29677 (fold_negate_expr): New wrapper.
29679 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
29681 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
29682 Correct terminology and de-emphasize pre-standard behavior.
29684 2017-02-16 Alan Modra <amodra@gmail.com>
29686 PR rtl-optimization/79286
29687 * ira.c (def_dominates_uses): New function.
29688 (update_equiv_regs): Don't create an equivalence for insns that
29689 may trap where the register def does not dominate the use.
29691 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
29693 PR rtl-optimization/78127
29694 * lra.c (lra): Call lra_eliminate before finish the loop after
29697 2017-02-16 Richard Biener <rguenther@suse.de>
29699 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
29701 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
29702 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
29703 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
29704 (isl_val_int_from_wi): New function.
29705 (extract_affine_gmp): Rename to ...
29706 (extract_affine_wi): ... this, take a widest_int.
29707 (extract_affine_int): Just wrap extract_affine_wi.
29708 (add_param_constraints): Use isl_val_int_from_wi.
29709 (add_loop_constraints): Likewise, and extract_affine_wi.
29711 2017-02-15 Jeff Law <law@redhat.com>
29713 PR middle-end/79521
29714 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
29715 ira_init_register_move_cost_if_necessary.
29717 2017-02-15 Martin Sebor <msebor@redhat.com>
29719 PR middle-end/32003
29720 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
29721 removed in a prior commit.
29723 2017-02-15 Bin Cheng <bin.cheng@arm.com>
29725 PR tree-optimization/79347
29726 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
29727 counters during peeling.
29729 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
29731 * Makefile.in (site.exp): Remove "set ISLVER".
29733 2017-02-15 Jakub Jelinek <jakub@redhat.com>
29736 * real.c (real_from_integer): Call real_convert even for decimal.
29738 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
29741 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
29743 2017-02-14 Andrew Pinski <apinski@cavium.com>
29745 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
29746 cores and change the partno/implementer to be correct.
29747 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
29748 the 'B" as the implementer.
29749 * config/aarch64/aarch64-tune.md: Regenerate.
29751 2017-02-14 Carl Love <cel@us.ibm.com>
29753 * config/rs6000/rs6000.c: Add case statement entry to make the
29754 xvcvuxdsp built-in argument unsigned.
29755 * config/rs6000/vsx.md: Fix the source and return operand types so they
29756 match the instruction definitions from the ISA document. Fix typo
29757 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
29760 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
29763 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
29764 member early_clobber_alts.
29765 * lra-lives.c (reg_early_clobber_p): New.
29766 (process_bb_lives): Use it.
29767 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
29768 (debug_operand_data): Initialize early_clobber_alts.
29769 (setup_operand_alternative): Set up early_clobber_alts.
29770 (collect_non_operand_hard_regs): Ditto. Pass early clobber
29771 alternatives to new_insn_reg.
29772 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
29774 (lra_update_insn_regno_info): Pass the new arg.
29776 2017-02-14 Jakub Jelinek <jakub@redhat.com>
29778 PR middle-end/79505
29779 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
29780 (new_oacc_loop_raw): Don't clear already cleared fields.
29783 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
29784 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
29785 _mm512_prefetch_i64gather_ps): New inline functions and macros.
29787 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
29790 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
29792 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
29795 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
29796 the extra instruction to the right place to store 128-bit constant
29799 2017-02-14 Martin Sebor <msebor@redhat.com>
29801 PR middle-end/79448
29802 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
29803 warning for strings of unknown length.
29805 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
29807 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
29809 2017-02-14 Jeff Law <law@redhat.com>
29812 * ira-costs.c (scan_one_insn): Initialize register move costs
29813 for pseudos seen in USE/CLOBBER insns.
29815 PR tree-optimization/79095
29816 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
29817 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
29818 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
29819 if the operands are known to be not equal, then the resulting range
29821 (intersect_ranges): If the new range is ~[0,0] and the old range is
29822 wide, then prefer ~[0,0].
29823 * tree-vrp.c (overflow_comparison_p_1): New function.
29824 (overflow_comparison_p): New function.
29825 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
29826 if NAME is used in an overflow test.
29827 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
29828 overflow check that can be expressed as an equality test, then adjust
29829 ops to be that equality test.
29831 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29833 * config/s390/s390-builtin-types.def: Remove flags argument.
29834 * config/s390/s390.c (s390_init_builtins): Likewise.
29836 2017-02-14 Martin Liska <mliska@suse.cz>
29838 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
29839 vector. Fix trailing white spaces.
29841 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
29843 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
29846 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29848 PR rtl-optimization/68664
29849 * config/arm/arm.c (arm_sched_can_speculate_insn):
29850 New function. Declare prototype.
29851 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
29853 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29855 PR rtl-optimization/68664
29856 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
29858 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
29860 2017-02-14 Amit Pawar <amit.pawar@amd.com>
29862 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
29863 max skip bytes for function, loop and jump.
29865 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29867 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
29868 ABS_EXPR for gimple dump.
29870 2017-02-14 Jakub Jelinek <jakub@redhat.com>
29873 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
29875 PR tree-optimization/79408
29876 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
29877 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
29878 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
29879 also if rhs1 is INTEGER_CST.
29881 2017-02-14 Richard Biener <rguenther@suse.de>
29883 PR middle-end/79432
29884 * tree-into-ssa.c (insert_phi_nodes): When the function can
29885 have abnormal edges rewrite SSA names with broken use-def
29886 dominance out of SSA and register them for PHI insertion.
29888 2017-02-13 Martin Sebor <msebor@redhat.com>
29890 PR middle-end/79496
29891 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
29892 clearing info.nowrite flag when snprintf size argument is a range.
29894 2017-02-13 Jakub Jelinek <jakub@redhat.com>
29896 * cprop.c (cprop_jump): Add missing space in string literal.
29897 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
29898 (get_constraint_for_component_ref): Likewise.
29899 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
29900 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
29901 * lra-constraints.c (process_alt_operands): Likewise.
29902 * ipa-inline.c (inline_small_functions): Likewise.
29903 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
29904 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
29905 * trans-mem.c (diagnose_tm_1_op): Likewise.
29906 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
29907 (grid_parallel_clauses_gridifiable): Likewise.
29909 * config/nvptx/mkoffload.c (process): Add space in between
29912 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
29913 "MOD4_SSE_REGS" and "ALL_REGS".
29915 * spellcheck.c (test_data): Add , in between "foo" and "food".
29917 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29920 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
29921 boundary crossing check and subsequent code generation agree.
29923 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29925 * config/aarch64/aarch64.c (has_memory_op): Delete.
29926 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
29929 2017-02-13 Jakub Jelinek <jakub@redhat.com>
29931 PR rtl-optimization/79388
29932 PR rtl-optimization/79450
29933 * combine.c (distribute_notes): When removing TEM_INSN for which
29934 corresponding dest has last value recorded, invalidate that last
29937 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29939 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
29940 of explicit '@'. Add missing assembly comment marker on branch costs
29943 2017-02-13 Nathan Sidwell <nathan@acm.org>
29945 * gengtype-lex.l (<in_struct>): Add '/'.
29947 2017-02-13 Martin Liska <mliska@suse.cz>
29950 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
29952 2017-02-13 Richard Biener <rguenther@suse.de>
29954 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
29956 * configure: Re-generate.
29957 * config.in: Likewise.
29958 * graphite-dependences.c: Simplify as if
29959 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
29960 * graphite-isl-ast-to-gimple.c: Likewise.
29961 * graphite-optimize-isl.c: Likewise.
29962 * graphite-poly.c: Likewise.
29963 * graphite-sese-to-poly.c: Likewise.
29964 * graphite.h: Likewise.
29965 * toplev.c: Include isl/version.h and use isl_version () for
29966 printing the ISL version.
29967 * doc/install.texi: Update ISL requirement.
29969 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
29971 * doc/standards.texi (Standards): Update reference to
29974 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
29976 * doc/extend.texi (Named Address Spaces): sourceware.org now
29978 * doc/install.texi (Binaries): Ditto.
29981 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
29983 * doc/cpp.texi: Replace "stringify"/"stringification" with C
29984 standard terminology "stringize"/"stringizing" throughout.
29985 * doc/cppinternals.texi: Likewise.
29987 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
29989 * doc/extend.texi: Fix some spelling mistakes and typos.
29990 * doc/invoke.texi: Likewise.
29992 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
29995 * params.def (inline-min-speedup) Change from 10 to 8.
29997 2017-02-11 Jakub Jelinek <jakub@redhat.com>
29999 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
30002 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30005 * ipa-inline-analysis.c (get_minimal_bb): New function.
30006 (record_modified): Use it.
30007 (remap_edge_change_prob): Handle also ancestor functions.
30009 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
30011 * doc/contrib.texi (Contributors): Remove broken link into
30012 the Mauve CVS repository.
30014 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30016 PR middle-end/79454
30017 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
30018 result computation whenever lhs doesn't have vector mode, not
30019 just when it has BLKmode.
30021 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30023 * doc/makefile.texi (profiledbootstrap): Refer to the
30024 installation instructions only in textual form.
30026 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30029 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
30031 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30033 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
30034 (Specific): Update mingw-w64 reference.
30036 (Specific): Remove broken link to Renesas RX processor.
30038 2017-02-10 Richard Biener <rguenther@suse.de>
30040 * toplev.c (process_options): Do not mention obsolete graphite
30041 options when printing sorry message about missing graphite support.
30042 Mention -floop-nest-optimize.
30044 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
30046 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
30047 (vtst_p16): Likewise.
30048 (vtstq_p8): Likewise.
30049 (vtstq_p16): Likewise.
30051 (vtstq_p64): Likewise.
30052 * config/arm/arm_neon.h (vgetq_lane_p64): New.
30053 (vset_lane_p64): New.
30054 (vsetq_lane_p64): New.
30056 2017-02-10 Jakub Jelinek <jakub@redhat.com>
30058 PR tree-optimization/79411
30059 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
30060 stmt operands are SSA_NAMEs used in abnormal phis.
30061 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
30064 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30067 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
30070 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30072 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
30074 2017-02-09 Jakub Jelinek <jakub@redhat.com>
30076 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
30080 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
30081 not arbitrary TREE_CONSTANT.
30084 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
30085 "omp declare target link" attribute unless is_global_var.
30086 * omp-offload.c (find_link_var_op): Likewise.
30088 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
30089 Chung-Lin Tang <cltang@codesourcery.com>
30091 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
30093 (gimplify_adjust_omp_clauses): Don't delete TILE.
30094 (gimplify_omp_for): Deal with TILE.
30095 * internal-fn.c (expand_GOACC_TILE): New function.
30096 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
30098 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
30099 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
30101 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
30102 avoid DIV for outermost collapse var.
30103 (expand_oacc_for): Insert tile element loop as needed. Adjust.
30104 Remove out of date comments, fix whitespace.
30105 * omp-general.c (omp_extract_for_data): Deal with tiling.
30106 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
30107 adjust OLF_DIM_BASE value.
30108 (struct omp_for_data): Add tiling field.
30109 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
30110 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
30111 for auto loops. Remove default auto determining, moved to
30112 oacc_loop_fixed_partitions.
30113 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
30114 stmts, add e_mask field.
30115 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
30116 (oacc_thread_numbers): Use oacc_dim_call.
30117 (oacc_xform_tile): New.
30118 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
30119 (finish_oacc_loop): Adjust for ifns vector.
30120 (oacc_loop_discover_walk): Append loop abstraction sites to list,
30121 add case for GOACC_TILE fns.
30122 (oacc_loop_xform_loop): Delete.
30123 (oacc_loop_process): Iterate over call list directly, and add
30124 handling for GOACC_TILE fns.
30125 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
30127 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
30128 vector partitioning to outer loops. Assign 2 partitions to loops
30129 when available. Add TILE handling.
30130 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
30131 (execite_oacc_device_lower): Process GOACC_TILE fns,
30132 ignore unknown specs.
30133 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
30134 * tree.c (omp_clause_num_ops): Adjust TILE ops.
30135 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
30137 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
30139 * configure.ac (ACX_BUGURL): Update.
30140 * configure: Regenerate.
30142 2017-02-09 Richard Biener <rguenther@suse.de>
30144 PR tree-optimization/69823
30145 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
30146 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
30148 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30150 * config/arc/arc-c.def: Add __NPS400__ definition.
30151 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
30152 (TARGET_NPS400): Define.
30154 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30156 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
30158 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
30159 pointer, arch_info.
30160 (arc_cpu_types): Fill the arch_info field with a pointer into the
30161 arc_arch_types table.
30162 (arc_selected_cpu): Declare.
30163 * config/arc/arc.c (arc_selected_cpu): Make global.
30164 (arc_selected_arch): Delete.
30165 (arc_base_cpu): Delete.
30166 (arc_override_options): Remove references to deleted variables,
30167 update access to arch information.
30168 (ARC_OPT): Update access to arch information.
30169 (ARC_OPTX): Likewise.
30170 * config/arc/arc.h (arc_base_cpu): Remove declaration.
30171 (TARGET_ARC600): Update access to arch information.
30172 (TARGET_ARC601): Likewise.
30173 (TARGET_ARC700): Likewise.
30174 (TARGET_EM): Likewise.
30175 (TARGET_HS): Likewise.
30176 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
30179 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
30182 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
30183 condition/operands for integer GE/LE/GEU/LEU operations.
30185 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
30187 PR translation/79397
30188 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
30191 2017-02-08 Martin Jambor <mjambor@suse.cz>
30194 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
30195 whether allocation happened.
30196 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
30197 nothing was allocated.
30199 2017-02-08 Jakub Jelinek <jakub@redhat.com>
30201 PR tree-optimization/79408
30202 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
30203 constant, but SSA_NAME with a known integer range, use the minimum
30204 of that range instead of op1 to determine if modulo can be replaced
30205 with its first operand.
30207 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30209 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
30211 2017-02-08 Richard Biener <rguenther@suse.de>
30213 PR tree-optimization/71824
30214 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
30215 Check all loops contained in the merged region.
30217 2017-02-07 Andrew Pinski <apinski@cavium.com>
30219 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
30221 2017-02-07 Andrew Pinski <apinski@cavium.com>
30223 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
30224 (thunderxt88): Likewise.
30225 (thunderxt81): Disable LSE and change v8.1 to v8.
30226 (thunderxt83): Likewise.
30228 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30229 Richard Biener <rguenther@suse.de>
30231 PR middle-end/79399
30232 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
30233 type from int to size_t.
30234 * ira-costs.c (struct_costs_size): Change type from int to size_t.
30236 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30238 PR rtl-optimization/79386
30239 * cprop.c (bypass_conditional_jumps): Initialize
30240 bypass_last_basic_block already before splitting bbs after
30241 unconditional traps...
30242 (bypass_conditional_jumps): ... rather than here.
30245 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
30246 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
30247 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
30248 fix -masm=intel patterns.
30250 2017-02-07 Richard Biener <rguenther@suse.de>
30252 PR tree-optimization/79256
30253 PR middle-end/79278
30254 * builtins.c (get_object_alignment_2): Use min_align_of_type
30255 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
30256 and ADJUST_FIELD_ALIGN.
30258 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
30260 * doc/tm.texi: Regenerate.
30261 * stor-layout.c (layout_decl): Adjust.
30262 (update_alignment_for_field): Likewise.
30263 (place_field): Likewise.
30264 (min_align_of_type): Likewise.
30265 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
30266 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
30267 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
30268 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
30269 * config/frv/frv.c (frv_adjust_field_align): Likewise.
30270 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
30271 * config/i386/i386.c (x86_field_alignment): Likewise.
30272 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
30273 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
30274 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
30275 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
30276 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
30277 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
30281 2017-01-30 Richard Biener <rguenther@suse.de>
30283 PR tree-optimization/79256
30284 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
30285 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
30288 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
30290 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
30291 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
30292 builtins to SImode and emit a zero-extend, if necessary.
30294 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30296 * docs/invoke.texi (RISC-V Options): Alphabetize.
30298 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30300 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
30303 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30305 * config/riscv/riscv.c: New file.
30306 * common/config/riscv/riscv-common.c: Likewise.
30307 * config.gcc: Likewise.
30308 * config/riscv/constraints.md: Likewise.
30309 * config/riscv/elf.h: Likewise.
30310 * config/riscv/generic.md: Likewise.
30311 * config/riscv/linux.h: Likewise.
30312 * config/riscv/multilib-generator: Likewise.
30313 * config/riscv/peephole.md: Likewise.
30314 * config/riscv/pic.md: Likewise.
30315 * config/riscv/predicates.md: Likewise.
30316 * config/riscv/riscv-builtins.c: Likewise.
30317 * config/riscv/riscv-c.c: Likewise.
30318 * config/riscv/riscv-ftypes.def: Likewise.
30319 * config/riscv/riscv-modes.def: Likewise.
30320 * config/riscv/riscv-opts.h: Likewise.
30321 * config/riscv/riscv-protos.h: Likewise.
30322 * config/riscv/riscv.h: Likewise.
30323 * config/riscv/riscv.md: Likewise.
30324 * config/riscv/riscv.opt: Likewise.
30325 * config/riscv/sync.md: Likewise.
30326 * config/riscv/t-elf-multilib: Likewise.
30327 * config/riscv/t-linux: Likewise.
30328 * config/riscv/t-linux-multilib: Likewise.
30329 * config/riscv/t-riscv: Likewise.
30330 * configure.ac: Likewise.
30331 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
30332 Waterman as RISC-V maintainers.
30333 * doc/install.texi: Add RISC-V entries.
30334 * doc/invoke.texi: Add RISC-V options section.
30335 * doc/md.texi: Add RISC-V constraints section.
30336 * configure: Regenerated.
30338 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
30341 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
30342 false values to be constant vectors with all 0 or all 1 bits set.
30343 (vcondu<mode><mode>): Likewise.
30344 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
30346 (fpmask_comparison_operator): Update comment.
30347 (vecint_comparison_operator): New predicate.
30348 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
30349 vector conditionals when the true and false values are constant
30350 vectors with all 0 bits or all 1 bits set.
30352 2017-02-06 Martin Sebor <msebor@redhat.com>
30354 PR tree-optimization/79376
30355 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
30357 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
30359 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
30360 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
30361 to simplify split condition.
30363 2017-02-06 Jakub Jelinek <jakub@redhat.com>
30365 * omp-expand.c (oxpand_omp_atomic_fetch_op,
30366 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
30369 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
30371 PR rtl-optimization/68664
30372 * target.def (can_speculate_insn): New hook.
30373 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
30374 * doc/tm.texi: Regenerate.
30375 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
30376 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
30377 (rs6000_sched_can_speculate_insn): New function.
30379 2017-02-06 Jakub Jelinek <jakub@redhat.com>
30381 PR tree-optimization/79284
30382 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
30383 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
30384 vectorizable_mask_load_store, vectorizable_operation,
30385 vect_is_simple_cond, get_same_sized_vectype): Use it instead
30386 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
30387 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
30388 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
30389 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
30390 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
30391 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
30392 is_gimple_assign (stmt). Replace another such test with
30393 is_gimple_assign (stmt).
30395 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
30398 * config/avr/avr.c (rtl-iter.h): Include it.
30399 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
30400 (avr_legitimate_combined_insn): ...and implementation.
30402 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
30404 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
30405 * config/s390/s390.c (s390_const_operand_ok)
30406 (s390_canonicalize_comparison, s390_extract_part)
30407 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
30408 (s390_contiguous_bitmask_p, s390_rtx_costs)
30409 (legitimize_pic_address): Likewise.
30410 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
30411 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
30412 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
30413 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
30414 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
30416 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
30418 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
30419 REGNO($0) == REGNO($1).
30421 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30423 * config/s390/linux.h(SIZE_TYPE): Add comment.
30425 2017-02-06 Julian Brown <julian@codesourcery.com>
30426 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30427 Virendra Pathak <virendra.pathak@broadcom.com>
30429 * config/aarch64/aarch64-cores.def: Change the scheduler
30431 * config/aarch64/aarch64.md: Include thunderx2t99.md.
30432 * config/aarch64/thunderx2t99.md: New file.
30434 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30436 * doc/standards.texi (Go Language): Update link to language
30439 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30441 * tree-eh.c (lower_resx): Sanitize profile.
30442 (cleanup_empty_eh_move_lp): Likewise.
30444 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30447 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
30449 * cfgloopmanip.h (loop_version): Update prototype.
30450 * modulo-sched.c (sms_schedule): Update call of loop_version.
30451 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
30452 * tree-parloops.c (gen_parallel_loop): Likewise.
30453 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
30454 * tree-ssa-loop-split.c (split_loop): Likewise.
30455 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
30456 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
30458 2017-02-05 Martin Liska <mliska@suse.cz>
30461 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
30463 (print_operand_address): Initialize a struct to zero.
30465 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30467 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
30468 garbage collector only in textual form.
30470 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30472 * doc/extend.texi (x86 specific memory model extensions for
30473 transactional memory): Simplify a phrase.
30475 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
30478 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
30479 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
30480 (atomic_storedi_1): Likewise.
30482 2017-02-04 Jakub Jelinek <jakub@redhat.com>
30484 PR tree-optimization/79338
30485 * tree-parloops.c (gather_scalar_reductions): Don't call
30486 vect_analyze_loop_form for loop->inner before destroying loop's
30489 2017-02-03 Martin Sebor <msebor@redhat.com>
30491 PR tree-optimization/79327
30492 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
30493 when precision has resulted in leading zeros.
30494 (format_integer): Adjust the likely counter to assume an unknown
30495 argument that may be zero is non-zero.
30497 2017-02-03 Jason Merrill <jason@redhat.com>
30500 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
30501 avoid copying non-taken branch.
30503 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30505 PR tree-optimization/79340
30506 * tree-vect-loop.c (vectorizable_reduction): Release
30507 vec_defs elements after safe_splicing them into other vectors.
30510 PR tree-optimization/79327
30511 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
30512 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
30514 (format_integer): Use wide_int_to_tree instead of build_int_cst
30515 + to_?hwi. If argmin is NULL, just set argmin and argmax to
30516 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
30517 of shortest and longest sequence.
30519 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
30521 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
30522 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
30524 2017-02-03 Walter Lee <walt@tilera.com>
30527 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
30528 after initial stackframe link reg save.
30529 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
30531 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30534 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
30535 wu for stxssp alternative.
30537 2017-02-03 Martin Sebor <msebor@redhat.com>
30539 PR tree-optimization/79352
30540 * gimple-fold.c (get_range_strlen): Add argument.
30541 (get_range_strlen): Change return type to bool.
30542 (get_maxval_strlen): Pass in a dummy argument.
30543 * gimple-fold.h (get_range_strlen): Change return type to bool.
30544 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
30545 * tree.h (array_at_struct_end_p): Add argument.
30546 * tree.c (array_at_struct_end_p): Handle it.
30548 2017-02-03 Martin Liska <mliska@suse.cz>
30551 * multiple_target.c (create_dispatcher_calls): Redirect edge
30552 from a caller of a dispatcher.
30553 (expand_target_clones): Make the clones local.
30554 (ipa_target_clone): Do both target clones and resolvers.
30555 (ipa_dispatcher_calls): Remove the pass.
30556 (pass_dispatcher_calls::gate): Likewise.
30557 (make_pass_dispatcher_calls): Likewise.
30558 * passes.def (pass_target_clone): Put as very first IPA early
30561 2017-02-03 Martin Liska <mliska@suse.cz>
30563 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
30564 in case of a function with ifunc attribute.
30566 2017-02-03 Martin Liska <mliska@suse.cz>
30568 * cgraph.c (cgraph_node::dump): Dump function version info.
30569 * symtab.c (symtab_node::dump_base): Add missing new line.
30571 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
30573 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
30574 (ifcombine_ifandif): Use it.
30576 2017-02-03 Martin Liska <mliska@suse.cz>
30578 * doc/invoke.texi: Document default value for
30579 use-after-scope-direct-emission-threshold.
30581 2017-02-03 Martin Liska <mliska@suse.cz>
30583 PR tree-optimization/79339
30584 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
30585 (format_floating): Likewise.
30587 2017-02-03 Martin Liska <mliska@suse.cz>
30590 * ipa-prop.c (ipa_node_params_t::insert): Remove current
30592 (ipa_node_params_t::remove): Likewise.
30593 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
30594 initialization from removed ipa_node_params_t::insert.
30595 (ipa_node_params::~ipa_node_params): Move from removed
30596 ipa_node_params_t::release.
30597 * symbol-summary.h (symbol_summary::m_released): New member.
30598 Do not release a summary twice. Do not allow to call finalizer
30599 for types of a summary that live in GGC memory.
30601 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30603 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
30606 2017-02-02 Martin Sebor <msebor@redhat.com>
30608 PR middle-end/79275
30609 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
30610 (format_string): Tighten up the range of output for non-constant
30611 strings and correct the expected range for wide non-constant strings.
30613 2017-02-02 Martin Sebor <msebor@redhat.com>
30615 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
30617 PR middle-end/32003
30618 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
30620 (-fdump-tree-@var): Add to index and document how to come up
30621 with pass-specific option and dump file names.
30622 (-fdump-passes): Clarify where to look for output.
30624 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
30626 PR middle-end/77445
30627 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
30628 statistics of the analyzed path; allow threading for speed when
30629 any of BBs along the path are optimized for speed.
30631 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
30633 PR middle-end/78468
30634 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
30635 settings of the virtual registers.
30638 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
30640 * explow.c (get_dynamic_stack_size): Take known alignment of stack
30641 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
30644 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30646 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
30647 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
30649 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30651 * config/s390/s390.md: Add missing comments with the expanded
30653 * config/s390/vector.md: Likewise.
30654 * config/s390/vx-builtins.md: Likewise.
30656 2017-02-02 Jakub Jelinek <jakub@redhat.com>
30659 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
30660 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
30661 conditions on a single line.
30663 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30665 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
30666 __S390_VX__ to __VX__.
30668 2017-02-01 Andrew Pinski <apinski@cavium.com>
30670 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
30671 stmt_info to record_stmt_cost.
30672 (vect_get_known_peeling_cost): Pass stmt_info if known to
30674 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
30675 cpu_vector_cost field into
30676 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
30677 field into vec_int_stmt_cost and vec_fp_stmt_cost.
30678 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
30679 splitting of scalar_stmt_cost and vec_stmt_cost.
30680 (thunderx_vector_cost): Likewise.
30681 (cortexa57_vector_cost): LIkewise.
30682 (exynosm1_vector_cost): Likewise.
30683 (xgene1_vector_cost): Likewise.
30684 (thunderx2t99_vector_cost): Improve after the splitting of the two
30686 (aarch64_builtin_vectorization_cost): Update for the splitting of
30687 scalar_stmt_cost and vec_stmt_cost.
30689 2017-02-01 Torvald Riegel <triegel@redhat.com>
30690 Richard Henderson <rth@redhat.com>
30692 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
30693 conditional on existance of a fast atomic load.
30694 * optabs-query.c (can_atomic_load_p): New function.
30695 * optabs-query.h (can_atomic_load_p): Declare it.
30696 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
30697 no fast atomic load is available for the particular size of access.
30698 (expand_atomic_compare_and_swap): Likewise.
30699 (expand_atomic_load): Likewise.
30700 (expand_atomic_store): Likewise.
30701 (expand_atomic_fetch_op): Likewise.
30702 * testsuite/lib/target-supports.exp
30703 (check_effective_target_sync_int_128): Remove x86 because it provides
30704 no fast atomic load.
30705 (check_effective_target_sync_int_128_runtime): Likewise.
30707 2017-02-01 Richard Biener <rguenther@suse.de>
30709 * graphite.c: Include tree-vectorizer.h for find_loop_location.
30710 (graphite_transform_loops): Provide opt-info for optimized nests.
30711 * tree-parloop.c (parallelize_loops): Provide opt-info for
30712 parallelized loops.
30714 2017-02-01 Richard Biener <rguenther@suse.de>
30716 PR middle-end/79315
30717 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
30718 was not set before.
30720 2017-02-01 Richard Biener <rguenther@suse.de>
30722 PR tree-optimization/71824
30723 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
30724 Verify the loops are valid in the merged SESE region.
30725 (scop_detection::can_represent_loop_1): Check analyzing the
30726 evolution of the number of iterations in the region succeeds.
30728 2017-01-31 Ian Lance Taylor <iant@golang.org>
30730 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
30731 REG_ARGS_SIZE note to 32-bit push insns and call insn.
30733 2017-01-31 David Malcolm <dmalcolm@redhat.com>
30735 PR preprocessor/79210
30736 * input.c (get_substring_ranges_for_loc): Replace line_width
30737 assertion with error-handling.
30739 2017-01-31 Richard Biener <rguenther@suse.de>
30741 PR tree-optimization/77318
30742 * graphite-sese-to-poly.c (extract_affine): Fix assert.
30743 (create_pw_aff_from_tree): Take loop parameter.
30744 (add_condition_to_pbb): Pass loop of the condition to
30745 create_pw_aff_from_tree.
30747 2017-01-31 Jakub Jelinek <jakub@redhat.com>
30749 * config/s390/s390.c (s390_asan_shadow_offset): New function.
30750 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
30752 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
30756 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
30758 (convert_int_to_float128): Likewise.
30759 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
30760 (convert_int_to_float128): Likewise.
30761 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
30762 (UNSPEC_IEEE128_CONVERT): Likewise.
30763 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
30764 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
30765 Use local variables for IBM extended format.
30766 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
30767 (fix_trunc<mode>si2_fprs): Likewise.
30768 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
30769 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
30770 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
30771 to know that we can now have integers of all sizes in vector
30773 (fix<uns>_<mode>di2_hw): Likewise.
30774 (float<uns>_<mode>si2_hw): Likewise.
30775 (fix_<mode>si2_hw): Likewise.
30776 (fixuns_<mode>si2_hw): Likewise.
30777 (float<uns>_<mode>di2_hw): Likewise.
30778 (float_<mode>di2_hw): Likewise.
30779 (float_<mode>si2_hw): Likewise.
30780 (floatuns_<mode>di2_hw): Likewise.
30781 (floatuns_<mode>si2_hw): Likewise.
30782 (xscvqp<su>wz_<mode>): Delete, no longer used.
30783 (xscvqp<su>dz_<mode>): Likewise.
30784 (xscv<su>dqp_<mode>): Likewise.
30785 (ieee128_mfvsrd_64bit): Likewise.
30786 (ieee128_mfvsrd_32bit): Likewise.
30787 (ieee128_mfvsrwz): Likewise.
30788 (ieee128_mtvsrw): Likewise.
30789 (ieee128_mtvsrd_64bit): Likewise.
30790 (ieee128_mtvsrd_32bit): Likewise.
30792 2017-01-31 Martin Liska <mliska@suse.cz>
30795 * ipa-prop.c (ipa_free_all_node_params): Call release method
30796 instead of ~sumbol_summary to not to trigger double times
30799 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
30801 PR tree-optimization/71691
30802 * bitmap.h (class auto_bitmap): New.
30803 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
30804 is_maybe_undefined instead of ssa_undefined_value_p.
30806 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30808 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
30809 __S390_ARCH_LEVEL__ to __ARCH__.
30811 2017-01-31 Jakub Jelinek <jakub@redhat.com>
30813 PR tree-optimization/79267
30814 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
30815 if should_remove_lhs_p is true.
30817 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
30820 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
30821 (add_alignment_attribute): New.
30822 (base_type_die): Add alignment attribute.
30823 (subrange_type_die): Likewise.
30824 (modified_type_die): Likewise.
30825 (gen_array_type_die): Likewise.
30826 (gen_descr_array_type_die: Likewise.
30827 (gen_enumeration_type_die): Likewise.
30828 (gen_subprogram_die): Likewise.
30829 (gen_variable_die): Likewise.
30830 (gen_field_die): Likewise.
30831 (gen_ptr_to_mbr_type_die): Likewise.
30832 (gen_struct_or_union_type_die): Likewise.
30833 (gen_subroutine_type_die): Likewise.
30834 (gen_typedef_die): Likewise.
30835 (base_type_cmp): Compare alignment attribute.
30837 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30840 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
30841 (setb_unsigned) New pattern for setb with CCUNS.
30842 * config/rs6000/rs6000.c (expand_block_compare): Use a different
30843 subfc./subfe sequence to avoid overflow problems. Generate a
30844 shorter sequence with cmpld/setb for power9.
30845 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
30846 for generating subfc. instruction.
30847 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
30848 now uses this instruction.
30850 2017-01-30 Ian Lance Taylor <iant@google.com>
30853 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
30854 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
30856 2017-01-30 Martin Sebor <msebor@redhat.com>
30858 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
30859 Move constant to the right of a relational operator.
30860 (get_mpfr_format_length, format_character, format_string): Ditto.
30861 (should_warn_p, maybe_warn): Same.
30863 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
30865 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
30868 * asan.c (get_translation_unit_decl): Remove function.
30869 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
30871 2017-01-30 Martin Liska <mliska@suse.cz>
30873 PR gcov-profile/79259
30874 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
30875 -fprofile-generate.
30877 2017-01-30 Martin Liska <mliska@suse.cz>
30880 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
30881 Initialize variables with NULL value.
30883 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
30886 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
30888 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
30890 2017-01-30 Richard Biener <rguenther@suse.de>
30892 PR tree-optimization/79276
30893 * tree-vrp.c (process_assert_insertions): Properly adjust common
30894 when removing a duplicate.
30896 2017-01-30 Richard Biener <rguenther@suse.de>
30898 PR tree-optimization/79256
30899 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
30900 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
30902 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
30904 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
30907 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
30908 ("*r<noxa>sbg_<mode>_sll_bitmask")
30909 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
30910 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
30911 Use contiguous_bitmask_nowrap_operand.
30913 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30916 * config/rs6000/altivec.h (vec_xl): Revise #define.
30917 (vec_xst): Likewise.
30919 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
30921 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
30923 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
30925 PR rtl-optimization/79194
30926 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
30927 traps before call to bypass_conditional_jumps.
30929 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
30931 PR tree-optimization/71374
30932 * lra-constraints.c (check_conflict_input_operands): New.
30933 (match_reload): Use it.
30935 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
30938 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
30939 account to calculate conflict_set.
30941 2017-01-27 Bin Cheng <bin.cheng@arm.com>
30943 PR rtl-optimization/78559
30944 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
30945 other_insn in combine.
30947 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
30949 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
30950 uint16_type_node for BT_UINT16.
30952 2017-01-27 David Malcolm <dmalcolm@redhat.com>
30954 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
30955 "RTL Tests" to menu.
30956 (GIMPLE Tests): New node.
30957 (RTL Tests): New node.
30959 2017-01-27 Richard Biener <rguenther@suse.de>
30961 PR tree-optimization/79245
30962 * tree-loop-distribution.c (distribute_loop): Apply cost
30963 modeling also to detected patterns.
30965 2017-01-27 Richard Biener <rguenther@suse.de>
30967 PR tree-optimization/71433
30968 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
30969 (compare_assert_loc): New function.
30970 (process_assert_insertions): Sort and optimize assert locations
30971 to remove duplicates and push down identical assertions on
30972 edges to their destination block.
30974 2017-01-27 Richard Biener <rguenther@suse.de>
30976 PR tree-optimization/79244
30977 * tree-vrp.c (remove_range_assertions): Forcefully propagate
30978 out SSA names even if abnormal.
30980 2017-01-27 Jakub Jelinek <jakub@redhat.com>
30982 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
30983 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
30984 instead of MPFR_RNDN.
30986 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
30989 * arm.c (arm_option_override): Don't call build_target_option_node
30990 until after doing all option overrides.
30991 (arm_valid_target_attribute_tree): Likewise.
30993 2017-01-27 Martin Liska <mliska@suse.cz>
30995 * doc/invoke.texi (-fprofile-arcs): Document profiling support
30996 for {cd}tors and C++ {cd}tors.
30998 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
31000 * config/s390/s390.md ("*setmem_long_and")
31001 ("*setmem_long_and_31z"): Use zero_extend instead of and.
31003 2017-01-26 Martin Sebor <msebor@redhat.com>
31005 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
31008 2017-01-26 Martin Sebor <msebor@redhat.com>
31010 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
31011 HAVE_DFmode before using XFmode or DFmode.
31012 (parse_directive): Avoid using the z length modifier to avoid
31013 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
31015 PR middle-end/78703
31016 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
31017 to accept adjustment as an array.
31018 (get_int_range): New function.
31019 (struct directive): Make width and prec arrays.
31020 (directive::set_width, directive::set_precision): Call get_int_range.
31021 (format_integer, format_floating): Handle width and precision ranges.
31022 (format_string, parse_directive): Same.
31024 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31027 * dwarf2out.c (generate_skeleton_bottom_up): For children with
31028 comdat_type_p set, just clone them, but keep the children in the
31032 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
31033 which have direct callers with -fvar-tracking-assignments enabled
31035 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
31036 inside of type units.
31038 2017-01-26 Martin Sebor <msebor@redhat.com>
31040 PR middle-end/78703
31041 * gimple-ssa-sprintf.c (struct result_range): Add likely and
31043 (struct format_result): Replace number_chars, number_chars_min,
31044 and number_chars_max with a single member of struct result_range.
31046 (format_result::operator+=): Adjust.
31047 (struct fmtresult): Remove bounded. Handle likely and unlikely
31049 (fmtresult::adjust_for_width_or_precision): New function.
31050 (fmtresult:type_max_digits): New function.
31051 (bytes_remaining): Handle likely and unlikely counters.
31052 (min_bytes_remaining): Remove.
31053 (format_percent): Simplify.
31054 (format_integer, format_floating): Set likely and unlikely counters.
31055 (get_string_length, format_character, format_string): Same.
31056 (format_plain, should_warn_p): New function.
31057 (maybe_warn): Call should_warn_p. Update diagnostic messages
31058 and handle those for all directives, including plain strings.
31059 (format_directive): Handle likely and unlikely counters.
31060 Remove unnecessary quoting from diagnostics. Add an informational
31062 (add_bytes): Remove.
31063 (pass_sprintf_length::compute_format_length): Simplify.
31064 (try_substitute_return_value): Handle likely and unlikely counters.
31066 2017-01-26 Carl Love <cel@us.ibm.com>
31068 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
31069 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
31071 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
31074 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
31075 endianess for subregs into account.
31076 * lra-constraints.c (lra_constraints): Do risky transformations
31077 always on the first iteration.
31078 * lra-lives.c (check_pseudos_live_through_calls): Add arg
31079 last_call_used_reg_set.
31080 (process_bb_lives): Define and use last_call_used_reg_set.
31081 * lra.c (lra): Always continue after lra_constraints on the first
31084 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
31086 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
31088 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
31090 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31092 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
31093 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
31094 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
31095 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
31096 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
31097 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
31098 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
31099 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
31100 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
31102 2017-01-26 Marek Polacek <polacek@redhat.com>
31105 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
31106 for the third operand.
31108 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31110 PR middle-end/79236
31111 * omp-low.c (struct omp_context): Add simt_stmt field.
31112 (scan_omp_for): Return omp_context *.
31113 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
31114 context to the _simt_ SIMD stmt.
31115 (lower_omp_for): For combined SIMD with sibling _simt_
31116 SIMD, make sure to use the same decls in _looptemp_
31117 clauses as in the sibling.
31119 2017-01-26 David Sherwood <david.sherwood@arm.com>
31121 PR middle-end/79212
31122 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
31125 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31128 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
31129 emit fld b; fld a; if possible.
31131 * brig-builtins.def: Update copyright years.
31132 * config/arm/arm_acle_builtins.def: Update copyright years.
31134 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
31137 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
31138 constraint instead of o for the stxsd instruction.
31140 2017-01-25 Carl Love <cel@us.ibm.com>
31142 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
31143 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
31145 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
31147 * doc/invoke.texi (C++ Dialect Options): Fix typo.
31149 2017-01-25 Richard Biener <rguenther@suse.de>
31151 PR tree-optimization/69264
31152 * target.def (vector_alignment_reachable): Improve documentation.
31153 * doc/tm.texi: Regenerate.
31154 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
31156 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
31157 earlier changes with respect to TYPE_USER_ALIGN.
31158 (vector_alignment_reachable_p): Likewise. Improve dumping.
31160 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31163 * config/arm/arm.md (xordi3): Force constant operand into a register
31166 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31168 * doc/invoke.texi (-fstore-merging): Correct default optimization
31169 levels at which it is enabled.
31170 (-O): Move -fstore-merging from list to...
31173 2017-01-25 Richard Biener <rguenther@suse.de>
31176 * omp-expand.c: Include debug.h.
31177 (expand_omp_taskreg): Make sure to generate early debug before
31178 outlining anything from a function.
31179 (expand_omp_target): Likewise.
31180 (grid_expand_target_grid_body): Likewise.
31182 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
31185 * asan.c (get_translation_unit_decl): New function.
31186 (asan_add_global): Extract modules file name from globals
31187 TRANSLATION_UNIT_DECL name.
31189 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
31192 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
31193 for long calls with APCS frame and VFP.
31195 2017-01-24 David Malcolm <dmalcolm@redhat.com>
31197 * cfg.c (original_copy_tables_initialized_p): New function.
31198 * cfg.h (original_copy_tables_initialized_p): New decl.
31199 * cfgrtl.c (relink_block_chain): Guard the call to
31200 free_original_copy_tables with a call to
31201 original_copy_tables_initialized_p.
31202 * cgraph.h (symtab_node::native_rtl_p): New decl.
31203 * cgraphunit.c (symtab_node::native_rtl_p): New function.
31204 (symtab_node::needed_p): Don't assert for early assembly output
31205 for __RTL functions.
31206 (cgraph_node::finalize_function): Set "force_output" for __RTL
31208 (cgraph_node::analyze): Bail out early for __RTL functions.
31209 (analyze_functions): Update assertion to support __RTL functions.
31210 (cgraph_node::expand): Bail out early for __RTL functions.
31211 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
31213 * function.h (struct function): Update comment for field
31215 * gimple-expr.c: Include "tree-pass.h".
31216 (gimple_has_body_p): Return false for __RTL functions.
31217 * Makefile.in (OBJS): Add run-rtl-passes.o.
31218 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
31220 (gcc::pass_manager::get_clean_slate): New accessor.
31221 * passes.c: Include "insn-addr.h".
31222 (should_skip_pass_p): Add logging. Update logic for running
31223 "expand" to be compatible with both __GIMPLE and __RTL. Guard
31224 property-provider override so it is only done for gimple passes.
31226 (skip_pass): New function.
31227 (execute_one_pass): Call skip_pass when skipping passes.
31228 * read-md.c (md_reader::read_char): Support filtering
31229 the input to a subset of line numbers.
31230 (md_reader::md_reader): Initialize fields
31231 m_first_line and m_last_line.
31232 (md_reader::read_file_fragment): New function.
31233 * read-md.h (md_reader::read_file_fragment): New decl.
31234 (md_reader::m_first_line): New field.
31235 (md_reader::m_last_line): New field.
31236 * read-rtl-function.c (function_reader::create_function): Only
31237 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
31238 curr_properties. Set DECL_INITIAL to a dummy block.
31239 (read_rtl_function_body_from_file_range): New function.
31240 * read-rtl-function.h (read_rtl_function_body_from_file_range):
31242 * run-rtl-passes.c: New file.
31243 * run-rtl-passes.h: New file.
31245 2017-01-24 Jeff Law <law@redhat.com>
31247 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
31250 2017-01-24 Bin Cheng <bin.cheng@arm.com>
31252 PR tree-optimization/79159
31253 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
31254 (record_nonwrapping_iv): Improve boundary using above function if no
31255 value range information.
31257 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
31258 Martin Jambor <mjambor@suse.cz>
31260 * brig-builtins.def: New file.
31261 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
31262 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
31263 (DEF_HSAIL_SAT_BUILTIN): Likewise.
31264 (DEF_HSAIL_INTR_BUILTIN): Likewise.
31265 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
31266 * builtin-types.def (BT_INT8): New.
31267 (BT_INT16): Likewise.
31268 (BT_UINT8): Likewise.
31269 (BT_UINT16): Likewise.
31270 (BT_FN_ULONG): Likewise.
31271 (BT_FN_UINT_INT): Likewise.
31272 (BT_FN_UINT_ULONG): Likewise.
31273 (BT_FN_UINT_LONG): Likewise.
31274 (BT_FN_UINT_PTR): Likewise.
31275 (BT_FN_ULONG_PTR): Likewise.
31276 (BT_FN_INT8_FLOAT): Likewise.
31277 (BT_FN_INT16_FLOAT): Likewise.
31278 (BT_FN_UINT32_FLOAT): Likewise.
31279 (BT_FN_UINT16_FLOAT): Likewise.
31280 (BT_FN_UINT8_FLOAT): Likewise.
31281 (BT_FN_UINT64_FLOAT): Likewise.
31282 (BT_FN_UINT16_UINT32): Likewise.
31283 (BT_FN_UINT32_UINT16): Likewise.
31284 (BT_FN_UINT16_UINT16_UINT16): Likewise.
31285 (BT_FN_INT_PTR_INT): Likewise.
31286 (BT_FN_UINT_PTR_UINT): Likewise.
31287 (BT_FN_LONG_PTR_LONG): Likewise.
31288 (BT_FN_ULONG_PTR_ULONG): Likewise.
31289 (BT_FN_VOID_UINT64_UINT64): Likewise.
31290 (BT_FN_UINT8_UINT8_UINT8): Likewise.
31291 (BT_FN_INT8_INT8_INT8): Likewise.
31292 (BT_FN_INT16_INT16_INT16): Likewise.
31293 (BT_FN_INT_INT_INT): Likewise.
31294 (BT_FN_UINT_FLOAT_UINT): Likewise.
31295 (BT_FN_FLOAT_UINT_UINT): Likewise.
31296 (BT_FN_ULONG_UINT_UINT): Likewise.
31297 (BT_FN_ULONG_UINT_PTR): Likewise.
31298 (BT_FN_ULONG_ULONG_ULONG): Likewise.
31299 (BT_FN_UINT_UINT_UINT): Likewise.
31300 (BT_FN_VOID_UINT_PTR): Likewise.
31301 (BT_FN_UINT_UINT_PTR: Likewise.
31302 (BT_FN_UINT32_UINT64_PTR): Likewise.
31303 (BT_FN_INT_INT_UINT_UINT): Likewise.
31304 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
31305 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
31306 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
31307 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
31308 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
31309 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
31310 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
31311 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
31312 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
31313 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
31314 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
31315 * doc/frontends.texi: List BRIG FE.
31316 * doc/install.texi (Testing): Add BRIG tesring requirements.
31317 * doc/invoke.texi (Overall Options): Mention BRIG.
31318 * doc/standards.texi (Standards): Doucment BRIG HSA version.
31320 2017-01-24 Richard Biener <rguenther@suse.de>
31322 PR translation/79208
31323 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
31325 2017-01-24 Martin Jambor <mjambor@suse.cz>
31328 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
31329 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
31330 and known_contexts.
31332 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
31334 PR middle-end/79123
31335 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
31336 casts from signed to unsigned really don't have a range.
31338 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
31340 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
31341 GMP_RNDx for compatiblity.
31343 2017-01-24 Martin Liska <mliska@suse.cz>
31346 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
31347 that would prevent us to call alloca with -1 as argument.
31349 2017-01-24 Jakub Jelinek <jakub@redhat.com>
31351 * dwarf2out.c (output_compilation_unit_header, output_file_names):
31352 Avoid -Wformat-security warning.
31354 2017-01-23 Andrew Pinski <apinski@cavium.com>
31356 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
31359 2017-01-23 Martin Sebor <msebor@redhat.com>
31361 PR middle-end/78703
31362 * gimple-ssa-sprintf.c (warn_level): New global.
31363 (format_integer): Use it here and throughout the rest of the file.
31364 Use the same switch to compute sign as base.
31365 (maybe_warn): New function.
31366 (format_directive): Factor out warnings into maybe_warn.
31367 Add debugging output. Use warn_level.
31368 (add_bytes): Use warn_level.
31369 (pass_sprintf_length::compute_format_length): Add debugging output.
31370 (try_substitute_return_value): Same.
31371 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
31373 PR middle-end/78703
31374 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
31375 (struct fmtresult, format_integer, format_floating): Adjust.
31376 (fmtresult::fmtresult): Set max correctly in two argument ctor.
31377 (get_string_length, format_string,format_directive): Same.
31378 (pass_sprintf_length::compute_format_length): Same.
31379 (try_substitute_return_value): Simplify slightly.
31381 PR middle-end/78703
31382 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
31383 (fmtresult::operator+=): Outlined.
31384 (struct fmtresult): Add ctors.
31385 (struct conversion_spec): Rename...
31386 (struct directive): ...to this. Add and remove data members.
31387 (directive::set_width, directive::set_precision): New functions.
31388 (format_percent): Use fmtresult ctor.
31389 (get_width_and_precision): Remove.
31390 (format_integer): Make naming changes. Avoid computing width and
31392 (format_floating): Same. Adjust indentation.
31393 (format_character, format_none): New functions.
31394 (format_string): Moved character handling to format_character.
31395 (format_directive): Remove arguments, change return type.
31396 (parse_directive): New function.
31397 (pass_sprintf_length::compute_format_length): Move directive
31398 parsing to parse_directive.
31400 2017-01-23 Jakub Jelinek <jakub@redhat.com>
31402 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
31403 (assign_assembler_name_if_needed): ... this.
31404 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
31405 (assign_assembler_name_if_needed): ... this.
31406 (free_lang_data_in_cgraph): Adjust callers.
31407 * cgraphunit.c (cgraph_node::analyze): Likewise.
31408 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
31411 2017-01-23 Richard Biener <rguenther@suse.de>
31413 PR tree-optimization/79088
31414 PR tree-optimization/79188
31415 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
31416 resetting loop bounds after last path deletion. Reset loop
31417 bounds of the target loop, make code match the comments.
31418 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
31419 Make sure loops need no fixups.
31421 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
31423 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
31424 exponent support with double type for first argument.
31425 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
31426 type returned by __builtin_vec_extract_sig,
31427 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
31428 functions from "vector int" to "vector unsigned int" or from
31429 "vector long long int" to "vector unsigned long long int".
31430 Changed type returned by __builtin_vec_extract_exp,
31431 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
31432 functions from "vector int" to "vector unsigned int" or from
31433 "vector long long int" to "vector unsigned long long int".
31434 Changed return type of __builtin_vec_test_data_class,
31435 __builtin_vec_test_data_class_sp, and
31436 __builtin_vec_test_data_class_dp from "vector int" to
31437 "vector bool int" or from "vector long long int" to "vector bool
31438 long long int" and changed second argument type from "unsigned
31439 int" to "int". Added new overloaded function forms "vector float
31440 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
31441 "vector float __builtin_vec_insert_exp_sp (vector float, vector
31442 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
31443 double, vector unsigned long long int)" and "vector double
31444 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
31445 long int)". Changed return type of
31446 __builtin_scalar_test_data_class and
31447 __builtin_scalar_test_data_class_sp and
31448 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
31449 int" and changed second argument from "unsigned int" to "int".
31450 Changed type returned by __builtin_scalar_test_neg,
31451 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
31452 from "int" to "bool int". Added new overloaded function form
31453 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
31454 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
31455 exponent double-precision with floating point first argument.
31456 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
31457 documentation of scalar_test_data_class, scalar_test_neg,
31458 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
31459 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
31460 vec_test_data_class built-in functions to reflect refinements in
31461 their type signatures.
31463 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
31465 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
31467 (aarch64_elf_asm_destructor): Likewise.
31469 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
31471 PR rtl-optimization/78634
31472 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
31473 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
31474 * ifcvt.c (noce_try_cmove): Add missing cost check.
31476 PR rtl-optimization/71724
31477 * combine.c (if_then_else_cond): Look for situations where it is
31478 beneficial to undo the work of one of the recursive calls.
31480 2017-01-23 Bin Cheng <bin.cheng@arm.com>
31482 PR tree-optimization/70754
31483 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
31484 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
31485 combined stmt before it if not NULL.
31486 (combine_chains): Process refs reversely and compute dominance point
31489 2017-01-23 Martin Liska <mliska@suse.cz>
31491 PR tree-optimization/79196
31492 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
31493 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
31495 (strlen_optimize_stmt): Call the renamed function.
31497 2017-01-23 Michael Matz <matz@suse.de>
31499 PR tree-optimization/78384
31500 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
31502 2017-01-23 Richard Biener <rguenther@suse.de>
31504 PR tree-optimization/79186
31505 * tree-vrp.c (register_new_assert_for): Make sure we've seen
31506 both incoming edges before moving an assert.
31508 2017-01-23 Martin Jambor <mjambor@suse.cz>
31510 * ipa-prop.c (load_from_param_1): Removed.
31511 (load_from_unmodified_param): Bits from load_from_param_1 put back
31513 (load_from_param): Removed.
31514 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
31515 with stmt. Reverted back to use of load_from_unmodified_param.
31517 2017-01-23 Martin Jambor <mjambor@suse.cz>
31520 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
31521 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
31522 field a pointer to garbage collected vector, mark lattices and
31523 ipcp_orig_node with GTY((skip)).
31524 (ipa_get_param_count): Adjust to descriptors being a pointer.
31525 (ipa_get_param): Likewise.
31526 (ipa_get_type): Likewise.
31527 (ipa_get_param_move_cost): Likewise.
31528 (ipa_set_param_used): Likewise.
31529 (ipa_get_controlled_uses): Likewise.
31530 (ipa_set_controlled_uses): Likewise.
31531 (ipa_is_param_used): Likewise.
31532 (ipa_node_params_t): Move into garbage collector. New methods insert
31534 (ipa_node_params_sum): Annotate wth GTY(()).
31535 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
31537 (ipa_load_from_parm_agg): Adjust declaration.
31538 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
31539 * ipa-profile.c (ipa_profile): Likewise.
31540 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
31541 (ipa_populate_param_decls): Make descriptors parameter garbage
31543 (ipa_dump_param): Adjust to descriptors being a pointer.
31544 (ipa_alloc_node_params): Likewise.
31545 (ipa_initialize_node_params): Likewise.
31546 (load_from_param_1): Make descriptors parameter garbage collected.
31547 (load_from_unmodified_param): Likewise.
31548 (load_from_param): Likewise.
31549 (ipa_load_from_parm_agg): Likewise.
31550 (ipa_node_params::~ipa_node_params): Removed.
31551 (ipa_free_all_node_params): Remove call to delete operator.
31552 (ipa_node_params_t::insert): New.
31553 (ipa_node_params_t::remove): Likewise.
31554 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
31555 copy known_csts and known_contexts vectors.
31556 (ipa_read_node_info): Adjust to descriptors being a pointer.
31557 (ipcp_modif_dom_walker): Make m_descriptors field garbage
31559 (ipcp_transform_function): Make descriptors variable garbage
31562 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
31564 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
31565 * config/i386/avx512dqintrin.h: Ditto.
31566 * config/i386/avx512fintrin.h: Ditto.
31567 * config/i386/i386.c: Handle new builtins.
31568 * config/i386/i386-builtin.def: Add new builtins.
31569 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
31570 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
31572 2017-01-23 Jakub Jelinek <jakub@redhat.com>
31573 Martin Liska <mliska@suse.cz>
31575 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
31576 * asan.c (asan_expand_poison_ifn): Support stores and use
31577 appropriate ASAN report function.
31578 * internal-fn.c (expand_ASAN_POISON_USE): New function.
31579 * internal-fn.def (ASAN_POISON_USE): Declare.
31580 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
31581 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
31582 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
31583 ASAN_POISON calls w/o LHS.
31584 * tree-ssa.c (execute_update_addresses_taken): Create clobber
31585 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
31586 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
31587 * gimplify.c (asan_poison_variables): Add attribute
31588 use_after_scope_memory to variables that really needs to live
31590 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
31591 having the attribute.
31593 2017-01-23 Martin Liska <mliska@suse.cz>
31595 * asan.c (create_asan_shadow_var): New function.
31596 (asan_expand_poison_ifn): Likewise.
31597 * asan.h (asan_expand_poison_ifn): New declaration.
31598 * internal-fn.c (expand_ASAN_POISON): Likewise.
31599 * internal-fn.def (ASAN_POISON): New builtin.
31600 * sanopt.c (pass_sanopt::execute): Expand
31601 asan_expand_poison_ifn.
31602 * tree-inline.c (copy_decl_for_dup_finish): Make function
31604 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
31605 * tree-ssa.c (is_asan_mark_p): New function.
31606 (execute_update_addresses_taken): Rewrite local variables
31607 (identified just by use-after-scope as addressable) into SSA.
31609 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
31611 * doc/install.texi (Specific): opensource.apple.com uses https
31612 now. Remove trailing slash.
31614 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
31616 * README.Portability: Remove note on an Irix compatibility issue.
31618 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
31620 * gcov.c (INCLUDE_ALGORITHM): Define.
31621 (INCLUDE_VECTOR): Define.
31622 No longer include <vector> and <algorithm> directly.
31624 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
31626 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
31628 * doc/invoke.texi (Code Gen Options): Ditto.
31630 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
31633 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
31635 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
31637 rtl-optimization/79125
31638 * cprop.c (local_cprop_pass): Handle cases where we make an
31639 unconditional trap.
31641 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
31645 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
31646 read from, for big endian.
31648 2017-01-20 Jiong Wang <jiong.wang@arm.com>
31650 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
31651 register pauth builtins for LP64 only.
31653 2017-01-20 Marek Polacek <polacek@redhat.com>
31656 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
31659 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
31661 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
31663 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
31664 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
31665 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
31667 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31670 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
31671 in big-endian mode when they are not a single duplicated value.
31673 2017-01-20 Richard Biener <rguenther@suse.de>
31675 * BASE-VER: Bump to 7.0.1.
31677 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
31679 * omp-low.c (omplow_simd_context): New struct. Use it...
31680 (lower_rec_simd_input_clauses): ...here and...
31681 (lower_rec_input_clauses): ...here to hold common data. Adjust all
31682 references to idx, lane, max_vf, is_simt.
31684 2017-01-20 Graham Markall <graham.markall@embecosm.com>
31686 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
31689 2017-01-20 Martin Jambor <mjambor@suse.cz>
31691 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
31692 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
31694 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
31695 (GTFILES): Rename hsa.c to hsa-common.c.
31696 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
31697 * hsa-dump.c: Likewise.
31698 * hsa-gen.c: Likewise.
31699 * hsa-regalloc.c: Likewise.
31700 * ipa-hsa.c: Likewise.
31701 * omp-expand.c: Likewise.
31702 * omp-low.c: Likewise.
31703 * toplev.c: Likewise.
31705 2017-01-20 Marek Polacek <polacek@redhat.com>
31708 * doc/invoke.texi: Document -Wduplicated-branches.
31709 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
31710 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
31711 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
31712 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
31713 return 0 only when not OEP_LEXICOGRAPHIC.
31714 (fold_build_cleanup_point_expr): Use the expression
31715 location when building CLEANUP_POINT_EXPR.
31716 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
31717 * tree.c (add_expr): Handle error_mark_node.
31719 2017-01-20 Martin Liska <mliska@suse.cz>
31722 * tree-profile.c (init_ic_make_global_vars): Do not call
31724 (gimple_init_gcov_profiler): Likewise.
31726 2017-01-20 Martin Liska <mliska@suse.cz>
31729 * cgraph.h (maybe_create_reference): Remove argument and
31731 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
31733 * ipa-cp.c (create_specialized_node): Likewise.
31734 * symtab.c (symtab_node::maybe_create_reference): Handle
31735 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
31737 2017-01-20 Martin Liska <mliska@suse.cz>
31739 * read-rtl-function.c (function_reader::create_function): Use
31740 build_decl instread of build_decl_stat.
31742 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
31744 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
31745 * config/i386/avx512dqintrin.h: Ditto.
31746 * config/i386/avx512fintrin.h: Ditto.
31747 * config/i386/i386-builtin-types.def: Add new types.
31748 * config/i386/i386.c: Handle new types.
31749 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
31750 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
31751 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
31752 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
31753 (__builtin_ia32_kshiftridi): New.
31754 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
31756 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
31760 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
31761 define to rs6000_init_stack_protect_guard.
31762 (rs6000_init_stack_protect_guard): New function.
31764 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
31765 Yunqiang Su <yunqiang.su@imgtec.com>
31767 * config.gcc (supported_defaults): Add madd4.
31768 (with_madd4): Add validation.
31769 (all_defaults): Add madd4.
31770 * config/mips/mips.opt (mmadd4): New option.
31771 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
31773 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
31775 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
31776 (ISA_HAS_FUSED_MADD4): Likewise.
31777 * doc/invoke.texi (-mmadd4): Document the new option.
31778 * doc/install.texi (--with-madd4): Document the new option.
31780 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31782 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
31783 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
31784 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
31785 (aarch64_init_pauth_hint_builtins): New.
31786 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
31787 (aarch64_expand_builtin): Expand new builtins.
31789 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31791 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
31792 * combine-stack-adj.c (no_unhandled_cfa): Handle
31793 REG_CFA_TOGGLE_RA_MANGLE.
31794 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
31795 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
31796 info for return address signing.
31797 (aarch64_expand_epilogue): Likewise.
31799 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31801 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
31802 * config/aarch64/aarch64-protos.h
31803 (aarch64_return_address_signing_enabled): New declaration.
31804 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
31806 (aarch64_expand_prologue): Sign return address before it's pushed onto
31808 (aarch64_expand_epilogue): Authenticate return address fetched from
31810 (aarch64_override_options): Sanity check for ILP32 and ISA level.
31811 (aarch64_attributes): New function attributes for "sign-return-address".
31812 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
31813 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
31814 ("*do_return"): Generate combined instructions according to key index.
31815 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
31816 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
31818 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
31819 * config/aarch64/aarch64.opt (msign-return-address=): New.
31820 * doc/extend.texi (AArch64 Function Attributes): Documents
31821 "sign-return-address=".
31822 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
31824 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
31826 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
31827 overall option summary.
31829 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31831 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
31832 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
31833 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
31834 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
31836 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
31838 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
31839 -mpower9-minmax by default for -mcpu=power9.
31840 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
31841 128-bit floating point.
31843 2017-01-20 Alan Modra <amodra@gmail.com>
31845 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
31846 optimizing for size.
31848 2017-01-20 Alan Modra <amodra@gmail.com>
31851 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
31852 for strcmp and strncmp from corresponding builtin decl.
31854 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
31856 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
31857 instead of i386/rtems-64.h.
31858 * config/i386/rtems-64.h: Remove.
31860 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
31864 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
31866 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
31868 2017-01-19 Tamar Christina <tamar.christina@arm.com>
31870 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
31871 Change int to HOST_WIDE_INT.
31872 * config/aarch64/aarch64-protos.h
31873 (aarch64_simd_gen_const_vector_dup): Likewise.
31874 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
31876 2017-01-19 David Malcolm <dmalcolm@redhat.com>
31878 * langhooks-def.h (lhd_type_for_size): New decl.
31879 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
31880 * langhooks.c (lhd_type_for_size): New function, taken from
31883 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
31885 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
31886 define_bypass for CR latency.
31887 (power9-cracked-alu): Update bypass latency and remove power9-branch.
31888 (power9-alu2): Add define_bypass for CR latency.
31890 (power9-mul): Update insn latency.
31891 (power9-mul-compare): Update insn latency, bypass latency and remove
31894 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31896 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
31898 * config/aarch64/aarch64.md
31899 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
31900 aarch64_nopcrelative_literal_loads.
31901 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
31903 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
31905 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
31906 TARGET_LOONGSON_3A.
31907 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
31909 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
31912 * config.gcc (supported_defaults): Add lxc1-sxc1.
31913 (with_lxc1_sxc1): Add validation.
31914 (all_defaults): Add lxc1-sxc1.
31915 * config/mips/mips.opt (mlxc1-sxc1): New option.
31916 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
31918 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
31919 __mips_no_lxc1_sxc1.
31920 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
31921 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
31922 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
31924 2017-01-19 Richard Biener <rguenther@suse.de>
31926 PR tree-optimization/72488
31927 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
31928 sure to restore SSA info.
31929 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
31931 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
31933 PR rtl-optimization/79121
31934 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
31935 of the inner type when shifting an extended value.
31937 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
31940 * symtab.c (symtab_node::equal_address_to): Fix comparing of
31941 interposable aliases.
31943 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
31946 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
31947 Use the evmergelohi instruction.
31948 (mov_si<mode>_e500_subreg4_2_le): Likewise.
31949 (mov_sitf_e500_subreg8_2_be): Likewise.
31950 (mov_sitf_e500_subreg12_2_le): Likewise.
31951 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
31952 (mov_si<mode>_e500_subreg4_2_be): Likewise.
31953 (mov_sitf_e500_subreg8_2_le): Likewise.
31954 (mov_sitf_e500_subreg12_2_be): Likewise.
31956 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31958 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
31959 attribute from vecsimple to vecperm.
31960 (altivec_vbpermq2): Likewise.
31962 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31965 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
31967 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31968 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
31969 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
31970 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
31971 case where N arg is SIZE_MAX.
31972 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
31973 (cmpstrsi): Add pattern.
31975 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
31977 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
31978 __builtin_vec_revb builtins.
31979 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
31980 built-in functions to support generation of the ISA 3.0 XXBR<x>
31981 vector byte reverse instructions.
31982 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
31983 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
31984 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
31985 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
31986 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
31987 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
31988 (P9V_BUILTIN_VEC_REVB): Likewise.
31989 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
31990 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
31991 (p9_xxbrq_v16qi): Likewise.
31992 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
31993 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
31994 (p9_xxbrh_v8hi): Likewise.
31995 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
31996 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
31997 vec_revb built-in functions.
31999 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
32001 PR rtl-optimization/78952
32002 * config/i386/i386.md (any_extract): New code iterator.
32003 (*insvqi_2): Use any_extract for source operand.
32004 (*insvqi_3): Use any_shiftrt for source operand.
32006 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
32008 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
32010 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
32012 2017-01-18 Matthias Klose <doko@ubuntu.com>
32014 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
32016 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32018 * config/rs6000/altivec.h (vec_bperm): Change #define.
32019 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
32020 (altivec_vbpermq2): New define_insn.
32021 (altivec_vbpermd): Likewise.
32022 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
32023 function interface.
32024 (VBPERMD): Likewise.
32025 (VBPERM): New polymorphic function interface.
32026 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
32027 Add entries for P9V_BUILTIN_VEC_VBPERM.
32028 * doc/extend.texi: Add interfaces for vec_bperm.
32030 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32032 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
32033 first letter of error messages.
32034 (s390_resolve_overloaded_builtin): Likewise.
32035 * config/s390/s390.c (s390_expand_builtin): Likewise.
32036 (s390_invalid_arg_for_unprototyped_fn): Likewise.
32037 (s390_valid_target_attribute_inner_p): Likewise.
32038 * config/s390/s390.md ("tabort"): Likewise.
32040 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
32042 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
32043 (ISA_AVOID_DIV_HILO): New macro.
32044 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
32045 (ISA_HAS_DDIV): Likewise.
32047 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32049 * doc/invoke.texi (fabi-version): Correct number of occurrences.
32051 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32053 * doc/invoke.texi (fabi-version): Spelling fix.
32055 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32058 * doc/invoke.texi (fabi-version): Mention mangling fix for
32061 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32064 * doc/invoke.texi (fabi-version): Document discriminator mangling.
32066 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
32069 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
32070 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
32072 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
32074 (stack_protect_test): Ditto.
32075 * config/rs6000/rs6000.opt (mstack-protector-guard=,
32076 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
32078 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
32079 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
32080 -mstack-protector-guard-offset=.
32081 (RS/6000 and PowerPC Options): Ditto.
32083 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32085 * config/i386/i386.h (MASK_CLASS_P): New define.
32086 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
32087 there are no registers from different register sets also when
32088 mask registers are used. Update function comment.
32089 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
32090 to (*k/*r) and (*k/*km) alternatives.
32092 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
32094 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
32095 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
32096 (EH_RETURN_HANDLER_RTX): New define.
32097 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
32098 Force frame pointer in EH return functions.
32099 (aarch64_expand_epilogue): Add barrier for eh_return.
32100 (aarch64_final_eh_return_addr): Remove.
32101 (aarch64_eh_return_handler_rtx): New function.
32102 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
32104 (aarch64_eh_return_handler_rtx): New prototype.
32106 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32108 * config/rs6000/altivec.h (vec_rlmi): New #define.
32109 (vec_vrlnm): Likewise.
32110 (vec_rlnm): Likewise.
32111 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
32112 (UNSPEC_VRLNM): Likewise.
32113 (VIlong): New mode iterator.
32114 (altivec_vrl<VI_char>mi): New define_insn.
32115 (altivec_vrl<VI_char>nm): Likewise.
32116 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
32118 (VRLDNM): Likewise.
32119 (RLNM): New polymorphic function entry.
32120 (VRLWMI): New monomorphic function entry.
32121 (VRLDMI): Likewise.
32122 (RLMI): New polymorphic function entry.
32123 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
32124 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
32125 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
32128 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32131 * dwarf2out.c (field_byte_offset): Restore the
32132 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
32133 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
32134 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
32137 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32140 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
32142 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32145 * dwarf2out.c (add_data_member_location_attribute): For constant
32146 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
32147 instead of DW_AT_data_member_location, DW_AT_bit_offset and
32148 DW_AT_byte_size attributes.
32150 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32152 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
32153 after forcing to constant memory when the code model is medium.
32155 2017-01-17 Julia Koval <julia.koval@intel.com>
32158 * config/i386/avx512fintrin.h
32159 (_mm512_i32gather_ps): Change __addr type to void const*.
32160 (_mm512_mask_i32gather_ps): Ditto.
32161 (_mm512_i32gather_pd): Ditto.
32162 (_mm512_mask_i32gather_pd): Ditto.
32163 (_mm512_i64gather_ps): Ditto.
32164 (_mm512_mask_i64gather_ps): Ditto.
32165 (_mm512_i64gather_pd): Ditto.
32166 (_mm512_mask_i64gather_pd): Ditto.
32167 (_mm512_i32gather_epi32): Ditto.
32168 (_mm512_mask_i32gather_epi32): Ditto.
32169 (_mm512_i32gather_epi64): Ditto.
32170 (_mm512_mask_i32gather_epi64): Ditto.
32171 (_mm512_i64gather_epi32): Ditto.
32172 (_mm512_mask_i64gather_epi32): Ditto.
32173 (_mm512_i64gather_epi64): Ditto.
32174 (_mm512_mask_i64gather_epi64): Ditto.
32175 (_mm512_i32scatter_ps): Change __addr type to void*.
32176 (_mm512_mask_i32scatter_ps): Ditto.
32177 (_mm512_i32scatter_pd): Ditto.
32178 (_mm512_mask_i32scatter_pd): Ditto.
32179 (_mm512_i64scatter_ps): Ditto.
32180 (_mm512_mask_i64scatter_ps): Ditto.
32181 (_mm512_i64scatter_pd): Ditto.
32182 (_mm512_mask_i64scatter_pd): Ditto.
32183 (_mm512_i32scatter_epi32): Ditto.
32184 (_mm512_mask_i32scatter_epi32): Ditto.
32185 (_mm512_i32scatter_epi64): Ditto.
32186 (_mm512_mask_i32scatter_epi64): Ditto.
32187 (_mm512_i64scatter_epi32): Ditto.
32188 (_mm512_mask_i64scatter_epi32): Ditto.
32189 (_mm512_i64scatter_epi64): Ditto.
32190 (_mm512_mask_i64scatter_epi64): Ditto.
32191 * config/i386/avx512pfintrin.h
32192 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
32193 (_mm512_mask_prefetch_i32gather_ps): Ditto.
32194 (_mm512_mask_prefetch_i64gather_pd): Ditto.
32195 (_mm512_mask_prefetch_i64gather_ps): Ditto.
32196 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
32197 (_mm512_prefetch_i32scatter_ps): Ditto.
32198 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
32199 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
32200 (_mm512_prefetch_i64scatter_pd): Ditto.
32201 (_mm512_prefetch_i64scatter_ps): Ditto.
32202 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
32203 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
32204 * config/i386/avx512vlintrin.h
32205 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
32206 (_mm_mmask_i32gather_ps): Ditto.
32207 (_mm256_mmask_i32gather_pd): Ditto.
32208 (_mm_mmask_i32gather_pd): Ditto.
32209 (_mm256_mmask_i64gather_ps): Ditto.
32210 (_mm_mmask_i64gather_ps): Ditto.
32211 (_mm256_mmask_i64gather_pd): Ditto.
32212 (_mm_mmask_i64gather_pd): Ditto.
32213 (_mm256_mmask_i32gather_epi32): Ditto.
32214 (_mm_mmask_i32gather_epi32): Ditto.
32215 (_mm256_mmask_i32gather_epi64): Ditto.
32216 (_mm_mmask_i32gather_epi64): Ditto.
32217 (_mm256_mmask_i64gather_epi32): Ditto.
32218 (_mm_mmask_i64gather_epi32): Ditto.
32219 (_mm256_mmask_i64gather_epi64): Ditto.
32220 (_mm_mmask_i64gather_epi64): Ditto.
32221 (_mm256_i32scatter_ps): Change __addr type to void*.
32222 (_mm256_mask_i32scatter_ps): Ditto.
32223 (_mm_i32scatter_ps): Ditto.
32224 (_mm_mask_i32scatter_ps): Ditto.
32225 (_mm256_i32scatter_pd): Ditto.
32226 (_mm256_mask_i32scatter_pd): Ditto.
32227 (_mm_i32scatter_pd): Ditto.
32228 (_mm_mask_i32scatter_pd): Ditto.
32229 (_mm256_i64scatter_ps): Ditto.
32230 (_mm256_mask_i64scatter_ps): Ditto.
32231 (_mm_i64scatter_ps): Ditto.
32232 (_mm_mask_i64scatter_ps): Ditto.
32233 (_mm256_i64scatter_pd): Ditto.
32234 (_mm256_mask_i64scatter_pd): Ditto.
32235 (_mm_i64scatter_pd): Ditto.
32236 (_mm_mask_i64scatter_pd): Ditto.
32237 (_mm256_i32scatter_epi32): Ditto.
32238 (_mm256_mask_i32scatter_epi32): Ditto.
32239 (_mm_i32scatter_epi32): Ditto.
32240 (_mm_mask_i32scatter_epi32): Ditto.
32241 (_mm256_i32scatter_epi64): Ditto.
32242 (_mm256_mask_i32scatter_epi64): Ditto.
32243 (_mm_i32scatter_epi64): Ditto.
32244 (_mm_mask_i32scatter_epi64): Ditto.
32245 (_mm256_i64scatter_epi32): Ditto.
32246 (_mm256_mask_i64scatter_epi32): Ditto.
32247 (_mm_i64scatter_epi32): Ditto.
32248 (_mm_mask_i64scatter_epi32): Ditto.
32249 (_mm256_i64scatter_epi64): Ditto.
32250 (_mm256_mask_i64scatter_epi64): Ditto.
32251 (_mm_i64scatter_epi64): Ditto.
32252 (_mm_mask_i64scatter_epi64): Ditto.
32253 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
32254 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
32255 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
32256 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
32257 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
32258 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
32259 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
32260 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
32261 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
32262 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
32263 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
32264 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
32265 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
32266 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
32267 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
32268 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
32269 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
32270 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
32271 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
32272 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
32273 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
32274 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
32275 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
32276 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
32277 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
32278 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
32279 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
32280 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
32281 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
32282 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
32283 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
32284 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
32285 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
32286 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
32287 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
32288 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
32289 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
32290 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
32291 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
32292 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
32293 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
32294 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
32295 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
32296 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
32297 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
32298 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
32299 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
32300 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
32301 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
32302 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
32303 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
32304 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
32305 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
32306 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
32307 definitions accordingly.
32309 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
32310 Kuan-Lin Chen <kuanlinchentw@gmail.com>
32313 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
32316 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
32319 * ira-conflicts.c (ira_build_conflicts): Update total conflict
32320 hard regs for inner regno.
32322 2017-01-17 Martin Liska <mliska@suse.cz>
32325 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
32326 assumption and add comment.
32328 2017-01-17 Nathan Sidwell <nathan@acm.org>
32330 * ipa-visibility.c (localize_node): New function, broken out of ...
32331 (function_and_variable_visibility): ... here. Call it.
32333 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
32335 PR middle-end/77445
32336 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
32337 correctly set frequency of oudgoing edge.
32338 (duplicate_thread_path): Fix profile updating.
32340 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32343 * configure.ac: Add GCC_BASE_VER.
32344 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
32345 version from BASE-VER file.
32346 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
32347 (gcc.o): Depend on $(BASEVER).
32348 * common.opt (dumpfullversion): New option.
32349 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
32350 * doc/invoke.texi: Document -dumpfullversion.
32351 * doc/install.texi: Document --with-gcc-major-version-only.
32352 * configure: Regenerated.
32354 2017-01-17 Richard Biener <rguenther@suse.de>
32356 PR tree-optimization/71433
32357 * tree-vrp.c (register_new_assert_for): Merge same asserts
32358 on all incoming edges.
32359 (process_assert_insertions_for): Handle insertions at the
32362 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
32364 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
32365 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
32367 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
32370 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
32373 2017-01-17 Alan Modra <amodra@gmail.com>
32376 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
32377 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
32378 symbolic stack limit when pic.
32380 2017-01-16 Martin Sebor <msebor@redhat.com>
32382 PR tree-optimization/78608
32383 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
32385 2017-01-16 Jeff Law <law@redhat.com>
32388 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
32389 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
32390 for several include directories that may be relative to sysroot.
32391 * config/i386/x-mingw32 (gplus_includedir): Define.
32392 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
32393 (native_system_includedir): Likewise.
32394 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
32395 override if TARGET_SYSTEM_ROOT is defined.
32396 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
32398 PR tree-optimization/79090
32399 PR tree-optimization/33562
32400 PR tree-optimization/61912
32401 PR tree-optimization/77485
32402 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
32403 and computed trims into the dump file.
32405 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32407 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
32409 2017-01-16 Jakub Jelinek <jakub@redhat.com>
32412 * gimplify.c (gimplify_init_constructor): If want_value and
32413 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
32417 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
32418 sequence. Formatting fixes.
32419 (doloop_optimize): Formatting fixes.
32422 * gcc.c (debug_level_greater_than_spec_func): New function.
32423 (static_spec_functions): Add debug-level-gt spec function.
32424 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
32426 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32427 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32428 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
32429 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
32430 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
32431 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
32433 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
32435 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
32436 QImode fixups to general and mask registers only.
32438 2017-01-16 Carl Love <cel@us.ibm.com>
32440 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
32441 for built-in functions
32442 vector signed char vec_nabs (vector signed char)
32443 vector signed short vec_nabs (vector signed short)
32444 vector signed int vec_nabs (vector signed int)
32445 vector signed long long vec_nabs (vector signed long long)
32446 vector float vec_nabs (vector float)
32447 vector double vec_nabs (vector double)
32448 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
32450 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
32451 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
32452 * doc/extend.texi: Update the documentation file for the new built-in
32455 2017-01-16 Martin Sebor <msebor@redhat.com>
32457 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
32460 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32462 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
32463 UNSPEC_VSX__XXSPLTD to require special splat handling.
32465 2017-01-16 David Malcolm <dmalcolm@redhat.com>
32468 * system.h: Poison strndup.
32470 2017-01-16 Alan Modra <amodra@gmail.com>
32473 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
32476 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
32478 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
32480 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
32482 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
32483 call recog here. Assert that INSN_CODE (insn) is non-negative.
32485 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32488 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
32490 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
32491 in the currently scheduled RTL fragment.
32493 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32495 PR rtl-optimization/78751
32496 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
32499 2017-01-14 Jeff Law <law@redhat.com>
32501 PR tree-optimization/79090
32502 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
32503 variable length stores.
32504 (compute_trims): Delete dead assignment to *trim_tail.
32505 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
32508 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
32510 PR rtl-optimization/78626
32511 PR rtl-optimization/78727
32512 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
32513 of a block, and split such blocks after everything else is finished.
32515 2017-01-14 Alan Modra <amodra@gmail.com>
32518 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
32519 target legitimate_combined_insn.
32520 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
32521 (rs6000_legitimate_combined_insn): New function.
32522 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
32524 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
32525 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
32526 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
32528 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
32530 * doc/frontends.texi (G++ and GCC): Remove references to Java.
32532 2017-01-13 Jeff Law <law@redhat.com>
32534 PR tree-optimization/33562
32535 PR tree-optimization/61912
32536 PR tree-optimization/77485
32537 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
32539 (delete_dead_assignment): Likewise.
32540 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
32541 statement to delete_dead_call and delete_dead_assignment.
32543 2017-01-13 David Malcolm <dmalcolm@redhat.com>
32546 * substring-locations.c (format_warning_va): Strengthen case 1 so
32547 that both endpoints of the substring must be within the format
32548 range for just the substring to be printed.
32550 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
32552 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
32553 * config/i386/i386.c (ix86_target_string): Add missing options
32554 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
32555 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
32556 flags_other and ix86_target_other to flags2_other. Display unknown
32558 (ix86_valid_target_attribute_inner_p): Add missing options and
32559 reorder options by implied ISAs, as in ix86_target_string.
32561 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
32563 * hash-table.h (hash_table::too_empty_p): New function.
32564 (hash_table::expand): Use it.
32565 (hash_table::traverse): Likewise.
32566 (hash_table::empty_slot): Use sizeof (value_type) instead of
32567 sizeof (PTR) to convert bytes to elements. Shrink the table
32568 if the current size is excessive for the current number of
32571 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
32573 * ira-costs.c (record_reg_classes): Break from the inner loop
32574 early once alt_fail is known to be true. Update outer loop
32575 handling accordingly.
32577 2017-01-13 Jeff Law <law@redhat.com>
32579 * tree-ssa-dse.c (decrement_count): New function.
32580 (increment_start_addr, maybe_trim_memstar_call): Likewise.
32581 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
32582 when we know the partially dead statement is a mem* function.
32584 PR tree-optimization/61912
32585 PR tree-optimization/77485
32586 * tree-ssa-dse.c: Include expr.h.
32587 (maybe_trim_constructor_store): New function.
32588 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
32590 PR tree-optimization/33562
32591 PR tree-optimization/61912
32592 PR tree-optimization/77485
32593 * doc/invoke.texi: Document new dse-max-object-size param.
32594 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
32595 * tree-ssa-dse.c: Include params.h.
32596 (dse_store_status): New enum.
32597 (initialize_ao_ref_for_dse): New, partially extracted from
32599 (valid_ao_ref_for_dse, normalize_ref): New.
32600 (setup_live_bytes_from_ref, compute_trims): Likewise.
32601 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
32602 (maybe_trim_partially_dead_store): Likewise.
32603 (maybe_trim_complex_store): Likewise.
32604 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
32605 Track what bytes live from the original store. Return tri-state
32606 for dead, partially dead or live.
32607 (dse_dom_walker): Add constructor, destructor and new private members.
32608 (delete_dead_call, delete_dead_assignment): New extracted from
32610 (dse_optimize_stmt): Make a member of dse_dom_walker.
32611 Use initialize_ao_ref_for_dse.
32613 PR tree-optimization/33562
32614 PR tree-optimization/61912
32615 PR tree-optimization/77485
32616 * sbitmap.h (bitmap_count_bits): Prototype.
32617 (bitmap_clear_range, bitmap_set_range): Likewise.
32618 * sbitmap.c (bitmap_clear_range): New function.
32619 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
32621 2017-01-13 Martin Liska <mliska@suse.cz>
32624 * function.c (set_cfun): Add new argument force.
32625 * function.h (set_cfun): Likewise.
32626 * ipa-inline-transform.c (inline_call): Use the function when
32627 strict alising from is dropped for function we inline to.
32629 2017-01-13 Richard Biener <rguenther@suse.de>
32631 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
32632 for dumping GIMPLE INTEGER_CSTs.
32634 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
32636 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
32637 to 201112L since C++17.
32639 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
32642 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
32643 if -fsanitize=kernel-address is present.
32645 2017-01-13 Richard Biener <rguenther@suse.de>
32647 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
32648 as _Literal ( type ) number in case usual suffixes do not
32649 preserve all information.
32651 2017-01-13 Richard Biener <rguenther@suse.de>
32653 PR tree-optimization/77283
32654 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
32655 and ssa-iterators.h.
32656 (is_feasible_trace): Implement a cost model based on joiner
32659 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
32662 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
32663 char or short to __float128/_Float128 directly.
32665 2017-01-12 Martin Sebor <msebor@redhat.com>
32667 to -Wformat-overflow.
32668 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
32669 (min_bytes_remaining): Same.
32670 (get_string_length): Same.
32671 (format_string): Same.
32672 (format_directive): Same.
32674 (pass_sprintf_length::handle_gimple_call): Same.
32676 2017-01-12 Jakub Jelinek <jakub@redhat.com>
32678 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
32679 info.nowrite calls with no lhs that can't throw. Return bool
32680 whether gsi_remove has been called or not.
32681 (pass_sprintf_length::handle_gimple_call): Return bool whether
32682 try_substitute_return_value called gsi_remove. Formatting fix.
32683 (pass_sprintf_length::execute): Don't use gsi_remove if
32684 handle_gimple_call returned true.
32687 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
32688 be removed due to side-effects, don't remove following barrier nor
32689 turn the successor edge into fallthru edge.
32691 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32694 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
32695 element-reversing loads and stores as not swappable.
32697 2017-01-12 Nathan Sidwell <nathan@acm.org>
32698 Nicolai Stange <nicstange@gmail.com>
32700 * combine.c (try_combine): Don't ignore result of overlap checking
32701 loop. Combine overlap & asm check into single loop.
32703 2017-01-12 Richard Biener <rguenther@suse.de>
32705 * tree-pretty-print.c (dump_generic_node): Provide -gimple
32706 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
32708 2017-01-12 Richard Biener <rguenther@suse.de>
32710 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
32711 and TS_TARGET_OPTION directly derive from TS_BASE.
32712 * tree-core.h (tree_optimization_option): Derive from tree_base.
32713 (tree_target_option): Likewise.
32715 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
32717 * config/i386/i386.c (memory_address_length): Increase len
32718 only when rip_relative_addr_p returns false.
32720 2017-01-11 Julia Koval <julia.koval@intel.com>
32722 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
32723 (OPTION_MASK_ISA_SGX_SET): New.
32724 (ix86_handle_option): Handle OPT_msgx.
32725 * config.gcc: Added sgxintrin.h.
32726 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
32727 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
32728 * config/i386/i386.c (ix86_target_string): Add -msgx.
32730 (ix86_option_override_internal): Handle new options.
32731 (ix86_valid_target_attribute_inner_p): Add sgx.
32732 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
32733 * config/i386/i386.opt: Add msgx.
32734 * config/i386/sgxintrin.h: New file.
32735 * config/i386/x86intrin.h: Add sgxintrin.h.
32737 2017-01-11 Jakub Jelinek <jakub@redhat.com>
32740 * fold-const.c (maybe_nonzero_address): Return 1 for function
32742 (tree_single_nonzero_warnv_p): Don't handle function local objects
32746 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
32749 2017-01-11 David Malcolm <dmalcolm@redhat.com>
32752 * opts.c: Include "spellcheck.h"
32753 (struct string_fragment): New struct.
32754 (struct edit_distance_traits<const string_fragment &>): New
32756 (get_closest_sanitizer_option): New function.
32757 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
32759 2017-01-11 Jakub Jelinek <jakub@redhat.com>
32761 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
32763 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
32764 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
32765 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
32766 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
32767 for initial die_offset if dwarf_split_debug_info.
32768 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
32769 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
32771 (output_skeleton_debug_sections): Formatting fix. Use
32772 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
32773 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
32775 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
32777 * config/arm/cortex-a53.md: Add bypasses for
32778 cortex_a53_r2f_cvt.
32779 (cortex_a53_r2f): Only use for transfers.
32780 (cortex_a53_f2r): Likewise.
32781 (cortex_a53_r2f_cvt): Add reservation for conversions.
32782 (cortex_a53_f2r_cvt): Likewise.
32784 2017-01-11 Tamar Christina <tamar.christina@arm.com>
32786 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
32787 to all inlined functions, change static to extern.
32789 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
32792 * config/arm/arm.c (legitimize_pic_address): Handle reference to
32794 (arm_assemble_integer): Likewise.
32796 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
32798 * config.gcc: Use new awk script to check CPU, FPU and architecture
32799 parameters for --with-... options.
32800 * config/arm/parsecpu.awk: New file
32801 * config/arm/arm-cpus.in: New file.
32802 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
32804 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
32806 * config/arm/t-arm: Update dependency rules.
32807 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
32808 of processing .def files.
32809 * config/arm/genopt.sh: Deleted.
32810 * config/arm/gentune.sh: Deleted.
32811 * config/arm/arm-cores.def: Deleted.
32812 * config/arm/arm-arches.def: Deleted.
32813 * config/arm/arm-fpus.def: Deleted.
32814 * config/arm/arm-tune.md: Regenerated.
32815 * config/arm/arm-tables.opt: Regenerated.
32816 * config/arm/arm-cpu.h: New generated file.
32817 * config/arm/arm-cpu-data.h: New generated file.
32818 * config/arm/arm-cpu-cdata.h: New generated file.
32820 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
32823 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
32825 (input_varpool_node): Unpack dynamically_initialized bit.
32827 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
32829 PR rtl-optimization/79032
32830 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
32831 the alignment of the adjusted memory reference against that of MODE,
32832 instead of the alignment of the original memory reference.
32834 2017-01-11 Martin Jambor <mjambor@suse.cz>
32836 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
32838 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
32839 decorated functions.
32841 2017-01-11 Richard Biener <rguenther@suse.de>
32843 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
32844 set range/nonnull info for PHI results. Do not set it on
32845 stmts marked for removal.
32847 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
32849 * expr.c (store_field): In the bitfield case, fetch the return value
32850 from the registers before applying a single big-endian adjustment.
32851 Always do a final load for a BLKmode value not larger than a word.
32853 2017-01-10 David Malcolm <dmalcolm@redhat.com>
32856 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
32857 that we correctly handle column numbers greater than
32858 LINE_MAP_MAX_COLUMN_NUMBER.
32860 2017-01-10 Martin Sebor <msebor@redhat.com>
32862 PR middle-end/78245
32863 * gimple-ssa-sprintf.c (get_destination_size): Call
32864 {init,fini}object_sizes.
32865 * tree-object-size.c (addr_object_size): Adjust.
32866 (pass_through_call): Adjust.
32867 (pass_object_sizes::execute): Adjust.
32868 * tree-object-size.h (fini_object_sizes): Declare.
32870 2017-01-10 Martin Sebor <msebor@redhat.com>
32872 PR tree-optimization/78775
32873 * builtins.c (get_size_range): Move...
32874 * calls.c: ...to here.
32875 (alloc_max_size): Accept zero argument.
32876 (operand_signed_p): Remove.
32877 (maybe_warn_alloc_args_overflow): Call get_size_range.
32878 * calls.h (get_size_range): Declare.
32880 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
32882 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
32883 from TI's devices.csv file as of September 2016.
32884 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
32886 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
32888 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
32889 * doc/invoke.texi: Likewise.
32890 * doc/md.texi: Likewise.
32891 * doc/objc.texi: Likewise.
32893 2017-01-10 Joshua Conner <joshconner@google.com>
32895 * config/arm/fuchsia-elf.h: New file.
32896 * config/fuchsia.h: New file.
32897 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
32898 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
32900 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
32902 2016-01-10 Richard Biener <rguenther@suse.de>
32904 PR tree-optimization/79034
32905 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
32906 Propagate out degenerate PHIs in the joiner.
32908 2017-01-10 Martin Liska <mliska@suse.cz>
32910 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
32911 (sort_congruence_classes_by_decl_uid): Likewise.
32912 (sort_congruence_class_groups_by_decl_uid): Likewise.
32913 (sem_item_optimizer::merge_classes): Sort class, groups in these
32914 classes and members in the groups by DECL_UID of declarations.
32915 This would make merge operations stable.
32917 2017-01-10 Martin Liska <mliska@suse.cz>
32919 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
32920 usage of m_classes_vec.
32921 (sem_item_optimizer::~sem_item_optimizer): Likewise.
32922 (sem_item_optimizer::get_group_by_hash): Likewise.
32923 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
32924 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
32925 (sem_item_optimizer::verify_classes): Likewise.
32926 (sem_item_optimizer::process_cong_reduction): Likewise.
32927 (sem_item_optimizer::dump_cong_classes): Likewise.
32928 (sem_item_optimizer::merge_classes): Likewise.
32929 * ipa-icf.h (congruence_class_hash): Rename from
32930 congruence_class_group_hash. Remove declaration of m_classes_vec.
32932 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
32934 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
32935 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
32936 * config.gcc: Add avx512vpopcntdqintrin.h.
32937 * config/i386/avx512vpopcntdqintrin.h: New.
32938 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
32939 * config/i386/i386-builtin-types.def: Add new types.
32940 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
32941 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
32942 __builtin_ia32_vpopcountq_v8di_mask): New.
32943 * config/i386/i386-c.c (ix86_target_macros_internal): Define
32944 __AVX512VPOPCNTDQ__.
32945 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
32946 (PTA_AVX512VPOPCNTDQ): Define.
32947 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
32948 TARGET_AVX512VPOPCNTDQ_P): Define.
32949 * config/i386/i386.opt: Add mavx512vpopcntdq.
32950 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
32951 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
32953 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32955 PR middle-end/77484
32956 * predict.def (PRED_CALL): Set to 67.
32958 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
32960 * expr.c (store_field): In the bitfield case, if the value comes from
32961 a function call and is of an aggregate type returned in registers, do
32962 not modify the field mode; extract the value in all cases if the mode
32963 is BLKmode and the size is not larger than a word.
32965 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
32968 * config/i386/cpuid.h: Fix undefined behavior.
32970 2017-01-04 Jeff Law <law@redhat.com>
32972 PR tree-optimization/79007
32973 PR tree-optimization/67955
32974 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
32975 conservative for pt.null when flag_non_call_exceptions is on.
32977 2017-01-09 Jakub Jelinek <jakub@redhat.com>
32979 PR translation/79019
32980 PR translation/79020
32981 * params.def (PARAM_INLINE_MIN_SPEEDUP,
32982 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
32983 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
32985 * config/avr/avr.opt (maccumulate-args): Likewise.
32986 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
32987 * common.opt (freport-bug): Likewise.
32988 * cif-code.def (CIF_FINAL_ERROR): Likewise.
32989 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
32990 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
32991 translatable string.
32992 * config/i386/i386.c (function_value_32): Likewise.
32993 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
32994 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
32996 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
32997 * common/config/msp430/msp430-common.c (msp430_handle_option):
32999 * symtab.c (symtab_node::verify_base): Likewise.
33000 * opts.c (set_debug_level): Likewise.
33001 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
33002 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
33003 missing whitespace to translatable strings.
33004 * config/avr/avr.md (bswapsi2): Fix typo in comment.
33005 * config/sh/superh.h: Likewise.
33006 * config/i386/xopintrin.h: Likewise.
33007 * config/i386/znver1.md: Likewise.
33008 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
33009 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
33010 * double-int.h (struct double_int): Likewise.
33011 * double-int.c (div_and_round_double): Likewise.
33012 * wide-int.cc: Likewise.
33013 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
33014 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
33015 * cfgcleanup.c (crossjumps_occured): Renamed to ...
33016 (crossjumps_occurred): ... this.
33017 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
33020 PR tree-optimization/78899
33021 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
33022 returning bool return struct loop *, NULL for failure and the new
33024 (versionable_outer_loop_p): Don't version outer loop if it has
33025 dont_vectorized bit set.
33026 (tree_if_conversion): When versioning outer loop, ensure
33027 tree_if_conversion is performed also on the inner loop of the
33028 non-vectorizable outer loop copy.
33029 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
33030 LOOP_VECTORIZED in inner loop of the scalar outer loop and
33031 prevent vectorization of it.
33032 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
33033 the outer loop vectorization of the non-scalar version is attempted
33034 before vectorization of the inner loop in scalar version. If
33035 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
33036 vectorization of its inner loop.
33037 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
33038 has 2 inner loops, rename also on edges from bb whose single pred
33039 is outer_loop->header. Fix typo in function comment.
33041 2017-01-09 Martin Sebor <msebor@redhat.com>
33044 * asan.c (asan_emit_stack_protection): Increase local buffer size
33045 to avoid snprintf truncation warning.
33047 2017-01-09 Andrew Pinski <apinski@cavium.com>
33049 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
33050 to reference thunderx2t99 for the tuning structure
33051 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
33053 (thunderx2t99_extra_costs): This.
33054 * config/aarch64/aarch64-tune.md: Regenerate.
33055 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
33056 (vulcan_addrcost_table): This.
33057 (vulcan_regmove_cost): Rename to ...
33058 (thunderx2t99_regmove_cost): This.
33059 (vulcan_vector_cost): Rename to ...
33060 (thunderx2t99_vector_cost): this.
33061 (vulcan_branch_cost): Rename to ...
33062 (thunderx2t99_branch_cost): This.
33063 (vulcan_tunings): Rename to ...
33064 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
33065 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
33067 2017-01-09 Martin Jambor <mjambor@suse.cz>
33071 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
33072 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
33073 (propagate_vr_accross_jump_function): Use the above function for all
33074 value range computations for pass-through jump functions and type
33075 converasion from explicit value range values.
33076 (ipcp_propagate_stage): Do not attempt to deduce types of formal
33077 parameters from TYPE_ARG_TYPES.
33078 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
33079 (ipa_write_node_info): Stream type of the actual argument.
33080 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
33082 2017-01-09 Martin Liska <mliska@suse.cz>
33085 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
33086 (lookup_compiler): Do not show error message with have_E.
33088 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33090 PR tree-optimization/78938
33091 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
33092 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
33093 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
33094 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
33097 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33099 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
33102 2017-01-09 Richard Biener <rguenther@suse.de>
33104 PR tree-optimization/78997
33105 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
33106 name condition properly.
33108 2017-01-09 Richard Biener <rguenther@suse.de>
33111 * dwarf2out.c (is_cxx): New overload with context.
33112 (is_naming_typedef_decl): Use it.
33114 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33116 * invoke.texi (Option Summary): Correct spacing in option lists
33117 and add line breaks to fix over-long lines.
33119 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33121 PR middle-end/17660
33123 * extend.texi (Common Variable Attributes): Add xref to GCC
33124 Internals manual to explain mode attribute keywords.
33126 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33129 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
33130 and Preprocessor Options.
33131 (Options for Linking): Document -pthread here....
33132 (RS/6000 and PowerPC Options): ...not here.
33133 (Solaris 2 Options): ...or here.
33134 * doc/cppopts.texi: Document -pthread.
33136 2017-01-08 Martin Sebor <msebor@redhat.com>
33138 PR middle-end/77708
33139 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
33140 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
33141 New member functions.
33142 (format_directive): Used them.
33144 (pass_sprintf_length::handle_gimple_call): Same.
33145 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
33146 to avoid truncation for any argument.
33147 (extract_affine_mul): Same.
33148 * tree.c (get_file_function_name): Same.
33150 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33152 PR middle-end/77484
33153 * predict.def (PRED_INDIR_CALL): Set to 86.
33155 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33157 PR preprocessor/54124
33158 * doc/cppopts.texi: Reformat -d subtable to list the full name
33159 of the options. Add cross-reference to the docs for the general
33160 compiler -d options.
33161 * doc/invoke.texi (Developer Options): Add cross-reference to the
33162 preprocessor-specific -d option documentation.
33164 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33166 PR preprocessor/13498
33167 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
33168 redudant material, and reflect new command-line options.
33169 (System Headers): Likewise.
33171 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33173 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
33174 -isystem, and -idirafter. Copy-edit.
33175 * doc/cppopts.texi: Copy-edit. Remove contradiction about
33176 default for -ftrack-macro-expansion. Delete obsolete and
33177 badly-formatted implementation details about -fdebug-cpp output.
33178 * doc/cppwarnopts.texi: Copy-edit.
33180 2017-01-07 David Malcolm <dmalcolm@redhat.com>
33183 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
33184 that the transition from a max line width >= 1<<10 to narrower
33185 lines works correctly.
33187 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
33189 * doc/options.texi (PerFunction): New.
33190 * opt-functions.awk (switch_flags): Map both Optimization and
33191 PerFunction to CL_OPTIMIZATION.
33192 * opth-gen.awk: Test for PerFunction flag along with
33194 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
33195 it only when the latter is present. Skip those that don't in
33196 the hash function generator.
33197 * common.opt (fvar-tracking): Mark as PerFunction instead of
33199 (fvar-tracking-assignments): Likewise.
33200 (fvar-tracking-assignments-toggle): Likewise.
33201 (fvar-tracking-uninit): Likewise.
33203 2017-01-07 Jakub Jelinek <jakub@redhat.com>
33205 PR translation/79018
33206 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
33209 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
33212 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
33213 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
33214 TARGET_LONG_JUMP_TABLE_OFFSETS.
33215 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
33216 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
33217 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
33218 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
33219 * config/m68k/m68k.md (tablejump expander): Likewise.
33220 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
33221 TARGET_LONG_JUMP_TABLE_OFFSETS.
33222 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
33223 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
33225 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
33226 David Holsgrove <david.holsgrove@xilinx.com>
33228 * common/config/microblaze/microblaze-common.c
33229 (TARGET_EXCEPT_UNWIND_INFO): Remove.
33230 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
33232 * config/microblaze/microblaze.c (microblaze_must_save_register)
33233 (microblaze_expand_epilogue, microblaze_return_addr): Handle
33235 (microblaze_eh_return): New function.
33236 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
33237 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
33238 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
33239 * config/microblaze/microblaze.md (eh_return): New pattern.
33241 2017-01-06 Jakub Jelinek <jakub@redhat.com>
33243 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
33244 GCC_DIAGNOSTIC_STRINGIFY): Define.
33246 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
33248 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33250 * config/arm/arm.md (<mcrr>): New.
33252 * config/arm/arm.c (arm_arch5te): New.
33253 (arm_option_override): Set arm_arch5te.
33254 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
33256 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
33257 (arm_mcrr_qualifiers): ... this. New.
33258 (MRRC_QUALIFIERS): Define to...
33259 (arm_mrrc_qualifiers): ... this. New.
33260 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
33262 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
33263 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
33264 (MRRCI, mrrc, MRRC): New.
33265 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
33266 VUNSPEC_MRRC2): New.
33268 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33270 * config/arm/arm.md (<mcr>): New.
33272 * config/arm/arm.c (arm_coproc_builtin_available): Add
33273 support for mcr, mrc, mcr2 and mrc2.
33274 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
33275 (arm_mcr_qualifiers): ... this. New.
33276 (MRC_QUALIFIERS): Define to ...
33277 (arm_mrc_qualifiers): ... this. New.
33278 (MCR_QUALIFIERS): Define to ...
33279 (arm_mcr_qualifiers): ... this. New.
33280 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
33282 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
33283 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
33284 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
33285 VUNSPEC_MRC2): New.
33287 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33289 * config/arm/arm.md (*ldc): New.
33293 * config/arm/arm.c (arm_coproc_builtin_available): Add
33294 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
33295 (arm_coproc_ldc_stc_legitimate_address): New.
33296 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
33297 'qualifier_const_pointer'.
33298 (LDC_QUALIFIERS): Define to...
33299 (arm_ldc_qualifiers): ... this. New.
33300 (STC_QUALIFIERS): Define to...
33301 (arm_stc_qualifiers): ... this. New.
33302 * config/arm/arm-protos.h
33303 (arm_coproc_ldc_stc_legitimate_address): New.
33304 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
33305 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
33306 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
33307 stc2, stcl, stc2l): New.
33308 * config/arm/constraints.md (Uz): New.
33309 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
33310 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
33311 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
33312 VUNSPEC_STC2L): New.
33314 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33316 * config/arm/arm.md (<cdp>): New.
33317 * config/arm/arm.c (neon_const_bounds): Rename this ...
33318 (arm_const_bounds): ... this.
33319 (arm_coproc_builtin_available): New.
33320 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
33321 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
33322 (CDP_QUALIFIERS): Define to...
33323 (arm_cdp_qualifiers): ... this. New.
33325 (arm_expand_builtin_args): Add case for 6 arguments.
33326 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
33327 (arm_const_bounds): ... this.
33328 (arm_coproc_builtin_available): New.
33329 * config/arm/arm_acle.h (__arm_cdp): New.
33331 * config/arm/arm_acle_builtins.def (cdp): New.
33333 * config/arm/iterators.md (CDPI,CDP,cdp): New.
33334 * config/arm/neon.md: Rename all 'neon_const_bounds' to
33335 'arm_const_bounds'.
33336 * config/arm/types.md (coproc): New.
33337 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
33338 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
33339 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
33340 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
33342 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33344 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
33345 (UBINOP_QUALIFIERS): New.
33347 (acle_builtin_data): New. Change comment.
33348 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
33349 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
33350 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
33351 arm_acle_builtins.def.
33352 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
33353 (arm_init_acle_builtins): New.
33354 (CRC32_BUILTIN): Remove.
33355 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
33356 crc32cb, crc32ch and crc32cw.
33357 (arm_init_crc32_builtins): Remove.
33358 (arm_init_builtins): Use arm_init_acle_builtins rather
33359 than arm_init_crc32_builtins.
33360 (arm_expand_acle_builtin): New.
33361 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
33362 * config/arm/arm_acle_builtins.def: New.
33364 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33366 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
33367 (arm_builtin_datum): ... this.
33368 (arm_init_neon_builtin): Rename to ...
33369 (arm_init_builtin): ... this. Add a new parameters PREFIX
33370 and USE_SIG_IN_NAME.
33371 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
33372 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
33373 'arm_builtin_datum'.
33374 (arm_init_vfp_builtins): Likewise.
33375 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
33376 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
33377 (arm_expand_neon_args): Rename to ...
33378 (arm_expand_builtin_args): ... this. Rename builtin_arg
33379 enum values and differentiate between ARG_BUILTIN_MEMORY
33380 and ARG_BUILTIN_NEON_MEMORY.
33381 (arm_expand_neon_builtin_1): Rename to ...
33382 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
33383 values, arm_expand_builtin_args and add bool parameter NEON.
33384 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
33385 (arm_expand_vfp_builtin): Likewise.
33386 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
33388 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33390 PR middle-end/77484
33391 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
33392 * predict.c (tree_estimate_probability_bb): Reverse direction of
33393 polymorphic call predictor.
33395 2017-01-06 David Malcolm <dmalcolm@redhat.com>
33397 * passes.c (execute_one_pass): Split out pass-skipping logic into...
33398 (determine_pass_name_match): ...this new function and...
33399 (should_skip_pass_p): ...this new function.
33401 2017-01-06 Nathan Sidwell <nathan@acm.org>
33403 * ipa-visibility.c (function_and_variable_visibility): Reformat
33404 comments and long lines. Remove extrneous if.
33405 * symtab.c (symtab_node::make_decl_local): Fix code format.
33406 (symtab_node::set_section_for_node): Fix comment typo.
33408 2017-01-06 Martin Liska <mliska@suse.cz>
33411 * lra-constraints.c: Rename invariant to lra_invariant.
33412 * predict.c (set_even_probabilities): Initialize e to NULL.
33414 2017-01-05 Martin Sebor <msebor@redhat.com>
33416 PR tree-optimization/78910
33417 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
33418 (format_integer): Correct off-by-one error in the handling
33419 of precision with negative numbers in signed conversions..
33421 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
33423 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
33425 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33427 PR tree-optimization/71016
33428 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
33429 factor_out_conditional_conversion. Formatting fix.
33430 (factor_out_conditional_conversion): Add cond_stmt argument.
33431 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
33432 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
33435 2017-01-05 David Malcolm <dmalcolm@redhat.com>
33437 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
33438 read-rtl-function.o, and selftest-rtl.o.
33439 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
33440 (selftest::aarch64_test_loading_full_dump): New function.
33441 (selftest::aarch64_run_selftests): New function.
33442 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
33443 selftest::aarch64_run_selftests.
33444 * config/i386/i386.c
33445 (selftest::ix86_test_loading_dump_fragment_1): New function.
33446 (selftest::ix86_test_loading_call_insn): New function.
33447 (selftest::ix86_test_loading_full_dump): New function.
33448 (selftest::ix86_test_loading_unspec): New function.
33449 (selftest::ix86_run_selftests): Call the new functions.
33450 * emit-rtl.c (maybe_set_max_label_num): New function.
33451 * emit-rtl.h (maybe_set_max_label_num): New decl.
33452 * function.c (instantiate_decls): Guard call to
33453 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
33454 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
33456 * gensupport.c (gen_reader::gen_reader): Pass "false"
33457 for new "compact" param of rtx_reader.
33458 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
33459 rather than an empty string for NULL strings.
33460 * read-md.c: Potentially include config.h rather than bconfig.h.
33461 Wrap include of errors.h with #ifdef GENERATOR_FILE.
33462 (have_error): New global, copied from errors.c.
33463 (md_reader::read_name): Rename to...
33464 (md_reader::read_name_1): ...this, adding "out_loc" param,
33465 and converting "missing name or number" to returning false, rather
33467 (md_reader::read_name): Reimplement in terms of read_name_1.
33468 (md_reader::read_name_or_nil): New function.
33469 (md_reader::read_string): Handle "(nil)" by returning NULL.
33470 (md_reader::md_reader): Add new param "compact".
33471 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
33472 (md_reader::read_file): New method.
33473 * read-md.h (md_reader::md_reader): Add new param "compact".
33474 (md_reader::read_file): New method.
33475 (md_reader::is_compact): New accessor.
33476 (md_reader::read_name): Convert return type from void to file_location.
33477 (md_reader::read_name_or_nil): New decl.
33478 (md_reader::read_name_1): New decl.
33479 (md_reader::m_compact): New field.
33480 (noop_reader::noop_reader): Pass "false" for new "compact" param
33482 (rtx_reader::rtx_reader): Add new "compact" param.
33483 (rtx_reader::read_rtx_operand): Make virtual and convert return
33484 type from void to rtx.
33485 (rtx_reader::read_until): New decl.
33486 (rtx_reader::handle_any_trailing_information): New virtual function.
33487 (rtx_reader::postprocess): New virtual function.
33488 (rtx_reader::finalize_string): New virtual function.
33489 (rtx_reader::m_in_call_function_usage): New field.
33490 (rtx_reader::m_reuse_rtx_by_id): New field.
33491 * read-rtl-function.c: New file.
33492 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
33493 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
33494 (selftest::verify_three_block_rtl_cfg): New decl.
33495 * read-rtl-function.h: New file.
33496 * read-rtl.c: Potentially include config.h rather than bconfig.h.
33497 For host, include function.h, memmodel.h, and emit-rtl.h.
33498 (one_time_initialization): New function.
33499 (struct compact_insn_name): New struct.
33500 (compact_insn_names): New array.
33501 (find_code): Handle insn codes in compact dumps.
33502 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
33503 (bind_subst_iter_and_attr): Likewise.
33504 (add_condition_to_string): Likewise.
33505 (add_condition_to_rtx): Likewise.
33506 (apply_attribute_uses): Likewise.
33507 (add_current_iterators): Likewise.
33508 (apply_iterators): Likewise.
33509 (initialize_iterators): Guard usage of apply_subst_iterator with
33510 #ifdef GENERATOR_FILE.
33511 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
33512 (md_reader::read_mapping): Likewise.
33513 (add_define_attr_for_define_subst): Likewise.
33514 (add_define_subst_attr): Likewise.
33515 (read_subst_mapping): Likewise.
33516 (check_code_iterator): Likewise.
33517 (rtx_reader::read_rtx): Likewise. Move one-time initialization
33519 (one_time_initialization): New function.
33520 (rtx_reader::read_until): New method.
33521 (read_flags): New function.
33522 (parse_reg_note_name): New function.
33523 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
33524 Handle reuse_rtx ids.
33525 Wrap iterator lookup within #ifdef GENERATOR_FILE.
33526 Add parsing support for RTL dumps, mirroring the special-cases in
33527 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
33528 values, and calling handle_any_trailing_information.
33529 (rtx_reader::read_rtx_operand): Convert return type from void
33530 to rtx, returning return_rtx. Handle case 'e'. Call
33531 finalize_string on XSTR and XTMPL fields.
33532 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
33533 "(nil)" values were omitted. Call the postprocess vfunc on the
33535 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
33536 class ctor. Initialize m_in_call_function_usage. Call
33537 one_time_initialization.
33538 * rtl-tests.c (selftest::test_uncond_jump): Call
33539 set_new_first_and_last_insn.
33540 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
33541 * selftest-rtl.c: New file.
33542 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
33543 (selftest::get_insn_by_uid): New decl.
33544 * selftest-run-tests.c (selftest::run_tests): Call
33545 read_rtl_function_c_tests.
33546 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
33547 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
33550 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
33552 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
33553 operands in a special way. Assert that pos+len <= mode precision.
33555 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33557 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
33558 3 argument Alias with unlimited for the negative form.
33559 (fno-vect-cost-model): Removed.
33561 2017-01-05 Martin Liska <mliska@suse.cz>
33563 * hsa-gen.c (gen_hsa_divmod): New function.
33564 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
33566 2017-01-05 Martin Liska <mliska@suse.cz>
33569 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
33572 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33574 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
33575 small constant length operands.
33577 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33579 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
33580 between loop iterations.
33582 2017-01-05 Martin Liska <mliska@suse.cz>
33585 * gimplify.c (gimplify_decl_expr): Compare to
33586 asan_poisoned_variables instread of checking flags.
33587 (gimplify_target_expr): Likewise.
33588 (gimplify_expr): Likewise.
33589 (gimplify_function_tree): Conditionally initialize
33590 asan_poisoned_variables.
33592 2017-01-04 Jeff Law <law@redhat.com>
33594 PR tree-optimizatin/78812
33595 * rtl.h (contains_mem_rtx_p): Prototype.
33596 * ifcvt.c (containts_mem_rtx_p): Move from here to...
33597 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
33598 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
33599 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
33600 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
33602 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33604 * input.c (assert_char_at_range): Default-initialize actual_range.
33606 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33608 * df-scan.c (df_ref_create_structure): Make regno unsigned,
33609 to match the caller.
33611 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33613 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
33614 insns after final jump in test to emit dummy move.
33616 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33618 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
33619 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
33621 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33623 * multiple_target.c (create_dispatcher_calls): Init e_next.
33624 * tree-ssa-loop-split.c (split_loop): Init border.
33625 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
33628 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
33633 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
33634 (altivec_register_operand): Do not return true if the operand
33635 contains a SUBREG mixing SImode and SFmode.
33636 (vsx_register_operand): Likewise.
33637 (vsx_reg_sfsubreg_ok): New predicate.
33638 (vfloat_operand): Do not return true if the operand contains a
33639 SUBREG mixing SImode and SFmode.
33640 (vint_operand): Likewise.
33641 (vlogical_operand): Likewise.
33642 (gpc_reg_operand): Likewise.
33643 (int_reg_operand): Likewise.
33644 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
33645 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
33646 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
33648 (rs6000_emit_move_si_sf_subreg): New helper function.
33649 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
33650 fixup SUBREGs involving SImode and SFmode.
33651 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
33652 numbers for the new peephole2 optimization.
33653 (peephole2 for SFmode unions): New peephole2 to optimize cases in
33654 the GLIBC math library that do AND/IOR/XOR operations on single
33655 precision floating point.
33656 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
33657 target macros to say whether we need to avoid SUBREGs mixing
33659 (TARGET_ALLOW_SF_SUBREG): Likewise.
33660 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
33661 (UNSPEC_SI_FROM_SF): Likewise.
33662 (iorxor): Change spacing.
33663 (and_ior_xor): New iterator for AND, IOR, and XOR.
33664 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
33665 (movdi_from_sf_zero_ext): Likewise.
33666 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
33667 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
33668 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
33669 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
33670 (fms<mode>4): Likewise.
33671 (fnma<mode>4): Likewise.
33672 (fnms<mode>4): Likewise.
33673 (nfma<mode>4): Likewise.
33674 (nfms<mode>4): Likewise.
33676 2017-01-04 Marek Polacek <polacek@redhat.com>
33679 * doc/invoke.texi: Document -Wpointer-compare.
33681 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33683 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
33686 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
33687 descriptions for -gdwarf-5 and emit them as uleb128 instead of
33690 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
33693 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
33694 documentation of the powerpc_popcntb_ok attribute.
33695 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
33696 code to issue warning messages if a requested CPU configuration is
33697 not supported by the binary (assembler and loader) toolchain.
33698 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
33699 made to define a built-in function that has been disabled.
33700 (paired_init_builtins): Add assertion to prevent ICE if attempt is
33701 made to define a built-in function that has been disabled.
33702 (altivec_init_builtins): Add comment explaining why definition
33703 of the DST built-in functions is not preceded by an assertion
33704 check. Add assertions to prevent ICE if attempts are made to
33705 define an altivec predicate or an abs* built-in function that has
33707 (htm_init_builtins): Add comment explaining why definition of the
33708 htm built-in functions is not preceded by an assertion check.
33710 2017-01-04 Jeff Law <law@redhat.com>
33712 PR tree-optimizatin/67955
33713 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
33714 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
33715 the points-to solution does not include pt_null. Use DECL_PT_UID
33718 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
33720 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
33721 Use gen_int_mode instead of gen_lopwart for const_int operands.
33723 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33725 PR tree-optimization/71563
33726 * match.pd: Simplify X << Y into X if Y is known to be 0 or
33727 out of range value - has low bits known to be zero.
33729 2017-01-04 Alan Modra <amodra@gmail.com>
33731 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
33732 * configure: Regenerate.
33733 * config.in: Regenerate.
33735 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33738 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
33739 a substring of the message, but strcmp with the whole message. Ifdef
33740 ENABLE_NLS, translate the message first using dgettext.
33742 2017-01-03 Jeff Law <law@redhat.com>
33744 PR tree-optimizatin/78856
33745 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
33746 (mark_threaded_blocks): Remove code to truncate thread paths that
33747 cross multiple loop headers. Instead invalidate the cached loop
33748 iteration information and handle case of a thread path walking
33749 into an irreducible region.
33751 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
33754 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
33755 assertions. Add support for doing the signbit if the IEEE 128-bit
33756 floating point value is in a GPR.
33757 * config/rs6000/rs6000.md (Fsignbit): Delete.
33758 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
33759 Update the length attribute if the value is in a GPR.
33760 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
33761 the sign or zero extension instruction, since the value is always 0/1.
33762 (signbit<mode>2_dm2): Delete using <Fsignbit>.
33765 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
33766 extracting SImode to a GPR register so that we can generate a
33767 store, limit the vector to be in a traditional Altivec register
33768 for the vextuwrx instruction.
33770 2017-01-03 Ian Lance Taylor <iant@google.com>
33772 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
33774 2017-01-03 Martin Sebor <msebor@redhat.com>
33776 PR tree-optimization/78696
33777 * gimple-ssa-sprintf.c (format_floating): Correct handling of
33778 precision. Use MPFR for %f for greater fidelity. Correct handling
33780 (pass_sprintf_length::compute_format_length): Set width and precision
33781 specified by asrerisk to void_node for vararg functions.
33782 (try_substitute_return_value): Adjust dump output.
33784 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
33786 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
33788 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
33790 * doc/invoke.texi (SPARC options): Document -mlra as the default.
33791 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
33792 -mlra/-mno-lra was passed to the compiler.
33794 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
33796 PR rtl-optimization/65618
33797 * emit-rtl.c (try_split): Move initialization of "before" and
33798 "after" to just before the call to emit_insn_after_setloc.
33800 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
33802 * doc/md.texi (Standard Names): Remove reference to Java frontend.
33804 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
33806 * dwarf2out.c (gen_enumeration_type_die): When
33807 -gno-strict-dwarf, add a DW_AT_encoding attribute.
33809 2017-01-03 Jakub Jelinek <jakub@redhat.com>
33811 PR tree-optimization/78965
33812 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
33813 Change first argument from const call_info & to call_info &. For %n
33814 set info.nowrite to false.
33816 PR middle-end/78901
33817 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
33818 possibly throwing calls.
33820 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
33821 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
33822 and fns handling, rather than in a separate case SSA_NAME.
33824 2017-01-02 Jeff Law <law@redhat.com>
33826 * config/darwin-driver.c (darwin_driver_init): Const-correctness
33827 fixes for first_period and second_period variables.
33829 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
33832 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
33833 (*insvqi_1): New insn pattern.
33834 (*insvqi_1_mem_rex64): Ditto.
33835 (*insvqi_2): Ditto.
33836 (*insvqi_3): Rename from *insvqi.
33838 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
33840 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
33842 * doc/cfg.texi (Edges): Remove reference to Java.
33843 (Maintaining the CFG): Ditto.
33845 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33847 PR middle-end/77674
33848 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
33849 transparent aliases.
33851 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33853 PR middle-end/77484
33854 * predict.def (PRED_CALL): Update hitrate.
33855 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
33856 * predict.c (tree_estimate_probability_bb): Split CALL predictor
33857 into direct/indirect/polymorphic variants.
33859 2017-01-01 Jakub Jelinek <jakub@redhat.com>
33861 Update copyright years.
33863 * gcc.c (process_command): Update copyright notice dates.
33864 * gcov-dump.c (print_version): Ditto.
33865 * gcov.c (print_version): Ditto.
33866 * gcov-tool.c (print_version): Ditto.
33867 * gengtype.c (create_file): Ditto.
33868 * doc/cpp.texi: Bump @copying's copyright year.
33869 * doc/cppinternals.texi: Ditto.
33870 * doc/gcc.texi: Ditto.
33871 * doc/gccint.texi: Ditto.
33872 * doc/gcov.texi: Ditto.
33873 * doc/install.texi: Ditto.
33874 * doc/invoke.texi: Ditto.
33876 Copyright (C) 2017 Free Software Foundation, Inc.
33878 Copying and distribution of this file, with or without modification,
33879 are permitted in any medium without royalty provided the copyright
33880 notice and this notice are preserved.