1 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
4 * stor-layout.c: Include attribs.h.
5 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
6 explicit lookup of "aligned" attribute.
8 2017-11-07 Andrew Waterman <andrew@sifive.com>
10 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New
12 (riscv_expand_block_move): Likewise.
13 gcc/config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
15 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
16 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
17 gcc/config/riscv/riscv.c (riscv_block_move_straight): New
19 (riscv_adjust_block_mem): Likewise.
20 (riscv_block_move_loop): Likewise.
21 (riscv_expand_block_move): Likewise.
22 gcc/config/riscv/riscv.md (movmemsi): New pattern.
24 2017-11-07 Michael Clark <michaeljclark@mac.com>
26 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
27 (MUSL_DYNAMIC_LINKER): Likewise.
29 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
31 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
32 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
35 2017-11-07 Richard Biener <rguenther@suse.de>
37 * match.pd: Fix build.
39 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
40 Jackson Woodruff <jackson.woodruff@arm.com>
42 PR tree-optimization/71026
43 * match.pd: Canonicalize negate in division.
45 2017-11-07 Sudakshina Das <sudi.das@arm.com>
48 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
50 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
52 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
53 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
55 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
57 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
58 non-scalar integral types.
59 * match.pd (negate_expr_p): Handle MINUS_EXPR.
60 (-(A-B), -(~A)): New transformations.
62 2017-11-07 Tom de Vries <tom@codesourcery.com>
64 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
65 semicolon after "do {} while (0)".
66 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
67 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
68 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
69 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
70 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
72 2017-11-07 Tom de Vries <tom@codesourcery.com>
74 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
75 after "do {} while (0)".
76 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
77 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
78 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
79 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
80 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
82 2017-11-07 Tom de Vries <tom@codesourcery.com>
85 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
87 (arm_rtx_costs_internal): Add missing semicolon after
88 HANDLE_NARROW_SHIFT_ARITH call.
90 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
92 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
93 disable isel if it was not set explicitly.
95 2017-11-06 James Bowman <james.bowman@ftdichip.com>
97 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
99 (add_type_attribute) likewise.
101 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
103 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
104 of struct ix86_frame.
105 (ix86_initial_elimination_offset): Likewise.
106 (ix86_expand_split_stack_prologue): Likewise.
108 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
110 * tree-vrp.h (enum value_range_type): Update stale comment.
112 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
113 Alan Hayward <alan.hayward@arm.com>
114 David Sherwood <david.sherwood@arm.com>
116 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
117 (aarch64_expand_vec_perm_const): Take the number of units too.
118 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
119 (aarch64_expand_vec_perm_const): Likewise.
120 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
121 (vec_perm<mode>): Update accordingly.
123 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
124 Alan Hayward <alan.hayward@arm.com>
125 David Sherwood <david.sherwood@arm.com>
127 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
128 Take the number of units too.
129 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
130 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
131 but check for a vector mode before rather than after the call.
132 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
133 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
134 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
135 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
136 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
137 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
138 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
139 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
140 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
141 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
142 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
143 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
144 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
145 (aarch64_sqdmull2_n<mode>): Update accordingly.
147 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
148 Alan Hayward <alan.hayward@arm.com>
149 David Sherwood <david.sherwood@arm.com>
151 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
152 the number of units too.
153 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
154 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
155 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
156 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
157 (vec_store_lanesxi<mode>): Update accordingly.
159 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
160 Alan Hayward <alan.hayward@arm.com>
161 David Sherwood <david.sherwood@arm.com>
163 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
164 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
165 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
166 of units rather than the mode.
167 * config/aarch64/iterators.md (nunits): New mode attribute.
168 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
169 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
170 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
171 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
172 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
173 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
174 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
175 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
176 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
177 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
178 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
179 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
180 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
181 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
182 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
183 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
184 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
185 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
186 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
187 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
188 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
189 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
190 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
191 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
192 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
193 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
194 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
195 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
196 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
197 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
198 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
199 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
200 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
201 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
202 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
203 (aarch64_simd_vec_setv2di): Likewise.
205 2017-11-06 Carl Love <cel@us.ibm.com>
207 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
209 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
210 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
212 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
214 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
215 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
216 Add power 8 macro expansions.
217 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
218 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
219 power 8 instructions. (VSX_XXBR): Add iterator.
221 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
223 * config/arm/arm.md (predicable_short_it): Change default to "no",
224 improve documentation, remove uses that are identical to the default.
225 (enabled_for_depr_it): Rename to enabled_for_short_it.
226 * gcc/config/arm/arm-fixed.md (predicable_short_it): Remove default uses.
227 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
228 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
229 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
230 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
232 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
235 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
236 float128 helper macros, which are no longer used after deleting
237 the old 'q' built-in functions, and moving the round to odd
238 built-in functions to being special built-in functions.
239 (BU_FLOAT128_2): Likewise.
240 (BU_FLOAT128_1_HW): Likewise.
241 (BU_FLOAT128_2_HW): Likewise.
242 (BU_FLOAT128_3_HW): Likewise.
243 (FABSQ): Delete old 'q' built-in functions.
244 (COPYSIGNQ): Likewise.
245 (SQRTF128_ODD): Move round to odd built-in functions to be
246 special built-in functions, so that we can handle
247 -mabi=ieeelongdouble.
248 (TRUNCF128_ODD): Likewise.
249 (ADDF128_ODD): Likewise.
250 (SUBF128_ODD): Likewise.
251 (MULF128_ODD): Likewise.
252 (DIVF128_ODD): Likewise.
253 (FMAF128_ODD): Likewise.
254 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
255 built-in names to 'f128'.
256 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
257 old 'q' built-in functions, as the machine independent code for
258 'f128' built-in functions handles this.
259 (rs6000_expand_builtin): Add expansion for float128 round to odd
260 functions, keying off on -mabi=ieeelongdouble of whether to use
261 the KFmode or TFmode variant.
262 (rs6000_init_builtins): Initialize the _Float128 round to odd
264 * doc/extend.texi (PowerPC Built-in Functions): Document the old
265 _Float128 'q' built-in functions are now mapped into the new
266 'f128' built-in functions.
268 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
270 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
271 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
273 2017-11-06 David Malcolm <dmalcolm@redhat.com>
276 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
277 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
278 * toplev.c: Include "ipa-fnsummary.h".
279 (toplev::finalize): Call ipa_fnsummary_c_finalize.
281 2017-11-06 Jakub Jelinek <jakub@redhat.com>
283 PR tree-optimization/82838
284 * gimple-ssa-store-merging.c
285 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
286 on a separate gimple_seq which is then appended to seq.
288 2017-11-06 Jeff Law <law@redhat.com>
291 * config/i386/i386.c (PROBE_INTERVAL): Remove.
292 (get_probe_interval): New functions.
293 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
294 (ix86_adjust_stack_and_probe): Likewise.
295 (output_adjust_stack_and_probe): Likewise.
296 (ix86_emit_probe_stack_range): Likewise.
297 (ix86_expand_prologue): Likewise.
299 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
301 PR tree-optimization/82816
302 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
303 if the modes of the two types are the same.
304 (convert_plusminus_to_widen): Likewise.
306 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
308 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
310 (usadv16qi): New define_expand.
311 (usadv8hi): New define_expand.
313 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
316 * ipa-inline-transform.c (update_noncloned_frequencies): Always
318 (inline_transform): Likewise.
319 * predict.c (counts_to_freqs): Remove useless conditional.
320 * profile-count.h (profile_count::apply_scale): Move sanity check.
321 * tree-inline.c (copy_bb): Always scale.
322 (copy_cfg_body): Likewise.
324 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
327 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
329 (*cmp_ite0): Add enabled_for_depr_it attribute.
330 (*cmp_ite1): Likewise.
332 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
334 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
337 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
339 * tree-vrp.c (vrp_int_const_binop): Return true on success and
340 return the value by pointer.
341 (extract_range_from_multiplicative_op_1): Update accordingly.
342 Return as soon as an operation fails.
344 2017-11-05 Tom de Vries <tom@codesourcery.com>
347 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
348 (DEF_SANITIZER_BUILTIN): ... here.
349 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
350 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
352 2017-11-05 Tom de Vries <tom@codesourcery.com>
355 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
357 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
358 ASM_OUTPUT_BEFORE_CASE_LABEL call.
359 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
361 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
362 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
363 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
365 2017-11-05 Tom de Vries <tom@codesourcery.com>
368 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
371 2017-11-04 Michael Clark <michaeljclark@mac.com>
373 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
374 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
376 (*addsi3_extended): Likewise.
377 (*addsi3_extended2): Likewise.
378 (<optab>si3): Likewise.
379 (<optab>di3): Likewise.
380 (<optab><mode>3): Likewise.
381 (<*optabe>si3_internal): Likewise.
382 (zero_extendqi<SUPERQI:mode>2): Likewise.
383 (*add<mode>hi3): Likewise.
384 (*xor<mode>hi3): Likewise.
385 (<optab>di3): Likewise.
386 (*<optab>si3_extend): Likewise.
387 (*sge<u>_<X:mode><GPR:mode>): Likewise.
388 (*slt<u>_<X:mode><GPR:mode>): Likewise.
389 (*sle<u>_<X:mode><GPR:mode>): Likewise.
391 2017-11-04 Andrew Waterman <andrew@sifive.com>
393 * config/riscv/riscv.c (riscv_option_override): Conditionally set
394 TARGET_STRICT_ALIGN based upon -mtune argument.
396 2017-11-04 Andrew Waterman <andrew@sifive.com>
398 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
400 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
402 config/i386/i386.c (choose_basereg): Use optional scratch
403 register and add assertion.
404 (x86_emit_outlined_ms2sysv_save): Use scratch register when
405 needed, and don't allocate stack.
406 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
407 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
408 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
410 2017-11-03 Jeff Law <law@redhat.com>
412 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
413 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
414 to probe at the start of a noreturn function.
416 2017-11-03 Jakub Jelinek <jakub@redhat.com>
418 PR tree-optimization/78821
419 * gimple-ssa-store-merging.c: Update the file comment.
420 (MAX_STORE_ALIAS_CHECKS): Define.
421 (struct store_operand_info): New type.
422 (store_operand_info::store_operand_info): New constructor.
423 (struct store_immediate_info): Add rhs_code and ops data members.
424 (store_immediate_info::store_immediate_info): Add rhscode, op0r
425 and op1r arguments to the ctor, initialize corresponding data members.
426 (struct merged_store_group): Add load_align_base and load_align
428 (merged_store_group::merged_store_group): Initialize them.
429 (merged_store_group::do_merge): Update them.
430 (merged_store_group::apply_stores): Pick the constant for
431 encode_tree_to_bitpos from one of the two operands, or skip
432 encode_tree_to_bitpos if neither operand is a constant.
433 (class pass_store_merging): Add process_store method decl. Remove
434 bool argument from terminate_all_aliasing_chains method decl.
435 (pass_store_merging::terminate_all_aliasing_chains): Remove
436 var_offset_p argument and corresponding handling.
437 (stmts_may_clobber_ref_p): New function.
438 (compatible_load_p): New function.
439 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
440 if there is overlap and rhs_code is not INTEGER_CST. For
441 non-overlapping stores terminate group if rhs is not mergeable.
442 (get_alias_type_for_stmts): Change first argument from
443 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
444 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
445 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
447 (get_location_for_stmts): Change first argument from
448 auto_vec<gimple *> & to vec<gimple *> &.
449 (struct split_store): Remove orig_stmts data member, add orig_stores.
450 (split_store::split_store): Create orig_stores rather than orig_stmts.
451 (find_constituent_stmts): Renamed to ...
452 (find_constituent_stores): ... this. Change second argument from
453 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
454 to info structures rather than the statements.
455 (split_group): Rename ALLOW_UNALIGNED argument to
456 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
457 it. Adjust find_constituent_stores caller.
458 (imm_store_chain_info::output_merged_store): Handle rhs_code other
459 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
460 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
461 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
462 (mem_valid_for_store_merging): New function.
463 (handled_load): New function.
464 (pass_store_merging::process_store): New method.
465 (pass_store_merging::execute): Use process_store method. Adjust
466 terminate_all_aliasing_chains caller.
468 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
470 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
471 Return true for more constants, symbols and label references.
472 (aarch64_valid_floating_const): Remove unused function.
474 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
477 * config/aarch64/aarch64.c (aarch64_layout_frame):
478 Undo forcing of LR at bottom of frame.
480 2017-11-03 Jeff Law <law@redhat.com>
483 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
484 for int_registers_saved.
486 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
487 extracted from tree-ssa-dom.c.
488 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
489 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
490 (record_equivalences_from_incoming_edge): Add additional argument
491 to single_pred_edge_ignoring_loop_edges call.
492 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
493 (uncprop_dom_walker::before_dom_children): Add additional argument
494 to single_pred_edge_ignoring_loop_edges call.
495 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
496 single_pred_edge_ignoring_loop_edges rather than open coding.
497 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
499 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
501 * match.pd (-(-A)): Rewrite.
503 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
505 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
506 (rs6000_emit_int_cmove): New declaration.
507 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
508 (rs6000_emit_sISEL): Delete.
509 (rs6000_emit_int_cmove): Make non-static.
510 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
511 instead of rs6000_emit_sISEL.
513 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
515 * asan.c (create_cond_insert_point): Maintain profile.
516 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
518 * basic-block.h (struct basic_block_def): Remove frequency.
519 (EDGE_FREQUENCY): Use to_frequency
520 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
522 (find_traces): Update to use to_frequency.
523 (find_traces_1_round): Likewise; use only IPA counts.
524 (bb_to_key): Likewise.
525 (connect_traces): Use IPA counts only.
526 (copy_bb_p): Update to use to_frequency.
527 (fix_up_crossing_landing_pad): Likewise.
528 (sanitize_hot_paths): Likewise.
529 * bt-load.c (basic_block_freq): Likewise.
530 * cfg.c (init_flow): Set count_max to uninitialized.
531 (check_bb_profile): Remove frequencies; check counts.
532 (dump_bb_info): Do not dump frequencies.
533 (update_bb_profile_for_threading): Update counts only.
534 (scale_bbs_frequencies_int): Likewise.
535 (MAX_SAFE_MULTIPLIER): Remove.
536 (scale_bbs_frequencies_gcov_type): Update counts only.
537 (scale_bbs_frequencies_profile_count): Update counts only.
538 (scale_bbs_frequencies): Update counts only.
539 * cfg.h (struct control_flow_graph): Add count-max.
540 (update_bb_profile_for_threading): Update prototype.
541 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
542 (find_many_sub_basic_blocks): Likewise.
543 * cfgcleanup.c (try_forward_edges): Likewise.
544 (try_crossjump_to_edge): Likewise.
545 * cfgexpand.c (expand_gimple_cond): Likewise.
546 (expand_gimple_tailcall): Likewise.
547 (construct_init_block): Likewise.
548 (construct_exit_block): Likewise.
549 * cfghooks.c (verify_flow_info): Check consistency of counts.
550 (dump_bb_for_graph): Do not dump frequencies.
551 (split_block_1): Do not update frequencies.
552 (split_edge): Do not update frequencies.
553 (make_forwarder_block): Do not update frequencies.
554 (duplicate_block): Do not update frequencies.
555 (account_profile_record): Do not update frequencies.
556 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
557 for global heuristics.
558 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
559 (expected_loop_iterations_unbounded): Use counts only.
560 * cfgloopmanip.c (scale_loop_profile): Simplify.
561 (create_empty_loop_on_edge): Simplify
563 (duplicate_loop_to_header_edge): Simplify
564 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
565 (update_br_prob_note): Take care of removing note when profile
567 (relink_block_chain): Do not dump frequency.
568 (rtl_account_profile_record): Use to_frequency.
569 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
570 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
571 (cgraph_update_edges_for_call_stmt_node): Likewise.
572 (cgraph_edge::verify_count_and_frequency): Update.
573 (cgraph_node::verify_node): Temporarily disable frequency verification.
574 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
576 (cgraph_edge::rebuild_edges): Convert to ipa counts.
577 * cgraphunit.c (init_lowered_empty_function): Do not initialize
579 (cgraph_node::expand_thunk): Update profile.
580 * except.c (dw2_build_landing_pads): Do not update frequency.
581 * final.c (compute_alignments): Use to_frequency.
582 (dump_basic_block_info): Do not dump frequency.
583 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
584 (dump_gimple_bb_header): Do not dump frequency.
585 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
587 * gimple-streamer-in.c (input_bb): Do not stream frequency.
588 * gimple-streamer-out.c (output_bb): Do not stream frequency.
589 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
590 (init_before_recovery): Do not update frequency.
591 (sched_create_recovery_edges): Do not update frequency.
592 * hsa-gen.c (convert_switch_statements): Do not update frequency.
593 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
594 (ipa_cp_c_finalize): Set max_count to uninitialized.
595 * ipa-fnsummary.c (get_minimal_bb): Use counts.
596 (param_change_prob): Use counts.
597 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
599 * ipa-split.c (consider_split): Use to_frequency.
600 (split_function): Use to_frequency.
601 * ira-build.c (loop_compare_func): Likewise.
602 (mark_loops_for_removal): Likewise.
603 (mark_all_loops_for_removal): Likewise.
604 * loop-doloop.c (doloop_modify): Do not update frequency.
605 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
607 * lto-streamer-in.c (input_function): Update count_max.
608 * omp-expand.c (expand_omp_taskreg): Update count_max.
609 * omp-simd-clone.c (simd_clone_adjust): Update profile.
610 * predict.c (maybe_hot_frequency_p): Use to_frequency.
611 (maybe_hot_count_p): Use ipa counts only.
612 (maybe_hot_bb_p): Simplify.
613 (maybe_hot_edge_p): Simplify.
614 (probably_never_executed): Do not take frequency argument.
615 (probably_never_executed_bb_p): Do not pass frequency.
616 (probably_never_executed_edge_p): Likewise.
617 (combine_predictions_for_bb): Check that profile is nonzero.
618 (propagate_freq): Do not set frequency.
619 (drop_profile): Simplify.
620 (counts_to_freqs): Simplify.
621 (expensive_function_p): Use to_frequency.
622 (propagate_unlikely_bbs_forward): Simplify.
623 (determine_unlikely_bbs): Simplify.
624 (estimate_bb_frequencies): Add hack to silence graphite issues.
625 (compute_function_frequency): Use ipa counts.
626 (pass_profile::execute): Update.
627 (rebuild_frequencies): Use counts only.
628 (force_edge_cold): Use counts only.
629 * profile-count.c (profile_count::dump): Dump new count types.
630 (profile_count::differs_from_p): Check compatiblity.
631 (profile_count::to_frequency): New function.
632 (profile_count::to_cgraph_frequency): New function.
633 * profile-count.h (struct function): Declare.
634 (enum profile_quality): Add profile_guessed_local and
635 profile_guessed_global0.
636 (class profile_proability): Decrease number of bits to 29;
637 update from_reg_br_prob_note and to_reg_br_prob_note.
638 (class profile_count: Update comment; decrease number of bits
639 to 61. Check compatibility.
640 (profile_count::compatible_p): New private member function.
641 (profile_count::ipa_p): New member function.
642 (profile_count::operator<): Handle global zero correctly.
643 (profile_count::operator>): Handle global zero correctly.
644 (profile_count::operator<=): Handle global zero correctly.
645 (profile_count::operator>=): Handle global zero correctly.
646 (profile_count::nonzero_p): New member function.
647 (profile_count::force_nonzero): New member function.
648 (profile_count::max): New member function.
649 (profile_count::apply_scale): Handle IPA scalling.
650 (profile_count::guessed_local): New member function.
651 (profile_count::global0): New member function.
652 (profile_count::ipa): New member function.
653 (profile_count::to_frequency): Declare.
654 (profile_count::to_cgraph_frequency): Declare.
655 * profile.c (OVERLAP_BASE): Delete.
656 (compute_frequency_overlap): Delete.
657 (compute_branch_probabilities): Do not use compute_frequency_overlap.
658 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
659 * sched-ebb.c (rank): Use counts only.
660 * shrink-wrap.c (handle_simple_exit): Use counts only.
661 (try_shrink_wrapping): Use counts only.
662 (place_prologue_for_one_component): Use counts only.
663 * tracer.c (find_best_predecessor): Use to_frequency.
664 (find_trace): Use to_frequency.
665 (tail_duplicate): Use to_frequency.
666 * trans-mem.c (expand_transaction): Do not update frequency.
667 * tree-call-cdce.c: Do not update frequency.
668 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
669 (gimple_merge_blocks): Likewise.
670 (gimple_split_edge): Likewise.
671 (gimple_duplicate_sese_region): Likewise.
672 (gimple_duplicate_sese_tail): Likewise.
673 (move_sese_region_to_fn): Likewise.
674 (gimple_account_profile_record): Likewise.
675 (insert_cond_bb): Likewise.
676 * tree-complex.c (expand_complex_div_wide): Likewise.
677 * tree-eh.c (lower_resx): Update profile.
678 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
680 (initialize_cfun): Do not initialize frequencies
681 (freqs_to_counts): Delete.
682 (copy_cfg_body): Ignore count parameter.
684 (expand_call_inline): Update count_max.
685 (optimize_inline_calls): Update count_max.
686 (tree_function_versioning): Update count_max.
687 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
688 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
690 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
691 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
692 (try_peel_loop): Likewise.
693 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
695 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
696 (tree_transform_and_unroll_loop): Do not use frequencies
697 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
698 Use reliable prediction only.
699 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
700 * tree-ssa-sink.c (select_best_block): Use to_frequency.
701 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
703 * tree-ssa-threadupdate.c (create_block_for_threading): Do
705 (any_remaining_duplicated_blocks): Likewise.
706 (update_profile): Likewise.
707 (estimated_freqs_path): Delete.
708 (freqs_to_counts_path): Delete.
709 (clear_counts_path): Delete.
710 (ssa_fix_duplicate_block_edges): Likewise.
711 (duplicate_thread_path): Likewise.
712 * tree-switch-conversion.c (gen_inbound_check): Use counts.
713 * tree-tailcall.c (decrease_profile): Do not update frequency.
714 (eliminate_tail_call): Likewise.
715 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
716 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
717 (optimize_mask_stores): Likewise.
718 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
719 * ubsan.c (ubsan_expand_null_ifn): Update profile.
720 (ubsan_expand_ptr_ifn): Update profile.
721 * value-prof.c (gimple_ic): Simplify.
722 * value-prof.h (gimple_ic): Update prototype.
723 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
724 * ipa-inline.c (compute_uninlined_call_time): Be sure that
726 (want_inline_self_recursive_call_p): Likewise.
727 (resolve_noninline_speculation): Only cummulate defined counts.
728 (inline_small_functions): Use nonzero_p.
729 (ipa_inline): Do not access freed node.
731 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
733 * config/aarch64/aarch64.c (aarch64_override_options_internal):
734 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
736 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
738 * config/riscv/riscv.c (riscv_legitimize_move): Handle
739 non-legitimate address.
741 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
743 * config/rs6000/rs6000.md (*lt0_disi): Delete.
744 (*lt0_<mode>di, *lt0_<mode>si): New.
746 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
748 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
751 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
752 Jim Wilson <jim.wilson@linaro.org>
754 * config/aarch64/aarch64-cores.def (saphira): New CPU.
755 * config/aarch64/aarch64-tune.md: Regenerated.
756 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
757 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
759 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
761 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
762 (arc_expand_prologue): Restore blink for millicode.
763 * config/arc/linux.h (LINK_EH_SPEC): Defined.
765 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
768 * config/i386/i386.c (ix86_vector_duplicate_value): Use
769 gen_vec_duplicate after forcing the scalar into a register.
771 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
773 * combine (try_combine): Print the insns input to try_combine to the
776 2017-11-02 Steve Ellcey <sellcey@cavium.com>
779 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
780 Remove second argument from aarch64_process_target_attr call.
781 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
783 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
785 (aarch64_handle_attr_arch): Remove second argument.
786 (aarch64_handle_attr_cpu): Ditto.
787 (aarch64_handle_attr_tune): Ditto.
788 (aarch64_handle_attr_isa_flags): Ditto.
789 (aarch64_process_one_target_attr): Ditto.
790 (aarch64_process_target_attr): Ditto.
791 (aarch64_option_valid_attribute_p): Remove second argument.
792 on aarch64_process_target_attr call.
794 2017-11-02 David Malcolm <dmalcolm@redhat.com>
796 * diagnostic.c: Include "selftest-diagnostic.h".
797 (selftest::assert_location_text): New function.
798 (selftest::test_diagnostic_get_location_text): New function.
799 (selftest::diagnostic_c_tests): Call it.
801 2017-11-02 David Malcolm <dmalcolm@redhat.com>
803 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
804 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
805 (class selftest::test_diagnostic_context): Move to...
806 * selftest-diagnostic.c: New file.
807 * selftest-diagnostic.h: New file.
809 2017-11-02 James Bowman <james.bowman@ftdichip.com>
811 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
812 offset range for FT32B.
813 * config/ft32/ft32.h: option "mcompress" enables relaxation.
814 * config/ft32/ft32.md: Add TARGET_NOPM.
815 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
816 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
818 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
820 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
822 2017-11-02 Jeff Law <law@redhat.com>
824 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
825 virtual keyword on FINAL OVERRIDE members.
827 * tree-ssa-propagate.h (ssa_propagation_engine): Group
828 virtuals together. Add virtual destructor.
829 (substitute_and_fold_engine): Similarly.
831 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
833 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
835 2017-11-02 Richard Biener <rguenther@suse.de>
837 PR tree-optimization/82795
838 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
840 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
842 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
844 (gcc_GAS_CHECK_FEATURE): Remove.
845 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
846 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
847 * configure: Regenerate.
849 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
851 * config/arc/arc.c (hwloop_optimize): Account for empty
854 2017-11-02 Richard Biener <rguenther@suse.de>
857 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
858 Truncate ARRAY_REF index and element size.
860 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
862 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
864 2017-11-01 Jeff Law <law@redhat.com>
866 * tree-ssa-ccp.c (ccp_folder): New class derived from
867 substitute_and_fold_engine.
868 (ccp_folder::get_value): New member function.
869 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
870 (ccp_fold_stmt): Remove prototype.
871 (ccp_finalize): Call substitute_and_fold from the ccp_class.
872 * tree-ssa-copy.c (copy_folder): New class derived from
873 substitute_and_fold_engine.
874 (copy_folder::get_value): Renamed from get_value.
875 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
876 * tree-vrp.c (vrp_folder): New class derived from
877 substitute_and_fold_engine.
878 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
879 (vrp_folder::get_value): New member function.
880 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
881 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
882 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
883 provide a class interface to folder/substitute routines.
884 (ssa_prop_fold_stmt_fn): Remove typedef.
885 (ssa_prop_get_value_fn): Likewise.
886 (subsitute_and_fold): Remove prototype.
887 (replace_uses_in): Likewise.
888 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
889 Renamed from replace_uses_in. Call the virtual member function
890 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
891 (substitute_and_fold_dom_walker): Remove initialization of
892 data member entries for calbacks. Add substitute_and_fold_engine
893 member and initialize it.
894 (substitute_and_fold_dom_walker::before_dom_children0: Use the
895 member functions for get_value, replace_phi_args_in c
896 replace_uses_in, and fold_stmt calls.
897 (substitute_and_fold_engine::substitute_and_fold): Renamed from
898 substitute_and_fold. Remove assert. Update ctor call.
900 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
901 (ssa_prop_visit_phi_fn): Likewise.
902 (class ssa_propagation_engine): New class to provide an interface
904 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
906 (ssa_prop_visit_phi): Likewise.
907 (ssa_propagation_engine::simulate_stmt): Moved into class.
908 Call visit_phi/visit_stmt from the class rather than via
909 file scoped static variables.
910 (ssa_propagation_engine::simulate_block): Moved into class.
911 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
912 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
913 set file scoped statics for the visit_stmt/visit_phi callbacks.
914 * tree-complex.c (complex_propagate): New class derived from
915 ssa_propagation_engine.
916 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
917 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
918 (tree_lower_complex): Call ssa_propagate via the complex_propagate
920 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
921 ssa_propagation_engine.
922 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
923 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
924 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
925 * tree-ssa-copy.c (copy_prop): New class derived from
926 ssa_propagation_engine.
927 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
928 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
929 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
930 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
931 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
932 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
933 (execute_vrp): Call ssa_propagate from the vrp_prop class.
935 2017-11-01 Jakub Jelinek <jakub@redhat.com>
937 PR rtl-optimization/82778
938 PR rtl-optimization/82597
939 * compare-elim.c (struct comparison): Add in_a_setter field.
940 (find_comparison_dom_walker::before_dom_children): Remove killed
941 bitmap and df_simulate_find_defs call, instead walk the defs.
942 Compute last_setter and initialize in_a_setter. Merge definitions
943 with first initialization for a few variables.
944 (try_validate_parallel): Use insn_invalid_p instead of
945 recog_memoized. Return insn rather than just the pattern.
946 (try_merge_compare): Fix up comment. Don't uselessly test if
947 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
949 (execute_compare_elim_after_reload): Remove df_chain_add_problem
952 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
953 Alan Hayward <alan.hayward@arm.com>
954 David Sherwood <david.sherwood@arm.com>
956 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
957 aarch64_hard_regno_nregs to get the number of registers
960 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
961 Alan Hayward <alan.hayward@arm.com>
962 David Sherwood <david.sherwood@arm.com>
964 * config/aarch64/constraints.md (Upl): Rename to...
966 * config/aarch64/aarch64.md
967 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
970 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
971 Alan Hayward <alan.hayward@arm.com>
972 David Sherwood <david.sherwood@arm.com>
974 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
975 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
978 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
979 Alan Hayward <alan.hayward@arm.com>
980 David Sherwood <david.sherwood@arm.com>
982 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
983 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
984 (aarch64_evpc_dup): Generate rtl direcly, rather than using
986 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
988 * config/aarch64/iterators.md: Add a comment above the permute
989 unspecs to say that they are generated directly by
990 aarch64_expand_vec_perm_const.
991 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
993 2017-11-01 Nathan Sidwell <nathan@acm.org>
995 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
997 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
999 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
1000 and medany code models, and describe what they do.
1002 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1004 Revert accidental duplicate:
1006 * combine.c (can_change_dest_mode): Reject changes in
1007 REGMODE_NATURAL_SIZE.
1009 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
1011 PR rtl-optimization/64682
1012 PR rtl-optimization/69567
1013 PR rtl-optimization/69737
1014 PR rtl-optimization/82683
1015 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
1016 register mentioned in the note, drop the note, unless it came from I3,
1017 in which case it should go to I3 again.
1019 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1021 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
1022 and return false if not.
1023 (clear_bytes_written_by, live_bytes_read): Update accordingly.
1025 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1027 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
1028 range is known to be empty.
1030 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1031 Alan Hayward <alan.hayward@arm.com>
1032 David Sherwood <david.sherwood@arm.com>
1034 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
1035 and CONST_VECTOR_NUNITS instead of computing the number of units from
1036 the byte sizes of the vector and element.
1037 (simplify_binary_operation_1): Likewise.
1038 (simplify_const_binary_operation): Likewise.
1039 (simplify_ternary_operation): Likewise.
1041 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1042 Alan Hayward <alan.hayward@arm.com>
1043 David Sherwood <david.sherwood@arm.com>
1045 * var-tracking.c (INT_MEM_OFFSET): Replace with...
1046 (int_mem_offset): ...this new function.
1047 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
1048 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
1049 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
1052 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1053 Alan Hayward <alan.hayward@arm.com>
1054 David Sherwood <david.sherwood@arm.com>
1056 * lower-subreg.c (interesting_mode_p): New function.
1057 (compute_costs, find_decomposable_subregs, decompose_register)
1058 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
1059 (resolve_clobber, dump_choices): Use it.
1061 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1062 Alan Hayward <alan.hayward@arm.com>
1063 David Sherwood <david.sherwood@arm.com>
1065 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
1067 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1068 Alan Hayward <alan.hayward@arm.com>
1069 David Sherwood <david.sherwood@arm.com>
1071 * alias.c (find_base_value, find_base_term): Only process integer
1072 truncations. Check the precision rather than the size.
1074 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1075 Alan Hayward <alan.hayward@arm.com>
1076 David Sherwood <david.sherwood@arm.com>
1078 * machmode.h (is_narrower_int_mode): New function
1079 * optabs.c (expand_float, expand_fix): Use it.
1080 * dwarf2out.c (rotate_loc_descriptor): Likewise.
1082 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1083 Alan Hayward <alan.hayward@arm.com>
1084 David Sherwood <david.sherwood@arm.com>
1086 * rtl.h (narrower_subreg_mode): New function.
1087 * ira-color.c (update_costs_from_allocno): Use it.
1089 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1090 Alan Hayward <alan.hayward@arm.com>
1091 David Sherwood <david.sherwood@arm.com>
1093 * optabs-query.h (convert_optab_p): New function, split out from...
1094 (convert_optab_handler): ...here.
1095 (widening_optab_handler): Delete.
1096 (find_widening_optab_handler): Remove permit_non_widening parameter.
1097 (find_widening_optab_handler_and_mode): Likewise. Provide an
1098 override that operates on mode class wrappers.
1099 * optabs-query.c (widening_optab_handler): Delete.
1100 (find_widening_optab_handler_and_mode): Remove permit_non_widening
1101 parameter. Assert that the two modes are the same class and that
1102 the "from" mode is narrower than the "to" mode. Use
1103 convert_optab_handler instead of widening_optab_handler.
1104 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
1105 instead of widening_optab_handler.
1106 * expr.c (expand_expr_real_2): Update calls to
1107 find_widening_optab_handler.
1108 * optabs.c (expand_widen_pattern_expr): Likewise.
1109 (expand_binop_directly): Take the insn_code as a parameter.
1110 (expand_binop): Only call find_widening_optab_handler for
1111 conversion optabs; use optab_handler otherwise. Update calls
1112 to find_widening_optab_handler and expand_binop_directly.
1113 Use convert_optab_handler instead of widening_optab_handler.
1114 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
1115 find_widening_optab_handler and use scalar_mode rather than
1117 (convert_plusminus_to_widen): Likewise.
1119 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1120 Alan Hayward <alan.hayward@arm.com>
1121 David Sherwood <david.sherwood@arm.com>
1123 * machmode.h (fixed_size_mode): New class.
1124 * rtl.h (get_pool_mode): Return fixed_size_mode.
1125 * gengtype.c (main): Add fixed_size_mode.
1126 * target.def (get_raw_result_mode): Return a fixed_size_mode.
1127 (get_raw_arg_mode): Likewise.
1128 * doc/tm.texi: Regenerate.
1129 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
1130 * targhooks.c (default_get_reg_raw_mode): Likewise.
1131 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
1132 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
1133 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
1134 (msp430_get_raw_result_mode): Likewise.
1135 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
1136 * dbxout.c (dbxout_parms): Require fixed-size modes.
1137 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
1138 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
1139 * omp-low.c (lower_oacc_reductions): Likewise.
1140 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
1141 (simplify_subreg): Update accordingly.
1142 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
1143 (force_const_mem): Update accordingly. Return NULL_RTX for modes
1144 that aren't fixed-size.
1145 (get_pool_mode): Return a fixed_size_mode.
1146 (output_constant_pool_2): Take a fixed_size_mode.
1148 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1149 Alan Hayward <alan.hayward@arm.com>
1150 David Sherwood <david.sherwood@arm.com>
1152 * doc/rtl.texi (vec_series): Document.
1153 (const): Say that the operand can be a vec_series.
1154 * rtl.def (VEC_SERIES): New rtx code.
1155 * rtl.h (const_vec_series_p_1): Declare.
1156 (const_vec_series_p): New function.
1157 * emit-rtl.h (gen_const_vec_series): Declare.
1158 (gen_vec_series): Likewise.
1159 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
1160 (gen_vec_series): Likewise.
1161 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
1162 * simplify-rtx.c (simplify_unary_operation): Handle negations
1164 (simplify_binary_operation_series): New function.
1165 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
1166 (test_vector_ops_series): New function.
1167 (test_vector_ops): Call it.
1168 * config/powerpcspe/altivec.md (altivec_lvsl): Use
1169 gen_const_vec_series.
1170 (altivec_lvsr): Likewise.
1171 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
1173 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1174 Alan Hayward <alan.hayward@arm.com>
1175 David Sherwood <david.sherwood@arm.com>
1177 * doc/rtl.texi (const): Update description of address constants.
1178 Say that vector constants are allowed too.
1179 * common.md (E, F): Use CONSTANT_P instead of checking for
1181 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
1182 checking for CONST_VECTOR.
1183 * expmed.c (make_tree): Use build_vector_from_val for a CONST
1185 * expr.c (expand_expr_real_2): Check for vector modes instead
1186 of checking for CONST_VECTOR.
1187 * rtl.h (const_vec_p): New function.
1188 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
1189 (unwrap_const_vec_duplicate): Handle them here too.
1191 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1192 David Malcolm <dmalcolm@redhat.com>
1193 Alan Hayward <alan.hayward@arm.com>
1194 David Sherwood <david.sherwood@arm.com>
1196 * rtl.h (vec_duplicate_p): New function.
1197 * selftest-rtl.c (assert_rtx_eq_at): New function.
1198 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
1199 (assert_rtx_eq_at): Declare.
1200 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
1201 * selftest-run-tests.c (selftest::run_tests): Call it.
1202 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
1203 (simplify_unary_operation_1): Recursively handle vector duplicates.
1204 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
1206 (simplify_subreg): Handle subregs of vector duplicates.
1207 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
1208 (selftest::simplify_rtx_c_tests): New functions.
1210 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1211 Alan Hayward <alan.hayward@arm.com>
1212 David Sherwood <david.sherwood@arm.com>
1214 * emit-rtl.h (gen_const_vec_duplicate): Declare.
1215 (gen_vec_duplicate): Likewise.
1216 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
1218 (gen_const_vector): ...here.
1219 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
1220 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
1221 whose elements are all equal.
1222 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
1223 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
1224 (simplify_relational_operation): Likewise.
1225 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
1227 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
1228 (aarch64_expand_vector_init): Likewise.
1229 * config/arm/arm.c (neon_vdup_constant): Likewise.
1230 (neon_expand_vector_init): Likewise.
1231 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
1232 (arm_block_set_unaligned_vect): Likewise.
1233 (arm_block_set_aligned_vect): Likewise.
1234 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
1235 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
1236 (expand_vec_perm_even_odd_pack): Likewise.
1237 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
1238 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
1239 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
1240 gen_const_vec_duplicate.
1241 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
1242 * config/mips/mips.c (mips_gen_const_int_vector): Use
1243 gen_const_vec_duplicate.
1244 (mips_expand_vector_init): Use CONST0_RTX.
1245 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
1246 (define_split): Use gen_const_vec_duplicate.
1247 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
1248 (define_split): Use gen_const_vec_duplicate.
1249 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
1250 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
1251 * config/spu/spu.c (spu_const): Likewise.
1253 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1254 Alan Hayward <alan.hayward@arm.com>
1255 David Sherwood <david.sherwood@arm.com>
1257 * combine.c (can_change_dest_mode): Reject changes in
1258 REGMODE_NATURAL_SIZE.
1260 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
1262 * configure.ac (--enable-libssp): New.
1263 (gcc_cv_libc_provides_ssp): Check for explicit setting before
1264 trying to determine target-specific default. Adjust indentation.
1265 * configure: Regenerated.
1266 * doc/install.texi (Configuration): Expand --disable-libssp
1269 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
1271 config/i386/i386.c (ix86_expand_epilogue): Correct stack
1274 2017-10-31 Martin Jambor <mjambor@suse.cz>
1277 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
1279 2017-10-31 David Malcolm <dmalcolm@redhat.com>
1281 * auto-profile.c (autofdo_source_profile::read): Use
1282 UNKNOWN_LOCATION rather than 0.
1283 * diagnostic-core.h (warning_at_rich_loc): Rename to...
1284 (warning_at): ...this overload.
1285 (warning_at_rich_loc_n): Rename to...
1286 (warning_n): ...this overload.
1287 (error_at_rich_loc): Rename to...
1288 (error_at): ...this overload.
1289 (pedwarn_at_rich_loc): Rename to...
1290 (pedwarn): ...this overload.
1291 (permerror_at_rich_loc): Rename to...
1292 (permerror): ...this overload.
1293 (inform_at_rich_loc): Rename to...
1294 (inform): ...this overload.
1295 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
1296 (diagnostic_n_impl_richloc): Rename to...
1297 (diagnostic_n_impl): ...this rich_location *-based decl.
1298 (inform_at_rich_loc): Rename to...
1299 (inform): ...this, and add an assertion.
1300 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
1301 (warning_at_rich_loc): Rename to...
1302 (warning_at): ...this, and add an assertion.
1303 (warning_at_rich_loc_n): Rename to...
1304 (warning_n): ...this, and add an assertion.
1305 (warning_n): Update location_t-based implementation for removal of
1306 location_t-based diagnostic_n_impl.
1307 (pedwarn_at_rich_loc): Rename to...
1308 (pedwarn): ...this, and add an assertion.
1309 (permerror_at_rich_loc): Rename to...
1310 (permerror): ...this, and add an assertion.
1311 (error_n): Update for removal of location_t-based diagnostic_n_impl.
1312 (error_at_rich_loc): Rename to...
1313 (error_at): ...this, and add an assertion.
1314 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
1315 (driver::do_spec_on_infiles): Likewise.
1316 * substring-locations.c (format_warning_va): Update for renaming
1317 of inform_at_rich_loc.
1319 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
1321 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
1322 _Float<N>X built-in functions so that the variant without the
1323 "__builtin_" prefix is only enabled for the GNU C and Objective C
1324 languages when they are in non-strict ANSI/ISO mode.
1325 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
1326 * target.def (floatn_builtin_p): Add a target hook to control
1327 whether _Float<N> and _Float<N>X built-in functions without the
1328 "__builtin_" prefix are enabled, and return true for C and
1329 Objective C in the default hook. Include langhooks.h in
1331 * targhooks.h (default_floatn_builtin_p): Likewise.
1332 * targhooks.c (default_floatn_builtin_p): Likewise.
1333 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
1334 floatn_builtin_p target hook.
1335 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
1337 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
1338 Eric Botcazou <ebotcazou@adacore.com>
1340 PR rtl-optimization/81803
1341 * lra-constraints.c (curr_insn_transform): Also reload the whole
1342 register for a strict subreg no wider than a word if this is for
1343 a WORD_REGISTER_OPERATIONS target.
1345 2017-10-31 Jason Merrill <jason@redhat.com>
1347 * gdbinit.in: Skip over inlines from timevar.h.
1349 2017-10-31 Martin Liska <mliska@suse.cz>
1351 * doc/gcov.texi: Document new option.
1352 * gcov.c (print_usage): Likewise print it.
1353 (process_args): Support the argument.
1354 (format_count): New function.
1355 (format_gcov): Use the function.
1357 2017-10-31 Martin Liska <mliska@suse.cz>
1359 * gcov.c (struct name_map): do not use typedef.
1360 Define operator== and operator<.
1361 (name_search): Remove.
1362 (name_sort): Remove.
1363 (main): Do not allocate names.
1364 (process_file): Add vertical space.
1365 (generate_results): Use std::find.
1366 (release_structures): Do not release memory.
1367 (find_source): Use std::find.
1369 2017-10-31 Martin Liska <mliska@suse.cz>
1371 * gcov.c (struct line_info): Remove it's typedef.
1372 (line_info::line_info): Add proper ctor.
1373 (line_info::has_block): Do not use a typedef.
1374 (struct source_info): Do not use typedef.
1375 (circuit): Likewise.
1376 (get_cycles_count): Likewise.
1377 (output_intermediate_file): Iterate via vector iterator.
1378 (add_line_counts): Use std::vector methods.
1379 (accumulate_line_counts): Likewise.
1380 (output_lines): Likewise.
1382 2017-10-31 Martin Liska <mliska@suse.cz>
1384 * gcov.c (struct source_info): Remove typedef.
1385 (source_info::source_info): Add proper ctor.
1386 (accumulate_line_counts): Use struct, not it's typedef.
1387 (output_gcov_file): Likewise.
1388 (output_lines): Likewise.
1389 (main): Do not allocate an array.
1390 (output_intermediate_file): Use size of vector container.
1391 (process_file): Resize the vector.
1392 (generate_results): Do not preallocate, use newly added vector
1394 (release_structures): Do not release sources.
1395 (find_source): Use vector methods.
1396 (add_line_counts): Do not use typedef.
1398 2017-10-31 Martin Liska <mliska@suse.cz>
1400 * doc/gcov.texi: Document that.
1401 * gcov.c (add_line_counts): Mark lines with a non-executed
1403 (output_line_beginning): Handle such lines.
1404 (output_lines): Pass new argument.
1405 (output_intermediate_file): Print it in intermediate format.
1407 2017-10-31 Martin Liska <mliska@suse.cz>
1409 * color-macros.h: New file.
1410 * diagnostic-color.c: Factor out color related to macros to
1412 * doc/gcov.texi: Document -k option.
1413 * gcov.c (INCLUDE_STRING): Include string.h.
1414 (print_usage): Add -k option.
1415 (process_args): Parse it.
1416 (pad_count_string): New function.
1417 (output_line_beginning): Likewise.
1418 (DEFAULT_LINE_START): New macro.
1419 (output_lines): Support color output.
1421 2017-10-31 Martin Liska <mliska@suse.cz>
1423 PR gcov-profile/82633
1424 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
1425 their interaction with GCOV infrastructure.
1426 * configure.ac: Add -fkeep-{inline,static}-functions to
1428 * configure: Regenerate.
1430 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
1433 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
1435 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
1438 * config/rs6000/rs6000.md (allocate_stack): Force update interval
1439 into a register if it does not fit into an immediate offset field.
1441 2017-10-31 Olivier Hainque <hainque@adacore.com>
1443 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
1445 2017-10-31 Julia Koval <julia.koval@intel.com>
1447 * config.gcc: Add gfniintrin.h.
1448 * config/i386/gfniintrin.h: New.
1449 * config/i386/i386-builtin-types.def
1450 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1451 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1452 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1453 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1454 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1455 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
1456 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
1457 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
1458 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
1459 V64QI_FTYPE_V64QI_V64QI_INT): New types.
1460 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1461 * config/i386/immintrin.h: Include gfniintrin.h.
1462 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
1464 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
1466 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
1468 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1470 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
1471 (arm_ashldi3_1bit): Remove pattern.
1472 (ashrdi3): Remove shift by 1 expansion.
1473 (arm_ashrdi3_1bit): Remove pattern.
1474 (lshrdi3): Remove shift by 1 expansion.
1475 (arm_lshrdi3_1bit): Remove pattern.
1476 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
1477 cost of ashldi3 by 1.
1478 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
1479 (<shift>di3_neon): Likewise.
1481 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
1483 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
1484 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
1485 and (*aarch64_simd_mov<VQ:mode>).
1486 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
1487 pattern alternative.
1488 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
1489 attributes to match pattern alternative.
1491 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
1493 * config.gcc (powerpc*-*-*): Add emmintrin.h.
1494 * config/rs6000/emmintrin.h: New file.
1495 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
1497 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1499 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
1500 * (movdi_vfp_cortexa8): Remove pattern.
1502 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1504 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
1506 (Specific, alpha*-dec-osf5.1): Remove.
1507 (Specific, mips-sgi-irix5): Remove.
1508 (Specific, mips-sgi-irix6): Remove.
1510 2017-10-30 Jakub Jelinek <jakub@redhat.com>
1513 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
1514 arguments to clear_bit_region_be.
1516 2017-10-30 Jim Wilson <wilson@tuliptree.org>
1518 * gimplify.c: Include memmodel.h.
1520 2017-10-30 Martin Jambor <mjambor@suse.cz>
1522 * omp-grid.c (grid_attempt_target_gridification): Also insert a
1523 condition whether loop should be executed at all.
1525 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
1527 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1528 gimple folding of vec_madd() intrinsics.
1529 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
1530 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
1531 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
1533 2017-10-30 Richard Biener <rguenther@suse.de>
1535 PR tree-optimization/82762
1537 2017-10-23 Richard Biener <rguenther@suse.de>
1539 PR tree-optimization/82129
1541 2017-08-01 Richard Biener <rguenther@suse.de>
1543 PR tree-optimization/81181
1544 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
1545 (compute_antic): ... end of iteration here.
1547 2017-10-30 Joseph Myers <joseph@codesourcery.com>
1549 * doc/invoke.texi (C Dialect Options): Document -std=c17,
1550 -std=iso9899:2017 and -std=gnu17.
1551 * doc/standards.texi (C Language): Document C17 support.
1552 * doc/cpp.texi (Overview): Mention -std=c17.
1553 (Standard Predefined Macros): Document C11 and C17 values of
1554 __STDC_VERSION__. Do not refer to C99 support as incomplete.
1555 * doc/extend.texi (Inline): Do not list individual options for
1556 standards newer than C99.
1557 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
1559 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
1562 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
1564 * asan.c (asan_finish_file): Align asan globals array by shadow
1567 2017-10-30 Jakub Jelinek <jakub@redhat.com>
1570 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
1571 (struct store_immediate_info): Add bitregion_start and bitregion_end
1573 (store_immediate_info::store_immediate_info): Add brs and bre
1574 arguments and initialize bitregion_{start,end} from those.
1575 (struct merged_store_group): Add bitregion_start, bitregion_end,
1576 align_base and mask fields. Drop unnecessary struct keyword from
1577 struct store_immediate_info. Add do_merge method.
1578 (clear_bit_region_be): Use memset instead of loop storing zeros.
1579 (merged_store_group::do_merge): New method.
1580 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
1581 stores as long as the surrounding bitregions have no gaps.
1582 (merged_store_group::merge_overlapping): Use do_merge.
1583 (merged_store_group::apply_stores): Test that bitregion_{start,end}
1584 is byte aligned, rather than requiring that start and width are
1585 byte aligned. Drop unnecessary struct keyword from
1586 struct store_immediate_info. Allocate and populate also mask array.
1587 Make start of the arrays relative to bitregion_start rather than
1588 start and size them according to bitregion_{end,start} difference.
1589 (struct imm_store_chain_info): Drop unnecessary struct keyword from
1590 struct store_immediate_info.
1591 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
1592 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
1593 struct keyword from struct store_immediate_info.
1594 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
1595 between stores as long as the surrounding bitregions have no gaps.
1597 (struct split_store): Add orig non-static data member.
1598 (split_store::split_store): Initialize orig to false.
1599 (find_constituent_stmts): Return store_immediate_info *, non-NULL
1600 if there is exactly a single original stmt. Change stmts argument
1601 to pointer from reference, if NULL, don't push anything to it. Add
1602 first argument, use it to optimize skipping over orig stmts that
1603 are known to be before bitpos already. Simplify.
1604 (split_group): Return unsigned int count how many stores are or
1605 would be needed rather than a bool. Add allow_unaligned argument.
1606 Change split_stores argument from reference to pointer, if NULL,
1607 only do a dry run computing how many stores would be produced.
1608 Rewritten algorithm to use both alignment and misalign if
1609 !allow_unaligned and handle bitfield stores with gaps.
1610 (imm_store_chain_info::output_merged_store): Set start_byte_pos
1611 from bitregion_start instead of start. Compute allow_unaligned
1612 here, if true, do 2 split_group dry runs to compute which one
1613 produces fewer stores and prefer aligned if equal. Punt if
1614 new count is bigger or equal than original before emitting any
1615 statements, rather than during that. Remove no longer needed
1616 new_ssa_names tracking. Replace num_stmts with
1617 split_stores.length (). Use 32-bit stack allocated entries
1618 in split_stores auto_vec. Try to reuse original store lhs/rhs1
1619 if possible. Handle bitfields with gaps.
1620 (pass_store_merging::execute): Ignore bitsize == 0 stores.
1621 Compute bitregion_{start,end} for the stores and construct
1622 store_immediate_info with that. Formatting fixes.
1624 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
1627 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
1628 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
1630 2017-10-29 Jim Wilson <wilson@tuliptree.org>
1632 * gimplify.c: Include tm_p.h.
1634 * common.opt (gcoff): Re-add as ignored option.
1635 (gcoff1, gcoff2, gcoff3): Likewise.
1637 * Makefile.in (OBJS): Delete sdbout.o.
1638 (GTFILES): Delete $(srcdir)/sdbout.c.
1639 * debug.h: Delete sdb_debug_hooks.
1640 * final.c: Delete sdbout.h include.
1641 (final_scan_insn): Delete SDB_DEBUG check.
1642 (rest_of_clean_state): Likewise.
1643 * output.h: Delete sdb_begin_function_line.
1646 * toplev.c: Delete sdbout.h include.
1647 (process_options): Delete SDB_DEBUG check.
1648 * tree-core.h (tree_type_common): Delete pointer field of
1650 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
1651 TYPE_SYMTAB_POINTER.
1652 * tree.h (TYPE_SYMTAB_POINTER): Delete.
1653 (TYPE_SYMTAB_IS_POINTER): Delete.
1654 (TYPE_SYMTAB_IS_DIE): Renumber.
1655 * xcoffout.c: Refer to former sdbout.c file.
1656 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
1658 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
1659 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
1660 (Debugging Options): Delete -gcoff.
1661 (-gstabs): Delete SDB reference.
1663 (-gcoff@var{level}): Delete.
1664 * doc/passes.texi (Debugging information output): Delete SDB and
1665 sdbout.c references.
1666 * doc/tm.texi: Regenerate.
1667 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
1668 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
1670 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
1671 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
1672 (SDB_DEBUGGING_INFO): Delete.
1673 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
1674 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
1675 * target.def (output_source_filename): Delete COFF reference.
1677 * common.opt (gcoff): Delete.
1678 (gxcoff+): Update Negative chain.
1679 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
1681 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
1682 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
1683 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
1685 (expand_function_start): Change sdb reference to past tense.
1686 (expand_function_end): Change sdb reference to past tense.
1687 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
1688 * opts.c (debug_type_names): Delete coff entry.
1689 (common_handle_option): Delete OPT_gcoff case.
1690 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
1692 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
1693 * config/cris/cris.h: Delete SDB reference in comment.
1694 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
1695 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
1696 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
1697 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
1699 (ix86_expand_prologue): Likewise.
1700 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
1701 * config/ia64/ia64.h: Likewise.
1702 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
1703 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
1705 * config/mmix/mmix.h: Likewise.
1706 * config/nds32/nds32.c: Likewise.
1707 * config/stormy/storym16.h: Likewise.
1708 * config/visium/visium.h: Likewise.
1709 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
1711 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
1713 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
1714 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
1715 FRAME_POINTER_REGNUM point at high end of local var area.
1717 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
1719 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
1720 Move comment around. Do not reset best_edge for a copiable
1721 destination if the copy would cause a partition change.
1722 (better_edge_p): Remove redundant check.
1724 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
1726 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
1728 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
1730 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
1731 for math functions that have _Float<N> and _Float<N>X variants.
1732 (mathfn_built_in_2): Add support for math functions that have
1733 _Float<N> and _Float<N>X variants.
1734 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
1735 (expand_builtin_mathfn_ternary): Add support for fma with
1736 _Float<N> and _Float<N>X variants.
1737 (expand_builtin): Likewise.
1738 (fold_builtin_3): Likewise.
1739 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
1740 create math function _Float<N> and _Float<N>X variants as external
1742 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
1743 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
1744 the __builtin_ prefix and if not strict ansi, without the prefix.
1745 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
1746 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
1747 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
1748 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
1749 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
1750 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
1751 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
1752 function signatures for fma _Float<N> and _Float<N>X variants.
1753 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
1754 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
1755 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
1756 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
1757 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
1758 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
1759 * gencfn-macros.c (print_case_cfn): Add support for math functions
1760 that have _Float<N> and _Float<N>X variants.
1761 (print_define_operator_list): Likewise.
1762 (fltfn_suffixes): Likewise.
1764 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
1765 for math functions that have _Float<N> and _Float<N>X variants.
1766 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
1767 and _Float<N>X variants.
1768 (COPYSIGN): Likewise.
1771 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
1772 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
1774 (integer_valued_read_call_p): Likewise.
1775 * fold-const-call.c (fold_const_call_ss): Likewise.
1776 (fold_const_call_sss): Add support for copysign, fmin, and fmax
1777 _Float<N> and _Float<N>X variants.
1778 (fold_const_call_ssss): Add support for fma _Float<N> and
1779 _Float<N>X variants.
1780 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
1781 support for copysign and fma _Float<N> and _Float<N>X variants.
1782 (backprop::process_builtin_call_use): Likewise.
1783 * tree-call-cdce.c (can_test_argument_range); Add support for
1784 sqrt _Float<N> and _Float<N>X variants.
1785 (edom_only_function): Likewise.
1786 (get_no_error_domain): Likewise.
1787 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
1788 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
1789 copysign _Float<N> and _Float<N>X variants.
1790 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
1791 handled by machine independent code.
1792 (FMAF128): Likewise.
1793 * doc/cpp.texi (Common Predefined Macros): Document defining
1794 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
1795 fma _Float<N> and _Float<N>X variants.
1797 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
1800 * config/i386/i386-modes.def (CCFPU): Remove definition.
1801 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
1802 (ix86_cc_modes_compatible): Ditto.
1803 (ix86_expand_carry_flag_compare): Ditto.
1804 (ix86_expand_int_movcc): Ditto.
1805 (ix86_expand_int_addcc): Ditto.
1806 (ix86_reverse_condition): Ditto.
1807 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
1808 Return true/false for unordered/ordered fp comparisons.
1809 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
1810 (ix86_prepare_fp_compare_args): Update for rename.
1811 (ix86_expand_fp_compare): Update for rename. Generate unordered
1812 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
1813 (ix86_expand_sse_compare_and_jump): Ditto.
1814 * config/i386/predicates.md (fcmov_comparison_operator):
1815 Remove CCFPU mode handling.
1816 (ix86_comparison_operator): Ditto.
1817 (ix86_carry_flag_operator): Ditto.
1818 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
1819 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
1820 (*cmpu<mode>_cc_i387): Ditto.
1821 (FPCMP): Remove mode iterator.
1822 (unord): Remove mode attribute.
1823 (unord_subst): New define_subst transformation
1824 (unord): New define_subst attribute.
1826 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
1827 (*cmpi<unord>xf_i387): Ditto.
1828 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
1829 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
1830 using unord_subst transformation.
1831 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
1832 (round_saeonly): Also handle CCFP mode.
1833 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
1834 Remove UNSPEC_SAHF unspec handling.
1836 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
1838 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
1840 2017-10-27 Jeff Law <law@redhat.com>
1842 * gimple-ssa-sprintf.c: Include domwalk.h.
1843 (class sprintf_dom_walker): New class, derived from dom_walker.
1844 (sprintf_dom_walker::before_dom_children): New function.
1845 (struct call_info): Moved into sprintf_dom_walker class
1846 (compute_formath_length, handle_gimple_call): Likewise.
1847 (sprintf_length::execute): Call the dominator walker rather
1848 than walking the statements.
1850 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
1851 gimple statement locations.
1852 (check_array_bounds): Corresponding changes. Get the statement's
1853 location directly from wi->stmt.
1855 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
1858 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
1860 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
1862 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
1863 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
1865 2017-10-27 Jakub Jelinek <jakub@redhat.com>
1868 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
1869 * config/i386/i386.c (maybe_get_pool_constant): Removed.
1870 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
1871 maybe_get_pool_constant.
1872 * config/i386/predicates.md (zero_extended_scalar_load_operand):
1875 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1877 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
1878 2.26 caveat. Update gas and gld versions.
1879 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
1882 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1884 * cgraph.h (set_malloc_flag): Declare.
1885 * cgraph.c (set_malloc_flag_1): New function.
1886 (set_malloc_flag): Likewise.
1887 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
1888 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
1890 (read_ipa_call_summary): Add support for reading is_return_callee.
1891 (write_ipa_call_summary): Stream is_return_callee.
1892 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
1893 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
1894 ipa-prop.h, ipa-fnsummary.h.
1895 (pure_const_names): Change to static.
1896 (malloc_state_e): Define.
1897 (malloc_state_names): Define.
1898 (funct_state_d): Add field malloc_state.
1899 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
1900 (check_retval_uses): New function.
1901 (malloc_candidate_p): Likewise.
1902 (analyze_function): Add support for malloc attribute.
1903 (pure_const_write_summary): Stream malloc_state.
1904 (pure_const_read_summary): Add support for reading malloc_state.
1905 (dump_malloc_lattice): New function.
1906 (propagate_malloc): New function.
1907 (warn_function_malloc): New function.
1908 (ipa_pure_const::execute): Call propagate_malloc and
1909 ipa_free_fn_summary.
1910 (pass_local_pure_const::execute): Add support for malloc attribute.
1911 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
1912 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
1914 2017-10-27 Martin Liska <mliska@suse.cz>
1916 PR gcov-profile/82457
1917 * doc/invoke.texi: Document that one needs a non-strict ISO mode
1918 for fork-like functions to be properly instrumented.
1920 2017-10-27 Richard Biener <rguenther@suse.de>
1923 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
1924 info when we redirected EH.
1926 2017-10-26 Michael Collison <michael.collison@arm.com>
1928 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
1930 (<optab>_trunchf<GPI:mode>2: New pattern.
1931 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
1932 * config/aarch64/iterators.md (wv): New mode attribute.
1933 (vf, VF): New mode attributes.
1934 (vgp, VGP): New mode attributes.
1935 (s): Update attribute with SImode and DImode prefixes.
1937 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
1939 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
1940 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
1941 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
1942 (nios2_option_overide): Initialize it. Don't allow R0-relative
1943 addressing with PIC.
1944 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
1945 (nios2_symbolic_constant_p): Likewise.
1946 (nios2_legitimate_address_p): Likewise.
1947 (nios2_r0rel_section_name_p): New.
1948 (nios2_symbol_ref_in_r0rel_data_p): New.
1949 (nios2_emit_move_sequence): Handle r0rel_constant_p.
1950 (r0rel_constant_p): New.
1951 (nios2_print_operand_address): Handle r0rel_constant_p.
1952 (nios2_cdx_narrow_form_p): Likewise.
1953 * config/nios2/nios2.opt (mr0rel-sec=): New option.
1954 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
1955 (Nios II Options): Document -mr0rel-sec.
1957 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
1959 * config/nios2/nios2.c: Include xregex.h.
1960 (nios2_gprel_sec_regex): New.
1961 (nios2_option_overide): Initialize it. Don't allow GP-relative
1962 addressing with PIC.
1963 (nios2_small_section_name_p): Check for regex match.
1964 * config/nios2/nios2.opt (mgprel-sec=): New option.
1965 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
1966 (Nios II Options): Document -mgprel-sec.
1968 2017-10-26 Jim Wilson <wilson@tuliptree.org>
1970 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
1972 2017-10-26 Tom de Vries <tom@codesourcery.com>
1974 PR tree-optimization/82707
1975 * gimple.c (gimple_copy): Fix unsharing of
1976 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
1978 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
1980 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
1981 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
1982 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
1983 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
1984 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
1985 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
1986 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
1987 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
1988 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
1989 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
1990 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
1991 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
1992 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
1993 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
1994 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
1995 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
1996 _mm512_mask_cmpunord_ps_mask): New intrinsics.
1998 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2000 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
2002 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
2003 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
2004 warning to rs6000.c. Remove the Undocumented flag, since it has
2006 (-mabi=ibmlongdouble): Likewise.
2007 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
2008 already set, set the default format for long double.
2009 (rs6000_debug_reg_global): Print whether long double is IBM or
2011 (rs6000_option_override_internal): Rework setting long double
2012 format. Only warn if the user is changing the long double default
2013 and they did not use -Wno-psabi.
2014 * doc/invoke.texi (PowerPC options): Update the documentation for
2015 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
2017 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2018 Alan Hayward <alan.hayward@arm.com>
2019 David Sherwood <david.sherwood@arm.com>
2021 * rtl.h (wider_subreg_mode): New function.
2022 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
2023 rather than an unsigned int *.
2024 * ira-color.c (regno_max_ref_width): Replace with...
2025 (regno_max_ref_mode): ...this new variable.
2026 (coalesced_pseudo_reg_slot_compare): Update accordingly.
2027 Use wider_subreg_mode.
2028 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
2029 rather than an unsigned int *.
2030 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
2031 (process_alt_operands): Likewise.
2032 (invariant_p): Likewise.
2033 * lra-spills.c (assign_mem_slot): Likewise.
2034 (add_pseudo_to_slot): Likewise.
2035 * lra.c (collect_non_operand_hard_regs): Likewise.
2036 (add_regs_to_insn_regno_info): Likewise.
2037 * reload1.c (regno_max_ref_width): Replace with...
2038 (regno_max_ref_mode): ...this new variable.
2039 (reload): Update accordingly. Update call to
2040 ira_sort_regnos_for_alter_reg.
2041 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
2042 (init_eliminable_invariants): Update to use regno_max_ref_mode.
2043 (scan_paradoxical_subregs): Likewise.
2045 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2047 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
2048 (aarch64_frame): Add emit_frame_chain boolean.
2049 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2050 Move eh_return case to aarch64_layout_frame.
2051 (aarch64_layout_frame): Initialize emit_frame_chain.
2052 (aarch64_expand_prologue): Use emit_frame_chain.
2054 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2056 * config/aarch64/aarch64.c (aarch64_layout_frame):
2057 Ensure LR is always stored at the bottom of the callee-saves.
2058 Remove rarely used frame layout which saves callee-saves at top of
2059 frame, so the store of LR can be used as a valid probe in all cases.
2061 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2063 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
2064 Improve unaligned TImode/TFmode base/offset split.
2066 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2067 Alan Hayward <alan.hayward@arm.com>
2068 David Sherwood <david.sherwood@arm.com>
2070 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
2071 * combine.c (find_single_use_1): Likewise.
2072 (expand_field_assignment): Likewise.
2073 (move_deaths): Likewise.
2074 * lra-constraints.c (simplify_operand_subreg): Likewise.
2075 (curr_insn_transform): Likewise.
2076 * lra.c (collect_non_operand_hard_regs): Likewise.
2077 (add_regs_to_insn_regno_info): Likewise.
2078 * rtlanal.c (reg_referenced_p): Likewise.
2079 (covers_regno_no_parallel_p): Likewise.
2081 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2083 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
2084 Don't print any bits outside the precision of the value.
2085 * wide-int.cc (test_printing): Add some new tests.
2087 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2089 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
2090 supports -xbrace_comment option.
2091 * configure: Regenerate.
2092 * config.in: Regenerate.
2093 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
2094 (ASM_CPU_SPEC): Use it.
2096 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2098 * target.def (static_rtx_alignment): New hook.
2099 * targhooks.h (default_static_rtx_alignment): Declare.
2100 * targhooks.c (default_static_rtx_alignment): New function.
2101 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
2102 * doc/tm.texi: Regenerate.
2103 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
2104 instead of targetm.constant_alignment. Remove call to
2106 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2107 (cris_preferred_mininum_alignment): New function, split out from...
2108 (cris_constant_alignment): ...here.
2109 (cris_static_rtx_alignment): New function.
2110 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
2112 (ix86_constant_alignment): ...here.
2113 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2114 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2115 (mmix_static_rtx_alignment): New function.
2116 * config/spu/spu.c (spu_static_rtx_alignment): New function.
2117 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2119 2017-10-26 Tamar Christina <tamar.christina@arm.com>
2122 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
2123 Add flag_trapping_math and flag_fp_int_builtin_inexact.
2125 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
2127 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
2128 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
2129 mark as a sign-extending load.
2130 (local_pic_load_u): Define.
2132 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
2135 * fold-const.c (operand_equal_for_comparison_p): Also return true
2136 if ARG0 is a simple variant of ARG1 with narrower precision.
2137 (fold_ternary_loc): Always pass unstripped operands to the predicate.
2139 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
2141 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
2143 * i386.h (processor_costs): Add gather_static, gather_per_elt,
2144 scatter_static, scatter_per_elt.
2145 * x86-tune-costs.h: Add new cost entries.
2147 2017-10-25 Richard Biener <rguenther@suse.de>
2149 * tree-ssa-sccvn.h (vn_eliminate): Declare.
2150 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
2151 class pass_fre): Move to ...
2152 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
2153 class pass_fre): ... here and adjust for statistics.
2155 2017-10-25 Jakub Jelinek <jakub@redhat.com>
2158 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
2159 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
2160 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
2161 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
2164 2017-10-25 Richard Biener <rguenther@suse.de>
2166 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
2167 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
2168 eliminate_push_avail, eliminate_insert): Move inside...
2169 (class eliminate_dom_walker): ... this class in preparation
2171 (fini_eliminate): Remove by merging with ...
2172 (eliminate): ... this function. Adjust for class changes.
2173 (pass_pre::execute): Remove fini_eliminate call.
2174 (pass_fre::execute): Likewise.
2176 2017-10-24 Jakub Jelinek <jakub@redhat.com>
2179 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
2180 (VPERMI2, VPERMI2I): New mode iterators.
2181 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
2182 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
2184 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
2185 mode iterator. Remove 3 old define_insn patterns.
2186 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
2187 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
2188 VPERMI2 mode iterator, remove the other two expanders.
2189 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
2190 to use VPERMI2 mode iterator, add another alternative for vpermi2*
2191 instructions, remove the other two patterns.
2192 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
2193 mode iterator, remove the other two patterns.
2194 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
2195 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
2196 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
2197 and adjust argument order accordingly.
2198 (ix86_expand_vec_perm): Adjust caller.
2199 (expand_vec_perm_1): Likewise.
2200 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
2201 (expand_vec_perm_vpermt2_vpshub2): ... this.
2202 (ix86_expand_vec_perm_const_1): Adjust caller.
2203 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
2206 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
2207 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
2208 (vec_shl_<mode>): Remove unused expander.
2209 (avx512bw_<shift_insn><mode>3): New define_insn.
2210 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
2211 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
2213 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
2216 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
2219 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
2221 PR rtl-optimization/82396
2222 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
2223 (autopref_multipass_init): Simplify initialization.
2224 (autopref_rank_data): Simplify sort order.
2225 * gcc/sched-int.h (autopref_multipass_data_): Remove
2226 multi_mem_insn_p, min_offset and max_offset.
2228 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
2231 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2232 Check special value of flag_omit_frame_pointer.
2233 (aarch64_can_eliminate): Likewise.
2234 (aarch64_override_options_after_change_1): Simplify handling of
2235 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
2237 2017-10-24 Richard Biener <rguenther@suse.de>
2239 PR tree-optimization/82697
2240 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
2241 zero for conditional load and unconditional store.
2243 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
2245 * doc/install.texi: Document bootstrap-cet.
2247 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
2250 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
2251 ENDBR instruction at function entrance if function is only
2254 2017-10-24 Jakub Jelinek <jakub@redhat.com>
2257 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
2258 patterns to better describe from which operation the CF is computed.
2259 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
2260 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
2261 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
2262 is 0, use _0 suffixed expanders instead of emitting a comparison
2265 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
2267 * config/i386/i386.md(*movsf_internal, *movdf_internal):
2268 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
2270 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
2273 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
2274 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
2275 * loop-iv.c (iv_get_reaching_def): Likewise.
2276 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
2277 variable is promoted and the partition contains undefined values.
2279 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2281 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
2283 (nios2_address_cost): Define.
2284 (nios2_legitimize_address): Recognize (exp + constant) directly.
2285 (TARGET_ADDRESS_COST): Define.
2287 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2289 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
2290 (nios2_symbolic_memory_operand_p): Declare.
2291 (nios2_split_large_constant): Declare.
2292 (nios2_split_symbolic_memory_operand): Declare.
2293 * config/nios2/nios2.c: Adjust includes.
2294 (nios2_symbolic_constant_allowed): New.
2295 (nios2_symbolic_constant_p): New.
2296 (nios2_plus_symbolic_constant_p): New.
2297 (nios2_valid_addr_expr_p): Recognize addresses involving
2299 (nios2_legitimate_address_p): Likewise, also LO_SUM.
2300 (nios2_symbolic_memory_operand_p): New.
2301 (nios2_large_constant_p): New.
2302 (nios2_split_large_constant): New.
2303 (nios2_split_plus_large_constant): New.
2304 (nios2_split_symbolic_memory_operand): New.
2305 (nios2_legitimize_address): Code refactoring. Handle addresses
2306 involving symbolic constants.
2307 (nios2_emit_move_sequence): Likewise.
2308 (nios2_print_operand): Improve error output.
2309 (nios2_print_operand_address): Handle LO_SUM.
2310 (nios2_cdx_narrow_form_p): Likewise.
2311 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
2312 operands involving symbolic constants.
2313 (movhi_internal, movsi_internal): Likewise.
2314 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
2315 (extendhisi2, extendqi<mode>2): Likewise.
2317 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2319 * tree-pass.h (PROP_rtl_split_insns): Define.
2320 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
2322 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2324 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
2326 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2329 * target.def (const_not_ok_for_debug_p): Default to
2330 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
2331 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
2332 * targhooks.c (default_const_not_ok_for_debug_p): New function.
2333 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
2334 which targetm.const_not_ok_for_debug_p returned true.
2335 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
2337 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
2339 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
2340 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
2341 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
2343 (ix86_const_not_ok_for_debug_p): New function.
2344 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
2345 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
2347 2017-10-23 David Malcolm <dmalcolm@redhat.com>
2350 * system.h: Conditionally include "unique-ptr.h" if
2351 INCLUDE_UNIQUE_PTR is defined.
2352 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
2353 of defining INCLUDE_UNIQUE_PTR before including "system.h".
2355 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
2357 * config/rl78/rl78.md: New define_expand "subdi3".
2359 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
2362 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
2363 DF_REF_INSN if DF_REF_INSN_INFO is false.
2365 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
2367 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
2368 xmm_move instead of sse_move.
2369 (sse_store_index): New function.
2370 (ix86_register_move_cost): Be more sensible about mismatch stall;
2371 model AVX moves correctly; make difference between sse->integer and
2373 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
2374 moves; make difference between SSE and AVX.
2375 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
2376 and zmm_move. Increase size of sse load and store tables;
2377 add unaligned load and store tables; add ssemmx_to_integer.
2378 * x86-tune-costs.h: Update all entries according to real
2379 move latencies from Agner Fog's manual and chip documentation.
2381 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2384 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
2385 * config/i386/constraints.md (Wf): New constraint.
2386 * config/i386/i386.md (UNSPEC_SBB): New unspec.
2387 (cmp<dwi>_doubleword): Removed.
2388 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
2389 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
2390 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
2391 expand with cmp<dwi>_doubleword. For LTU and GEU use
2392 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
2394 * common.opt (gcolumn-info): Enable by default.
2395 * doc/invoke.texi (gcolumn-info): Document new default.
2397 2017-10-23 Richard Biener <rguenther@suse.de>
2399 PR tree-optimization/82672
2400 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
2401 Fold the stmt if we propagated into it.
2403 2017-10-23 Richard Biener <rguenther@suse.de>
2405 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
2406 (bitmap_remove_expr_from_set): ... this. All callers call this
2407 for non-constant values.
2408 (bitmap_set_subtract): Rename to...
2409 (bitmap_set_subtract_expressions): ... this. Adjust and
2411 (bitmap_set_contains_value): Remove superfluous check.
2412 (bitmap_set_replace_value): Inline into single caller ...
2413 (bitmap_value_replace_in_set): ... here and simplify.
2414 (dependent_clean): Merge into ...
2415 (clean): ... this using an overload. Adjust.
2416 (prune_clobbered_mems): Adjust.
2417 (compute_antic_aux): Likewise.
2418 (compute_partial_antic_aux): Likewise.
2420 2017-10-23 Richard Biener <rguenther@suse.de>
2422 PR tree-optimization/82129
2424 2017-08-01 Richard Biener <rguenther@suse.de>
2426 PR tree-optimization/81181
2427 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
2428 (compute_antic): ... end of iteration here.
2430 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2432 * target.def (starting_frame_offset): New hook.
2433 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
2434 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
2435 * doc/tm.texi.in: Regenerate.
2436 * hooks.h (hook_hwi_void_0): Declare.
2437 * hooks.c (hook_hwi_void_0): New function.
2438 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
2439 STARTING_FRAME_OFFSET.
2440 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
2441 * reload1.c (reload): Likewise.
2442 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
2443 instead of STARTING_FRAME_OFFSET.
2444 * function.c (try_fit_stack_local): Likewise.
2445 (assign_stack_local_1): Likewise
2446 (instantiate_virtual_regs): Likewise.
2447 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
2448 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
2449 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
2450 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
2451 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
2452 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
2453 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
2454 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
2455 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
2456 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
2457 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
2458 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
2459 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
2460 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
2461 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
2462 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
2463 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
2464 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
2465 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
2466 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
2467 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
2468 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
2469 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
2470 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
2471 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
2472 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
2473 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
2474 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
2475 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
2476 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
2477 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
2478 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
2479 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
2480 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
2481 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
2482 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
2483 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
2484 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
2485 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
2486 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
2487 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
2488 * config/avr/avr.c (avr_starting_frame_offset): Make static and
2489 return a HOST_WIDE_INT.
2490 (avr_builtin_setjmp_frame_value): Use it instead of
2491 STARTING_FRAME_OFFSET.
2492 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2493 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
2494 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
2496 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2497 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
2498 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
2499 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2500 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
2501 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
2502 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2503 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
2504 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
2505 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2506 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
2507 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
2509 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2510 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
2511 * config/mips/mips.c (mips_compute_frame_info): Refer to
2512 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
2513 (mips_starting_frame_offset): New function.
2514 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2515 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
2516 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
2517 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
2518 and return a HOST_WIDE_INT.
2519 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2520 (mmix_initial_elimination_offset): Refer to
2521 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
2522 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
2523 * config/pa/pa.c (pa_starting_frame_offset): New function.
2524 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
2525 (pa_expand_prologue): Likewise.
2526 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2527 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
2528 !FRAME_GROWS_DOWNWARD handling to...
2529 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2530 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
2531 !FRAME_GROWS_DOWNWARD handling to...
2532 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2533 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
2534 !FRAME_GROWS_DOWNWARD handling to...
2535 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2536 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
2538 (rs6000_starting_frame_offset): New function.
2539 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
2540 !FRAME_GROWS_DOWNWARD handling to...
2541 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2542 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
2543 !FRAME_GROWS_DOWNWARD handling to...
2544 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2545 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
2546 !FRAME_GROWS_DOWNWARD handling to...
2547 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2548 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
2549 (rs6000_starting_frame_offset): New function.
2550 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
2551 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
2552 * config/vax/vax.c (vax_starting_frame_offset): New function.
2553 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
2554 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2555 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
2556 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
2557 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2558 * system.h (STARTING_FRAME_OFFSET): Poison.
2560 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2562 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
2563 SCALAR_TYPE_MODE instead of TYPE_MODE.
2565 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2566 Alan Hayward <alan.hayward@arm.com>
2567 David Sherwood <david.sherwood@arm.com>
2569 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
2571 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2572 Alan Hayward <alan.hayward@arm.com>
2573 David Sherwood <david.sherwood@arm.com>
2575 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
2577 2017-10-23 Richard Biener <rguenther@suse.de>
2579 PR tree-optimization/82129
2580 * tree-ssa-pre.c (bitmap_set_and): Remove.
2581 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
2582 canonicalizing expressions in the set to those with lowest
2583 ID rather than taking that from the first edge.
2585 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2587 * combine.c (rtx_equal_for_field_assignment_p): Use
2588 byte_lowpart_offset.
2590 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2591 Alan Hayward <alan.hayward@arm.com>
2592 David Sherwood <david.sherwood@arm.com>
2594 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
2595 to a SUBREG_PROMOTED_VAR.
2597 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2598 Alan Hayward <alan.hayward@arm.com>
2599 David Sherwood <david.sherwood@arm.com>
2601 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
2602 (expand_debug_source_expr): Likewise.
2603 * combine.c (combine_simplify_rtx): Likewise.
2604 * cse.c (fold_rtx): Likewise.
2605 * optabs.c (expand_float): Likewise.
2606 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2607 (simplify_binary_operation_1): Likewise.
2609 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2610 Alan Hayward <alan.hayward@arm.com>
2611 David Sherwood <david.sherwood@arm.com>
2613 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
2614 (record_promoted_value): Likewise.
2615 * expr.c (expand_expr_real_2): Likewise.
2616 * ree.c (update_reg_equal_equiv_notes): Likewise.
2617 (combine_set_extension): Likewise.
2618 * rtlanal.c (low_bitmask_len): Likewise.
2619 * simplify-rtx.c (neg_const_int): Likewise.
2620 (simplify_binary_operation_1): Likewise.
2622 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2623 Alan Hayward <alan.hayward@arm.com>
2624 David Sherwood <david.sherwood@arm.com>
2626 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
2627 * regcprop.c (maybe_mode_change): Likewise.
2628 * reload1.c (alter_reg): Likewise.
2630 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2632 * inchash.h (inchash::hash::add_wide_int): New function.
2633 * lto-streamer-out.c (hash_tree): Use it.
2635 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2637 * inchash.h (inchash::hash::add_wide_int): Rename to...
2638 (inchash::hash::add_hwi): ...this.
2639 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
2640 (polymorphic_call_target_hasher::hash): Likewise.
2641 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
2642 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
2643 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
2644 * lto-streamer-out.c (hash_tree): Likewise.
2645 * optc-save-gen.awk: Likewise.
2646 * tree.c (add_expr): Likewise.
2648 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
2651 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
2652 for ordered inequality comparisons even with TARGET_IEEE_FP.
2654 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
2657 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
2658 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
2659 Expand with cmp<dwi>_doubleword.
2661 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2663 * extend.texi: Add x86 specific to 'nocf_check' attribute.
2664 List CET intrinsics.
2665 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
2666 specific to -fcf-protection option.
2668 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2670 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
2671 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
2672 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
2673 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
2674 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
2675 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
2676 (extra_objs): Add cet.o for Linux/x86 targets.
2677 (tmake_file): Add i386/t-cet for Linux/x86 targets.
2678 * config/i386/cet.c: New file.
2679 * config/i386/cetintrin.h: Likewise.
2680 * config/i386/t-cet: Likewise.
2681 * config/i386/cpuid.h (bit_SHSTK): New.
2682 (bit_IBT): Likewise.
2683 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
2684 pass IBT and SHSTK bits.
2685 * config/i386/i386-builtin-types.def
2686 (VOID_FTYPE_UNSIGNED_PVOID): New.
2687 (VOID_FTYPE_UINT64_PVOID): Likewise.
2688 * config/i386/i386-builtin.def: Add CET intrinsics.
2689 * config/i386/i386-c.c (ix86_target_macros_internal): Add
2690 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
2691 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
2692 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
2694 * config/i386/i386.c (rest_of_insert_endbranch): New.
2695 (pass_data_insert_endbranch): Likewise.
2696 (pass_insert_endbranch): Likewise.
2697 (make_pass_insert_endbranch): Likewise.
2698 (ix86_notrack_prefixed_insn_p): Likewise.
2699 (ix86_target_string): Add -mibt, -mshstk flags.
2700 (ix86_option_override_internal): Add flag_cf_protection
2702 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
2703 (ix86_print_operand): Add 'notrack' prefix output.
2704 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
2705 (ix86_expand_builtin): Expand CET intrinsics.
2706 (x86_output_mi_thunk): Add 'endbranch' instruction.
2707 * config/i386/i386.h (TARGET_IBT): New.
2708 (TARGET_IBT_P): Likewise.
2709 (TARGET_SHSTK): Likewise.
2710 (TARGET_SHSTK_P): Likewise.
2711 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
2712 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
2713 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
2714 (builtin_setjmp_setup): New pattern.
2715 (builtin_longjmp): Likewise.
2716 (rdssp<mode>): Likewise.
2717 (incssp<mode>): Likewise.
2718 (saveprevssp): Likewise.
2719 (rstorssp): Likewise.
2720 (wrss<mode>): Likewise.
2721 (wruss<mode>): Likewise.
2722 (setssbsy): Likewise.
2723 (clrssbsy): Likewise.
2724 (nop_endbr): Likewise.
2725 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
2727 * config/i386/immintrin.h: Include <cetintrin.h>.
2728 * config/i386/linux-common.h
2729 (file_end_indicate_exec_stack_and_cet): New prototype.
2730 (TARGET_ASM_FILE_END): New.
2732 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
2734 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
2735 latencies instead of having separate table; make difference between
2736 integer and float costs.
2737 * i386.h (processor_costs): Remove scalar_stmt_cost,
2738 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
2739 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
2741 * x86-tune-costs.h: Remove entries which has been removed in
2742 procesor_costs from all tables; make cond_taken_branch_cost
2743 and cond_not_taken_branch_cost COST_N_INSNS based.
2745 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
2747 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
2749 2017-10-20 Jakub Jelinek <jakub@redhat.com>
2751 * config/i386/i386.md (isa): Remove fma_avx512f.
2752 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
2753 <avx512>_fmadd_<mode>_mask3<round_name>,
2754 <avx512>_fmsub_<mode>_mask<round_name>,
2755 <avx512>_fmsub_<mode>_mask3<round_name>,
2756 <avx512>_fnmadd_<mode>_mask<round_name>,
2757 <avx512>_fnmadd_<mode>_mask3<round_name>,
2758 <avx512>_fnmsub_<mode>_mask<round_name>,
2759 <avx512>_fnmsub_<mode>_mask3<round_name>,
2760 <avx512>_fmaddsub_<mode>_mask<round_name>,
2761 <avx512>_fmaddsub_<mode>_mask3<round_name>,
2762 <avx512>_fmsubadd_<mode>_mask<round_name>,
2763 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
2764 (*vec_widen_umult_even_v16si<mask_name>,
2765 *vec_widen_smult_even_v16si<mask_name>): Likewise.
2766 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
2768 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2770 * extend.texi: Add 'nocf_check' documentation.
2771 * gimple.texi: Add second parameter to
2772 gimple_build_call_from_tree.
2773 * invoke.texi: Add -fcf-protection documentation.
2774 * rtl.texi: Add REG_CALL_NOTRACK documenation.
2776 2017-10-20 Richard Biener <rguenther@suse.de>
2778 PR tree-optimization/82473
2779 * tree-vect-loop.c (vectorizable_reduction): Properly get at
2780 the largest input type.
2782 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2784 * c-attribs.c (handle_nocf_check_attribute): New function.
2785 (c_common_attribute_table): Add 'nocf_check' handling.
2786 * gimple-parser.c: Add second argument NULL to
2787 gimple_build_call_from_tree.
2788 * attrib.c (comp_type_attributes): Check nocf_check attribute.
2789 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
2791 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
2792 * common.opt: Add fcf-protection flag.
2793 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
2794 * flag-types.h: Add enum cf_protection_level.
2795 * gimple.c (gimple_build_call_from_tree): Add second parameter.
2796 Add 'nocf_check' attribute propagation to gimple call.
2797 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
2798 (gimple_build_call_from_tree): Update prototype.
2799 (gimple_call_nocf_check_p): New function.
2800 (gimple_call_set_nocf_check): Likewise.
2801 * gimplify.c: Add second argument to gimple_build_call_from_tree.
2802 * ipa-icf.c: Add nocf_check attribute in statement hash.
2803 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
2804 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
2805 * toplev.c (process_options): Add flag_cf_protection handling.
2807 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
2809 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
2811 2017-10-20 Richard Biener <rguenther@suse.de>
2813 PR tree-optimization/82603
2814 * tree-if-conv.c (predicate_mem_writes): Make sure to only
2815 remove false predicated stores.
2817 2017-10-20 Richard Biener <rguenther@suse.de>
2819 * graphite-isl-ast-to-gimple.c
2820 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
2821 Remove return value and simplify, dump copied stmt after lhs
2823 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
2824 Reduce dump verbosity.
2825 (gsi_insert_earliest): Likewise.
2826 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
2827 * graphite.c (print_global_statistics): Adjust dumping.
2828 (print_graphite_scop_statistics): Likewise.
2829 (print_graphite_statistics): Do not dump loops here.
2830 (graphite_transform_loops): But here.
2832 2017-10-20 Nicolas Roche <roche@adacore.com>
2834 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
2835 * configure: Regenerate.
2837 2017-10-20 Jakub Jelinek <jakub@redhat.com>
2840 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
2841 functions when optimizing replace GIMPLE_RETURN stmts with
2842 calls to __builtin_unreachable ().
2845 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
2846 for -fsanitize=thread link of executables.
2847 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
2848 link of executables.
2851 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
2853 (<shift_insn><mode>3<mask_name>): Change the last of the 3
2854 define_insns for logical vector shifts to use VI248_AVX512BW
2855 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
2856 condition, useless isa and prefix attributes. Change the first
2857 2 of these define_insns to ...
2858 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
2859 define_insn for avx512vl.
2860 (<shift_insn><mode>3): ... and this, new define_insn without
2861 masking for non-avx512vl.
2864 * config/i386/sse.md (*andnot<mode>3,
2865 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
2866 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
2867 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
2868 not applied use empty suffix even for TARGET_AVX512VL.
2869 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
2870 is applied, supply evex,evex or evex,evex,evex instead of just
2873 2017-10-20 Julia Koval <julia.koval@intel.com>
2875 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
2876 (OPTION_MASK_ISA_GFNI_UNSET): New.
2877 (ix86_handle_option): Handle OPT_mgfni.
2878 * config/i386/cpuid.h (bit_GFNI): New.
2879 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
2880 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
2881 * config/i386/i386.c (ix86_target_string): Add -mgfni.
2882 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
2883 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
2884 * config/i386/i386.opt: Add mgfni.
2886 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
2888 * config/msp430/msp430.c (msp430_option_override): Disable
2889 -fdelete-null-pointer-checks.
2890 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
2892 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
2894 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
2895 of x87 and SSE instructions.
2897 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
2899 * asan.c (create_cond_insert_point): Do not update edge count.
2900 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
2901 (afdo_propagate_circuit): Likewise.
2902 (afdo_calculate_branch_prob): Likewise.
2903 (afdo_annotate_cfg): Likewise.
2904 * basic-block.h (struct edge_def): Remove count.
2905 (edge_def::count): New accessor.
2906 * bb-reorder.c (rotate_loop): Update.
2907 (find_traces_1_round): Update.
2908 (connect_traces): Update.
2909 (sanitize_hot_paths): Update.
2910 * cfg.c (unchecked_make_edge): Update.
2911 (make_single_succ_edge): Update.
2912 (check_bb_profile): Update.
2913 (dump_edge_info): Update.
2914 (update_bb_profile_for_threading): Update.
2915 (scale_bbs_frequencies_int): Update.
2916 (scale_bbs_frequencies_gcov_type): Update.
2917 (scale_bbs_frequencies_profile_count): Update.
2918 (scale_bbs_frequencies): Update.
2919 * cfganal.c (connect_infinite_loops_to_exit): Update.
2920 * cfgbuild.c (compute_outgoing_frequencies): Update.
2921 (find_many_sub_basic_blocks): Update.
2922 * cfgcleanup.c (try_forward_edges): Update.
2923 (try_crossjump_to_edge): Update
2924 * cfgexpand.c (expand_gimple_cond): Update
2925 (expand_gimple_tailcall): Update
2926 (construct_exit_block): Update
2927 * cfghooks.c (verify_flow_info): Update
2928 (redirect_edge_succ_nodup): Update
2929 (split_edge): Update
2930 (make_forwarder_block): Update
2931 (duplicate_block): Update
2932 (account_profile_record): Update
2933 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
2934 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
2935 * cfgloopmanip.c (scale_loop_profile): Update.
2937 (lv_adjust_loop_entry_edge): Update.
2938 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
2939 (force_nonfallthru_and_redirect): Update.
2940 (purge_dead_edges): Update.
2941 (rtl_flow_call_edges_add): Update.
2942 * cgraphunit.c (init_lowered_empty_function): Update.
2943 (cgraph_node::expand_thunk): Update.
2944 * gimple-pretty-print.c (dump_probability): Update.
2945 (dump_edge_probability): Update.
2946 * gimple-ssa-isolate-paths.c (isolate_path): Update.
2947 * haifa-sched.c (sched_create_recovery_edges): Update.
2948 * hsa-gen.c (convert_switch_statements): Update.
2949 * ifcvt.c (dead_or_predicable): Update.
2950 * ipa-inline-transform.c (inline_transform): Update.
2951 * ipa-split.c (split_function): Update.
2952 * ipa-utils.c (ipa_merge_profiles): Update.
2953 * loop-doloop.c (add_test): Update.
2954 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
2955 * lto-streamer-in.c (input_cfg): Update.
2956 (input_function): Update.
2957 * lto-streamer-out.c (output_cfg): Update.
2958 * modulo-sched.c (sms_schedule): Update.
2959 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
2960 * predict.c (maybe_hot_edge_p): Update.
2961 (unlikely_executed_edge_p): Update.
2962 (probably_never_executed_edge_p): Update.
2963 (dump_prediction): Update.
2964 (drop_profile): Update.
2965 (propagate_unlikely_bbs_forward): Update.
2966 (determine_unlikely_bbs): Update.
2967 (force_edge_cold): Update.
2968 * profile.c (compute_branch_probabilities): Update.
2969 * reg-stack.c (better_edge): Update.
2970 * shrink-wrap.c (handle_simple_exit): Update.
2971 * tracer.c (better_p): Update.
2972 * trans-mem.c (expand_transaction): Update.
2973 (split_bb_make_tm_edge): Update.
2974 * tree-call-cdce.c: Update.
2975 * tree-cfg.c (gimple_find_sub_bbs): Update.
2976 (gimple_split_edge): Update.
2977 (gimple_duplicate_sese_region): Update.
2978 (gimple_duplicate_sese_tail): Update.
2979 (gimple_flow_call_edges_add): Update.
2980 (insert_cond_bb): Update.
2981 (execute_fixup_cfg): Update.
2982 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
2983 * tree-complex.c (expand_complex_div_wide): Update.
2984 * tree-eh.c (lower_resx): Update.
2985 (unsplit_eh): Update.
2986 (cleanup_empty_eh_move_lp): Update.
2987 * tree-inline.c (copy_edges_for_bb): Update.
2988 (freqs_to_counts): Update.
2989 (copy_cfg_body): Update.
2990 * tree-ssa-dce.c (remove_dead_stmt): Update.
2991 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
2992 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
2993 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
2994 (unloop_loops): Update.
2995 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
2996 * tree-ssa-loop-split.c (connect_loops): Update.
2997 (split_loop): Update.
2998 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
2999 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
3000 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
3001 * tree-ssa-reassoc.c (branch_fixup): Update.
3002 * tree-ssa-tail-merge.c (replace_block_by): Update.
3003 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
3004 (compute_path_counts): Update.
3005 (update_profile): Update.
3006 (recompute_probabilities): Update.
3007 (update_joiner_offpath_counts): Update.
3008 (estimated_freqs_path): Update.
3009 (freqs_to_counts_path): Update.
3010 (clear_counts_path): Update.
3011 (ssa_fix_duplicate_block_edges): Update.
3012 (duplicate_thread_path): Update.
3013 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
3014 (case_bit_test_cmp): Update.
3015 (collect_switch_conv_info): Update.
3016 (gen_inbound_check): Update.
3017 (do_jump_if_equal): Update.
3018 (emit_cmp_and_jump_insns): Update.
3019 * tree-tailcall.c (decrease_profile): Update.
3020 (eliminate_tail_call): Update.
3021 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
3022 (vect_do_peeling): Update.
3023 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3024 * ubsan.c (ubsan_expand_null_ifn): Update.
3025 (ubsan_expand_ptr_ifn): Update.
3026 * value-prof.c (gimple_divmod_fixed_value): Update.
3027 (gimple_mod_pow2): Update.
3028 (gimple_mod_subtract): Update.
3029 (gimple_ic): Update.
3030 (gimple_stringop_fixed_value): Update.
3032 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3035 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
3037 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
3039 PR rtl-optimization/82395
3040 * ira-color.c (allocno_priority_compare_func): Fix comparison step
3041 based on non_spilled_static_chain_regno_p.
3043 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3045 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
3046 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
3047 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
3049 2017-10-19 Martin Sebor <msebor@redhat.com>
3051 PR tree-optimization/82596
3052 * tree.c (array_at_struct_end_p): Handle STRING_CST.
3054 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3056 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
3057 (get_mem_refs_of_builtin_call): Likewise.
3058 * builtins.c (expand_builtin_apply): Adjust call to
3059 allocate_dynamic_stack_space.
3060 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
3061 the third argument to allocate_dynamic_stack_space, otherwise -1.
3062 (expand_builtin): Deal with all alloca variants.
3063 (is_inexpensive_builtin): Likewise.
3064 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
3065 * calls.c (special_function_p): Deal with all alloca variants.
3066 (initialize_argument_information): Adjust call to
3067 allocate_dynamic_stack_space.
3068 (expand_call): Likewise.
3069 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
3070 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
3071 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
3072 use it for the stack usage computation.
3073 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
3074 * function.c (gimplify_parameters): Call build_alloca_call_expr.
3075 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
3076 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
3077 (in_loop_p): Remove first argument and useless check.
3078 (pass_walloca::execute): Remove useless test and adjust call to above.
3079 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
3080 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
3081 (gimplify_call_expr): Deal with all alloca variants.
3082 * hsa-gen.c (gen_hsa_alloca): Likewise.
3083 (gen_hsa_insns_for_call): Likewise.
3084 * ipa-pure-const.c (special_builtin_state): Likewise.
3085 * tree-chkp.c (chkp_build_returned_bound): Likewise.
3086 * tree-object-size.c (alloc_object_size): Likewise.
3087 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
3088 (call_may_clobber_ref_p_1): Likewise.
3089 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
3090 (ccp_fold_stmt): Likewise.
3091 (optimize_stack_restore): Likewise.
3092 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
3093 (mark_all_reaching_defs_necessary_1): Likewise.
3094 (propagate_necessity): Likewise.
3095 (eliminate_unnecessary_stmts): Likewise.
3096 * tree.c (build_common_builtin_nodes): Build
3097 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
3098 (build_alloca_call_expr): New function.
3099 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
3100 (CASE_BUILT_IN_ALLOCA): Likewise.
3101 (build_alloca_call_expr): Declare.
3102 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
3104 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3107 * dwarf2out.c (new_die_raw): New static inline function.
3108 (new_die): Use it to create the DIE.
3109 (add_AT_external_die_ref): Likewise.
3110 (clone_die): Likewise.
3111 (clone_as_declaration): Likewise.
3112 (dwarf2out_vms_debug_main_pointer): Likewise.
3113 (base_type_die): Likewise. Remove early return for corner cases.
3114 Do not call add_pubtype on the DIE here.
3115 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
3116 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
3117 typedefs for base types with DW_AT_endianity. Make sure a DIE with
3118 native order exists for base types, attach the DIE manually and call
3119 add_pubtype on it. Do not equate a reverse order DIE to the type.
3121 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
3123 * config/arm/arm.c (align_ok_ldrd_strd): New function.
3124 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
3126 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
3128 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3130 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
3131 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
3132 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
3133 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
3134 * opts.c (sanitizer_opts): Add builtin.
3135 * ubsan.c (instrument_builtin): New function.
3136 (pass_ubsan::execute): Call it.
3137 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
3138 * doc/invoke.texi: Document -fsanitize=builtin.
3140 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
3141 builtins, store max (log2 (align), 0) into uchar field instead of
3142 align into uptr field.
3143 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
3144 store uchar 0 field instead of uptr 0 field.
3145 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
3146 instead of passing one address of struct with 2 locations pass
3147 two addresses of structs with 1 location each.
3148 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
3149 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
3150 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
3151 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
3152 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
3153 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
3154 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
3155 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
3157 2017-10-19 Martin Liska <mliska@suse.cz>
3160 * file-find.c (remove_prefix): Remove.
3161 * file-find.h (remove_prefix): Likewise.
3162 * gcc-ar.c: Remove smartness of lookup.
3164 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
3166 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
3167 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
3168 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
3170 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3173 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
3174 (setcc + and to xor + setcc): New peephole2.
3176 2017-10-19 Tom de Vries <tom@codesourcery.com>
3178 * doc/sourcebuild.texi (Test Directives, Variants of
3179 dg-require-support): Add dg-require-stack-size.
3181 2017-10-19 Martin Liska <mliska@suse.cz>
3184 * gimplify.c (gimplify_decl_expr): Do not instrument variables
3185 that have a large alignment.
3186 (gimplify_target_expr): Likewise.
3188 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
3190 PR rtl-optimization/82602
3191 * ira.c (rtx_moveable_p): Return false for volatile asm.
3193 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
3196 * config/i386/i386-modes.def (CCGZ): New CC mode.
3197 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
3198 * config/i386/predicates.md (ix86_comparison_operator):
3200 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
3201 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
3202 with double-word subtraction.
3203 (put_condition_code): Handle CCGZmode.
3205 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
3207 * wide-int.cc (debug (const wide_int &)): New.
3208 (debug (const wide_int *)): New.
3209 (debug (const widest_int &)): New.
3210 (debug (const widest_int *)): New.
3212 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
3215 * lra-constraints.c (curr_insn_transform): Use non-input operand
3216 instead of output one for matched reload.
3218 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3220 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
3221 (tree-ssa-loop-ivopts.h): New header file.
3222 (struct builtin_info): New fields.
3223 (classify_builtin_1): Compute and record base and offset parts for
3224 memset builtin partition by calling strip_offset.
3225 (offset_cmp, fuse_memset_builtins): New functions.
3226 (finalize_partitions): Fuse adjacent memset partitions by calling
3228 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
3229 Expose the interface.
3230 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
3232 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3234 PR tree-optimization/82574
3235 * tree-loop-distribution.c (find_single_drs): New parameter. Check
3236 that data reference must be executed exactly once per iteration
3237 against the outermost loop in nest.
3238 (classify_partition): Update call to above function.
3240 2017-10-18 Richard Biener <rguenther@suse.de>
3242 PR tree-optimization/82591
3243 * graphite.c (graphite_transform_loops): Move code gen message
3245 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3246 Here. Handle scop_to_isl_ast failing.
3247 (scop_to_isl_ast): Limit the number of ISL operations.
3249 2017-10-18 Richard Biener <rguenther@suse.de>
3251 * graphite-isl-ast-to-gimple.c
3252 (translate_isl_ast_to_gimple::set_rename): Simplify.
3253 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
3254 (graphite_copy_stmts_from_block): ... here.
3255 (copy_bb_and_scalar_dependences): Simplify.
3256 (add_parameters_to_ivs_params): Canonicalize.
3257 (generate_entry_out_of_ssa_copies): Simplify.
3258 * graphite-sese-to-poly.c (extract_affine_name): Simplify
3259 by passing in ISL dimension.
3260 (parameter_index_in_region_1): Rename to ...
3261 (parameter_index_in_region): ... this.
3262 (extract_affine): Adjust assert, pass down parameter index.
3263 (add_param_constraints): Use range-info when available.
3264 (build_scop_context): Adjust.
3265 * sese.c (new_sese_info): Adjust.
3266 (free_sese_info): Likewise.
3267 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
3268 Remove unused typedefs.
3269 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
3271 2017-10-18 Martin Liska <mliska@suse.cz>
3273 * combine.c (simplify_compare_const): Add gcc_fallthrough.
3275 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
3277 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
3278 (s390_sched_init): Do not reset s390_sched_state if we entered the
3279 current basic block via a fallthru edge and all others are unlikely.
3281 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
3283 * config/s390/s390.c (NUM_SIDES): New variable.
3284 (LONGRUNNING_THRESHOLD): New variable.
3285 (LATENCY_FACTOR): New variable.
3286 (s390_sched_score): Decrease score for long-running instructions on
3288 (s390_sched_variable_issue): Perform bookkeeping for long-running
3291 2017-10-18 Richard Biener <rguenther@suse.de>
3293 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
3294 Simplify with removal of the parameter rename map.
3295 (set_rename): Likewise.
3296 (should_copy_to_new_region): Likewise.
3297 (graphite_copy_stmts_from_block): Likewise.
3298 (copy_bb_and_scalar_dependences): Remove initialization of
3299 unused copied_bb_map.
3301 (copy_internal_parameters): Likewise.
3302 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
3303 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
3304 Use INTEGRAL_TYPE_P.
3305 (parameter_index_in_region_1): Rename to ...
3306 (assign_parameter_index_in_region): ... this. Assert we have
3307 a parameter we handle.
3308 (scan_tree_for_params): Adjust.
3309 * sese.h (parameter_rename_map_t): Remove.
3310 (struct sese_info_t): Remove unused parameter_rename_map and
3311 copied_bb_map members.
3312 * sese.c (new_sese_info): Adjust.
3313 (free_sese_info): Likewise.
3315 2017-10-18 Martin Liska <mliska@suse.cz>
3318 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
3319 on an abnormal edge.
3321 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
3323 * doc/invoke.texi (ffunction-sections and fdata-sections):
3326 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3328 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
3329 the use statement can throw internally.
3331 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3333 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
3334 any RTX present on the RHS of a SET.
3335 * compare-elim.c (try_eliminate_compare): Restore comment.
3337 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3339 * langhooks.h (struct lang_hooks): Document that tree_size langhook
3340 may be also called on tcc_type nodes.
3341 * langhooks.c (lhd_tree_size): Likewise.
3343 2017-10-17 David Malcolm <dmalcolm@redhat.com>
3345 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
3346 format_warning_at_substring.
3347 (maybe_warn): Convert source_range * param to a location_t. Pass
3348 UNKNOWN_LOCATION rather than NULL to fmtwarn.
3349 (format_directive): Remove code to extract source_ranges and
3350 source_range * in favor of just a location_t.
3351 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
3353 * substring-locations.c (format_warning_va): Convert
3354 source_range * param to a location_t.
3355 (format_warning_at_substring): Likewise.
3356 * substring-locations.h (format_warning_va): Likewise.
3357 (format_warning_at_substring): Likewise.
3359 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
3361 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
3363 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
3364 and scatter/gather ops.
3366 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
3367 vec_gather_load and vec_scatter_store.
3368 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
3369 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
3370 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
3371 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
3372 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
3373 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
3375 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
3377 * reg-stack.c (compare_for_stack_reg): Add bool argument.
3378 Detect FTST instruction and handle its register pops. Only pop
3379 second operand if can_pop_second_op is true.
3380 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
3381 set can_pop_second_op to false in the compare_for_stack_reg call.
3383 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
3384 output_fp_compare for stack register operands.
3385 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
3386 instructions here. Do not emit stack register pops here. Assert
3387 that FCOMPP pops next to top stack register. Rewrite function.
3389 2017-10-17 Nathan Sidwell <nathan@acm.org>
3392 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
3393 use DECL_ASSEMBLER_NAME_RAW.
3396 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
3399 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
3400 Wilco Dijkstra <wilco.dijkstra@arm.com>
3402 * builtins.c (expand_builtin_update_setjmp_buf): Add a
3403 converstion to Pmode from the buf_addr.
3405 2017-10-17 Richard Biener <rguenther@suse.de>
3407 * graphite-dependences.c (scop_get_reads_and_writes): Change
3408 output parameters to references.
3410 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
3412 PR 71026/tree-optimization
3413 * fold-const.c (distribute_real_division): Removed.
3414 (fold_binary_loc): Remove calls to distribute_real_divison.
3416 2017-10-17 Richard Biener <rguenther@suse.de>
3418 * graphite-scop-detection.c
3419 (scop_detection::stmt_has_simple_data_refs_p): Always use
3420 the full nest as region.
3421 (try_generate_gimple_bb): Likewise.
3422 * sese.c (scalar_evolution_in_region): Simplify now that
3423 SCEV can handle instantiation in regions.
3424 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
3425 in the non-loop part of a function if requested.
3427 2017-10-17 Richard Biener <rguenther@suse.de>
3429 PR tree-optimization/82563
3430 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
3432 (graphite_regenerate_ast_isl): Call it.
3433 * graphite-scop-detection.c (build_scops): Remove entry edge split.
3435 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3437 PR tree-optimization/82549
3438 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
3439 Formatting fixes. Instead of calling make_bit_field_ref with negative
3442 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
3444 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
3445 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
3446 _mm_maskz_reduce_ss): New.
3447 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
3448 __builtin_ia32_reducess_mask): Ditto..
3449 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
3450 * config/i386/sse.md (reduces<mode>): Renamed to ...
3451 (reduces<mode><mask_scalar_name>): ... this.
3452 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
3454 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
3455 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
3457 2017-10-16 David Malcolm <dmalcolm@redhat.com>
3459 * Makefile.in (OBJS): Add unique-ptr-tests.o.
3460 * selftest-run-tests.c (selftest::run_tests): Call
3461 selftest::unique_ptr_tests_cc_tests.
3462 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
3463 * unique-ptr-tests.cc: New file.
3465 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
3468 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
3470 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
3471 (make_hard_regno_born, make_hard_regno_dead): Update
3472 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
3474 2017-10-16 Jeff Law <law@redhat.com>
3476 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
3478 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
3480 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
3482 2017-10-16 Olivier Hainque <hainque@adacore.com>
3484 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
3485 with_cpu if we were configured for an e500v2 target cpu name.
3487 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
3489 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
3490 * doc/invoke.texi: Document +nodsp as a valid extension for
3493 2017-10-16 Martin Liska <mliska@suse.cz>
3495 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
3496 (test_set_range): Likewise.
3497 (test_range_functions): Rename to ...
3498 (test_bit_in_range): ... this.
3499 (sbitmap_c_tests): Add new test.
3501 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3503 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
3505 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
3506 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
3508 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3510 * config/aarch64/aarch64-builtins.c
3511 (aarch64_types_quadopu_lane_qualifiers): New.
3512 (TYPES_QUADOPU_LANE): New.
3513 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
3514 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
3515 (aarch64_<sur>dot_laneq<vsi2qi>): New.
3516 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
3517 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
3518 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
3519 (Vdottype, DOTPROD): New.
3520 (sur): Add SDOT and UDOT.
3522 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3524 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
3525 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
3526 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
3528 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
3529 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
3530 Enable TARGET_DOTPROD.
3531 (cortex-a75.cortex-a55): Likewise.
3532 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
3534 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3536 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
3537 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
3539 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
3541 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
3542 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
3543 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
3544 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
3545 * config/arm/types.md (neon_dot, neon_dot_q): New.
3546 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
3548 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3550 * config/arm/arm.h (TARGET_DOTPROD): New.
3551 * config/arm/arm.c (arm_arch_dotprod): New.
3552 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
3553 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
3554 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
3555 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
3556 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
3557 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
3558 * doc/invoke.texi (armv8.2-a): Document dotprod
3560 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
3562 * i386.c (ix86_vec_cost): New function.
3563 (ix86_rtx_costs): Handle vector operations better.
3564 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
3565 * x86-tune-costs.h: Add new costs to all tables.
3567 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
3569 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
3571 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
3572 divsd, sqrtss and sqrtsd
3573 * x86-tune-costs.h: Add new entries to all costs.
3574 (znver1_cost): Fix to match real instruction latencies.
3576 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3577 Michael Collison <michael.collison@arm.com>
3579 * compare-elim.c: Include emit-rtl.h.
3580 (can_merge_compare_into_arith): New function.
3581 (try_validate_parallel): Likewise.
3582 (try_merge_compare): Likewise.
3583 (try_eliminate_compare): Call the above when no previous clobber
3585 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
3588 2017-10-14 Jakub Jelinek <jakub@redhat.com>
3592 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
3593 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
3597 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
3598 to be any operand_equal_p operands. For & (B - 1) require
3599 B to be power of 2. Recognize
3600 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
3602 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3605 * optabs.c (expand_memory_blockage): Fix call of
3606 targetm.have_memory_blockage.
3608 2017-10-14 Jakub Jelinek <jakub@redhat.com>
3611 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
3612 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
3613 objects to extra_objs instead of overwriting it.
3615 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3617 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
3618 Use any_fp_register_operand as operand[3] predicate. Simplify
3619 equality test for operands[2] and operands[4] memory location.
3620 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
3621 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
3622 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
3623 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
3624 any_fp_register_operand as operand[1] predicate. Simplify
3625 equality test for operands[0] and operands[3] memory location.
3626 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
3627 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
3628 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
3630 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3632 * target-insns.def: Add memory_blockage.
3633 * optabs.c (expand_memory_blockage): New function.
3634 (expand_asm_memory_barrier): Rename ...
3635 (expand_asm_memory_blockage): ... to this.
3636 (expand_mem_thread_fence): Call expand_memory_blockage
3637 instead of expand_asm_memory_barrier.
3638 (expand_mem_singnal_fence): Ditto.
3639 (expand_atomic_load): Ditto.
3640 (expand_atomic_store): Ditto.
3641 * doc/md.texi (Standard Pattern Names For Generation):
3642 Document memory_blockage instruction pattern.
3644 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
3646 * config/rl78/rl78.c (rl78_emit_libcall): New function.
3647 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
3648 * config/rl78/rl78.md: New define_expand "adddi3".
3650 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
3652 * cfghooks.c (verify_flow_info): Disable check that all probabilities
3655 2017-10-13 Jeff Law <law@redhat.com>
3657 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
3659 2017-10-13 Jakub Jelinek <jakub@redhat.com>
3662 * internal-fn.c (expand_mul_overflow): If both operands have
3663 the same highpart of -1 or 0 and the topmost bit of lowpart
3664 is different, overflow is if res <= 0 rather than res < 0.
3666 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
3668 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
3669 TARGET_P9_VECTOR code for unaligned_load case.
3671 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
3673 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
3675 2017-10-13 Nathan Sidwell <nathan@acm.org>
3677 * tree-core.h (tree_contains_struct): Make bool.
3678 * tree.c (tree_contains_struct): Likewise.
3679 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
3680 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
3681 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
3682 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
3683 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
3685 2017-10-13 Richard Biener <rguenther@suse.de>
3687 * graphite-isl-ast-to-gimple.c
3688 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
3689 parameters and dominance check.
3690 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
3691 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
3692 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
3693 Do not update SSA form here or do intermediate IL verification.
3694 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
3695 (graphite_initialize): Remove check on the number of loops in
3696 the function and inline into graphite_transform_loops.
3697 (graphite_finalize): Inline into graphite_transform_loops.
3698 (graphite_transform_loops): Perform SSA update and IL verification
3700 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
3702 2017-10-13 Richard Biener <rguenther@suse.de>
3704 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
3705 graphite_expression_type_precision): Avoid global constructor
3707 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
3708 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
3709 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
3710 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
3711 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
3712 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
3714 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
3717 * config/i386/i386.h (ix86_red_zone_size): New.
3718 * config/i386/i386.md (push peephole2s): Replace
3719 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
3721 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3722 Alan Hayward <alan.hayward@arm.com>
3723 David Sherwood <david.sherwood@arm.com>
3725 * combine.c (can_change_dest_mode): Reject changes in
3726 REGMODE_NATURAL_SIZE.
3728 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3729 Alan Hayward <alan.hayward@arm.com>
3730 David Sherwood <david.sherwood@arm.com>
3732 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
3733 (expand_debug_source_expr): Likewise.
3734 * combine.c (combine_simplify_rtx): Likewise.
3735 * cse.c (fold_rtx): Likewise.
3736 * fwprop.c (canonicalize_address): Likewise.
3737 * targhooks.c (default_shift_truncation_mask): Likewise.
3739 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3740 Alan Hayward <alan.hayward@arm.com>
3741 David Sherwood <david.sherwood@arm.com>
3743 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
3744 (widened_mode): Likewise.
3745 (expand_unop): Likewise.
3746 * ree.c (transform_ifelse): Likewise.
3747 (merge_def_and_ext): Likewise.
3748 (combine_reaching_defs): Likewise.
3749 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3751 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3752 Alan Hayward <alan.hayward@arm.com>
3753 David Sherwood <david.sherwood@arm.com>
3755 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
3756 * combine.c (gen_lowpart_for_combine): Likewise.
3757 * dwarf2out.c (rtl_for_decl_location): Likewise.
3758 * final.c (alter_subreg): Likewise.
3759 * rtlhooks.c (gen_lowpart_general): Likewise.
3760 (gen_lowpart_if_possible): Likewise.
3762 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3763 Alan Hayward <alan.hayward@arm.com>
3764 David Sherwood <david.sherwood@arm.com>
3766 * calls.c (expand_call): Use subreg_lowpart_offset.
3767 * cse.c (cse_insn): Likewise.
3768 * regcprop.c (copy_value): Likewise.
3769 (copyprop_hardreg_forward_1): Likewise.
3771 2017-10-13 Jakub Jelinek <jakub@redhat.com>
3774 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
3775 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
3776 =Q constraints to +Q and into insn condition add check
3777 that operands[0] and operands[1] are equal.
3778 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
3779 =Q constraints to +Q and into insn condition add check
3780 that operands[0] is equal to either operands[1] or operands[2].
3783 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
3784 instead of handling MINUS_EXPR twice (once for each argument),
3785 canonicalize operand order and handle just once, use rtype where
3786 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
3789 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
3790 any values of __C while still being pattern recognizable as a simple
3793 2017-10-13 Richard Biener <rguenther@suse.de>
3795 PR tree-optimization/82451
3797 2017-10-02 Richard Biener <rguenther@suse.de>
3799 PR tree-optimization/82355
3800 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
3801 a mapping for the enclosing loop but avoid generating one for
3803 (copy_bb_and_scalar_dependences): Remove premature codegen
3804 error on PHIs in blocks duplicated into multiple places.
3805 * graphite-scop-detection.c
3806 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
3807 in the region use it as loop and nest to analyze the DR in.
3808 (try_generate_gimple_bb): Likewise.
3809 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
3810 (add_loop_constraints): For blocks in a loop not in the region
3811 create a dimension with a single iteration.
3812 * sese.h (gbb_loop_at_index): Remove assert.
3814 * cfgloop.c (loop_preheader_edge): For the loop tree root
3815 return the single successor of the entry block.
3816 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3817 Reset the SCEV hashtable and niters.
3818 * graphite-scop-detection.c
3819 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
3820 assert that we only have POLYNOMIAL_CHREC that vary in loops
3821 contained in the region.
3822 (scop_detection::graphite_can_represent_expr): Adjust.
3823 (scop_detection::stmt_has_simple_data_refs_p): For loops
3824 not in the region set loop to NULL. The nest is now the
3825 entry edge to the region.
3826 (try_generate_gimple_bb): Likewise.
3827 * sese.c (scalar_evolution_in_region): Adjust for
3828 instantiate_scev change.
3829 * tree-data-ref.h (graphite_find_data_references_in_stmt):
3830 Make nest parameter the edge into the region.
3831 (create_data_ref): Likewise.
3832 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
3833 entry edge into a region and adjust instantiate_scev calls.
3834 (create_data_ref): Likewise.
3835 (graphite_find_data_references_in_stmt): Likewise.
3836 (find_data_references_in_stmt): Pass the loop preheader edge
3837 from the nest argument.
3838 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
3839 parameter the edge into the region.
3840 (instantiate_parameters): Use the loop preheader edge as entry.
3841 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
3843 (get_instantiated_value_entry): Make instantiate_below parameter
3844 the edge into the region.
3845 (instantiate_scev_name): Likewise. Adjust dominance checks,
3846 when we cannot use loop-based instantiation instantiate by
3847 walking use-def chains.
3848 (instantiate_scev_poly): Adjust.
3849 (instantiate_scev_binary): Likewise.
3850 (instantiate_scev_convert): Likewise.
3851 (instantiate_scev_not): Likewise.
3852 (instantiate_array_ref): Remove.
3853 (instantiate_scev_3): Likewise.
3854 (instantiate_scev_2): Likewise.
3855 (instantiate_scev_1): Likewise.
3856 (instantiate_scev_r): Do not blindly handle N-operand trees.
3857 Do not instantiate array-refs. Handle all constants and invariants.
3858 (instantiate_scev): Make instantiate_below parameter
3859 the edge into the region.
3860 (resolve_mixers): Use the loop preheader edge for the region
3861 parameter to instantiate_scev_r.
3862 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
3864 2017-10-13 Richard Biener <rguenther@suse.de>
3866 PR tree-optimization/82525
3867 * graphite-isl-ast-to-gimple.c
3868 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
3870 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
3871 Fail code generation when we cannot represent the isl integer.
3872 (binary_op_to_tree): Elide modulo operations that are no-ops
3873 in the type we code generate. Remove now superfluous code
3876 2017-10-13 Richard Biener <rguenther@suse.de>
3878 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
3879 (scop_detection::harmful_loop_in_region): Remove premature
3880 IV type restriction.
3881 (scop_detection::graphite_can_represent_scev): We can handle
3882 pointer IVs just fine.
3884 2017-10-13 Alan Modra <amodra@gmail.com>
3886 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
3887 "Clobbers and Scratch Registers". Add paragraph on
3888 alternative to clobbers for scratch registers and OpenBLAS
3891 2017-10-13 Alan Modra <amodra@gmail.com>
3893 * doc/extend.texi (Clobbers): Correct vax example. Delete old
3894 example of a memory input for a string of known length. Move
3895 commentary out of table. Add a number of new examples
3896 covering array memory inputs.
3898 2017-10-12 Martin Liska <mliska@suse.cz>
3900 PR tree-optimization/82493
3901 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
3902 (test_range_functions): New function.
3903 (sbitmap_c_tests): Likewise.
3904 * selftest-run-tests.c (selftest::run_tests): Run new tests.
3905 * selftest.h (sbitmap_c_tests): New function.
3907 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
3909 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
3911 * config/rs6000/amo.h: Fix spacing issue.
3913 2017-10-12 Jakub Jelinek <jakub@redhat.com>
3916 * config/i386/i386.md (*ashl<mode>3_mask_1,
3917 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
3918 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
3921 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
3923 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
3924 (profile_probability): Set max_probability
3925 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
3927 * profile-count.c (profile_probability::differs_from_p): Do not
3928 rely on max_probaiblity == 10000
3930 2017-10-12 Jeff Law <law@redhat.com>
3932 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
3935 2017-10-12 Martin Sebor <msebor@redhat.com>
3939 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
3940 (handle_alias_pairs): Call it.
3941 * common.opt (-Wattribute-alias): New option.
3942 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
3943 * doc/invoke.texi (-Wattribute-alias): Document.
3945 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
3948 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
3950 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
3952 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
3953 (make_hard_regno_born, make_hard_regno_dead): Update
3954 bb_killed_pseudos and bb_gen_pseudos.
3956 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
3958 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
3960 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
3962 * config/alpha/alpha.c (alpha_split_conditional_move):
3963 Use std::swap instead of manually swapping.
3964 (alpha_stdarg_optimize_hook): Ditto.
3965 (alpha_canonicalize_comparison): Ditto.
3967 2017-10-12 Bin Cheng <bin.cheng@arm.com>
3969 * tree-loop-distribution.c (struct builtin_info): New struct.
3970 (struct partition): Refactor fields into struct builtin_info.
3971 (partition_free): Free struct builtin_info.
3972 (build_size_arg_loc, build_addr_arg_loc): Delete.
3973 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
3974 information from struct builtin_info.
3975 (find_single_drs): New function refactored from classify_partition.
3976 Also moved builtin validity checks to this function.
3977 (compute_access_range, alloc_builtin): New functions.
3978 (classify_builtin_st, classify_builtin_ldst): New functions.
3979 (classify_partition): Refactor code into functions find_single_drs,
3980 classify_builtin_st and classify_builtin_ldst.
3981 (distribute_loop): Don't do runtime alias check when distributing
3983 (find_seed_stmts_for_distribution): New function.
3984 (pass_loop_distribution::execute): Refactor code finding seed
3985 stmts into above function. Support distribution for the innermost
3986 two-level loop nest. Adjust dump information.
3988 2017-10-12 Bin Cheng <bin.cheng@arm.com>
3990 * tree-loop-distribution.c: Adjust the general comment.
3991 (NUM_PARTITION_THRESHOLD): New macro.
3992 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
3993 (classify_partition): Skip builtin pattern of loop nest's inner loop.
3994 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
3995 in call to build_partition_graph.
3996 (finalize_partitions): New parameter. Make loop distribution more
3997 conservative by fusing more partitions.
3998 (distribute_loop): Don't do runtime alias check in case of loop nest
4000 (find_seed_stmts_for_distribution): New function.
4001 (prepare_perfect_loop_nest): New function.
4002 (pass_loop_distribution::execute): Refactor code finding seed stmts
4003 and loop nest into above functions. Support loop nest distribution.
4004 Adjust dump information accordingly.
4006 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4008 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
4009 and set PTYPE_SEQUENTIAL for merged partition.
4011 2017-10-12 Richard Biener <rguenther@suse.de>
4013 PR tree-optimization/69728
4015 2017-09-19 Richard Biener <rguenther@suse.de>
4017 PR tree-optimization/69728
4018 * graphite-sese-to-poly.c (schedule_error): New global.
4019 (add_loop_schedule): Handle empty domain by failing the
4021 (build_original_schedule): Handle schedule_error.
4023 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
4024 domain by returning an unchanged schedule.
4026 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4028 * genrecog.c (validate_pattern): For VEC_SELECT verify that
4029 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
4031 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
4033 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
4036 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4039 * expr.c (store_field): Don't optimize away bitsize == 0 store
4040 from CALL_EXPR with addressable return type.
4042 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4044 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
4045 * config/rs6000/rs6000.md (sel): Delete mode attribute.
4046 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
4047 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
4048 TARGET_ISEL instead of TARGET_ISEL<sel>.
4050 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
4052 * config/rs6000/rs6000.c
4053 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
4055 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4057 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
4059 (reg_or_cint_operand): Fix comment.
4060 (reg_or_zero_operand): New predicate.
4061 * config/rs6000/rs6000-protos.h (output_isel): Delete.
4062 * config/rs6000/rs6000.c (output_isel): Delete.
4063 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
4064 instead of reg_or_cint_operand. Output instruction directly (not via
4066 (isel_unsigned_<mode>): Ditto.
4067 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
4068 gpc_reg_operand. Add an instruction alternative for this. Output
4069 instruction directly.
4070 (*isel_reversed_unsigned_<mode>): Ditto.
4072 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4074 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
4075 (TARGET_CANONICALIZE_COMPARISON): Define.
4077 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
4080 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
4081 Check whether the dest is REG before adding REG_EQUIV note.
4083 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4086 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4088 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4089 (make_hard_regno_born, make_hard_regno_dead): Update
4090 bb_killed_pseudos and bb_gen_pseudos.
4092 2017-10-11 Nathan Sidwell <nathan@acm.org>
4094 * incpath.h (enum incpath_kind): Name enum, prefix values.
4095 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
4096 * incpath.c (heads, tails): Use INC_MAX.
4097 (add_env_var_paths, add_standard_paths): Use incpath_kind.
4098 (merge_include_chains, split_quote_chain,
4099 register_include_chains): Update incpath_kind names.
4100 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
4101 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
4103 (add_framework_path, darwin_register_objc_includes): Likewise.
4104 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
4106 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4108 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
4109 Do not use float_operator operator predicate.
4110 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
4111 * config/i386/predicates.md (float_operator): Remove predicate.
4113 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4115 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
4116 (*jccxf_i387): Ditto.
4117 (*jcc<mode>_i387): Ditto.
4118 (*jccu<mode>_i387): Ditto.
4119 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
4120 (*jcc_*_i387 splitters): Remove.
4121 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
4122 * config/i386/i386.c (ix86_split_fp_branch): Remove.
4123 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
4126 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4128 * profile-count.h (slow_safe_scale_64bit): New function.
4129 (safe_scale_64bit): New inline.
4130 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
4131 * profile-count.c: Include wide-int.h
4132 (slow_safe_scale_64bit): New.
4134 2017-10-11 Nathan Sidwell <nathan@acm.org>
4136 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
4137 HAS_DECL_ASSEMBLER_NAME_P.
4138 * gimple-expr.c (gimple_decl_printable_name: Check
4139 HAS_DECL_ASSEMBLER_NAME_P too.
4140 * ipa-utils.h (type_in_anonymous_namespace_p): Check
4141 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
4142 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
4143 * passes.c (rest_of_decl_compilation): Check
4144 HAS_DECL_ASSEMBLER_NAME_P too.
4145 * recog.c (verify_changes): Likewise.
4146 * tree-pretty-print.c (dump_decl_name): Likewise.
4147 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
4149 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
4150 (SET_DECL_ASSEMBLER_NAME): Use it.
4151 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
4152 (COPY_DECL_ASSEMBLER_NAME): Likewise.
4153 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
4155 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4157 * config.gcc (i386, x86_64): Add extra objects.
4158 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
4159 (ix86_min_insn_size): Declare.
4160 (ix86_issue_rate): Declare.
4161 (ix86_adjust_cost): Declare.
4162 (ia32_multipass_dfa_lookahead): Declare.
4163 (ix86_macro_fusion_p): Declare.
4164 (ix86_macro_fusion_pair_p): Declare.
4165 (ix86_bd_has_dispatch): Declare.
4166 (ix86_bd_do_dispatch): Declare.
4167 (ix86_core2i7_init_hooks): Declare.
4168 (ix86_atom_sched_reorder): Declare.
4169 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
4170 (COSTS_N_BYTES): Move to x86-tune-costs.h.
4171 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
4172 (rip_relative_addr_p): Rename to ...
4173 (ix86_rip_relative_addr_p): ... this one; export.
4174 (memory_address_length): Update.
4175 (ix86_issue_rate): Move to x86-tune-sched.c.
4176 (ix86_flags_dependent): Move to x86-tune-sched.c.
4177 (ix86_agi_dependent): Move to x86-tune-sched.c.
4178 (exact_dependency_1): Move to x86-tune-sched.c.
4179 (exact_store_load_dependency): Move to x86-tune-sched.c.
4180 (ix86_adjust_cost): Move to x86-tune-sched.c.
4181 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
4182 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
4183 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
4184 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
4185 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
4186 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
4187 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
4188 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
4189 (min_insn_size): Rename to ...
4190 (ix86_min_insn_size): ... this one; export.
4191 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
4192 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
4193 (core2i7_first_cycle_multipass_backtrack): Move to
4194 x86-tune-sched-core.c.
4195 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
4196 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
4197 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
4198 (ix86_avoid_jump_mispredicts): Update.
4199 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
4200 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
4201 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
4202 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
4203 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
4204 (MAX_INSN): Move to ix86-tune-sched-bd.c.
4205 (MAX_IMM): Move to ix86-tune-sched-bd.c.
4206 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
4207 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
4208 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
4209 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
4210 (MAX_STORE): Move to ix86-tune-sched-bd.c.
4211 (BIG): Move to ix86-tune-sched-bd.c.
4212 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
4213 (enum insn_path): Move to ix86-tune-sched-bd.c.
4214 (get_mem_group): Move to ix86-tune-sched-bd.c.
4215 (is_cmp): Move to ix86-tune-sched-bd.c.
4216 (dispatch_violation): Move to ix86-tune-sched-bd.c.
4217 (is_branch): Move to ix86-tune-sched-bd.c.
4218 (is_prefetch): Move to ix86-tune-sched-bd.c.
4219 (init_window): Move to ix86-tune-sched-bd.c.
4220 (allocate_window): Move to ix86-tune-sched-bd.c.
4221 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
4222 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
4223 (process_end_window): Move to ix86-tune-sched-bd.c.
4224 (allocate_next_window): Move to ix86-tune-sched-bd.c.
4225 (find_constant): Move to ix86-tune-sched-bd.c.
4226 (get_num_immediates): Move to ix86-tune-sched-bd.c.
4227 (has_immediate): Move to ix86-tune-sched-bd.c.
4228 (get_insn_path): Move to ix86-tune-sched-bd.c.
4229 (get_insn_group): Move to ix86-tune-sched-bd.c.
4230 (count_num_restricted): Move to ix86-tune-sched-bd.c.
4231 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
4232 (add_insn_window): Move to ix86-tune-sched-bd.c.
4233 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
4234 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
4235 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
4236 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
4237 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
4238 (do_dispatch): Move to ix86-tune-sched-bd.c.
4239 (has_dispatch): Move to ix86-tune-sched-bd.c.
4240 * config/i386/t-i386: Add new object files.
4241 * config/i386/x86-tune-costs.h: New file.
4242 * config/i386/x86-tune-sched-atom.c: New file.
4243 * config/i386/x86-tune-sched-bd.c: New file.
4244 * config/i386/x86-tune-sched-core.c: New file.
4245 * config/i386/x86-tune-sched.c: New file.
4247 2017-10-11 Liu Hao <lh_mouse@126.com>
4249 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
4250 the generic version below instead.
4251 (should_colorize): Recognize Windows consoles as terminals
4253 * pretty-print.c [__MINGW32__] (write_all): New function.
4254 [__MINGW32__] (find_esc_head): Likewise.
4255 [__MINGW32__] (find_esc_terminator): Likewise.
4256 [__MINGW32__] (eat_esc_sequence): Likewise.
4257 [__MINGW32__] (mingw_ansi_fputs): New function that handles
4259 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
4262 2017-10-11 Richard Biener <rguenther@suse.de>
4264 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
4265 Properly call analyze_scalar_evolution with the loop of the stmt.
4267 2017-10-11 Richard Biener <rguenther@suse.de>
4269 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
4270 * tree-core.h (tree_base): Add chrec_var union member.
4271 * tree.h (CHREC_VAR): Remove.
4272 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
4273 * tree-chrec.h (build_polynomial_chrec): Adjust.
4274 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
4275 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
4277 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
4279 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
4280 * match.pd: ... here.
4281 ((T) X == (T) Y): Relax condition.
4283 2017-10-11 Bin Cheng <bin.cheng@arm.com>
4285 PR tree-optimization/82472
4286 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
4288 (break_alias_scc_partitions): Update postorder number.
4290 2017-10-11 Martin Liska <mliska@suse.cz>
4293 * opts.c (parse_no_sanitize_attribute): Do not use error_value
4295 * opts.h (parse_no_sanitize_attribute): Remove last argument.
4297 2017-10-11 Martin Liska <mliska@suse.cz>
4299 * print-rtl.c (print_insn): Move declaration of idbuf
4300 to same scope as name.
4302 2017-10-11 Martin Liska <mliska@suse.cz>
4307 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
4310 2017-10-11 Martin Liska <mliska@suse.cz>
4313 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
4316 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
4318 * config.gcc (default_gnu_indirect_function): Default to yes for
4319 arm*-*-linux* with glibc.
4321 2017-10-11 Richard Biener <rguenther@suse.de>
4323 * tree-scalar-evolution.c (get_scalar_evolution): Handle
4324 default-defs and types we do not want to analyze.
4325 (interpret_loop_phi): Replace unreachable code with an assert.
4326 (compute_scalar_evolution_in_loop): Remove and inline ...
4327 (analyze_scalar_evolution_1): ... here, replacing condition with
4328 what makes the intent clearer. Remove handling of cases
4329 get_scalar_evolution now handles.
4331 2017-10-10 Jim Wilson <wilson@tuliptree.org>
4333 PR rtl-optimization/81434
4334 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
4335 comment for main loop. In sched_group_found if, also add checks for
4336 pass and min_cost_group.
4338 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
4340 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
4341 (rs6000_insn_cost): New function.
4342 * config/rs6000/rs6000.md (cost): New attribute.
4344 2017-10-10 Jakub Jelinek <jakub@redhat.com>
4345 H.J. Lu <hongjiu.lu@intel.com>
4349 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
4350 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
4351 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
4352 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
4353 to OPTION_MASK_ISA_AVX512VL - builtins that have both
4354 OPTION_MASK_ISA_MMX and some other bit set require both
4355 mmx and the ISAs without the mmx bit.
4356 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
4357 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
4358 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
4359 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
4360 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
4361 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
4362 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
4363 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
4364 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
4365 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
4366 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
4367 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
4368 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
4369 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
4370 Add OPTION_MASK_ISA_MMX.
4372 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
4374 * config.gcc (armv7*-*-freebsd*): New target.
4375 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
4377 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
4379 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
4380 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
4383 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
4385 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
4386 a template parameter.
4387 (WIDE_INT_REF_FOR): Update accordingly.
4388 * tree.h (wi::int_traits <const_tree>): Delete.
4389 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
4390 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
4391 (wi::tree_to_wide_ref): New typedef.
4392 (wi::to_wide): New function.
4393 * calls.c (get_size_range): Use wi::to_wide when operating on
4395 * cgraph.c (cgraph_node::create_thunk): Likewise.
4396 * config/i386/i386.c (ix86_data_alignment): Likewise.
4397 (ix86_local_alignment): Likewise.
4398 * dbxout.c (stabstr_O): Likewise.
4399 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
4400 * expr.c (const_vector_from_tree): Likewise.
4401 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
4402 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
4403 (fold_negate_expr_1, int_const_binop_1, const_binop)
4404 (fold_convert_const_int_from_real, optimize_bit_field_compare)
4405 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
4406 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
4407 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
4408 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
4409 (fold_not_const, round_up_loc): Likewise.
4410 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
4411 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
4412 (alloca_call_type): Likewise.
4413 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
4414 * godump.c (go_output_typedef): Likewise.
4415 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
4416 * internal-fn.c (get_min_precision): Likewise.
4417 * ipa-cp.c (ipcp_store_vr_results): Likewise.
4418 * ipa-polymorphic-call.c
4419 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
4420 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
4421 (ipa_modify_call_arguments): Likewise.
4422 * match.pd: Likewise.
4423 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
4424 * print-tree.c (print_node_brief, print_node): Likewise.
4425 * stmt.c (expand_case): Likewise.
4426 * stor-layout.c (layout_type): Likewise.
4427 * tree-affine.c (tree_to_aff_combination): Likewise.
4428 * tree-cfg.c (group_case_labels_stmt): Likewise.
4429 * tree-data-ref.c (dr_analyze_indices): Likewise.
4430 (prune_runtime_alias_test_list): Likewise.
4431 * tree-dump.c (dequeue_and_dump): Likewise.
4432 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
4433 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
4434 * tree-pretty-print.c (dump_generic_node): Likewise.
4435 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
4436 (simple_iv_with_niters): Likewise.
4437 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
4438 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
4439 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
4440 * tree-ssa-loop-niter.c (split_to_var_and_offset)
4441 (refine_value_range_using_guard, number_of_iterations_ne_max)
4442 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
4443 (get_cst_init_from_scev, record_nonwrapping_iv)
4444 (scev_var_range_cant_overflow): Likewise.
4445 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
4446 * tree-ssa-pre.c (compute_avail): Likewise.
4447 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
4448 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
4449 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
4450 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
4451 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
4452 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
4453 (dump_case_nodes, try_switch_expansion): Likewise.
4454 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
4455 (vect_do_peeling): Likewise.
4456 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
4457 * tree-vect-stmts.c (vectorizable_load): Likewise.
4458 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
4459 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
4460 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
4461 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
4462 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
4463 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
4464 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
4465 (evrp_dom_walker::before_dom_children): Likewise.
4466 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
4467 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
4468 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
4469 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
4470 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
4471 (get_range_pos_neg): Likewise.
4472 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
4473 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
4474 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
4475 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
4476 * config/avr/avr.c (avr_fold_builtin): Likewise.
4477 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
4478 * config/msp430/msp430.c (msp430_attr): Likewise.
4479 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
4480 * config/powerpcspe/powerpcspe-c.c
4481 (altivec_resolve_overloaded_builtin): Likewise.
4482 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
4483 (rs6000_expand_ternop_builtin): Likewise.
4484 * config/rs6000/rs6000-c.c
4485 (altivec_resolve_overloaded_builtin): Likewise.
4486 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
4487 (rs6000_expand_ternop_builtin): Likewise.
4488 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
4490 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4492 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
4493 when copying loop nest with only one inner loop.
4495 2017-10-10 Richard Biener <rguenther@suse.de>
4497 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
4498 blocks if SCEV is active.
4499 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
4501 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
4502 (scev_initialize): Assert we are not yet initialized.
4504 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4506 * tree-loop-distribution.c (generate_loops_for_partition): Remove
4507 inner loop's exit stmt by making it always exit the loop, otherwise
4508 we would generate an infinite empty loop.
4510 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4512 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
4513 renaming variables in new preheader if it's deleted.
4515 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4517 * tree-loop-distribution.c (struct partition): Remove unused field
4518 loops of the structure.
4519 (partition_alloc, partition_free): Ditto.
4520 (build_rdg_partition_for_vertex): Ditto.
4522 2017-10-09 Jeff Law <law@redhat.com>
4524 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
4525 return type to match prototype and documentation.
4527 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4529 * config/rs6000/rs6000.c (processor_costs): Move to ...
4530 * config/rs6000/rs6000.h: ... here.
4531 (rs6000_cost): Declare.
4533 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
4535 * except.c (setjmp_fn): New global variable.
4536 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
4537 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
4538 if DONT_USE_BUILTIN_SETJMP is defined.
4540 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4542 * target.def (insn_cost): New hook.
4543 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
4544 * doc/tm.texi: Regenerate.
4545 * rtlanal.c (insn_cost): Use the new hook.
4547 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4549 * combine.c (combine_validate_cost): Compute the new insn_cost,
4550 not just pattern_cost.
4551 (try_combine): Adjust comment.
4553 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4555 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
4557 * combine.c (uid_insn_cost): Adjust comment.
4558 (combine_validate_cost): Adjust comment. Use pattern_cost instead
4560 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
4561 * dse.c (find_shift_sequence): Ditto.
4562 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
4563 (bb_valid_for_noce_process_p): Use pattern_cost.
4564 * rtl.h (insn_rtx_cost): Delete.
4565 (pattern_cost): New prototype.
4566 (insn_cost): New prototype.
4567 * rtlanal.c (insn_rtx_cost): Rename to...
4568 (pattern_cost): ... this.
4571 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
4573 * config/i386/i386.md (*jcc_2): Remove insn pattern.
4574 (*jcc<mode>_0_r_i387): Ditto.
4575 (*jccxf_r_i387): Ditto.
4576 (*jcc<mode>_r_i387): Ditto.
4577 (*jccu<mode>_r_i387): Ditto.
4578 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
4579 (*jcc): Rename from *jcc_1.
4581 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4583 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
4584 deferred rescans after the lvx/stvx recombination pre-pass.
4586 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
4588 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
4589 memory operation instruction support.
4590 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
4591 (rs6000-ibm-aix[789]*): Likewise.
4592 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
4593 Document new functions.
4595 2017-10-09 Richard Biener <rguenther@suse.de>
4597 PR tree-optimization/82397
4598 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
4599 equality only for semantically equal trees.
4601 2017-10-09 Richard Biener <rguenther@suse.de>
4603 PR tree-optimization/82449
4604 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
4605 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
4606 allow constant addresses.
4607 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
4610 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4612 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
4615 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4618 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
4621 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4624 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
4626 2017-10-09 Jakub Jelinek <jakub@redhat.com>
4629 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
4630 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
4632 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
4634 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
4635 (WI_BINARY_PREDICATE_RESULT): Likewise.
4636 (wi::binary_traits::operator_result): New type.
4637 (wi::binary_traits::predicate_result): Likewise.
4638 (generic_wide_int::operator~, unary generic_wide_int::operator-)
4639 (generic_wide_int::operator==, generic_wide_int::operator!=)
4640 (generic_wide_int::operator&, generic_wide_int::and_not)
4641 (generic_wide_int::operator|, generic_wide_int::or_not)
4642 (generic_wide_int::operator^, generic_wide_int::operator+
4643 (binary generic_wide_int::operator-, generic_wide_int::operator*):
4645 (operator~, unary operator-, operator==, operator!=, operator&)
4646 (operator|, operator^, operator+, binary operator-, operator*): New
4648 * expr.c (get_inner_reference): Use wi::bit_and_not.
4649 * fold-const.c (fold_binary_loc): Likewise.
4650 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
4651 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
4652 (bit_value_binop): Likewise.
4653 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
4654 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
4655 (extract_range_from_binary_expr_1): Likewise.
4656 (masked_increment): Likewise.
4657 (simplify_bit_ops_using_ranges): Likewise.
4659 2017-10-09 Martin Jambor <mjambor@suse.cz>
4662 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
4663 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
4664 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
4666 (hsa_fixup_mov_insn_type): New function.
4667 (hsa_op_with_type::get_in_type): Use it.
4668 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
4669 immediates in an assert.
4670 (hsa_op_with_type::extend_int_to_32bit): New method.
4671 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
4672 types. Convert to dest type if necessary.
4673 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
4674 (reg_for_gimple_ssa): Pass false as min32int to
4675 hsa_type_for_scalar_tree_type.
4676 (gen_hsa_addr): Fixup type when creating addresable temporary.
4677 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
4678 (gen_hsa_unary_operation): Extend operands and convert to dest type if
4679 necessary. Call hsa_fixup_mov_insn_type.
4680 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
4681 extend operands and convert to dest type if necessary.
4682 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
4683 to dest type if necessary.
4684 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
4685 if conversion nt necessary and size matches.
4686 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
4687 to dest type if necessary.
4688 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
4689 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
4691 (gen_hsa_clrsb): Likewise.
4692 (gen_hsa_ffs): Likewise.
4693 (gen_hsa_divmod): Extend operands and convert to dest type if
4695 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
4697 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
4699 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
4700 Remove empty default arguments. Use a brace block as output
4702 (conditional return): Ditto.
4704 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
4705 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
4707 (group_ending_nop): Ditto.
4708 (doloop_end): Ditto.
4709 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
4710 (splitters for those): Ditto.
4712 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
4714 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
4715 a conditional jump (and the compare for it) so that pc_rtx is the
4717 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
4718 for the deleted and renamed ctr<mode>_internal[234] patterns.
4719 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
4720 Delete second conditional return pattern.
4721 (ctr<mode>_internal2): Delete this second bdnz pattern.
4722 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
4723 (ctr<mode>_internal4): Delete this second bdz pattern.
4725 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
4727 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
4728 (always_initialized_rtx_for_ssa_name_p): New predicate.
4729 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
4730 (finish_out_of_ssa): Free new field of SA.
4731 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
4732 * tree-ssa-coalesce.c: Include tree-ssa.h.
4733 (get_parm_default_def_partitions): Remove extern keyword.
4734 (get_undefined_value_partitions): New function.
4735 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
4736 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
4738 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
4740 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
4742 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
4744 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
4746 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
4747 for targets that preffer 128bit.
4749 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
4751 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
4753 2017-10-08 Olivier Hainque <hainque@adacore.com>
4755 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
4756 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
4758 (thumb_set_return_address): Likewise.
4760 2017-10-08 Olivier Hainque <hainque@adacore.com>
4762 * common/config/arm/arm-common.c (arm_except_unwind_info):
4763 Handle DWARF2_UNWIND_INFO.
4765 2017-10-07 Michael Collison <michael.collison@arm.com>
4767 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
4770 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
4772 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
4773 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
4774 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
4775 defined, force the creation of a new block for a dispatch label.
4777 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
4779 * invoke.texi (Wsuggest-attribute=cold): Document.
4780 * common.opt (Wsuggest-attribute=cold): New
4781 * ipa-pure-const.c (warn_function_cold): New function.
4782 * predict.c (compute_function_frequency): Use it.
4783 * predict.h (warn_function_cold): Declare.
4785 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
4787 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
4790 2017-10-06 Martin Liska <mliska@suse.cz>
4792 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
4793 keyword for member functions.
4794 (struct sanopt_tree_couple): New struct.
4795 (struct sanopt_tree_couple_hash): New function.
4796 (struct sanopt_ctx): Add new hash_map.
4797 (has_dominating_ubsan_ptr_check): New function.
4798 (record_ubsan_ptr_check_stmt): Likewise.
4799 (maybe_optimize_ubsan_ptr_ifn): Likewise.
4800 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
4801 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
4803 2017-10-06 Sudakshina Das <sudi.das@arm.com>
4806 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
4807 aarch64_simd_valid_immediate on CONST_VECTORs.
4808 (aarch64_reg_or_bic_imm): Likewise.
4810 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
4812 PR rtl-optimization/82396
4813 * haifa-sched.c (ready_sort_real): Disable qsort checking.
4815 2017-10-06 Sebastian Pop <sebpop@gmail.com>
4817 * graphite-dependences.c (scop_get_reads): Move code to...
4818 (scop_get_must_writes): Move code to...
4819 (scop_get_may_writes): Move code to...
4820 (scop_get_reads_and_writes): ... here.
4821 (scop_get_dependences): Call scop_get_reads_and_writes.
4823 2017-10-06 Jakub Jelinek <jakub@redhat.com>
4825 PR tree-optimization/82434
4826 * fold-const.h (can_native_encode_type_p,
4827 can_native_encode_string_p): Remove.
4828 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
4829 don't encode anything, just return what would be otherwise returned.
4830 (native_encode_fixed, native_encode_complex, native_encode_vector):
4832 (native_encode_string): Likewise. Inline by hand
4833 can_native_encode_string_p.
4834 (can_native_encode_type_p): Remove.
4835 (can_native_encode_string_p): Remove.
4836 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
4837 STRING_CSTs using can_native_encode_string_p, test all
4838 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
4839 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
4840 argument from native_encode_expr.
4841 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
4842 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
4845 2017-10-06 Richard Biener <rguenther@suse.de>
4847 PR tree-optimization/82397
4848 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
4849 operand_equal_p but rely on data_ref_compare_tree for detecting
4851 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
4852 to match up with dr_group_sort_cmp.
4854 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4857 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
4859 * config/s390/s390-builtin-types.def: Regenerate.
4861 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4864 * config/s390/s390-builtin-types.def: Regenerate.
4865 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
4866 Change flag from B_VXE to B_VX.
4867 (s390_vec_min_dbl): Remove B_VXE flag.
4869 2017-10-06 Richard Biener <rguenther@suse.de>
4871 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
4872 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
4873 translate_isl_ast_to_gimple::is_valid_rename,
4874 translate_isl_ast_to_gimple::get_rename,
4875 translate_isl_ast_to_gimple::get_def_bb_for_const,
4876 translate_isl_ast_to_gimple::get_new_name,
4877 translate_isl_ast_to_gimple::collect_all_ssa_names,
4878 translate_isl_ast_to_gimple::copy_loop_phi_args,
4879 translate_isl_ast_to_gimple::collect_all_ssa_names,
4880 translate_isl_ast_to_gimple::copy_loop_phi_args,
4881 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
4882 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
4883 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
4884 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
4885 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
4886 translate_isl_ast_to_gimple::copy_cond_phi_args,
4887 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
4888 translate_isl_ast_to_gimple::edge_for_new_close_phis,
4889 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
4890 translate_isl_ast_to_gimple::rename_uses,
4891 translate_isl_ast_to_gimple::rename_all_uses): Remove.
4892 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
4893 (set_rename_for_each_def): Likewise.
4894 (graphite_copy_stmts_from_block): Handle debug stmt resetting
4895 here. Handle rewriting SCEV analyzable uses here.
4896 (copy_bb_and_scalar_dependences): Generate code for PHI
4898 (graphite_regenerate_ast_isl): Adjust.
4899 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
4900 (add_write, add_read): New functions.
4901 (build_cross_bb_scalars_def): Use it and simplify.
4902 (build_cross_bb_scalars_use): Likewise.
4903 (graphite_find_cross_bb_scalar_vars): Inline into...
4904 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
4905 simulating out-of-SSA. Compute liveout and add dependencies.
4906 (build_scops): Force an empty entry block.
4907 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
4909 (sese_build_liveouts): Declare.
4910 (sese_trivially_empty_bb_p): Likewise.
4911 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
4912 compute liveout and debug_liveout.
4913 (sese_bad_liveouts_use): Remove.
4914 (sese_reset_debug_liveouts_bb): Likewise.
4915 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
4916 (sese_build_liveouts): Build liveout and debug_liveout and store
4918 (new_sese_info): Adjust.
4919 (free_sese_info): Likewise.
4920 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
4921 do not build liveout here.
4922 (move_sese_in_condition): Adjust region entry.
4923 (scev_analyzable_p): Match up with chrec_apply requirements.
4924 (sese_trivially_empty_bb_p): New.
4925 * tree-into-ssa.c (get_reaching_def): Properly support generating
4926 default-defs for incremental rewrite of anonymous names.
4928 2017-10-06 Richard Biener <rguenther@suse.de>
4930 * graphite-sese-to-poly.c (extract_affine): For casts increasing
4931 precision do not perform modulo reduction.
4933 2017-10-06 Richard Biener <rguenther@suse.de>
4935 PR tree-optimization/82436
4936 * tree-vect-slp.c (vect_supported_load_permutation_p): More
4937 conservatively choose the vectorization factor when checking
4938 whether we can perform the required load permutation.
4939 (vect_transform_slp_perm_load): Assert when we may not fail.
4941 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
4943 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
4944 message for incompatible -msdata=* and -mcall-* options.
4946 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
4948 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
4949 rate for post-reload scheduling.
4951 2017-10-05 Tamar Christina <tamar.christina@arm.com>
4953 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
4955 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
4957 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
4958 to improve monte carlo in scimark.
4960 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
4962 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
4963 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
4964 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
4965 pentium4_cost, nocona_cost): Set reassociation width to 1.
4966 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
4967 width to 2 for fp operations and 1 otherwise.
4968 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
4970 (atom_cost): Set reassociation width to 2.
4971 (slm_cost, generic_cost): Set fp reassociation width
4972 to 2 and 1 otherwise.
4973 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
4974 (core_cost): Set fp reassociation width to 4 and vector to 2.
4975 (ix86_reassociation_width): Rewrite using cost table; special case
4976 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
4977 and TARGET_AVX128_OPTIMAL.
4978 * config/i386/i386.h (processor_costs): Add
4979 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
4980 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
4981 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
4982 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
4983 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
4984 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
4986 2017-10-05 Nathan Sidwell <nathan@acm.org>
4988 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
4990 2017-10-05 Tamar Christina <tamar.christina@arm.com>
4992 * config/arm/arm.c (arm_test_fpu_data): New.
4993 (arm_run_selftests): Call arm_test_fpu_data.
4995 2017-10-04 Nathan Sidwell <nathan@acm.org>
4997 * toplev.c (toplev::main): Remove excess parens on pretty_printer
4999 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
5001 2017-10-04 Sudakshina Das <sudi.das@arm.com>
5003 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
5004 check type for aarch64_simd_valid_immediate.
5005 (aarch64_output_simd_mov_immediate): Update prototype.
5006 (aarch64_simd_valid_immediate): Update prototype.
5007 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
5008 support for ORR-immediate.
5009 (and<mode>3): modified pattern to add support for BIC-immediate.
5010 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
5011 now checks for valid immediate for BIC and ORR based on new enum
5013 (aarch64_output_simd_mov_immediate): Function now used to output
5014 BIC/ORR imm as well based on new enum argument.
5015 * config/aarch64/constraints.md (Do): New vector immediate constraint.
5017 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
5018 (aarch64_reg_or_bic_imm): Likewise.
5020 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5022 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
5023 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
5025 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5029 PR rtl-optimization/82396
5030 * haifa-sched.c (autopref_multipass_init): Simplify
5032 (autopref_rank_data): Simplify sort order.
5033 * sched-int.h (autopref_multipass_data_): Remove
5034 multi_mem_insn_p, min_offset and max_offset.
5036 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5038 * doc/sourcebuild.texi: Document vect_peeling_profitable.
5040 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5042 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
5045 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5047 * doc/sourcebuild.texi: Document vect_long_mult.
5049 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
5051 PR tree-optimization/82413
5052 * fold-const.c (build_range_check): Use widest_int when comparing
5053 the maximum ETYPE value with HIGH.
5055 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5057 PR rtl-optimization/82396
5058 * haifa-sched.c (autopref_multipass_init): Simplify
5060 (autopref_rank_data): Simplify sort order.
5061 * sched-int.h (autopref_multipass_data_): Remove
5062 multi_mem_insn_p, min_offset and max_offset.
5064 2017-10-04 Jakub Jelinek <jakub@redhat.com>
5066 PR tree-optimization/82381
5067 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
5068 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
5071 PR tree-optimization/82374
5072 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
5073 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
5074 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
5075 current_function_decl to the new decl.
5077 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
5079 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
5080 helper macro for IEEE float128 hardware built-in functions.
5081 (SQRTF128_ODD): Add built-in functions with the round-to-odd
5083 (TRUNCF128_ODD): Likewise.
5084 (ADDF128_ODD): Likewise.
5085 (SUBF128_ODD): Likewise.
5086 (MULF128_ODD): Likewise.
5087 (DIVF128_ODD): Likewise.
5088 (FMAF128_ODD): Likewise.
5089 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
5090 UNSPEC_TRUNC_ROUND_TO_ODD.
5091 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
5092 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
5093 floating point round to odd instructions.
5094 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
5095 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
5096 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
5097 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
5098 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
5099 (trunc<mode>sf2_hw): Change the truncate with round to odd
5100 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
5101 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
5102 to odd hardware instructions.
5103 (sub<mode>3_odd): Likewise.
5104 (mul<mode>3_odd): Likewise.
5105 (div<mode>3_odd): Likewise.
5106 (sqrt<mode>2_odd): Likewise.
5107 (fma<mode>4_odd): Likewise.
5108 (fms<mode>4_odd): Likewise.
5109 (nfma<mode>4_odd): Likewise.
5110 (nfms<mode>4_odd): Likewise.
5111 (trunc<mode>df2_odd): Change the truncate with round to odd
5112 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
5114 * doc/extend.texi (PowerPC built-in functions): Update documentation
5115 for existing IEEE float128-bit built-in functions. Add built-in
5116 functions that generate the IEEE 128-bit floating point round to
5119 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
5121 PR rtl-optimization/77729
5122 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
5123 to (X&(C1&~C2))|C2 transformations.
5125 2017-10-03 Martin Jambor <mjambor@suse.cz>
5127 PR tree-optimization/82363
5128 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
5129 mismatch, mark lacc written regardless of racc.
5131 2017-10-03 Jakub Jelinek <jakub@redhat.com>
5133 PR tree-optimization/82381
5134 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
5135 stmt_to_insert nor wheather SSA_NAMEs are default defs.
5136 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
5137 fallthrough into reassoc_stmt_dominates_stmt_p.
5140 * combine.c (combine_instructions): Don't combine in unreachable
5143 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
5146 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
5147 function to not use the have_cpu variable. Do not set cpu_index,
5148 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
5150 (rs6000_valid_attribute_p): Remove duplicate initializations of
5151 old_optimize and func_optimize.
5152 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
5153 (rs6000_activate_target_options): Make global.
5154 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
5157 2017-10-02 Jakub Jelinek <jakub@redhat.com>
5159 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
5160 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
5161 Set *poffset to 0 and *psize and *pmax_size to -1 if
5162 *poffset + *psize overflows in HOST_WIDE_INT.
5164 PR tree-optimization/82387
5165 PR tree-optimization/82388
5166 PR tree-optimization/82389
5167 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
5168 instead of live_bytes non-NULL.
5170 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
5173 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
5176 2017-10-02 Richard Biener <rguenther@suse.de>
5178 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
5179 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
5180 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
5182 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
5184 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
5185 requested precision matches the type's.
5186 * calls.c (alloc_max_size): Calculate the new candidate size as
5187 a widest_int and use wi::to_widest when comparing it with the
5188 current candidate size.
5189 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
5190 zero rather than integer_zero_node.
5191 * match.pd: Check for a no-op conversion before using wi::add
5192 rather than after. Use tree_to_uhwi when summing small shift
5193 counts into an unsigned int.
5195 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
5196 Alan Hayward <alan.hayward@arm.com>
5197 David Sherwood <david.sherwood@arm.com>
5200 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
5201 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
5202 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
5203 POINTER_AND_FP_REGS.
5205 2017-10-02 Richard Biener <rguenther@suse.de>
5207 PR tree-optimization/82355
5208 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5209 a mapping for the enclosing loop but avoid generating one for
5211 (copy_bb_and_scalar_dependences): Remove premature codegen
5212 error on PHIs in blocks duplicated into multiple places.
5213 * graphite-scop-detection.c
5214 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5215 in the region use it as loop and nest to analyze the DR in.
5216 (try_generate_gimple_bb): Likewise.
5217 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5218 (add_loop_constraints): For blocks in a loop not in the region
5219 create a dimension with a single iteration.
5220 * sese.h (gbb_loop_at_index): Remove assert.
5222 2017-10-01 Kevin Buettner <kevinb@redhat.com>
5224 * omp-expand.c (adjust_context_scope): New function.
5225 (expand_parallel_call): Call adjust_context_scope.
5227 2017-10-01 Jeff Law <law@redhat.com>
5229 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
5230 dom_opt_dom_walker class with direct access to private members.
5231 Add comments. Call test_for_singularity.
5232 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
5233 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
5234 m_dummy_cond anymore.
5235 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
5237 (pass_dominator:execute): Build the dummy_cond here and pass it
5238 to the dom_opt_dom_walker ctor.
5239 (test_for_singularity): New function.
5241 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
5242 Maya Rashish <coypu@sdf.org>
5244 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
5245 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
5246 (alpha*-*-netbsd*) Use nbsd_tm_file.
5247 (arm*-*-netbsdelf*) Likewise.
5248 (i[34567]86-*-netbsdelf*) Likewise.
5249 (x86_64-*-netbsd*) Likewise.
5250 (mips*-*-netbsd*) Likewise.
5251 (powerpc-*-netbsd*) Likewise.
5252 (sh*-*-netbsd*) Likewise.
5253 (sparc-*-netbsdelf*) Likewise.
5254 (sparc64-*-netbsd*) Likewise.
5255 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
5257 (vax-*-netbsdelf*) Likewise.
5258 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
5259 (UINT_FAST8_TYPE) Likewise.
5260 (INT_FAST16_TYPE) Check CHAR_FAST16.
5261 (UINT_FAST16_TYPE) Likewise.
5263 2017-09-30 Jakub Jelinek <jakub@redhat.com>
5266 * config/i386/i386.md
5267 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
5268 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
5269 *divmodsi4_zext_2): New define_insn_and_split.
5270 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
5271 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
5272 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
5273 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
5274 New define_insn_and_split.
5275 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
5276 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
5277 operands[1] having DImode when mode is SImode.
5279 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
5280 always SImode for DIV and MOD in REG_EQUAL notes.
5282 2017-09-29 Yury Gribov <tetra2005@gmail.com>
5285 * match.pd: Fix handling of NaNs in pattern.
5287 2017-09-29 Jeff Law <law@redhat.com>
5289 * sbitmap.c (bitmap_bit_in_range_p): New function.
5290 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
5291 * tree-ssa-dse.c (live_bytes_read): New function.
5292 (dse_classify_store): Ignore reads of dead bytes.
5294 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
5295 typos and whitespace errors.
5296 * config/i386/predicates.md (address_no_seg_operand): Likewise.
5297 * config/s390/s390.c (s390_emit_prologue): Likewise.
5299 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
5302 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
5303 with a symbol for LRA.
5305 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
5307 PR rtl-optimization/82338
5308 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
5310 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
5312 * genmodes.c (calc_wider_mode): Suppress qsort macro.
5313 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
5314 (qsort_chk): Declare.
5315 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
5316 (qsort_chk): New function.
5318 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5320 PR tree-optimization/82337
5321 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
5322 phi definition if the PHI result appears in an abnormal PHI.
5323 (find_basis_for_base_expr): Don't record a basis if the LHS of the
5324 basis appears in an abnormal PHI.
5326 2017-09-29 Richard Biener <rguenther@suse.de>
5328 * graphite-isl-ast-to-gimple.c
5329 (translate_isl_ast_to_gimple::set_codegen_error): New function.
5330 (binary_op_to_tree): Use it.
5331 (get_rename_from_scev): Likewise.
5332 (copy_loop_phi_nodes): Likewise.
5333 (copy_bb_and_scalar_dependences): Likewise.
5334 (translate_pending_phi_nodes): Likewise.
5336 2017-09-29 Jakub Jelinek <jakub@redhat.com>
5339 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
5340 for movabsq $(i32 << shift), r64.
5342 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
5344 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
5345 index when encoding %esp as %rsp to avoid 0x67 prefix.
5347 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
5349 * config/i386/i386.md (*movsf_internal, *movdf_internal):
5350 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
5352 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
5354 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
5355 Extensions with more than 16 double VFP registers.
5356 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
5357 to_clear_mask and all code related to it. Replace the remaining
5358 entry by a sbitmap and adapt code accordingly.
5360 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
5362 * brig-builtins.def: Change pure attributes to const.
5364 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5366 * config.gcc (default_gnu_indirect_function): Default to yes for
5367 sparc*-*-linux* with glibc.
5369 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5371 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
5372 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
5373 when creating .init_array and .fini_array sections with priority
5376 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
5379 * config/aarch64/aarch64.c
5380 (aarch64_builtin_support_vector_misalignment): Always return false
5381 when misalignment is unknown.
5383 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
5385 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
5386 this function to return false if the definition used by the swap
5387 instruction is artificial, or if the memory address from which the
5388 constant value is loaded is not represented by a base address held
5389 in a register or if the base address register is a frame or stack
5390 pointer. Additionally, return false if the base address of the
5391 loaded constant is a SYMBOL_REF but is not considered to be a
5393 (replace_swapped_load_constant): New function.
5394 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
5395 loaded constant vector with a load of a swapped constant vector.
5397 2017-09-27 Carl Love <cel@us.ibm.com>
5399 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
5400 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
5401 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
5404 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
5406 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
5407 first, always call autopref_rank_data otherwise.
5409 2017-09-27 Richard Biener <rguenther@suse.de>
5411 * graphite-scop-detection.c (find_scop_parameters): Move
5412 loop bound handling ...
5413 (gather_bbs::before_dom_children): ... here, avoiding the need
5414 to build scop_info->loop_nest.
5415 (record_loop_in_sese): Remove.
5416 * sese.h (sese_info_t::loop_nest): Remove.
5417 * sese.c (new_sese_info): Do not allocate loop_nest.
5418 (free_sese_info): Do not free loop_nest.
5420 2017-09-27 Jakub Jelinek <jakub@redhat.com>
5423 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
5424 lhs from calls if the lhs has addressable type.
5426 2017-09-27 Richard Biener <rguenther@suse.de>
5428 * graphite.h (scop::max_alias_set): New member.
5429 * graphite-scop-detection.c: Remove references to non-existing
5430 --param in comments.
5431 (build_alias_sets): Record the maximum alias set used for drs.
5432 (build_scops): Support zero as unlimited for
5433 --param graphite-max-arrays-per-scop.
5434 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
5436 (build_poly_sr_1): ... here. Compute alias set based on the
5437 maximum alias set used for drs rather than
5438 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
5440 2017-09-27 Richard Biener <rguenther@suse.de>
5442 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
5443 --param loop-block-tile-size=0 to disable tiling.
5445 2017-09-27 Richard Biener <rguenther@suse.de>
5447 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
5448 (graphite-max-nb-scop-params): Document special value zero.
5449 * domwalk.h (dom_walker::STOP): New symbolical constant.
5450 (dom_walker::dom_walker): Add optional parameter for bb to
5452 (dom_walker::~dom_walker): Declare.
5453 (dom_walker::before_dom_children): Document STOP return value.
5454 (dom_walker::m_user_bb_to_rpo): New member.
5455 (dom_walker::m_bb_to_rpo): Likewise.
5456 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
5457 mapping here if not provided by the user.
5458 (dom_walker::~dom_walker): Free bb to RPO mapping if not
5459 provided by the user.
5460 (dom_walker::STOP): Define.
5461 (dom_walker::walk): Do not compute bb to RPO mapping here.
5462 Support STOP return value from before_dom_children to stop
5464 * graphite-optimize-isl.c (optimize_isl): If the schedule
5465 is the same still generate code if -fgraphite-identity
5466 or -floop-parallelize-all are given.
5467 * graphite-scop-detection.c: Include cfganal.h.
5468 (gather_bbs::gather_bbs): Get and pass through bb to RPO
5470 (gather_bbs::before_dom_children): Return STOP for BBs
5472 (build_scops): Compute bb to RPO mapping and pass it to
5473 the domwalk. Treat --param graphite-max-nb-scop-params=0
5474 as not limiting the number of params.
5475 * graphite.c (graphite_initialize): Remove limit on the
5476 number of basic-blocks in a function.
5477 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
5478 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
5479 default value of 10.
5481 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5483 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
5484 Adjust code to eliminate needing to do the shift right 32-bits
5485 operation after XSCVDPSPN.
5487 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5489 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
5490 ((X / Y) != 0 -> X >= Y): Likewise.
5492 2017-09-26 Carl Love <cel@us.ibm.com>
5494 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
5495 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
5496 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
5497 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
5498 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
5499 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
5500 definitions and overloading.
5501 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
5502 statement for P9V_BUILTIN_XST_LEN_R.
5503 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
5504 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
5505 define_expand and define_insn for the instructions and builtins.
5506 * doc/extend.texi: Update the built-in documentation file for the new
5508 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
5509 define_insn for the instructions
5511 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
5514 * gcc/config/netbsd-protos.h: New file.
5515 * gcc/config/netbsd.c: New file.
5516 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
5517 * gcc/config/t-netbsd: New file.
5518 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
5519 (tmake_file) Add t-netbsd.
5520 (extra_objs) Add netbsd.o.
5522 2017-09-26 Janus Weil <janus@gcc.gnu.org>
5526 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
5528 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5530 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
5531 sign extension from a vector register to a GPR by doing a 32-bit
5532 direct move and then an EXTSW.
5533 (extendsi<mode>2 splitter): Likewise.
5534 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
5535 right or vector extract after doing XSCVDPSPN. Use
5536 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
5538 (movdi_from_sf_zero_ext): Likewise.
5539 (reload_gpr_from_vsxsf): Likewise.
5540 (p8_mfvsrd_4_disf): Delete, no longer used.
5541 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
5542 then needing to move the SFmode to a GPR to use the XSCVDPSP
5543 instruction instead of FRSP and XSCVDPSPN.
5544 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
5545 it is adjacent to the other XSCVSPDP insns.
5546 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
5547 SFmode to be in traditional Altivec registers.
5548 (vsx_xscvdpspn): Eliminate useless alternative constraint.
5549 (vsx_xscvspdpn): Likewise.
5550 (vsx_xscvspdpn_scalar): Likewise.
5552 2017-09-26 Martin Jambor <mjambor@suse.cz>
5554 * tree-sra.c (compare_access_positions): Put integral types first,
5555 stabilize sorting of integral types, remove conditions putting
5556 non-full-precision integers last.
5557 (sort_and_splice_var_accesses): Disable scalarization if a
5558 non-integert would be represented by a non-full-precision integer.
5560 2017-09-26 Joseph Myers <joseph@codesourcery.com>
5562 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
5563 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
5564 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
5565 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
5566 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
5567 conditionals inside the function instead of around it. Call
5568 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
5569 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
5571 2017-09-26 Richard Biener <rguenther@suse.de>
5573 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
5575 (scop_detection::build_scop_breadth): ... this. Removed.
5576 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
5577 (scop_detection::harmful_stmt_in_bb): Likewise.
5578 (scop_detection::graphite_can_represent_stmt): Likewise.
5579 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
5580 (scop_detection::can_represent_loop): Remove recursion, fold in ...
5581 (scop_detection::can_represent_loop_1): ... this. Removed.
5582 (scop_detection::harmful_loop_in_region): Simplify after inlining
5583 the above and remove more quadraticness.
5584 (build_scops): Adjust.
5585 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
5588 2017-09-26 Jakub Jelinek <jakub@redhat.com>
5591 * config/i386/i386.c (ix86_print_operand_address_as): Only test
5592 REGNO (base) == SP_REG if base is a REG.
5595 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
5596 if it is different SSA_NAME.
5597 (optimize_range_tests_cmp_bitwise): New function.
5598 (optimize_range_tests): Call it.
5600 2017-09-26 Richard Biener <rguenther@suse.de>
5602 PR tree-optimization/82321
5603 * graphite.c (canonicalize_loop_closed_ssa): Properly check
5604 for the def being inside the loop.
5606 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5608 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
5610 * config/s390/s390-builtins.def: Fix constraint on op4.
5612 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5614 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
5615 independent expanders.
5616 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
5617 ("vec_ordered", "vec_unordered"): New expanders.
5619 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5621 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
5624 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5626 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
5627 vec_unpacks_lo_v16qi.
5628 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
5630 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5632 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
5633 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
5634 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
5636 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5638 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
5640 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
5642 ("*vec_slb<mode>"): New insn pattern.
5643 ("vec_shr_<mode>"): New expander.
5644 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
5645 and force the shift count operand to V16QImode.
5646 ("vec_srb<mode>"): Set shift count mode to V16QI.
5648 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5650 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
5651 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
5652 ("vec_widen_smult_hi_<mode>"): New expander definitions.
5654 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
5657 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
5659 2017-09-26 Richard Biener <rguenther@suse.de>
5661 PR tree-optimization/82320
5662 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
5665 2017-09-25 Jeff Law <law@redhat.com>
5667 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
5668 prototype for new argument.
5669 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
5670 mostly extracted from rs6000_emit_allocate_stack.
5671 (rs6000_emit_probe_stack_range_stack_clash): New function.
5672 (rs6000_emit_allocate_stack): Call
5673 rs6000_emit_probe_stack_range_stack_clash as needed.
5674 (rs6000_emit_probe_stack_range): Add additional argument
5675 to call to gen_probe_stack_range{si,di}.
5676 (output_probe_stack_range): New.
5677 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
5678 (output_probe_stack_range_stack_clash): New.
5679 (rs6000_emit_prologue): Emit notes into dump file as requested.
5680 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
5681 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
5682 Add additional operand and pass it to output_probe_stack_range.
5684 2017-09-25 Bin Cheng <bin.cheng@arm.com>
5686 PR tree-optimization/82163
5687 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
5688 (checking_verify_loop_closed_ssa): New parameter.
5689 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
5690 (check_loop_closed_ssa_stmt): Delete.
5691 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
5692 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
5693 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
5696 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
5698 * brig-builtins.def: Treat HSAIL barrier builtins as
5699 setjmp/longjump style functions.
5701 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
5703 * target.def (constant_alignment): New hook.
5704 * defaults.h (CONSTANT_ALIGNMENT): Delete.
5705 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
5706 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
5707 * doc/tm.texi: Regenerate.
5708 * targhooks.h (default_constant_alignment): Declare.
5709 (constant_alignment_word_strings): Likewise.
5710 * targhooks.c (default_constant_alignment): New function.
5711 (constant_alignment_word_strings): Likewise.
5712 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
5713 instead of CONSTANT_ALIGNMENT.
5714 * varasm.c (align_variable, get_variable_align, build_constant_desc)
5715 (force_const_mem): Likewise.
5716 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
5717 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
5718 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
5719 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5720 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
5722 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
5723 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5724 constant_alignment_word_strings.
5725 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
5726 (CONSTANT_ALIGNMENT): Likewise.
5727 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5728 (arm_constant_alignment): New function.
5729 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
5730 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5731 constant_alignment_word_strings.
5732 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
5733 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5734 constant_alignment_word_strings.
5735 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
5736 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5737 (cris_constant_alignment): New function.
5738 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
5739 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5740 (epiphany_constant_alignment): New function.
5741 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
5742 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5743 constant_alignment_word_strings.
5744 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
5745 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5746 constant_alignment_word_strings.
5747 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
5748 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5749 constant_alignment_word_strings.
5750 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
5751 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
5752 * config/i386/i386.c (ix86_constant_alignment): Make static.
5753 Use the same interface as the target hook.
5754 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5755 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
5756 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5757 constant_alignment_word_strings.
5758 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
5759 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
5760 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5761 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
5762 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5763 constant_alignment_word_strings.
5764 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
5765 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5766 constant_alignment_word_strings.
5767 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
5768 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5769 constant_alignment_word_strings.
5770 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
5771 * config/microblaze/microblaze.c (microblaze_constant_alignment):
5773 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5774 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
5775 * config/mips/mips.c (mips_constant_alignment): New function.
5776 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5777 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
5778 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
5779 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5780 (mmix_constant_alignment): Make static. Use the same interface
5782 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
5783 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5784 constant_alignment_word_strings.
5785 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
5786 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5787 constant_alignment_word_strings.
5788 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
5789 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5790 constant_alignment_word_strings.
5791 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
5792 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5793 (rs6000_constant_alignment): New function.
5794 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
5795 * config/riscv/riscv.c (riscv_constant_alignment): New function.
5796 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5797 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
5798 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5799 (rs6000_constant_alignment): New function.
5800 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
5801 * config/s390/s390.c (s390_constant_alignment): New function.
5802 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5803 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
5804 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5805 constant_alignment_word_strings.
5806 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
5807 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5808 (sparc_constant_alignment): New function.
5809 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
5810 * config/spu/spu.c (spu_constant_alignment): New function.
5811 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5812 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
5813 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5814 constant_alignment_word_strings.
5815 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
5816 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5817 constant_alignment_word_strings.
5818 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
5819 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5820 constant_alignment_word_strings.
5821 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
5822 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5823 (visium_constant_alignment): New function.
5824 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
5825 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5826 (xtensa_constant_alignment): New function.
5827 * system.h (CONSTANT_ALIGNMENT): Poison.
5829 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
5831 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
5832 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
5833 (rs6000_builtin_valid_without_lhs): New helper function.
5834 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5835 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
5837 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
5839 * target.h (vec_perm_indices): Use unsigned short rather than
5841 (auto_vec_perm_indices): Likewise.
5842 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
5843 Use unsigned int rather than unsigned char.
5844 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
5846 2017-09-25 Richard Biener <rguenther@suse.de>
5848 * cfgloop.h (sort_sibling_loops): Declare.
5849 * cfgloop.c (sort_sibling_loops_cmp): New helper.
5850 (sort_sibling_loops): New function sorting the sibling loop list
5852 * graphite.c (graphite_transform_loops): Sort sibling loops.
5854 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
5856 * target.def (vec_perm_const_ok): Change sel parameter to
5858 * optabs-query.c (can_vec_perm_p): Update accordingly.
5859 * doc/tm.texi: Regenerate.
5860 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
5861 auto_vec_perm_indices and remove separate nelt field.
5862 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
5863 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
5864 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
5865 (aarch64_expand_vec_perm_const): Update accordingly.
5866 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
5867 to vec_perm_indices.
5868 * config/arm/arm.c (expand_vec_perm_d): Change perm to
5869 auto_vec_perm_indices and remove separate nelt field.
5870 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
5871 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
5872 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
5874 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
5875 to vec_perm_indices.
5876 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
5877 sel to vec_perm_indices.
5878 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
5879 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
5880 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
5882 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
5885 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
5888 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
5889 on the FUNCTION_DECL function context if it has a DIE that is a
5892 2017-09-25 Richard Biener <rguenther@suse.de>
5894 PR tree-optimization/82285
5895 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
5898 2017-09-25 Tom de Vries <tom@codesourcery.com>
5902 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
5905 2017-09-25 Richard Biener <rguenther@suse.de>
5907 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
5908 ISL errors other than isl_error_quota happen. Dump if the
5909 schedule is the same.
5910 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
5911 errors instead of aborting inside ISL.
5913 2017-09-25 Iain Sandoe <iain@codesourcery.com>
5916 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
5917 of libgcc_eh for m64.
5918 * config/i386/darwin64.h: Likewise.
5920 2017-09-25 Richard Biener <rguenther@suse.de>
5923 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
5924 attribute for incomplete types nor twice for complete ones.
5926 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
5929 * config/i386/i386.c (ix86_print_operand_address_as): Encode
5930 %esp as %rsp to avoid 0x67 prefix if there is no index or base
5933 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
5936 * config/i386/i386.opt (mprefer-avx256): Use
5937 ix86_target_flags variable.
5938 * config/i386/i386.c (ix86_target_string): Move
5939 -mprefer-avx256 to flag2_opts.
5941 2017-09-22 Jakub Jelinek <jakub@redhat.com>
5944 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
5945 and x != -1 | y != -1 into (x & y) != -1.
5947 2017-09-22 Steve Ellcey <sellcey@cavium.com>
5949 * config.gcc: Add new case statement to set
5950 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
5951 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
5952 s390x-*-linux* case statements. Added aarch64 to the list of
5953 supported architectures.
5955 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
5957 PR tree-optimization/82289
5958 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
5959 STMT_VINFO_RELEVANT_P.
5961 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
5962 Alan Hayward <alan.hayward@arm.com>
5963 David Sherwood <david.sherwood@arm.com>
5965 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
5966 for VR_RANGE only; don't allow VR_ANTI_RANGE.
5967 (extract_range_from_binary_expr_1): Don't call
5968 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
5970 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
5971 Alan Hayward <alan.hayward@arm.com>
5972 David Sherwood <david.sherwood@arm.com>
5974 * target.def (preferred_vector_alignment): New hook.
5975 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
5977 * doc/tm.texi: Regenerate.
5978 * targhooks.h (default_preferred_vector_alignment): Declare.
5979 * targhooks.c (default_preferred_vector_alignment): New function.
5980 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
5982 (DR_TARGET_ALIGNMENT): New macro.
5983 (aligned_access_p): Update commentary.
5984 (vect_known_alignment_in_bytes): New function.
5985 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
5987 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
5988 Calculate the misalignment based on the target alignment rather than
5990 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
5991 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
5992 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
5993 the target alignment, rather than masking the element misalignment
5994 with the number of elements in a vector. Also use the target
5995 alignment when calculating the maximum number of peels.
5996 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
5997 instead of TYPE_ALIGN_UNIT.
5998 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
5999 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
6000 (vect_create_addr_base_for_vector_ref): Update call accordingly.
6001 (vect_create_data_ref_ptr): Likewise.
6002 (vect_setup_realignment): Realign by ANDing with
6003 -DR_TARGET_MISALIGNMENT.
6004 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
6005 the number of peels based on DR_TARGET_ALIGNMENT.
6006 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
6007 with the guaranteed alignment boundary when deciding whether
6009 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
6010 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
6011 (ensure_base_align): Remove stmt_info parameter. Get the
6012 target base alignment from DR_TARGET_ALIGNMENT.
6013 (vectorizable_store): Update call accordingly. Interpret
6014 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
6016 (vectorizable_load): Likewise.
6018 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6019 Alan Hayward <alan.hayward@arm.com>
6020 David Sherwood <david.sherwood@arm.com>
6022 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
6023 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
6024 (vect_enhance_data_refs_alignment): Likewise.
6026 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6028 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
6029 error. Only quit immediately if parsing is complete.
6030 (BEGIN): Initialize fatal_err and parse_done.
6031 (begin fpu, end fpu): Check number of arguments.
6032 (begin arch, end arch): Likewise.
6033 (begin cpu, end cpu): Likewise.
6034 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
6035 (optalias): Likewise.
6037 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6039 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
6040 * config/arm/arm-isa.h: Delete. Move definitions to ...
6041 * arm-cpus.in: ... here. Use new feature and fgroup values.
6042 * config/arm/arm.c (arm_option_override): Use lower case for feature
6044 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
6045 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
6046 * config/arm/parsecpu.awk (END): Add new command 'isa'.
6048 (print_isa_bits_for): New function.
6049 (gen_isa): New function.
6050 (gen_comm_data): Use print_isa_bits_for.
6051 (define feature): New keyword.
6052 (define fgroup): New keyword.
6053 * config/arm/t-arm (TM_H): Remove.
6054 (GTM_H): Add arm-isa.h.
6055 (arm-isa.h): Add rule to generate file.
6056 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
6057 case for feature bit names.
6059 2017-09-22 Richard Biener <rguenther@suse.de>
6061 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
6063 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
6064 print of no dependency loops ...
6065 * graphite.c (graphite_transform_loops): ... here.
6066 (canonicalize_loop_closed_ssa_form): Work from inner to outer
6068 (same_close_phi_node, remove_duplicate_close_phi,
6069 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
6070 (canonicalize_loop_closed_ssa): ... here and simplify.
6071 * graphite-optimize-isl.c: Include tree-vectorizer.h.
6072 (optimize_isl): Use dump_printf_loc to tell when we stopped
6073 optimizing because of an ISL timeout.
6075 2017-09-22 Richard Biener <rguenther@suse.de>
6077 PR tree-optimization/82291
6078 * tree-if-conv.c (predicate_mem_writes): Make sure to
6079 remove writes in blocks predicated with false.
6081 2017-09-22 Richard Biener <rguenther@suse.de>
6083 * sese.c: Include cfganal.h.
6084 (if_region_set_false_region): Remove.
6085 (create_if_region_on_edge): Likewise.
6086 (move_sese_in_condition): Re-implement without destroying
6089 2017-09-22 Richard Biener <rguenther@suse.de>
6091 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
6092 Verify both BBs contain loop PHI nodes before dispatching to
6094 (graphite_regenerate_ast_isl): Do not recompute dominators,
6095 do not verify three times. Restructure for clarity.
6096 * graphite-scop-detection.c (same_close_phi_node,
6097 remove_duplicate_close_phi, make_close_phi_nodes_unique,
6098 defined_in_loop_p, canonicalize_loop_closed_ssa,
6099 canonicalize_loop_closed_ssa_form): Simplify, remove excess
6100 checking and SSA rewrite, move to ...
6101 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
6102 (graphite_initialize): Do not pass in ctx, do not reset the
6103 SCEV cache, compute only dominators.
6104 (graphite_transform_loops): Allocate ISL ctx after
6105 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
6106 Maintain post-dominators only around build_scops.
6107 * sese.c (if_region_set_false_region): Make static. Free
6108 and recompute dominators.
6109 (move_sese_in_condition): Assert we don't get called with
6110 post-dominators computed.
6111 * sese.h (if_region_set_false_region): Remove.
6113 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
6115 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
6116 mode attribute for TARGET_AVX512VL.
6118 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
6120 * config/i386/i386.opt (mprefer-avx256): New option.
6121 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
6123 (ix86_preferred_simd_mode): Return 256-bit AVX modes
6124 for TARGET_PREFER_AVX256.
6125 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
6127 2017-09-21 Jeff Law <law@redhat.com>
6129 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
6130 Fix dump output if the only stack space is for pushed registers.
6132 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6134 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
6137 2017-09-21 Martin Sebor <msebor@redhat.com>
6140 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
6141 code (in C++) or code that triggers warnings.
6143 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
6145 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
6147 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
6149 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
6150 * sched-rgn.c: Ditto.
6151 * sel-sched-ir.c: Ditto.
6153 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
6155 * toplev.h (set_random_seed): Adjust return type.
6156 * toplev.c (init_local_tick): Move eager initialization of random_seed
6157 to get_random_seed. Adjust comment.
6158 (init_random_seed): Inline to get_random_seed, delete.
6159 (get_random_seed): Initialize random_seed lazily.
6160 (set_random_seed): Do not return previous value.
6161 (print_switch_value): Do not call get_random_seed.
6163 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
6165 * cgraph.c (delete_function_version): New, broken out from...
6166 (cgraph_node::delete_function_version): ...here. Rename to
6167 cgraph_node::delete_function_version_by_decl. Update all uses.
6168 (cgraph_node::remove): Call delete_function_version.
6170 2017-09-21 Jakub Jelinek <jakub@redhat.com>
6173 * tree-inline.c (expand_call_inline): Emit clobber stmts for
6174 VAR_DECLs to which addressable non-volatile parameters are mapped
6175 and for id->retvar after the return value assignment. Clear
6176 id->retval and id->retbnd after inlining.
6178 2017-09-21 Richard Biener <rguenther@suse.de>
6180 PR tree-optimization/82276
6181 PR tree-optimization/82244
6182 * tree-vrp.c (build_assert_expr_for): Set
6183 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
6185 (remove_range_assertions): Revert earlier change.
6187 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
6190 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
6192 2017-09-21 Richard Biener <rguenther@suse.de>
6194 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
6195 Restore valid IL after code generation errors.
6196 * graphite.c (graphite_transform_loops): Diagnose code
6197 generation issues as MSG_MISSED_OPTIMIZATION and continue
6198 with processing SCOPs.
6200 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6201 Alan Hayward <alan.hayward@arm.com>
6202 David Sherwood <david.sherwood@arm.com>
6204 * calls.c (compute_argument_addresses): Use simplify_gen_binary
6205 rather than choosing between plus_constant and gen_rtx_<CODE>.
6206 * expr.c (emit_push_insn): Likewise.
6207 (expand_expr_real_2): Likewise.
6209 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6210 Alan Hayward <alan.hayward@arm.com>
6211 David Sherwood <david.sherwood@arm.com>
6213 * loop-unroll.c (split_iv): Call copy_rtx on the step.
6215 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6216 Alan Hayward <alan.hayward@arm.com>
6217 David Sherwood <david.sherwood@arm.com>
6219 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
6220 calling tree_to_uhwi.
6222 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6223 Alan Hayward <alan.hayward@arm.com>
6224 David Sherwood <david.sherwood@arm.com>
6226 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
6227 INTEGER_CST rather than a negative test for ADDR_EXPR.
6229 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6230 Alan Hayward <alan.hayward@arm.com>
6231 David Sherwood <david.sherwood@arm.com>
6233 * tree-vrp.c (extract_range_from_binary_expr_1): Check
6234 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
6236 2017-09-21 Richard Biener <rguenther@suse.de>
6238 PR tree-optimization/71351
6239 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
6240 graphite_create_new_loop_guard): Remove, fold remaining parts
6242 (translate_isl_ast_node_for): ... here and simplify.
6244 2017-09-21 Jakub Jelinek <jakub@redhat.com>
6247 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
6248 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
6249 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
6250 alternative always use QI mode, for -Os imov (=R,R) alternative
6251 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
6254 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6255 Jeff Law <law@redhat.com>
6257 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
6258 (allocate_stack_space): New function, partially extracted from
6260 (s390_emit_prologue): Track offset to most recent stack probe.
6261 Code to allocate space moved into allocate_stack_space.
6262 Dump actions when no stack is allocated.
6263 (s390_prologue_plus_offset): New function.
6264 (s390_emit_stack_probe): Likewise.
6266 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
6268 * common.opt (Wa, Wl, Wp, g, gz=): Add
6270 (gno-column-info): Remove.
6271 (gcolumn-info): Drop RejectNegative.
6273 (gno-record-gcc-switches): Remove.
6274 (grecord-gcc-switches): Drop RejectNegative.
6275 (gno-split-dwarf): Remove.
6276 (gsplit-dwarf): Drop RejectNegative.
6277 (gno-strict-dwarf): Remove.
6278 (gstrict-dwarf): Drop RejectNegative.
6279 * config/darwin.opt (gfull, gused): Add RejectNegative.
6280 * dwarf2out.c (gen_producer_string): Drop
6281 gno-record-gcc-switches handler.
6282 * optc-gen.awk: Add g to prefixes with negative forms.
6283 * opts-common.c (remapping_prefix_p): New.
6284 (find_opt): Check it.
6285 (generate_canonical_option): Test g prefix.
6286 (option_map): Add -gno- mapping.
6287 (add_misspelling_candidates): Check remapping_prefix_p.
6289 2017-09-20 Jeff Law <law@redhat.com>
6291 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
6292 thinko in stack clash protection support.
6294 * explow.c (compute_stack_clash_protection_loop_data): Use
6295 CONST_INT_P instead of explicit test. Verify object is a
6296 CONST_INT_P before looking at INTVAL.
6297 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
6298 instead of explicit test.
6300 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
6303 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
6304 address instead of to r1 and r11.
6306 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
6308 * config.gcc: Support "knm".
6309 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
6310 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
6312 * config/i386/i386.c (m_KNM): Define.
6313 (processor_target_table): Add "knm".
6315 (ix86_option_override_internal): Add "knm".
6316 (ix86_issue_rate): Add PROCESSOR_KNM.
6317 (ix86_adjust_cost): Ditto.
6318 (ia32_multipass_dfa_lookahead): Ditto.
6319 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
6320 (fold_builtin_cpu): Add M_INTEL_KNM.
6321 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
6322 (processor_type): Add PROCESSOR_KNM.
6323 * config/i386/x86-tune.def: Add m_KNM.
6324 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
6326 2017-09-20 Richard Biener <rguenther@suse.de>
6328 PR tree-optimization/80213
6329 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
6330 are allowed in empty BBs as well.
6331 (canonicalize_loop_closed_ssa): Also look for other complex
6333 (scop_detection::get_sese): Include the loop-closed PHI block
6335 (scop_detection::merge_sese): Remove code adding extra blocks.
6336 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
6337 (build_scops): Assert the final returned scop is invalid.
6339 2017-09-20 Richard Biener <rguenther@suse.de>
6341 PR tree-optimization/82264
6342 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
6344 (vn_phi_lookup): Likewise.
6345 (vn_phi_insert): Likewise.
6347 2017-09-20 Jakub Jelinek <jakub@redhat.com>
6349 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
6350 that fits into uhwi or shwi, add DW_AT_const_value regardless
6351 of early_dwarf without going through RTL, using add_AT_unsigned
6354 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
6355 (DEBUG_LTO_ABBREV_SECTION): Likewise.
6356 (DEBUG_LTO_MACINFO_SECTION): Likewise.
6357 (DEBUG_MACRO_SECTION): Likewise.
6358 (DEBUG_LTO_MACRO_SECTION): Likewise.
6359 (DEBUG_STR_DWO_SECTION): Likewise.
6360 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
6361 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
6362 (DEBUG_LTO_DWO_LINE_SECTION): Define.
6363 (DEBUG_LTO_LINE_STR_SECTION): Define.
6364 (init_sections_and_labels): Initialize debug_line_str_section
6365 variable. Initialize debug_loc_section for -gdwarf-5 to
6366 DEBUG_LOCLISTS_SECTION. Formatting fixes.
6368 2017-09-20 Richard Biener <rguenther@suse.de>
6370 * graphite-sese-to-poly.c (extract_affine): Properly handle
6371 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
6373 2017-09-20 Richard Biener <rguenther@suse.de>
6375 PR tree-optimization/81373
6376 * graphite-scop-detection.c (build_cross_bb_scalars_def):
6377 Force SESE live-out defs to be handled even if they are
6380 2017-09-19 Jeff Law <law@redhat.com>
6382 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
6383 nothing for stack adjustments with REG_STACK_CHECK.
6384 * sched-deps.c (parse_add_or_inc): Reject insns with
6385 REG_STACK_CHECK from dependency breaking.
6386 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
6387 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
6388 * reg-notes.def (STACK_CHECK): New note.
6390 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
6391 (ix86_expand_prologue): Dump stack clash info as needed.
6392 Call ix86_adjust_stack_and_probe_stack_clash as needed.
6394 * function.c (dump_stack_clash_frame_info): New function.
6395 * function.h (dump_stack_clash_frame_info): Prototype.
6396 (enum stack_clash_probes): New enum.
6398 * config/alpha/alpha.c (alpha_expand_prologue): Also check
6399 flag_stack_clash_protection.
6400 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
6401 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
6402 (arm_frame_pointer_required): Likewise.
6403 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
6404 (ia64_expand_prologue): Likewise.
6405 * config/mips/mips.c (mips_expand_prologue): Likewise.
6406 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
6407 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6408 (sparc_flat_expand_prologue): Likewise.
6409 * config/spu/spu.c (spu_expand_prologue): Likewise.
6411 * explow.c: Include "params.h".
6412 (anti_adjust_stack_and_probe_stack_clash): New function.
6413 (get_stack_check_protect): Likewise.
6414 (compute_stack_clash_protection_loop_data): Likewise.
6415 (emit_stack_clash_protection_loop_start): Likewise.
6416 (emit_stack_clash_protection_loop_end): Likewise.
6417 (allocate_dynamic_stack_space): Use get_stack_check_protect.
6418 Use anti_adjust_stack_and_probe_stack_clash.
6419 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
6420 (emit_stack_clash_protection_loop_start): Likewise.
6421 (emit_stack_clash_protection_loop_end): Likewise.
6422 * rtl.h (get_stack_check_protect): Prototype.
6423 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
6424 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
6425 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
6427 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
6429 * doc/tm.texi: Rebuilt.
6430 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
6431 get_stack_check_protect.
6432 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
6433 * config/arm/arm.c (arm_expand_prologue): Likewise.
6434 (arm_frame_pointer_required): Likewise.
6435 * config/i386/i386.c (ix86_expand_prologue): Likewise.
6436 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
6437 * config/mips/mips.c (mips_expand_prologue): Likewise.
6438 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
6439 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
6440 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6441 (sparc_flat_expand_prologue): Likewise.
6443 * common.opt (-fstack-clash-protection): New option.
6444 * flag-types.h (enum stack_check_type): Note difference between
6445 -fstack-check= and -fstack-clash-protection.
6446 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
6447 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
6448 * toplev.c (process_options): Issue warnings/errors for cases
6449 not handled with -fstack-clash-protection.
6450 * doc/invoke.texi (-fstack-clash-protection): Document new option.
6451 (-fstack-check): Note additional problem with -fstack-check=generic.
6452 Note that -fstack-check is primarily for Ada and refer users
6453 to -fstack-clash-protection for stack-clash-protection.
6454 Document new params for stack clash protection.
6456 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
6458 * config/i386/i386.c (ix86_split_long_move): Do not handle
6459 address used for LEA in a special way.
6461 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
6463 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
6465 2017-09-19 Martin Sebor <msebor@redhat.com>
6468 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
6469 of incompatible types.
6471 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
6473 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6474 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
6475 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6476 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
6478 2017-09-19 Richard Biener <rguenther@suse.de>
6480 PR tree-optimization/82244
6481 * tree-vrp.c (remove_range_assertions): Do not propagate
6482 a constant to abnormals but replace the assert with a copy.
6484 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
6486 PR rtl-optimization/57878
6487 PR rtl-optimization/68988
6488 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
6489 avoidance test involving non_reload_pseudos. Move frequency test
6490 below the general fragmentation avoidance test.
6492 2017-09-19 Richard Biener <rguenther@suse.de>
6494 PR tree-optimization/69728
6495 * graphite-sese-to-poly.c (schedule_error): New global.
6496 (add_loop_schedule): Handle empty domain by failing the
6498 (build_original_schedule): Handle schedule_error.
6500 2017-09-19 Richard Biener <rguenther@suse.de>
6502 * graphite-scop-detection.c (scop_detection::can_represent_loop):
6503 Do not iterate to sibling loops but only to siblings of inner
6506 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
6509 * config/m68k/m68k.md (moveq feeding equality comparison): Check
6510 that the registers are different.
6512 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
6514 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
6515 to processor_model and "amdfam17h" to arch_names_table.
6516 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
6518 2017-09-18 Jakub Jelinek <jakub@redhat.com>
6521 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
6523 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6524 Alan Hayward <alan.hayward@arm.com>
6525 David Sherwood <david.sherwood@arm.com>
6527 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
6529 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
6531 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
6532 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
6533 vect_schedule_slp_instance.
6534 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
6535 Update call to vect_slp_analyze_node_operations. Simplify return
6537 (vect_slp_analyze_bb_1): Update call accordingly.
6538 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
6539 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
6540 (vect_schedule_slp): Update call accordingly.
6542 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6543 Alan Hayward <alan.hayward@arm.com>
6544 David Sherwood <david.sherwood@arm.com>
6546 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
6547 with types that aren't in fact scalar.
6549 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6551 * tree-vect-slp.c (vect_record_max_nunits): New function,
6553 (vect_build_slp_tree_1): ...here.
6554 (vect_build_slp_tree_2): Call it for phis too.
6556 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6558 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
6559 to vect_get_vec_def_for_operand when getting the mask operand.
6561 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6562 Alan Hayward <alan.hayward@arm.com>
6563 David Sherwood <david.sherwood@arm.com>
6565 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
6568 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6569 Alan Hayward <alan.hayward@arm.com>
6570 David Sherwood <david.sherwood@arm.com>
6572 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
6573 calculation for vector booleans.
6575 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6576 Alan Hayward <alan.hayward@arm.com>
6577 David Sherwood <david.sherwood@arm.com>
6579 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
6581 (vect_transform_stmt): ...here.
6582 (vect_analyze_stmt): Use it instead of calling
6583 vectorizable_live_operation directly.
6585 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
6587 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
6588 non-SIMT targets in acc vector loops.
6590 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
6592 * configure.ac: Add arc and check if assembler supports gdwarf2.
6593 * configure: Regenerate.
6595 2017-09-18 Richard Biener <rguenther@suse.de>
6597 PR tree-optimization/82220
6598 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
6599 epilogue niters from the min_profitable_iters compute.
6601 2017-09-18 Jakub Jelinek <jakub@redhat.com>
6604 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
6605 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
6606 (ix86_init_pic_reg): Revert 2017-09-01 changes.
6608 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
6611 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
6612 switching to a new text section.
6614 2017-09-18 Richard Biener <rguenther@suse.de>
6616 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
6618 (build_alias_set): Reject aliases with no access function.
6620 2017-09-18 Richard Biener <rguenther@suse.de>
6622 PR tree-optimization/79622
6623 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
6625 (build_cross_bb_scalars_use): Likewise.
6627 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
6629 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
6631 2017-09-18 Alan Modra <amodra@gmail.com>
6634 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
6635 stack_pointer_rtx for count 0. Update comments. Break up
6636 large rtl expression.
6638 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
6640 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
6641 Increase to 20 bytes.
6642 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
6643 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
6644 or avx version of the stub.
6646 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
6649 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
6650 compute the minimum stack alignment. Also update preferred stack
6651 boundary for leaf functions.
6653 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
6655 PR tree-optimization/82228
6656 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
6659 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
6661 * common/config/nds32/nds32-common.c
6662 (nds32_option_optimization_table): Refine formatting.
6663 (nds32_option_optimization_table): Use -fsched-pressure and
6664 -fomit-frame-pointer for specific optimization level.
6666 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
6668 * config/nds32/nds32.c: Refine formatting and comments.
6669 * config/nds32/nds32.h: Likewise.
6670 * config/nds32/nds32.md: Likewise.
6671 * config/nds32/nds32-cost.c: Likewise.
6672 * config/nds32/nds32-isr.c: Likewise.
6673 * config/nds32/nds32-md-auxiliary.c: Likewise.
6674 * config/nds32/nds32-multiple.md: Likewise.
6675 * config/nds32/nds32-predicates.c: Likewise.
6677 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
6678 Jakub Jelinek <jakub@redhat.com>
6680 Add support for -std=c++2a.
6681 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
6683 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
6685 2017-09-15 Steve Ellcey <sellcey@cavium.com>
6688 * doc/extend.texi (Common Function Attributes): Add
6689 references to ARM, AArch64, and S/390 specific attributes.
6690 (Function Specific Option Pragmas): Add AArch64 and S/390
6691 to list of back ends that support the target pragma.
6693 2017-09-15 Nathan Sidwell <nathan@acm.org>
6695 * doc/standards.texi: Fix C++17 description. Update URLs for
6698 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
6700 * common.opt (Wcast-align=strict): New warning option.
6701 * doc/invoke.texi: Document -Wcast-align=strict.
6703 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
6705 * cgraph.h (cgraph_thunk_info): Add comments.
6706 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
6707 assert for VIRTUAL_* arguments stricter.
6709 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
6711 PR tree-optimization/71026
6712 * match.pd: Move RDIV patterns from fold-const.c
6713 * fold-const.c (distribute_real_division): Removed.
6714 (fold_binary_loc): Remove calls to distribute_real_divison.
6716 2017-09-15 Jakub Jelinek <jakub@redhat.com>
6718 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
6719 c++1z and gnu++1z as deprecated. Change other references to
6720 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
6721 Change -Wc++1z-compat to -Wc++17-compat.
6722 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
6723 * dwarf2out.c (highest_c_language): Handle C++17.
6724 (gen_compile_unit_die): Likewise.
6726 2017-09-15 Jakub Jelinek <jakub@redhat.com>
6728 PR rtl-optimization/82192
6729 * combine.c (make_extraction): Don't look through non-paradoxical
6730 SUBREGs or TRUNCATE if pos + len is or might be bigger than
6733 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
6734 Alan Hayward <alan.hayward@arm.com>
6735 David Sherwood <david.sherwood@arm.com>
6737 * target.def (function_arg_offset): New hook.
6738 * targhooks.h (default_function_arg_offset): Declare.
6739 * targhooks.c (default_function_arg_offset): New function.
6740 * function.c (locate_and_pad_parm): Use
6741 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
6742 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
6743 (TARGET_FUNCTION_ARG_OFFSET): ...this.
6744 * doc/tm.texi: Regenerate.
6745 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
6746 * config/spu/spu.c (spu_function_arg_offset): New function.
6747 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
6748 * system.h (FUNCTION_ARG_OFFSET): Poison.
6750 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
6751 Alan Hayard <alan.hayward@arm.com>
6752 David Sherwood <david.sherwood@arm.com>
6754 * target.def (truly_noop_truncation): New hook.
6755 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
6756 than TRULY_NOOP_TRUNCATION.
6757 * hooks.h (hook_bool_uint_uint_true): Declare.
6758 * hooks.c (hook_bool_uint_uint_true): New function.
6759 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
6760 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
6761 * doc/tm.texi: Regenerate.
6762 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
6763 rather than TRULY_NOOP_TRUNCATION in comments.
6764 (simplify_comparison): Likewise.
6765 (record_truncated_value): Likewise.
6766 * expmed.c (extract_bit_field_1): Likewise.
6767 (extract_split_bit_field): Likewise.
6768 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
6769 instead of TRULY_NOOP_TRUNCATION.
6770 * function.c (assign_parm_setup_block): Likewise.
6771 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
6772 * rtlhooks.c: Include target.h.
6773 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
6774 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
6775 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
6776 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
6777 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
6778 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
6779 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
6780 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
6781 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
6782 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
6783 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
6784 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
6785 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
6786 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
6787 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
6788 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
6789 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
6790 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
6791 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
6792 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
6793 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
6794 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
6795 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
6796 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
6797 * config/mips/mips.c (mips_truly_noop_truncation): New function.
6798 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
6799 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
6800 rather than TRULY_NOOP_TRUNCATION in comments.
6801 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
6802 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
6803 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
6804 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
6805 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
6806 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
6807 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
6808 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
6809 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
6810 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
6811 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
6812 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
6813 rather than TRULY_NOOP_TRUNCATION in comments.
6814 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
6815 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
6816 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
6817 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
6818 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
6819 TRULY_NOOP_TRUNCATION condition.
6820 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
6821 (TRULY_NOOP_TRUNCATION): Delete.
6822 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
6823 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
6824 * config/spu/spu.c (spu_truly_noop_truncation): New function.
6825 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
6826 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
6827 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
6828 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
6829 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
6830 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
6831 rather than TRULY_NOOP_TRUNCATION in comments.
6832 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
6833 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
6834 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
6835 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
6836 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
6837 * system.h (TRULY_NOOP_TRUNCATION): Poison.
6839 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
6842 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
6843 (*cmp_ior): Likewise.
6844 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
6845 (*ior_scc_scc_cmp): Likewise.
6846 (*and_scc_scc): Likewise.
6847 (*and_scc_scc_cmp): Likewise.
6849 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
6850 Alan Hayard <alan.hayward@arm.com>
6851 David Sherwood <david.sherwood@arm.com>
6853 * target.def (can_change_mode_class): New hook.
6854 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
6855 (hard_regno_nregs): Likewise.
6856 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
6857 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
6858 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
6859 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
6860 (LOAD_EXTEND_OP): Update accordingly.
6861 * doc/tm.texi: Regenerate.
6862 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
6863 CANNOT_CHANGE_MODE_CLASS.
6864 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
6865 (REG_CAN_CHANGE_MODE_P): ...this new macro.
6866 * combine.c (simplify_set): Update accordingly.
6867 * emit-rtl.c (validate_subreg): Likewise.
6868 * recog.c (general_operand): Likewise.
6869 * regcprop.c (mode_change_ok): Likewise.
6870 * reload1.c (choose_reload_regs): Likewise.
6871 (inherit_piecemeal_p): Likewise.
6872 * rtlanal.c (simplify_subreg_regno): Likewise.
6873 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
6874 instead of CANNOT_CHANGE_MODE_CLASS.
6875 (reload_cse_simplify_operands): Likewise.
6876 * reload.c (push_reload): Use targetm.can_change_mode_class
6877 instead of CANNOT_CHANGE_MODE_CLASS.
6878 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
6879 REG_CANNOT_CHANGE_MODE_P.
6880 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6881 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
6882 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6883 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6884 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6885 (arm_can_change_mode_class): New function.
6886 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
6887 than CANNOT_CHANGE_MODE_CLASS in comments.
6888 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6889 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
6890 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
6891 (ix86_can_change_mode_class): ...this new function, inverting the
6892 sense of the return value.
6893 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6894 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6895 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6896 (ia64_can_change_mode_class): New function.
6897 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6898 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
6899 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
6900 (m32c_can_change_mode_class): ...this new function, inverting the
6901 sense of the return value.
6902 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6903 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6904 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
6905 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
6906 (mips_can_change_mode_class): ...this new function, inverting the
6907 sense of the return value.
6908 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6909 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6910 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6911 (msp430_can_change_mode_class): New function.
6912 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6913 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
6914 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6915 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6916 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6917 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
6918 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6919 (pa_cannot_change_mode_class): Replace with...
6920 (pa_can_change_mode_class): ...this new function, inverting the
6921 sense of the return value.
6922 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
6923 than CANNOT_CHANGE_MODE_CLASS in comments.
6924 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6925 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
6926 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6927 (pdp11_cannot_change_mode_class): Replace with...
6928 (pdp11_can_change_mode_class): ...this new function, inverting the
6929 sense of the return value.
6930 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6931 * config/powerpcspe/powerpcspe-protos.h
6932 (rs6000_cannot_change_mode_class_ptr): Delete.
6933 * config/powerpcspe/powerpcspe.c
6934 (rs6000_cannot_change_mode_class_ptr): Delete.
6935 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6936 (rs6000_option_override_internal): Assign to
6937 targetm.can_change_mode_class instead of
6938 rs6000_cannot_change_mode_class_ptr.
6939 (rs6000_cannot_change_mode_class): Replace with...
6940 (rs6000_can_change_mode_class): ...this new function, inverting the
6941 sense of the return value.
6942 (rs6000_debug_cannot_change_mode_class): Replace with...
6943 (rs6000_debug_can_change_mode_class): ...this new function.
6944 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6945 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
6946 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6947 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6948 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
6950 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
6951 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6952 (rs6000_option_override_internal): Assign to
6953 targetm.can_change_mode_class instead of
6954 rs6000_cannot_change_mode_class_ptr.
6955 (rs6000_cannot_change_mode_class): Replace with...
6956 (rs6000_can_change_mode_class): ...this new function, inverting the
6957 sense of the return value.
6958 (rs6000_debug_cannot_change_mode_class): Replace with...
6959 (rs6000_debug_can_change_mode_class): ...this new function.
6960 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6961 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
6962 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
6963 (s390_can_change_mode_class): ...this new function, inverting the
6964 sense of the return value.
6965 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6966 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6967 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
6968 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6969 (sh_cannot_change_mode_class): Replace with...
6970 (sh_can_change_mode_class): ...this new function, inverting the
6971 sense of the return value.
6972 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6973 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6974 (sparc_can_change_mode_class): New function.
6975 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6976 * config/spu/spu.c (spu_can_change_mode_class): New function.
6977 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6978 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6979 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6980 (visium_can_change_mode_class): New function.
6981 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
6983 2017-09-15 Richard Biener <rguenther@suse.de>
6985 PR tree-optimization/82217
6986 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
6987 but not undefined case.
6989 2017-09-15 Jakub Jelinek <jakub@redhat.com>
6992 * postreload.c (reload_cse_simplify_operands): Skip
6993 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
6995 2017-09-15 Richard Biener <rguenther@suse.de>
6997 PR tree-optimization/68823
6998 * graphite-scop-detection.c (build_alias_set): If we have a
6999 possible dependence check whether we can handle them by just
7000 looking at the DRs DR_ACCESS_FNs.
7001 (build_scops): If build_alias_set fails, fail the SCOP.
7003 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
7005 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
7006 to support float128 built-in functions that require the ISA 3.0
7008 (BU_FLOAT128_3_HW): Likewise.
7009 (SQRTF128): Add support for the IEEE 128-bit square root and fma
7011 (FMAF128): Likewise.
7013 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
7014 support for built-in functions that need the ISA 3.0 IEEE 128-bit
7015 floating point instructions.
7016 (rs6000_invalid_builtin): Likewise.
7017 (rs6000_builtin_mask_names): Likewise.
7018 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
7019 (RS6000_BTM_FLOAT128_HW): Likewise.
7020 (RS6000_BTM_COMMON): Likewise.
7021 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
7023 * doc/extend.texi (RS/6000 built-in functions): Document the
7024 IEEE 128-bit floating point square root and fused multiply-add
7027 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
7029 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
7030 reg (r2) isn't in the set of registers defined in the prologue.
7032 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7033 Alan Hayward <alan.hayward@arm.com>
7034 David Sherwood <david.sherwood@arm.com>
7036 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
7037 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
7038 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
7039 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
7040 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
7042 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
7043 max_vectorization_factor.
7044 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
7046 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7047 Alan Hayward <alan.hayward@arm.com>
7048 David Sherwood <david.sherwood@arm.com>
7050 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
7051 (vect_worthwhile_without_simd_p): Declare.
7052 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
7053 (vectorizable_reduction): Use it.
7054 * tree-vect-stmts.c (vectorizable_shift): Likewise.
7055 (vectorizable_operation): Likewise.
7057 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7058 Alan Hayward <alan.hayward@arm.com>
7059 David Sherwood <david.sherwood@arm.com>
7061 * tree-vectorizer.h (vect_get_num_copies): New function.
7062 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
7063 * tree-vect-loop.c (vectorizable_reduction): Likewise.
7064 (vectorizable_induction): Likewise.
7065 (vectorizable_live_operation): Likewise.
7066 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7067 (vectorizable_bswap): Likewise.
7068 (vectorizable_call): Likewise.
7069 (vectorizable_conversion): Likewise.
7070 (vectorizable_assignment): Likewise.
7071 (vectorizable_shift): Likewise.
7072 (vectorizable_operation): Likewise.
7073 (vectorizable_store): Likewise.
7074 (vectorizable_load): Likewise.
7075 (vectorizable_condition): Likewise.
7076 (vectorizable_comparison): Likewise.
7077 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
7079 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7080 Alan Hayward <alan.hayward@arm.com>
7081 David Sherwood <david.sherwood@arm.com>
7083 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
7084 of vect_init_vector.
7086 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7087 Alan Hayward <alan.hayward@arm.com>
7088 David Sherwood <david.sherwood@arm.com>
7090 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
7091 an inline wrapper that provides a location.
7092 (gimple_build_vector): Likewise.
7093 * gimple-fold.c (gimple_build_vector_from_val): New function.
7094 (gimple_build_vector): Likewise.
7095 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
7096 functions to build the initial value. Always return a gimple value.
7097 (get_initial_defs_for_reduction): Likewise. Only compute
7099 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
7100 vect_init_vector on the results from get_initial_def(s)_for_reduction.
7101 (vectorizable_induction): Use gimple_build_vector rather than
7104 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7105 Alan Hayward <alan.hayward@arm.com>
7106 David Sherwood <david.sherwood@arm.com>
7108 * target.h (vec_perm_indices): New typedef.
7109 (auto_vec_perm_indices): Likewise.
7110 * optabs-query.h: Include target.h
7111 (can_vec_perm_p): Take a vec_perm_indices *.
7112 * optabs-query.c (can_vec_perm_p): Likewise.
7113 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
7114 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7115 * tree-vect-generic.c (lower_vec_perm): Likewise.
7116 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
7117 (vect_grouped_load_supported): Likewise.
7118 (vect_shift_permute_load_chain): Likewise.
7119 (vect_permute_store_chain): Use auto_vec_perm_indices.
7120 (vect_permute_load_chain): Likewise.
7121 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
7122 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
7123 Update uses of can_vec_perm_p.
7124 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
7125 mode with a number of elements. Take a vec_perm_indices *.
7126 (vect_create_epilog_for_reduction): Update accordingly.
7127 Use auto_vec_perm_indices.
7128 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
7129 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
7130 (vect_transform_slp_perm_load): Likewise.
7131 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
7132 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
7133 (vect_gen_perm_mask_checked): Likewise.
7134 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
7135 (vect_gen_perm_mask_checked): Likewise.
7136 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
7137 (vectorizable_store): Likewise.
7138 (vectorizable_load): Likewise.
7139 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
7140 (vectorizable_bswap): Likewise.
7142 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7143 Alan Hayward <alan.hayward@arm.com>
7144 David Sherwood <david.sherwood@arm.com>
7146 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
7147 * tree.c (build_vector): Likewise.
7148 (build_vector_from_ctor): Update accordingly.
7149 (build_vector_from_val): Likewise.
7150 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
7151 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7152 * tree-vect-generic.c (add_rshift): Likewise.
7153 (expand_vector_divmod): Likewise.
7154 (optimize_vector_constructor): Likewise.
7155 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
7156 (vect_transform_slp_perm_load): Likewise.
7157 (vect_schedule_slp_instance): Likewise.
7158 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
7159 (vectorizable_call): Likewise.
7160 (vect_gen_perm_mask_any): Likewise. Add elements in order.
7161 * expmed.c (make_tree): Likewise.
7162 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
7163 a vector passed to build_vector.
7164 (fold_convert_const): Likewise.
7165 (exact_inverse): Likewise.
7166 (fold_ternary_loc): Likewise.
7167 (fold_relational_const): Likewise.
7168 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
7169 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
7170 (const_unop): Likewise. Store the reduction accumulator in a
7171 variable rather than an array.
7172 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
7173 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
7174 the new vector, rather than constructing it after the input arrays.
7175 (native_interpret_vector): Use auto_vec<tree> when building
7176 a vector passed to build_vector. Add elements in order.
7177 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
7178 auto_vec<tree> when building a vector passed to build_vector.
7179 (vect_create_epilog_for_reduction): Likewise.
7180 (vectorizable_induction): Likewise.
7181 (get_initial_def_for_reduction): Likewise. Fix indentation of
7183 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
7185 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
7186 passed to build_vector.
7188 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7189 Alan Hayward <alan.hayward@arm.com>
7190 David Sherwood <david.sherwood@arm.com>
7192 * tree-core.h (tree_base::u): Add an "nelts" field.
7193 (tree_vector): Use VECTOR_CST_NELTS as the length.
7194 * tree.c (tree_size): Likewise.
7195 (make_vector): Initialize VECTOR_CST_NELTS.
7196 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
7197 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
7198 TYPE_VECTOR_SUBPARTS.
7199 * expr.c (const_vector_mask_from_tree): Consistently use "units"
7200 as the number of units, setting it from VECTOR_CST_NELTS.
7201 (const_vector_from_tree): Likewise.
7202 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
7203 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
7204 (fold_negate_expr_1): Likewise.
7205 (fold_convert_const): Likewise.
7206 (const_binop): Likewise. Differentiate the number of output and
7208 (const_unop): Likewise.
7209 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
7210 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
7211 in cases that did the opposite.
7213 2017-09-14 Richard Biener <rguenther@suse.de>
7215 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
7218 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
7220 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
7222 2017-09-14 Jakub Jelinek <jakub@redhat.com>
7225 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
7226 if and where to split a bb, except for splitting before debug insn
7227 sequences followed by non-label real insn. Delete debug insns
7228 in between basic blocks.
7230 * combine.c (make_compound_operation_int): Formatting fixes.
7232 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
7233 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7234 * config/netbsd.h (LINK_EH_SPEC): Likewise.
7235 * config/sol2.h (LINK_EH_SPEC): Likewise.
7236 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7237 * config/s390/linux.h (LINK_SPEC): Likewise.
7238 * config/freebsd.h (LINK_EH_SPEC): Likewise.
7239 * config/openbsd.h (LINK_EH_SPEC): Likewise.
7240 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7241 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
7242 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
7243 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7244 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
7245 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
7247 2017-09-13 Jakub Jelinek <jakub@redhat.com>
7249 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
7251 (ENDFILE_LINUX_SPEC): Likewise.
7252 (LINK_EH_SPEC): Likewise.
7253 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
7254 (LINK_OS_LINUX_SPEC32): Likewise.
7255 (LINK_OS_LINUX_SPEC64): Likewise.
7256 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
7257 (LINK_OS_LINUX_SPEC): Likewise.
7259 2017-09-13 Martin Liska <mliska@suse.cz>
7262 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
7263 CASE_HIGH is NULL_TREE.
7265 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
7266 Alan Hayward <alan.hayward@arm.com>
7267 David Sherwood <david.sherwood@arm.com>
7269 * target.def (secondary_memory_needed): New hook.
7270 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
7271 instead of SECONDARY_MEMORY_NEEDED.
7272 (secondary_memory_needed_mode): Likewise.
7273 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
7274 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
7275 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
7276 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
7277 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
7278 * doc/tm.texi: Regenerate.
7279 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
7280 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
7281 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7282 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
7283 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
7284 * config/i386/i386.c (inline_secondary_memory_needed): Put the
7285 mode argument first and change the reg_class arguments to reg_class_t.
7286 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
7287 Make static. Update the call to inline_secondary_memory_needed.
7288 (ix86_register_move_cost): Update the call to
7289 inline_secondary_memory_needed.
7290 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7291 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
7293 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
7294 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7296 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
7297 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
7298 * config/mips/mips.c (mips_secondary_memory_needed): Make static
7299 and match hook interface. Add comment from mips.h.
7300 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7301 * config/mmix/mmix.md (truncdfsf2): Refer to
7302 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7304 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
7305 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
7306 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7307 (pa_secondary_memory_needed): New function.
7308 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
7309 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
7310 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7311 (pdp11_secondary_memory_needed): Make static and match hook interface.
7312 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
7313 * config/powerpcspe/powerpcspe-protos.h
7314 (rs6000_secondary_memory_needed_ptr): Delete.
7315 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
7317 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7318 (rs6000_option_override_internal): Assign to
7319 targetm.secondary_memory_needed rather than
7320 rs6000_secondary_memory_needed_ptr.
7321 (rs6000_secondary_memory_needed): Match hook interface.
7322 (rs6000_debug_secondary_memory_needed): Likewise.
7323 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
7324 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
7325 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
7326 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7327 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
7328 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
7330 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
7331 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7332 (rs6000_option_override_internal): Assign to
7333 targetm.secondary_memory_needed rather than
7334 rs6000_secondary_memory_needed_ptr.
7335 (rs6000_secondary_memory_needed): Match hook interface.
7336 (rs6000_debug_secondary_memory_needed): Likewise.
7337 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
7338 * config/s390/s390.c (s390_secondary_memory_needed): New function.
7339 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7340 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
7341 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7342 (sparc_secondary_memory_needed): New function.
7343 * lra-constraints.c (check_and_process_move): Refer to
7344 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7346 (curr_insn_transform): Likewise.
7347 (process_alt_operands): Use targetm.secondary_memory_needed
7348 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7349 (check_secondary_memory_needed_p): Likewise.
7350 (choose_split_class): Likewise.
7351 * reload.c: Unconditionally include code that was previously
7352 conditional on SECONDARY_MEMORY_NEEDED.
7353 (push_secondary_reload): Use targetm.secondary_memory_needed
7354 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7355 (push_reload): Likewise.
7356 * reload1.c: Unconditionally include code that was previously
7357 conditional on SECONDARY_MEMORY_NEEDED.
7358 (choose_reload_regs): Use targetm.secondary_memory_needed
7359 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7360 (gen_reload): Likewise.
7361 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
7363 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
7364 Alan Hayward <alan.hayward@arm.com>
7365 David Sherwood <david.sherwood@arm.com>
7367 * target.def (secondary_memory_needed_mode): New hook:
7368 * targhooks.c (default_secondary_memory_needed_mode): Declare.
7369 * targhooks.h (default_secondary_memory_needed_mode): New function.
7370 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
7371 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
7372 * doc/tm.texi: Regenerate.
7373 * lra-constraints.c (check_and_process_move): Use
7374 targetm.secondary_memory_needed_mode instead of
7375 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
7376 (curr_insn_transform): Likewise.
7377 * reload.c (get_secondary_mem): Likewise.
7378 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7379 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
7381 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7382 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7383 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
7384 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7385 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
7387 * config/powerpcspe/powerpcspe-protos.h
7388 (rs6000_secondary_memory_needed_mode): Delete.
7389 * config/powerpcspe/powerpcspe.c
7390 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7391 (rs6000_secondary_memory_needed_mode): Make static.
7392 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7393 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
7395 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7397 (rs6000_secondary_memory_needed_mode): Make static.
7398 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7399 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
7400 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7401 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7402 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7404 (sparc_secondary_memory_needed_mode): New function.
7405 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
7407 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
7409 * config/aarch64/constraints.md (Umq): New constraint.
7410 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
7412 (mov<mode>): Update condition.
7414 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7416 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
7417 when bitposition is the same.
7419 2017-09-13 Richard Biener <rguenther@suse.de>
7421 * dwarf2out.c (output_die_symbol): Remove.
7422 (output_die): Do not output a DIEs symbol.
7424 2017-09-13 Richard Biener <rguenther@suse.de>
7427 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
7428 default-def to avoid breaking iterator update with the weird
7429 interaction with cgraph_update_edges_for_call_stmt_node.
7431 2017-09-13 Richard Biener <rguenther@suse.de>
7433 * tree-cfg.c (verify_gimple_assign_binary): Add verification
7434 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
7435 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
7436 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
7438 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
7440 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
7441 Disable pc relative literal load irrespective of
7442 TARGET_FIX_ERR_A53_84341 for default.
7444 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
7446 * config/sparc/sparc.c (output_return): Output the source location of
7447 the insn in the delay slot, if any.
7448 (output_sibcall): Likewise.
7450 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
7453 * common.opt (-static-pie): New alias.
7454 (shared): Negate static-pie.
7455 (-no-pie): Update help text.
7457 (static-pie): New option.
7458 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
7459 -static-pie support.
7460 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
7461 (LINK_EH_SPEC): Likewise.
7462 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7463 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7464 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7465 * gcc.c (LINK_COMMAND_SPEC): Likewise.
7466 (init_gcc_specs): Likewise.
7467 (init_spec): Likewise.
7468 (display_help): Update help message for -pie.
7469 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
7472 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
7474 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
7475 (movdi_aarch64): Likewise.
7476 (movti_aarch64): Likewise.
7478 2017-09-12 Simon Wright <simon@pushface.org>
7481 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
7482 calculation of the minor version, always output as 0.
7484 2017-09-12 Jakub Jelinek <jakub@redhat.com>
7487 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
7488 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
7489 on it early, rather than manual conversion late. For
7490 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
7491 instead of performing manual conversion.
7493 2017-09-12 Carl Love <cel@us.ibm.com>
7495 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
7496 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
7498 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7499 VMULOSW): Add definitions.
7500 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7501 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7502 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7503 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7504 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7506 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
7508 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
7510 (movti_aarch64): Likewise.
7511 (movdf_aarch64): Likewise.
7512 (movtf_aarch64): Likewise.
7513 (load_pairdi): Likewise.
7514 (store_pairdi): Likewise.
7515 (load_pairdf): Likewise.
7516 (store_pairdf): Likewise.
7517 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
7518 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
7519 (ldr_got_small_<mode>): Likewise.
7520 (ldr_got_small_28k_<mode>): Likewise.
7521 (ldr_got_tiny): Likewise.
7522 * config/aarch64/iterators.md (ldst_sz): New.
7523 (ldpstp_sz): Likewise.
7524 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
7526 (thunderx_load): Split load_8 to load_16.
7527 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
7529 (thunderx2t99_storepair_basic): Split store_8 to store_16.
7530 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
7531 (xgene1_store_pair): Split store_8 to store_16.
7532 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
7533 (falkor_st_0_st_sd): Split store_8 to store_16.
7535 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
7537 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
7538 and store1/2/3/4 to store_4/8/12/16.
7539 * config/aarch64/aarch64.md: Update for rename.
7540 * config/arm/arm.md: Likewise.: Likewise.
7541 * config/arm/arm.c: Likewise.
7542 * config/arm/thumb1.md: Likewise.
7543 * config/arm/thumb2.md: Likewise.
7544 * config/arm/vfp.md: Likewise.
7545 * config/arm/arm-generic.md: Likewise.
7546 * config/arm/arm1020e.md: Likewise.
7547 * config/arm/arm1026ejs.md: Likewise.
7548 * config/arm/arm1136jfs.md: Likewise.
7549 * config/arm/arm926ejs.md: Likewise.
7550 * config/arm/cortex-a15.md: Likewise.
7551 * config/arm/cortex-a17.md: Likewise.
7552 * config/arm/cortex-a5.md: Likewise.
7553 * config/arm/cortex-a53.md: Likewise.
7554 * config/arm/cortex-a57.md: Likewise.
7555 * config/arm/cortex-a7.md: Likewise.
7556 * config/arm/cortex-a8.md: Likewise.
7557 * config/arm/cortex-a9.md: Likewise.
7558 * config/arm/cortex-m4.md: Likewise.
7559 * config/arm/cortex-m7.md: Likewise.
7560 * config/arm/cortex-r4.md: Likewise.
7561 * config/arm/exynos-m1.md: Likewise.
7562 * config/arm/fa526.md: Likewise.
7563 * config/arm/fa606te.md: Likewise.
7564 * config/arm/fa626te.md: Likewise.
7565 * config/arm/fa726te.md: Likewise.
7566 * config/arm/fmp626.md: Likewise.
7567 * config/arm/iwmmxt.md: Likewise.
7568 * config/arm/ldmstm.md: Likewise.
7569 * config/arm/marvell-pj4.md: Likewise.
7570 * config/arm/xgene1.md: Likewise.
7571 * config/aarch64/thunderx.md: Likewise.
7572 * config/aarch64/thunderx2t99.md: Likewise.
7573 * config/aarch64/falkor.md: Likewise.
7575 2017-09-12 Martin Liska <mliska@suse.cz>
7577 * attribs.c (private_lookup_attribute): New function.
7578 * attribs.h (private_lookup_attribute): Declared here.
7579 (lookup_attribute): Called from this place.
7581 2017-09-12 Richard Biener <rguenther@suse.de>
7583 PR tree-optimization/82157
7584 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
7585 stmts with side-effects.
7587 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7588 Alan Hayward <alan.hayward@arm.com>
7589 David Sherwood <david.sherwood@arm.com>
7591 * target.def (hard_regno_nregs): New hook.
7592 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
7593 * targhooks.h (default_hard_regno_nregs): Declare.
7594 * targhooks.c (default_hard_regno_nregs): New function.
7595 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
7596 (TARGET_HARD_REGNO_NREGS): ...this hook.
7597 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
7598 (CLASS_MAX_NREGS): Likewise.
7599 * doc/tm.texi: Regenerate.
7600 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
7601 instead of HARD_REGNO_NREGS.
7602 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
7603 HARD_REGNO_NREGS in the comment.
7604 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
7605 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
7606 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
7607 Return an unsigned int.
7608 (TARGET_HARD_REGNO_NREGS): Redefine.
7609 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
7610 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
7611 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
7612 (arc_hard_regno_nregs): New function.
7613 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
7614 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
7615 (arm_hard_regno_nregs): New function.
7616 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
7617 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
7618 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
7619 (TARGET_HARD_REGNO_NREGS): Redefine.
7620 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
7621 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
7622 (HARD_REGNO_NREGS): Delete.
7623 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
7624 (cr16_hard_regno_nregs): New function.
7625 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
7626 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
7627 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
7628 (cris_hard_regno_nregs): New function.
7629 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
7630 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
7631 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
7632 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
7633 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
7634 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
7635 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
7636 (frv_hard_regno_nregs): Make static. Take and return an
7638 (frv_class_max_nregs): Remove outdated copy of documentation.
7639 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
7640 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
7641 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
7642 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
7643 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
7644 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
7645 (TARGET_HARD_REGNO_NREGS): Redefine.
7646 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
7647 (CLASS_MAX_NREGS): Update comment.
7648 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
7649 (ia64_hard_regno_nregs): New function.
7650 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
7651 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
7652 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
7653 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
7654 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
7656 (m32c_hard_regno_nregs): Likewise. Make static.
7657 (TARGET_HARD_REGNO_NREGS): Redefine.
7658 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
7659 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
7660 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
7661 (m68k_hard_regno_nregs): New function.
7662 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
7663 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
7664 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
7665 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
7666 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
7667 Take and return an unsigned int.
7668 (TARGET_HARD_REGNO_NREGS): Redefine.
7669 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
7670 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
7671 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
7672 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
7673 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
7674 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
7675 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
7676 (msp430_hard_regno_nregs): Make static. Take and return an
7678 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
7679 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
7680 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
7681 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
7682 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
7683 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
7684 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
7685 (TARGET_HARD_REGNO_NREGS): Redefine.
7686 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
7687 (PA_HARD_REGNO_NREGS): ...this.
7688 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
7689 (PA_HARD_REGNO_NREGS): ...this.
7690 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
7691 (pa_hard_regno_nregs): New function.
7692 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
7693 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
7694 (pdp11_hard_regno_nregs): New function.
7695 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
7696 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
7697 (rs6000_hard_regno_nregs_hook): New function.
7698 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
7699 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
7700 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
7701 Take and return an unsigned int. Move earlier in file.
7702 (TARGET_HARD_REGNO_NREGS): Redefine.
7703 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
7704 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
7705 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
7706 (rl78_hard_regno_nregs): Make static. Take and return an
7708 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
7709 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
7710 (rs6000_hard_regno_nregs_hook): New function.
7711 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
7712 * config/rx/rx.c (rx_hard_regno_nregs): New function.
7713 (TARGET_HARD_REGNO_NREGS): Redefine.
7714 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
7715 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
7716 instead of HARD_REGNO_NREGS.
7717 (s390_hard_regno_nregs): New function.
7718 (s390_hard_regno_mode_ok): Add comment from s390.h.
7719 (TARGET_HARD_REGNO_NREGS): Redefine.
7720 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
7721 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
7722 (sh_hard_regno_nregs): New function.
7723 (sh_pass_in_reg_p): Use it.
7724 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
7725 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
7726 (sparc_hard_regno_nregs): New function.
7727 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
7728 * config/spu/spu.c (spu_hard_regno_nregs): New function.
7729 (spu_function_arg_advance): Use it, supplying a valid register number.
7730 (TARGET_HARD_REGNO_NREGS): Redefine.
7731 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
7732 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
7733 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
7734 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
7735 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
7736 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
7737 (CLASS_MAX_NREGS): Remove copy of old documentation.
7738 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
7739 (visium_hard_regno_nregs): New function.
7740 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
7741 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
7742 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
7743 xtensa_hard_regno_nregs): New function.
7744 * system.h (HARD_REGNO_NREGS): Poison.
7746 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7748 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
7749 hard_regno_nregs instead of HARD_REGNO_NREGS.
7750 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
7751 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
7752 (c6x_expand_epilogue): Likewise.
7753 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
7754 (frv_read_iacc_argument): Likewise.
7755 * config/sh/sh.c: Include regs.h.
7756 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
7757 (regs_used): Likewise.
7758 (output_stack_adjust): Likewise.
7759 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
7760 * expmed.c: Include regs.h.
7761 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
7762 * ree.c: Include regs.h.
7763 (combine_reaching_defs): Use hard_regno_nregs instead of
7765 (add_removable_extension): Likewise.
7767 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7769 * regs.h (hard_regno_nregs): Turn into a function.
7770 (end_hard_regno): Update accordingly.
7771 * caller-save.c (setup_save_areas): Likewise.
7772 (save_call_clobbered_regs): Likewise.
7773 (replace_reg_with_saved_mem): Likewise.
7774 (insert_restore): Likewise.
7775 (insert_save): Likewise.
7776 * combine.c (can_change_dest_mode): Likewise.
7777 (move_deaths): Likewise.
7778 (distribute_notes): Likewise.
7779 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
7780 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
7781 (rs6000_split_multireg_move): Likewise.
7782 (rs6000_register_move_cost): Likewise.
7783 (rs6000_memory_move_cost): Likewise.
7784 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
7785 (rs6000_split_multireg_move): Likewise.
7786 (rs6000_register_move_cost): Likewise.
7787 (rs6000_memory_move_cost): Likewise.
7788 * cselib.c (cselib_reset_table): Likewise.
7789 (cselib_lookup_1): Likewise.
7790 * emit-rtl.c (set_mode_and_regno): Likewise.
7791 * function.c (aggregate_value_p): Likewise.
7792 * ira-color.c (setup_profitable_hard_regs): Likewise.
7793 (check_hard_reg_p): Likewise.
7794 (calculate_saved_nregs): Likewise.
7795 (assign_hard_reg): Likewise.
7796 (improve_allocation): Likewise.
7797 (calculate_spill_cost): Likewise.
7798 * ira-emit.c (modify_move_list): Likewise.
7799 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
7800 (ira_hard_reg_in_set_p): Likewise.
7801 * ira.c (setup_reg_mode_hard_regset): Likewise.
7802 (clarify_prohibited_class_mode_regs): Likewise.
7803 (check_allocation): Likewise.
7804 * lra-assigns.c (find_hard_regno_for_1): Likewise.
7805 (lra_setup_reg_renumber): Likewise.
7806 (setup_try_hard_regno_pseudos): Likewise.
7807 (spill_for): Likewise.
7808 (assign_hard_regno): Likewise.
7809 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
7810 * lra-constraints.c (in_class_p): Likewise.
7811 (lra_constraint_offset): Likewise.
7812 (simplify_operand_subreg): Likewise.
7813 (lra_constraints): Likewise.
7814 (split_reg): Likewise.
7815 (split_if_necessary): Likewise.
7816 (invariant_p): Likewise.
7817 (inherit_in_ebb): Likewise.
7818 * lra-lives.c (process_bb_lives): Likewise.
7819 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
7820 (get_hard_regs): Likewise.
7821 (do_remat): Likewise.
7822 * lra-spills.c (assign_spill_hard_regs): Likewise.
7823 * mode-switching.c (create_pre_exit): Likewise.
7824 * postreload.c (reload_combine_recognize_pattern): Likewise.
7825 * recog.c (peep2_find_free_register): Likewise.
7826 * regcprop.c (kill_value_regno): Likewise.
7827 (set_value_regno): Likewise.
7828 (copy_value): Likewise.
7829 (maybe_mode_change): Likewise.
7830 (find_oldest_value_reg): Likewise.
7831 (copyprop_hardreg_forward_1): Likewise.
7832 * regrename.c (check_new_reg_p): Likewise.
7833 (regrename_do_replace): Likewise.
7834 * reload.c (push_reload): Likewise.
7835 (combine_reloads): Likewise.
7836 (find_dummy_reload): Likewise.
7837 (operands_match_p): Likewise.
7838 (find_reloads): Likewise.
7839 (find_equiv_reg): Likewise.
7840 (reload_adjust_reg_for_mode): Likewise.
7841 * reload1.c (count_pseudo): Likewise.
7842 (count_spilled_pseudo): Likewise.
7843 (find_reg): Likewise.
7844 (clear_reload_reg_in_use): Likewise.
7845 (free_for_value_p): Likewise.
7846 (allocate_reload_reg): Likewise.
7847 (choose_reload_regs): Likewise.
7848 (reload_adjust_reg_for_temp): Likewise.
7849 (emit_reload_insns): Likewise.
7850 (delete_output_reload): Likewise.
7851 * rtlanal.c (subreg_get_info): Likewise.
7852 * sched-deps.c (sched_analyze_reg): Likewise.
7853 * sel-sched.c (init_regs_for_mode): Likewise.
7854 (mark_unavailable_hard_regs): Likewise.
7855 (choose_best_reg_1): Likewise.
7856 (verify_target_availability): Likewise.
7857 * valtrack.c (dead_debug_insert_temp): Likewise.
7858 * var-tracking.c (track_loc_p): Likewise.
7859 (emit_note_insn_var_location): Likewise.
7860 * varasm.c (make_decl_rtl): Likewise.
7861 * reginfo.c (choose_hard_reg_mode): Likewise.
7862 (init_reg_modes_target): Refer directly to
7863 this_target_regs->x_hard_regno_nregs.
7865 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7867 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
7868 instead of hard_regno_nregs.
7870 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7872 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
7873 end_hard_regno instead of hard_regno_nregs.
7874 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
7875 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
7876 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
7877 * ira-color.c (improve_allocation): Likewise.
7878 * lra-assigns.c (find_hard_regno_for_1): Likewise.
7879 * lra-lives.c (mark_regno_live): Likewise.
7880 (mark_regno_dead): Likewise.
7881 * lra-remat.c (operand_to_remat): Likewise.
7882 * lra.c (collect_non_operand_hard_regs): Likewise.
7883 * postreload.c (reload_combine_note_store): Likewise.
7884 (move2add_valid_value_p): Likewise.
7885 * reload.c (regno_clobbered_p): Likewise.
7887 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7889 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
7891 * config/v850/v850.c (v850_reorg): Likewise.
7892 * reload.c (refers_to_regno_for_reload_p): Likewise.
7893 (find_equiv_reg): Likewise.
7894 * reload1.c (reload_reg_reaches_end_p): Likewise.
7896 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7898 * caller-save.c (add_used_regs): Use REG_NREGS instead of
7900 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
7901 * config/arm/arm.c (output_move_neon): Likewise.
7902 (arm_attr_length_move_neon): Likewise.
7903 (neon_split_vcombine): Likewise.
7904 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
7905 (c6x_mark_reg_written): Likewise.
7906 (c6x_dwarf_register_span): Likewise.
7907 * config/i386/i386.c (ix86_save_reg): Likewise.
7908 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
7909 (rws_access_reg): Likewise.
7910 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
7911 * mode-switching.c (create_pre_exit): Likewise.
7912 * ree.c (combine_reaching_defs): Likewise.
7913 (add_removable_extension): Likewise.
7914 * regcprop.c (find_oldest_value_reg): Likewise.
7915 (copyprop_hardreg_forward_1): Likewise.
7916 * reload.c (reload_inner_reg_of_subreg): Likewise.
7917 (push_reload): Likewise.
7918 (combine_reloads): Likewise.
7919 (find_dummy_reload): Likewise.
7920 (reload_adjust_reg_for_mode): Likewise.
7921 * reload1.c (find_reload_regs): Likewise.
7922 (forget_old_reloads_1): Likewise.
7923 (reload_reg_free_for_value_p): Likewise.
7924 (reload_adjust_reg_for_temp): Likewise.
7925 (emit_reload_insns): Likewise.
7926 (delete_output_reload): Likewise.
7927 * sel-sched.c (choose_best_reg_1): Likewise.
7928 (choose_best_pseudo_reg): Likewise.
7930 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7931 Alan Hayward <alan.hayward@arm.com>
7932 David Sherwood <david.sherwood@arm.com>
7934 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
7935 * target.def (slow_unaligned_access): New hook.
7936 * targhooks.h (default_slow_unaligned_access): Declare.
7937 * targhooks.c (default_slow_unaligned_access): New function.
7938 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
7939 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
7940 * doc/tm.texi: Regenerate.
7941 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
7942 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
7943 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
7945 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
7946 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
7948 (rs6000_slow_unaligned_access): New function.
7949 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
7950 (expand_block_compare): Likewise.
7951 (expand_strn_compare): Likewise.
7952 (rs6000_rtx_costs): Likewise.
7953 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
7954 (riscv_slow_unaligned_access): Likewise.
7955 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
7956 (riscv_slow_unaligned_access_p): ...this and make static.
7957 (riscv_option_override): Update accordingly.
7958 (riscv_slow_unaligned_access): New function.
7959 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
7960 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
7961 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
7962 (rs6000_slow_unaligned_access): New function.
7963 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
7964 (rs6000_rtx_costs): Likewise.
7965 * config/rs6000/rs6000-string.c (expand_block_compare)
7966 (expand_strn_compare): Use targetm.slow_unaligned_access instead
7967 of SLOW_UNALIGNED_ACCESS.
7968 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
7969 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
7970 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
7971 of SLOW_UNALIGNED_ACCESS.
7972 * expmed.c (simple_mem_bitfield_p): Likewise.
7973 * expr.c (alignment_for_piecewise_move): Likewise.
7974 (emit_group_load_1): Likewise.
7975 (emit_group_store): Likewise.
7976 (copy_blkmode_from_reg): Likewise.
7977 (emit_push_insn): Likewise.
7978 (expand_assignment): Likewise.
7979 (store_field): Likewise.
7980 (expand_expr_real_1): Likewise.
7981 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
7982 * lra-constraints.c (simplify_operand_subreg): Likewise.
7983 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
7984 * gimple-ssa-store-merging.c: Likewise in block comment at start
7986 * tree-ssa-strlen.c: Include target.h.
7987 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
7988 of SLOW_UNALIGNED_ACCESS.
7989 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
7991 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7993 PR rtl-optimization/82185
7994 * expmed.c (emit_store_flag_int): Only test tem if it has been
7997 2017-09-12 Richard Biener <rguenther@suse.de>
8000 * match.pd ((FTYPE) N CMP CST): Fix typo.
8002 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
8004 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
8006 (mips_near_type_p): Add 'short_call' attribute as a synonym
8008 * doc/extend.texi (short_call): Document new function attribute.
8010 2017-09-12 Jakub Jelinek <jakub@redhat.com>
8013 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
8014 assertion check that in the condition.
8015 (get_atomic_generic_size): Likewise. Before testing if parameter
8016 has pointer type, if it has array type, call for C++
8017 default_conversion to perform array-to-pointer conversion.
8019 2017-09-12 Richard Biener <rguenther@suse.de>
8021 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
8022 for operations we cannot scalarize.
8024 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
8026 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
8027 vectors heap vectors. Clean up comments.
8028 Make visited_bbs a reference.
8029 (profitable_jump_thread_path): Make GC
8030 vectors heap vectors. Clean up comments.
8032 (convert_and_register_jump_thread_path): Make GC vectors heap
8034 (check_subpath_and_update_thread_path): Same. Clean up comments.
8035 Make visited_bbs a reference.
8036 (handle_phi): Abstract common code to to
8037 register_jump_thread_path_if_profitable.
8038 Rename VAR_BB to DEF_BB.
8040 Make GC vectors heap vectors.
8041 Make visited_bbs a reference.
8042 (handle_assignment): Same.
8043 (register_jump_thread_path_if_profitable): New.
8044 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
8046 Make GC vectors heap vectors. Clean up comments.
8047 Make visited_bbs a reference.
8048 (find_jump_threads_backwards): Make visited_bbs live in the stack.
8049 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
8052 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
8055 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
8056 words of E_DImode object are reachable by xtensa_uimm8x4 access.
8058 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
8060 Revert r251800 and r251799.
8062 2017-09-11 Martin Jambor <mjambor@suse.cz>
8065 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
8066 arguments in advance.
8067 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
8068 use it to find predecessor edges.
8069 (naive_outof_ssa): Collect vector of predecessors.
8071 2017-09-08 Jason Merrill <jason@redhat.com>
8073 PR c++/70029 - ICE with ref-qualifier and -flto
8074 * langhooks.h (struct lang_hooks_for_types): Add
8075 copy_lang_qualifiers.
8076 * attribs.c (build_type_attribute_qual_variant): Use it.
8077 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
8079 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
8080 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
8082 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
8085 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
8086 (*mulsi3_sp64): New instruction.
8087 (mulsi3): New expander.
8089 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
8091 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
8093 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8095 * sancov.c: Include memmodel.h.
8097 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
8100 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
8103 2017-09-07 Carl Love <cel@us.ibm.com>
8105 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
8106 the sldi instruction.
8108 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
8110 * sancov.c: Include tm_p.h.
8112 2017-09-07 Jakub Jelinek <jakub@redhat.com>
8115 * output.h (switch_to_other_text_partition): New declaration.
8116 * varasm.c (switch_to_other_text_partition): New function.
8117 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
8118 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
8119 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
8120 to the other text partition before emitting LCL label and switch back
8121 after emitting the word after it.
8123 2017-09-07 Richard Biener <rguenther@suse.de>
8125 * passes.def (pass_split_crit_edges): Remove instance before PRE.
8126 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
8127 critical edges here, after loop init.
8128 (pass_data_pre): Remove PROP_no_crit_edges flags.
8129 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
8130 for valueization of call args to avoid leaking VN_TOP.
8131 (visit_use): Assert we do not visit default defs.
8132 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
8133 Use error_mark_node to more easily detect leaking VN_TOP.
8134 All default-defs are varying, not VN_TOP. Mark them visited.
8135 (run_scc_vn): Make code match comment.
8137 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8139 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
8140 OPTION_MASK_FLOAT128_KEYWORD.
8141 (POWERPC_MASKS): Likewise.
8142 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
8143 support for the -mfloat128-type option, and make -mfloat128
8144 default on PowerPC Linux systems. Define or undefine
8145 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
8146 Define __float128 to be __ieee128 if IEEE 128-bit support is
8147 enabled, or undefine it.
8148 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
8149 Delete defining __FLOAT128_TYPE__.
8150 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
8151 -mfloat128-type option and make -mfloat128 default on PowerPC
8153 (TARGET_FLOAT128_TYPE): Likewise.
8154 (-mfloat128-type): Likewise.
8155 * config/rs6000/rs6000.c (rs6000_option_override_internal):
8156 Delete the -mfloat128-type option and make -mfloat128 default on
8157 PowerPC Linux systems. Always use __ieee128 to be the keyword for
8158 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
8159 128-bit floating point is enabled. Change tests from using
8160 -mfloat128-type to -mfloat128.
8161 (rs6000_mangle_type): Use the correct mangling for the __float128
8162 type even if normal long double is restricted to 64-bits.
8163 (floatn_mode): Enable the _Float128 type by default on VSX Linux
8165 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
8166 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
8167 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
8169 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
8170 documentation for -mfloat128.
8172 2017-09-06 Olivier Hainque <hainque@adacore.com>
8174 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
8176 2017-09-06 Wish Wu <wishwu007@gmail.com>
8177 Jakub Jelinek <jakub@redhat.com>
8179 * asan.c (initialize_sanitizer_builtins): Add
8180 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
8181 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
8182 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
8183 BT_FN_VOID_UINT64_PTR variables.
8184 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
8185 (BT_FN_VOID_UINT16_UINT16): Likewise.
8186 (BT_FN_VOID_UINT32_UINT32): Likewise.
8187 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
8188 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
8189 (BT_FN_VOID_UINT64_PTR): Likewise.
8190 * common.opt (flag_sanitize_coverage): New variable.
8191 (fsanitize-coverage=trace-pc): Remove.
8192 (fsanitize-coverage=): Add.
8193 * flag-types.h (enum sanitize_coverage_code): New enum.
8194 * fold-const.c (fold_range_test): Disable non-short-circuit
8195 optimization if flag_sanitize_coverage.
8196 (fold_truth_andor): Likewise.
8197 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
8198 * opts.c (COVERAGE_SANITIZER_OPT): Define.
8199 (coverage_sanitizer_opts): New array.
8200 (get_closest_sanitizer_option): Add OPTS argument, handle also
8201 OPT_fsanitize_coverage_.
8202 (parse_sanitizer_options): Adjusted to also handle
8203 OPT_fsanitize_coverage_.
8204 (common_handle_option): Add OPT_fsanitize_coverage_.
8205 * sancov.c (instrument_comparison, instrument_switch): New function.
8206 (sancov_pass): Add trace-cmp support.
8207 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
8208 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
8209 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
8210 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
8211 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
8212 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
8213 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
8214 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
8215 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
8216 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
8218 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
8220 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
8221 error. Only quit immediately if parsing is complete.
8222 (BEGIN): Initialize fatal_err and parse_done.
8223 (begin fpu, end fpu): Check number of arguments.
8224 (begin arch, end arch): Likewise.
8225 (begin cpu, end cpu): Likewise.
8226 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
8227 (optalias): Likewise.
8229 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
8231 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
8232 * config/arm/arm-isa.h: Delete. Move definitions to ...
8233 * arm-cpus.in: ... here. Use new feature and fgroup values.
8234 * config/arm/arm.c (arm_option_override): Use lower case for feature
8236 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
8237 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
8238 * config/arm/parsecpu.awk (END): Add new command 'isa'.
8240 (print_isa_bits_for): New function.
8241 (gen_isa): New function.
8242 (gen_comm_data): Use print_isa_bits_for.
8243 (define feature): New keyword.
8244 (define fgroup): New keyword.
8245 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
8246 (arm-isa.h): Add rule to generate file.
8247 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
8248 case for feature bit names.
8250 2017-09-06 Richard Biener <rguenther@suse.de>
8252 * tree-ssa-pre.c (NECESSARY): Remove.
8253 (create_expression_by_pieces): Do not touch pass-local flags.
8254 (insert_into_preds_of_block): Likewise.
8255 (do_pre_regular_insertion): Likewise.
8256 (eliminate_insert): Likewise.
8257 (eliminate_dom_walker::before_dom_children): Likewise.
8258 (fini_eliminate): Do not look at inserted_exprs.
8259 (mark_operand_necessary): Remove.
8260 (remove_dead_inserted_code): Replace with simple work-list
8261 algorithm based on inserted_exprs and SSA uses.
8262 (pass_pre::execute): Re-order fini_eliminate and
8263 remove_dead_inserted_code.
8265 2017-09-06 Olivier Hainque <hainque@adacore.com>
8267 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
8268 for VxWorks 7. Adjust surrounding comments.
8270 2017-09-06 Richard Biener <rguenther@suse.de>
8272 * gimple-ssa-strength-reduction.c
8273 (find_candidates_dom_walker::before_dom_children): Also allow
8276 2017-09-06 Richard Biener <rguenther@suse.de>
8278 PR tree-optimization/82108
8279 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
8280 for gap in the non-permutation SLP case.
8282 2017-09-06 Martin Jambor <mjambor@suse.cz>
8284 PR tree-optimization/82078
8285 * tree-sra.c (sort_and_splice_var_accesses): Move call to
8286 add_access_to_work_queue...
8287 (build_accesses_from_assign): ...here.
8288 (propagate_all_subaccesses): Make sure racc is the group
8289 representative, if there is one.
8291 2017-09-06 Jakub Jelinek <jakub@redhat.com>
8294 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
8297 2017-09-06 Richard Biener <rguenther@suse.de>
8299 * gimple-ssa-strength-reduction.c
8300 (find_candidates_dom_walker::before_doom_children): Use a
8301 type and not a mode check.
8303 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
8306 * config/arm/predicates.md (arm_general_adddi_operand): Create new
8308 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
8310 2017-09-05 Jeff Law <law@redhat.com>
8312 PR tree-optimization/64910
8313 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
8314 cases where we have 3 or more operands.
8316 2017-09-05 Jakub Jelinek <jakub@redhat.com>
8319 * omp-low.c (lower_omp_for): Recompute tree invariant if
8320 gimple_omp_for_initial/final is ADDR_EXPR.
8323 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
8324 into gimple val before gimplification fo the COND_EXPR.
8326 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
8328 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
8329 REGION_COPY argument.
8330 (thread_through_all_blocks): Remove unused argument to
8331 duplicate_thread_path.
8333 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8334 Alan Hayward <alan.hayward@arm.com>
8335 David Sherwood <david.sherwood@arm.com>
8337 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
8338 Take a scalar_mode rather than a machine_mode.
8339 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8340 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
8341 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8342 (aarch64_gen_adjusted_ldpstp): Likewise.
8343 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
8345 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8346 Alan Hayward <alan.hayward@arm.com>
8347 David Sherwood <david.sherwood@arm.com>
8349 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
8350 Take a scalar_int_mode instead of a machine_mode.
8351 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8352 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8353 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8354 (aarch64_simd_attr_length_rglist): Delete.
8355 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
8356 a scalar_int_mode instead of a machine_mode.
8357 (aarch64_add_offset): Likewise.
8358 (aarch64_internal_mov_immediate): Likewise
8359 (aarch64_add_constant_internal): Likewise.
8360 (aarch64_add_constant): Likewise.
8361 (aarch64_movw_imm): Likewise.
8362 (aarch64_rtx_arith_op_extract_p): Likewise.
8363 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8364 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8365 Remove assert that the mode isn't a vector.
8366 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8367 (aarch64_expand_mov_immediate): Update calls after above changes.
8368 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
8369 (aarch64_and_bitmask_imm): Check for scalar integer modes.
8370 (aarch64_move_imm): Likewise.
8371 (aarch64_can_const_movi_rtx_p): Likewise.
8372 (aarch64_strip_extend): Likewise.
8373 (aarch64_extr_rtx_p): Likewise.
8374 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
8375 a CONST_INT when the mode parameter is VOIDmode.
8376 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
8378 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8380 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
8381 * stor-layout.c (bitwise_mode_for_mode): Likewise.
8382 (bitwise_type_for_mode): Update accordingly.
8384 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8386 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
8387 * stor-layout.c (mode_for_size_tree): Likewise.
8388 (mode_for_array): Update accordingly.
8389 (layout_decl): Likewise.
8390 (compute_record_mode): Likewise. Only set the mode once.
8392 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8394 * target.def (get_mask_mode): Change return type to opt_mode.
8396 * doc/tm.texi: Regenerate.
8397 * targhooks.h (default_get_mask_mode): Return an opt_mode.
8398 * targhooks.c (default_get_mask_mode): Likewise.
8399 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
8400 * optabs-query.c (can_vec_mask_load_store_p): Update use of
8401 targetm.get_mask_mode.
8402 * tree.c (build_truth_vector_type): Likewise.
8404 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8406 * machmode.h (mode_for_vector): Return an opt_mode.
8407 * stor-layout.c (mode_for_vector): Likewise.
8408 (mode_for_int_vector): Update accordingly.
8409 (layout_type): Likewise.
8410 * config/i386/i386.c (emit_memmov): Likewise.
8411 (ix86_expand_set_or_movmem): Likewise.
8412 (ix86_expand_vector_init): Likewise.
8413 (ix86_get_mask_mode): Likewise.
8414 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
8416 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
8417 * expmed.c (extract_bit_field_1): Likewise.
8418 * expr.c (expand_expr_real_2): Likewise.
8419 * optabs-query.c (can_vec_perm_p): Likewise.
8420 (can_vec_mask_load_store_p): Likewise.
8421 * optabs.c (expand_vec_perm): Likewise.
8422 * targhooks.c (default_get_mask_mode): Likewise.
8423 * tree-vect-stmts.c (vectorizable_store): Likewise.
8424 (vectorizable_load): Likewise.
8425 (get_vectype_for_scalar_type_and_size): Likewise.
8427 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8429 * machmode.h (mode_for_int_vector): New function.
8430 * stor-layout.c (mode_for_int_vector): Likewise.
8431 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
8432 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
8433 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
8434 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
8435 (s390_expand_vcond): Likewise.
8437 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8439 * machmode.h (opt_machine_mode): New type.
8440 (opt_mode<T>): Allow construction from anything that can be
8442 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
8443 (mode_for_size): Return an opt_machine_mode.
8444 * stor-layout.c (mode_for_size): Likewise.
8445 (mode_for_size_tree): Update call accordingly.
8446 (bitwise_mode_for_mode): Likewise.
8447 (make_fract_type): Likewise.
8448 (make_accum_type): Likewise.
8449 * caller-save.c (replace_reg_with_saved_mem): Update call
8451 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8452 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8453 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8454 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8455 * expmed.c (extract_bit_field_1): Likewise.
8456 * reload.c (get_secondary_mem): Likewise.
8457 * varasm.c (assemble_integer): Likewise.
8458 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
8461 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8463 * machmode.h (decimal_float_mode_for_size): New function.
8464 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
8465 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
8466 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
8467 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
8468 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
8469 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
8471 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8473 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
8474 (get_builtin_sync_mode): Likewise.
8475 (expand_ifn_atomic_compare_exchange): Likewise.
8476 (expand_builtin_atomic_clear): Likewise.
8477 (expand_builtin_atomic_test_and_set): Likewise.
8478 (fold_builtin_atomic_always_lock_free): Likewise.
8479 * calls.c (compute_argument_addresses): Likewise.
8480 (emit_library_call_value_1): Likewise.
8481 (store_one_arg): Likewise.
8482 * combine.c (combine_instructions): Likewise.
8483 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
8484 * config/arm/arm.c (arm_function_value): Likewise.
8485 (aapcs_allocate_return_reg): Likewise.
8486 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
8487 * config/i386/i386.c (construct_container): Likewise.
8488 (ix86_gimplify_va_arg): Likewise.
8489 (ix86_expand_sse_cmp): Likewise.
8490 (emit_memmov): Likewise.
8491 (emit_memset): Likewise.
8492 (expand_small_movmem_or_setmem): Likewise.
8493 (ix86_expand_pextr): Likewise.
8494 (ix86_expand_pinsr): Likewise.
8495 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
8496 * config/microblaze/microblaze.c (microblaze_block_move_straight):
8498 * config/mips/mips.c (mips_function_value_1) Likewise.
8499 (mips_block_move_straight): Likewise.
8500 (mips_expand_ins_as_unaligned_store): Likewise.
8501 * config/powerpcspe/powerpcspe.c
8502 (rs6000_darwin64_record_arg_advance_flush): Likewise.
8503 (rs6000_darwin64_record_arg_flush): Likewise.
8504 * config/rs6000/rs6000.c
8505 (rs6000_darwin64_record_arg_advance_flush): Likewise.
8506 (rs6000_darwin64_record_arg_flush): Likewise.
8507 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
8508 (sparc_function_value_1): Likewise.
8509 * config/spu/spu.c (adjust_operand): Likewise.
8510 (spu_emit_branch_or_set): Likewise.
8511 (arith_immediate_p): Likewise.
8512 * emit-rtl.c (gen_lowpart_common): Likewise.
8513 * expr.c (expand_expr_real_1): Likewise.
8514 * function.c (assign_parm_setup_block): Likewise.
8515 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
8516 * reload1.c (alter_reg): Likewise.
8517 * stor-layout.c (mode_for_vector): Likewise.
8518 (layout_type): Likewise.
8520 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8522 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
8523 (spu_convert_move): Likewise.
8524 * lower-subreg.c (resolve_simple_move): Likewise.
8526 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8529 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
8530 define_insn to a define_expand.
8531 (altivec_vsum2sws_direct): New define_insn.
8532 (altivec_vsumsws): Convert from a define_insn to a define_expand.
8534 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
8536 * config/arm/arm.c (arm_option_params_internal): Improve setting of
8539 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
8544 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
8545 realign stack if stack alignment needed is less than incoming
8548 2017-09-05 Marek Polacek <polacek@redhat.com>
8551 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
8554 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
8556 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
8558 2017-09-05 Richard Biener <rguenther@suse.de>
8560 PR tree-optimization/82084
8561 * fold-const.c (can_native_encode_string_p): Handle wide characters.
8563 2017-09-05 Richard Biener <rguenther@suse.de>
8565 PR tree-optimization/82102
8566 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
8568 2017-09-05 Martin Liska <mliska@suse.cz>
8570 PR tree-optimization/82032
8571 * tree-cfg.c (generate_range_test): New function.
8572 * tree-cfg.h (generate_range_test): Declared here.
8573 * tree-cfgcleanup.c (convert_single_case_switch): New function.
8574 (cleanup_control_expr_graph): Use it.
8575 * tree-switch-conversion.c (try_switch_expansion): Remove
8577 (emit_case_nodes): Use generate_range_test.
8579 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
8582 * config/i386/i386.md (*<btsc><mode>_mask): Add
8583 TARGET_USE_BT to insn constraint.
8584 (*btr<mode>_mask): Ditto.
8586 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
8588 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
8589 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
8591 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
8594 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
8595 TARGET_NEON and TARGET_IWMMXT.
8596 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
8597 TARGET_NEON and TARGET_IWMMXT.
8598 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
8600 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
8602 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
8603 (ix86_rewrite_tls_address): Ditto.
8604 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
8605 (ix86_rewrite_tls_address_1): Ditto.
8606 (ix86_rewrite_tls_address): Ditto.
8607 * config/i386/predicates.md (tls_address_pattern): New predicate.
8608 * config/i386/i386.md (TLS address splitter): New splitter.
8610 2017-09-04 Richard Biener <rguenther@suse.de>
8612 PR tree-optimization/82084
8613 * fold-const.h (can_native_encode_string_p): Declare.
8614 * fold-const.c (can_native_encode_string_p): Factor out from ...
8615 (native_encode_string): ... here.
8616 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
8617 vectorizing stores from constants we later cannot handle.
8619 2017-09-04 Marek Polacek <polacek@redhat.com>
8622 * doc/invoke.texi: Update -Wtautological-compare documentation.
8624 2017-09-04 Jeff Law <law@redhat.com>
8626 PR tree-optimization/64910
8627 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
8628 swap the first and last operand if the last is a constant.
8630 2017-09-04 Marek Polacek <polacek@redhat.com>
8633 * convert.c (do_narrow): When sanitizing signed integer overflows,
8634 bail out for signed types.
8635 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
8637 2017-09-04 Richard Biener <rguenther@suse.de>
8639 PR tree-optimization/82060
8640 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
8641 Move devirtualization after stmt folding and before EH/AB/noreturn
8642 cleanup to get the stmt refs canonicalized. Use a bool instead
8643 of gimple_modified_p since that doesn't work for NOPs. Schedule
8644 NOPs generated by folding for removal.
8646 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8647 Alan Hayward <alan.hayward@arm.com>
8648 David Sherwood <david.sherwood@arm.com>
8650 * coretypes.h (pad_direction): New enum.
8651 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
8652 (FUNCTION_ARG_PADDING): Likewise.
8653 * target.def (function_arg_padding): New hook.
8654 * targhooks.h (default_function_arg_padding): Declare.
8655 * targhooks.c (default_function_arg_padding): New function.
8656 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
8657 (TARGET_FUNCTION_ARG_PADDING): ...this.
8658 * doc/tm.texi: Regenerate.
8659 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
8660 instead of direction.
8661 (compute_argument_addresses): Likewise.
8662 (load_register_parameters): Likewise.
8663 (emit_library_call_value_1): Likewise.
8664 (store_one_arg): Use targetm.calls.function_arg_padding instead
8665 of FUNCTION_ARG_PADDING.
8666 (must_pass_in_stack_var_size_or_pad): Likewise.
8667 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
8668 (emit_group_store): Likewise.
8669 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
8670 instead of FUNCTION_ARG_PADDING.
8671 (emit_push_insn): Likewise, and propagate enum change throughout
8673 * function.h (direction): Delete.
8674 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
8676 * function.c (assign_parm_find_stack_rtl): Likewise.
8677 (assign_parm_setup_block_p): Likewise.
8678 (assign_parm_setup_block): Likewise.
8679 (gimplify_parameters): Likewise.
8680 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
8681 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
8683 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
8684 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8685 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
8686 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
8687 (aarch64_function_arg_padding): ...this new function.
8688 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
8689 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8690 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
8691 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8692 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
8693 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8694 (arm_pad_arg_upward): Replace with...
8695 (arm_function_arg_padding): ...this new function.
8696 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
8698 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
8699 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
8700 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8701 (ia64_hpux_function_arg_padding): Replace with...
8702 (ia64_function_arg_padding): ...this new function. Use pad_direction
8703 instead of direction. Check for TARGET_HPUX.
8704 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
8705 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8706 (iq2000_function_arg_padding): New function.
8707 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
8708 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
8709 (mips_function_arg_padding): ...this new function.
8710 (mips_pad_reg_upward): Update accordingly.
8711 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8712 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
8713 targetm.calls.function_arg_padding.
8714 (FUNCTION_ARG_PADDING): Delete.
8715 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8716 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
8717 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8718 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
8719 (nios2_block_reg_padding): Return pad_direction instead of direction.
8720 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
8721 instead of direction.
8722 (nios2_function_arg_padding): Likewise. Make static.
8723 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8724 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
8725 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
8726 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
8727 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8728 (pa_function_arg_padding): Make static. Return pad_direction instead
8730 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
8731 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8732 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
8733 instead of direction. Use targetm.calls.function_arg_padding.
8734 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
8735 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
8736 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
8737 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
8738 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
8740 (function_arg_padding): Rename to...
8741 (rs6000_function_arg_padding): ...this. Make static. Return
8742 pad_direction instead of direction.
8743 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
8744 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
8745 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8746 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
8747 instead of direction. Use targetm.calls.function_arg_padding.
8748 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
8749 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
8750 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
8751 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
8752 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8753 (function_arg_padding): Rename to...
8754 (rs6000_function_arg_padding): ...this. Make static. Return
8755 pad_direction instead of direction.
8756 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
8757 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
8758 * config/s390/s390.c (s390_function_arg_padding): New function.
8759 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8760 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
8761 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
8762 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8763 (function_arg_padding): Rename to...
8764 (sparc_function_arg_padding): ...this. Make static. Return
8765 pad_direction instead of direction.
8766 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
8767 * config/spu/spu.c (spu_function_arg_padding): New function.
8768 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8769 * system.h (FUNCTION_ARG_PADDING): Poison.
8771 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8772 Alan Hayward <alan.hayward@arm.com>
8773 David Sherwood <david.sherwood@arm.com>
8775 * target.def (modes_tieable_p): New hook.
8776 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
8777 (TARGET_MODES_TIEABLE_P): ...this.
8778 * doc/tm.texi.in: Regenerate.
8779 * hooks.h (hook_bool_mode_mode_true): Declare.
8780 * hooks.c (hook_bool_mode_mode_true): New function.
8781 * combine.c (subst): Use targetm.modes_tieable_p instead of
8783 * dse.c (find_shift_sequence): Likewise.
8784 * expmed.c (extract_low_bits): Likewise.
8785 * lower-subreg.c: Include target.h.
8786 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
8788 * rtlanal.c (rtx_cost): Likewise.
8789 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
8790 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
8791 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
8792 (TARGET_MODES_TIEABLE_P): Redefine.
8793 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
8794 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
8795 (TARGET_MODES_TIEABLE_P): Redefine.
8796 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
8797 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
8798 (arc_modes_tieable_p): New function.
8799 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
8800 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
8801 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
8802 (arm_modes_tieable_p): Make static.
8803 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
8804 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
8805 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
8806 (TARGET_MODES_TIEABLE_P): Redefine.
8807 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
8808 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
8809 (TARGET_MODES_TIEABLE_P): Redefine.
8810 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
8811 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
8812 (cr16_modes_tieable_p): New function.
8813 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
8814 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
8815 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
8816 (TRULY_NOOP_TRUNCATION): Update comment.
8817 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
8818 (TRULY_NOOP_TRUNCATION): Update comment.
8819 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
8820 (frv_modes_tieable_p): New function.
8821 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
8822 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
8823 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
8824 (TARGET_MODES_TIEABLE_P): Redefine.
8825 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
8826 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
8827 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
8828 (TARGET_MODES_TIEABLE_P): Redefine.
8829 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
8830 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
8831 (ia64_modes_tieable_p): New function.
8832 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
8833 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
8834 (iq2000_modes_tieable_p): New function.
8835 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
8836 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
8837 (lm32_modes_tieable_p): New function.
8838 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
8839 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
8840 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
8841 (TARGET_MODES_TIEABLE_P): Redefine.
8842 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
8843 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
8844 (m32r_modes_tieable_p): New function.
8845 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
8846 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
8847 (m68k_modes_tieable_p): New function.
8848 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
8849 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
8850 (mcore_modes_tieable_p): New function.
8851 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
8852 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
8854 (TARGET_MODES_TIEABLE_P): Redefine.
8855 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
8856 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
8857 * config/mips/mips.c (mips_modes_tieable_p): Make static.
8858 (TARGET_MODES_TIEABLE_P): Redefine.
8859 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
8860 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
8861 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
8862 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
8863 (mn10300_modes_tieable_p): ...this and make static.
8864 (TARGET_MODES_TIEABLE_P): Redefine.
8865 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
8866 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
8867 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
8868 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
8869 (msp430_modes_tieable_p): Make static.
8870 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
8871 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
8872 (TARGET_MODES_TIEABLE_P): Redefine.
8873 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
8874 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
8875 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
8876 (TARGET_MODES_TIEABLE_P): Redefine.
8877 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
8878 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
8879 * config/pa/pa.c (pa_modes_tieable_p): Make static.
8880 (TARGET_MODES_TIEABLE_P): Redefine.
8881 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
8882 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
8883 (pdp11_modes_tieable_p): New function.
8884 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
8885 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
8886 (rs6000_modes_tieable_p): New function.
8887 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
8888 * config/powerpcspe/powerpcspe.md: Update comment.
8889 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
8890 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
8891 (TARGET_MODES_TIEABLE_P): Redefine.
8892 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
8893 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
8894 (rl78_modes_tieable_p): New function.
8895 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
8896 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
8897 (rs6000_modes_tieable_p): New function.
8898 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
8899 * config/rs6000/rs6000.md: Update comment.
8900 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
8901 * config/rx/rx.c (rx_modes_tieable_p): New function.
8902 (TARGET_MODES_TIEABLE_P): Redefine.
8903 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
8904 * config/s390/s390.c (s390_modes_tieable_p): New function.
8905 (TARGET_MODES_TIEABLE_P): Redefine.
8906 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
8907 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
8908 (sh_modes_tieable_p): New function.
8909 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
8910 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
8911 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
8912 (sparc_modes_tieable_p): Make static.
8913 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
8914 * config/spu/spu.c (spu_modes_tieable_p): New function.
8915 (TARGET_MODES_TIEABLE_P): Redefine.
8916 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
8917 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
8918 (TARGET_MODES_TIEABLE_P): Redefine.
8919 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
8920 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
8921 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
8922 * config/v850/v850.c (v850_modes_tieable_p): New function.
8923 (TARGET_MODES_TIEABLE_P): Redefine.
8924 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
8925 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
8926 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
8927 (visium_modes_tieable_p): New function.
8928 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
8929 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
8930 (xtensa_modes_tieable_p): New function.
8931 * system.h (MODES_TIEABLE_P): Poison.
8933 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8934 Alan Hayward <alan.hayward@arm.com>
8935 David Sherwood <david.sherwood@arm.com>
8937 * target.def (hard_regno_mode_ok): New hook.
8938 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
8939 (TARGET_HARD_REGNO_MODE_OK): ...this.
8940 * doc/tm.texi.in: Regenerate.
8941 * hooks.h (hook_bool_uint_mode_true): Declare.
8942 * hooks.c (hook_bool_uint_mode_true): New function.
8943 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
8945 * genpreds.c (write_insn_preds_c): Add an include of target.h.
8946 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
8947 instead of HARD_REGNO_MODE_OK.
8948 * caller-save.c: Include target.h.
8949 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
8951 * combine.c (can_combine_p): Likewise.
8952 (combinable_i3pat): Likewise.
8953 (can_change_dest_mode): Likewise.
8954 * expr.c (init_expr_target): Likewise.
8955 (convert_move): Likewise.
8956 (convert_modes): Likewise.
8957 * ira.c (setup_prohibited_class_mode_regs): Likewise.
8958 (setup_prohibited_mode_move_regs): Likewise.
8959 * ira.h (target_ira): Likewise.
8960 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8961 * lra-constraints.c (process_alt_operands): Likewise.
8962 (split_reg): Likewise.
8963 * recog.c (peep2_find_free_register): Likewise.
8964 * ree.c (combine_reaching_defs): Likewise.
8965 * regcprop.c (maybe_mode_change): Likewise.
8966 * reginfo.c (init_reg_sets_1): Likewise.
8967 (choose_hard_reg_mode): Likewise.
8968 (simplifiable_subregs): Likewise.
8969 * regrename.c (check_new_reg_p): Likewise.
8970 * reload.c (find_valid_class): Likewise.
8971 (find_valid_class_1): Likewise.
8972 (reload_inner_reg_of_subreg): Likewise.
8973 (push_reload): Likewise.
8974 (combine_reloads): Likewise.
8975 (find_dummy_reload): Likewise.
8976 (find_reloads): Likewise.
8977 * reload1.c (find_reg): Likewise.
8978 (set_reload_reg): Likewise.
8979 (allocate_reload_reg): Likewise.
8980 (choose_reload_regs): Likewise.
8981 (reload_adjust_reg_for_temp): Likewise.
8982 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
8983 (simplify_subreg_regno): Likewise.
8984 * sel-sched.c (init_regs_for_mode): Likewise.
8985 * varasm.c (make_decl_rtl): Likewise.
8986 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
8987 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
8989 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
8990 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
8991 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8992 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
8993 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
8994 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8995 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
8996 (arc_mode_class): Delete.
8997 (HARD_REGNO_MODE_OK): Delete.
8998 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8999 (arc_hard_regno_mode_ok): Rename old array to...
9000 (arc_hard_regno_mode_ok_modes): ...this.
9001 (arc_conditional_register_usage): Update accordingly.
9002 (arc_mode_class): Make static.
9003 (arc_hard_regno_mode_ok): New function.
9004 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
9005 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
9006 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9007 (arm_hard_regno_mode_ok): Make static.
9008 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
9010 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
9011 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
9012 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
9014 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9015 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
9016 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
9017 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
9018 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
9019 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9020 * config/bfin/predicates.md (valid_reg_operand): Use
9021 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
9022 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
9023 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
9024 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9025 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
9026 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
9027 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9028 (cr16_hard_regno_mode_ok): Make static and return a bool.
9029 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
9030 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9031 (cris_hard_regno_mode_ok): New function.
9032 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
9033 (epiphany_mode_class): Delete.
9034 (HARD_REGNO_MODE_OK): Delete.
9035 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
9036 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9037 (hard_regno_mode_ok): Rename to...
9038 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
9039 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
9040 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
9042 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
9043 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
9044 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9045 (frv_hard_regno_mode_ok): Make static and return a bool.
9046 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
9048 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
9049 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
9050 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
9051 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
9053 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9054 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
9055 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
9056 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
9058 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9059 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
9060 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9061 (ia64_hard_regno_mode_ok): New function.
9062 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
9063 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9064 (iq2000_hard_regno_mode_ok): New function.
9065 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
9066 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9067 (lm32_hard_regno_mode_ok): New function.
9068 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
9069 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
9070 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
9071 instead of HARD_REGNO_MODE_OK.
9072 (m32c_hard_regno_ok): Rename to...
9073 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
9074 (m32c_cannot_change_mode_class): Update accordingly.
9075 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9076 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
9077 (m32r_mode_class): Delete.
9078 (HARD_REGNO_MODE_OK): Delete.
9079 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9080 (m32r_hard_regno_mode_ok): Rename to...
9081 (m32r_hard_regno_modes): ...this.
9082 (m32r_mode_class): Make static.
9083 (m32r_hard_regno_mode_ok): New function.
9084 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
9085 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
9086 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9087 (m68k_hard_regno_mode_ok): Make static.
9088 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
9089 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9090 (mcore_hard_regno_mode_ok): New function.
9091 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
9092 (HARD_REGNO_MODE_OK): Delete.
9093 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
9095 (microblaze_hard_regno_mode_ok_p): ...this and make static.
9096 (microblaze_hard_regno_mode_ok): New function.
9097 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9098 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
9099 (mips_hard_regno_mode_ok): Delete.
9100 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
9101 (mips_hard_regno_mode_ok_p): ...this and make static.
9102 (mips_hard_regno_mode_ok_p): Rename to...
9103 (mips_hard_regno_mode_ok_uncached): ...this.
9104 (mips_hard_regno_mode_ok): New function.
9105 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
9106 of HARD_REGNO_MODE_OK.
9107 (mips_option_override): Update after above name changes.
9108 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9109 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
9110 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
9111 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
9112 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
9113 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9114 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
9115 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
9116 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
9117 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9118 (msp430_hard_regno_mode_ok): Make static and return a bool.
9119 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
9120 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
9121 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
9123 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9124 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
9125 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
9126 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
9127 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
9128 (PA_HARD_REGNO_MODE_OK): ...this
9129 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
9130 (PA_HARD_REGNO_MODE_OK): ...this.
9131 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9132 (pa_hard_regno_mode_ok): New function.
9133 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
9134 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9135 (pdp11_hard_regno_mode_ok): New function.
9136 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
9137 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
9139 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
9141 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9142 (rs6000_hard_regno_mode_ok): Rename to...
9143 (rs6000_hard_regno_mode_ok_uncached): ...this.
9144 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9145 (rs6000_hard_regno_mode_ok): New function.
9146 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
9147 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
9148 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
9149 (riscv_hard_regno_mode_ok): ...this and make static.
9150 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9151 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
9152 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
9153 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9154 (rl78_hard_regno_mode_ok): Make static and return bool.
9155 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
9156 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
9158 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
9159 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9160 (rs6000_hard_regno_mode_ok): Rename to...
9161 (rs6000_hard_regno_mode_ok_uncached): ...this.
9162 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9163 (rs6000_hard_regno_mode_ok): New function.
9164 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
9165 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
9166 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9167 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
9168 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
9169 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
9170 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9171 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
9172 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
9173 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9174 (sh_hard_regno_mode_ok): Make static.
9175 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
9176 instead of HARD_REGNO_MODE_OK.
9177 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
9178 (sparc_mode_class): Delete.
9179 (HARD_REGNO_MODE_OK): Delete.
9180 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9181 (hard_regno_mode_classes): Make static.
9182 (sparc_mode_class): Likewise.
9183 (sparc_hard_regno_mode_ok): New function.
9184 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
9185 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
9186 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
9188 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9189 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
9190 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
9191 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
9192 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
9193 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9194 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
9195 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
9196 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9197 (visium_hard_regno_mode_ok): New function.
9198 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
9199 instead of HARD_REGNO_MODE_OK.
9200 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
9201 (HARD_REGNO_MODE_OK): Delete.
9202 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
9203 (xtensa_hard_regno_mode_ok_p): ...this and make static.
9204 (xtensa_option_override): Update accordingly.
9205 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9206 (xtensa_hard_regno_mode_ok): New function.
9207 * system.h (HARD_REGNO_MODE_OK): Poison.
9209 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9210 Alan Hayward <alan.hayward@arm.com>
9211 David Sherwood <david.sherwood@arm.com>
9213 * target.def (hard_regno_call_part_clobbered): New hook.
9214 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
9215 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
9216 * doc/tm.texi: Regenerate.
9217 * hooks.h (hook_bool_uint_mode_false): Declare.
9218 * hooks.c (hook_bool_uint_mode_false): New function.
9219 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9220 * cselib.c (cselib_process_insn): Use
9221 targetm.hard_regno_call_part_clobbered instead of
9222 HARD_REGNO_CALL_PART_CLOBBERED.
9223 * ira-conflicts.c (ira_build_conflicts): Likewise.
9224 * ira-costs.c (ira_tune_allocno_costs): Likewise.
9225 * lra-constraints.c (need_for_call_save_p): Likewise.
9226 * lra-lives.c: Include target.h.
9227 (check_pseudos_live_through_calls): Use
9228 targetm.hard_regno_call_part_clobbered instead of
9229 HARD_REGNO_CALL_PART_CLOBBERED.
9230 * regcprop.c: Include target.h.
9231 (copyprop_hardreg_forward_1): Use
9232 targetm.hard_regno_call_part_clobbered instead of
9233 HARD_REGNO_CALL_PART_CLOBBERED.
9234 * reginfo.c (choose_hard_reg_mode): Likewise.
9235 * regrename.c (check_new_reg_p): Likewise.
9236 * reload.c (find_equiv_reg): Likewise.
9237 * reload1.c (emit_reload_insns): Likewise.
9238 * sched-deps.c (deps_analyze_insn): Likewise.
9239 * sel-sched.c (init_regs_for_mode): Likewise.
9240 (mark_unavailable_hard_regs): Likewise.
9241 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
9242 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9243 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
9245 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9246 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9247 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
9249 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
9251 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9252 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9253 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
9255 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9256 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9257 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
9259 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9260 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
9262 * config/powerpcspe/powerpcspe.c
9263 (rs6000_hard_regno_call_part_clobbered): New function.
9264 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9265 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9266 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
9268 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9269 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9270 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
9272 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9273 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9274 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
9276 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9277 Alan Hayward <alan.hayward@arm.com>
9278 David Sherwood <david.sherwood@arm.com>
9280 * rtl.h (subreg_memory_offset): Declare.
9281 * emit-rtl.c (subreg_memory_offset): New function.
9282 * expmed.c (store_bit_field_1): Use it.
9283 * expr.c (undefined_operand_subword_p): Likewise.
9284 * simplify-rtx.c (simplify_subreg): Likewise.
9286 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
9288 PR rtl-optimization/57448
9291 * optabs.c (expand_atomic_load): Place compiler memory barriers if
9292 using atomic_load pattern.
9293 (expand_atomic_store): Likewise.
9295 2017-09-04 Jakub Jelinek <jakub@redhat.com>
9298 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
9299 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
9300 handling. Use replace_call_with_value with NULL instead of
9301 gsi_replace, unlink_stmt_vdef and release_defs.
9303 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
9306 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
9308 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9311 * rtl.h (emit_library_call_value_1): Declare.
9312 (emit_library_call): Replace declaration with a series of overloads.
9313 Remove the parameter count argument.
9314 (emit_library_call_value): Likewise.
9315 * calls.c (emit_library_call_value_1): Make global. Replace varargs
9316 with an "rtx_mode_t *".
9317 (emit_library_call_value): Delete.
9318 (emit_library_call): Likewise.
9319 * asan.c (asan_emit_stack_protection): Update calls accordingly.
9320 (asan_emit_allocas_unpoison): Likewise.
9321 * builtins.c (expand_builtin_powi): Likewise.
9322 (expand_asan_emit_allocas_unpoison): Likewise.
9323 * cfgexpand.c (expand_main_function): Likewise.
9324 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
9325 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
9326 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
9327 * config/arm/arm.c (arm_trampoline_init): Likewise.
9328 (arm_call_tls_get_addr): Likewise.
9329 (arm_expand_divmod_libfunc): Likewise.
9330 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
9331 (smulsi3_highpart): Likewise.
9332 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
9333 (c6x_expand_compare): Likewise.
9334 (c6x_expand_movmem): Likewise.
9335 * config/frv/frv.c (frv_trampoline_init): Likewise.
9336 * config/i386/i386.c (ix86_trampoline_init): Likewise.
9337 (ix86_expand_divmod_libfunc): Likewise.
9338 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
9339 (ia64_expand_compare): Likewise.
9340 (ia64_profile_hook): Likewise.
9341 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
9342 (nonlocal_goto): Likewise.
9343 (restore_stack_nonlocal): Likewise.
9344 * config/m32r/m32r.c (block_move_call): Likewise.
9345 (m32r_trampoline_init): Likewise.
9346 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
9347 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
9348 (m68k_call_m68k_read_tp): Likewise.
9349 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
9350 (microblaze_expand_divide): Likewise.
9351 * config/mips/mips.h (mips_args): Likewise.
9352 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
9353 (MIPS_ICACHE_SYNC): Likewise.
9354 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
9355 (nios2_trampoline_init): Likewise.
9356 * config/pa/pa.c (hppa_tls_call): Likewise.
9357 (pa_trampoline_init): Likewise.
9358 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
9359 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
9360 (expand_strn_compare): Likewise.
9361 (rs6000_generate_compare): Likewise.
9362 (rs6000_expand_float128_convert): Likewise.
9363 (output_profile_hook): Likewise.
9364 (rs6000_trampoline_init): Likewise.
9365 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
9366 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
9367 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
9368 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
9369 (rs6000_generate_compare): Likewise.
9370 (rs6000_expand_float128_convert): Likewise.
9371 (output_profile_hook): Likewise.
9372 (rs6000_trampoline_init): Likewise.
9373 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
9374 * config/sh/sh.c (sh_trampoline_init): Likewise.
9375 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
9376 (sparc_emit_float_lib_cmp): Likewise.
9377 (sparc32_initialize_trampoline): Likewise.
9378 (sparc64_initialize_trampoline): Likewise.
9379 (sparc_profile_hook): Likewise.
9380 * config/spu/spu.c (ea_load_store): Likewise.
9381 * config/spu/spu.md (floatunssidf2): Likewise.
9382 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
9383 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
9384 * config/visium/visium.c (expand_block_move_4): Likewise.
9385 (expand_block_move_2): Likewise.
9386 (expand_block_move_1): Likewise.
9387 (expand_block_set_4): Likewise.
9388 (expand_block_set_2): Likewise.
9389 (expand_block_set_1): Likewise.
9390 (visium_trampoline_init): Likewise.
9391 (visium_profile_hook): Likewise.
9392 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
9393 (xtensa_setup_frame_addresses): Likewise.
9394 (xtensa_trampoline_init): Likewise.
9395 * except.c (sjlj_emit_function_enter): Likewise.
9396 (sjlj_emit_function_exit): Likewise.
9397 * explow.c (allocate_dynamic_stack_space): Likewise.
9398 (probe_stack_range): Likewise.
9399 * expr.c (convert_mode_scalar): Likewise.
9400 * optabs.c (expand_binop): Likewise.
9401 (expand_twoval_binop_libfunc): Likewise.
9402 (expand_unop): Likewise.
9403 (prepare_cmp_insn): Likewise.
9404 (prepare_float_lib_cmp): Likewise.
9405 (expand_float): Likewise.
9406 (expand_fix): Likewise.
9407 (expand_fixed_convert): Likewise.
9408 (maybe_emit_sync_lock_test_and_set): Likewise.
9409 (expand_atomic_compare_and_swap): Likewise.
9410 (expand_mem_thread_fence): Likewise.
9411 (expand_atomic_fetch_op): Likewise.
9413 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
9415 * doc/generic.texi (OpenACC): Adjust URL.
9416 * doc/invoke.texi (C Dialect Options): Ditto.
9418 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
9420 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
9421 predicate for operand 1. Add (m,<S>) constraint.
9422 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
9423 Prevent memory operand 1 with register operand 2.
9425 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
9427 PR rtl-optimization/82024
9428 * combine.c (try_combine): If the combination result is a PARALLEL,
9429 and we only need to retain the SET in there that would be placed
9430 at I2, check that we can place that at I3 instead, before doing so.
9432 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9435 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
9437 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
9438 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
9441 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
9442 Jeff Law <law@redhat.com>
9444 * varasm.c (bss_initializer_p): Do not put constants into .bss
9445 (categorize_decl_for_section): Handle bss_initializer_p returning
9446 false when DECL_INITIAL is NULL.
9448 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9451 * config/s390/s390.c (s390_can_inline_p): New function.
9453 2017-09-01 Jeff Law <law@redhat.com>
9455 PR tree-optimization/82052
9456 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
9457 Always initialize the returned slot after a hash table miss
9458 when INSERT is true.
9460 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
9462 * config/s390/s390.md (mem_signal_fence): Remove.
9463 * doc/md.texi (mem_signal_fence): Remove.
9464 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
9466 * target-insns.def (mem_signal_fence): Remove.
9468 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9471 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
9474 * asan.c (create_odr_indicator): Strip name encoding from assembler
9475 name before appending it after __odr_asan_.
9477 2017-09-01 Martin Liska <mliska@suse.cz>
9479 PR tree-optimization/82059
9480 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
9481 frequency only when an edge is redirected.
9483 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9485 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
9486 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
9487 (arc_conditional_register_usage): Remove ARC600 lp_count
9489 (arc_file_start): Emit Tag_ARC_CPU_variation.
9490 (arc_can_use_doloop_p): New conditions to use ZOLs.
9491 (hwloop_fail): New function.
9492 (hwloop_optimize): Likewise.
9493 (hwloop_pattern_reg): Likewise.
9494 (arc_doloop_hooks): New struct, to be used with reorg_loops.
9495 (arc_reorg_loops): New function, calls reorg_loops.
9496 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
9497 (arc600_corereg_hazard): Remove ZOL checking, case handled by
9499 (arc_loop_hazard): Remove function, functionality moved into
9501 (arc_hazard): Remove arc_loop_hazard call.
9502 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
9503 into hwloop_optimize.
9504 (arc_label_align): Remove ZOL handling.
9505 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
9506 * config/arc/arc.md (doloop_begin): Remove pattern.
9507 (doloop_begin_i): Likewise.
9508 (doloop_end_i): Likewise.
9509 (doloop_fallback): Likewise.
9510 (doloop_fallback_m): Likewise.
9511 (doloop_end): Reimplement expand.
9512 (arc_lp): New pattern for LP instruction.
9513 (loop_end): New pattern.
9514 (loop_fail): Likewise.
9515 (decrement_and_branch_until_zero): Likewise.
9516 * config/arc/arc.opt (mlpc-width): New option.
9517 * doc/invoke.texi (mlpc-width): Document option.
9519 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9521 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
9522 (arc_ccfsm_advance): Fix checking for delay slots.
9523 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
9525 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9527 * config/arc/arc.md (movqi_insn): Add stores to save constant long
9529 (movhi_insn): Update store instruction constraint which are saving
9530 6-bit short immediates.
9531 (movsi_insn): Consider also short scaled load operations.
9532 (zero_extendhisi2_i): Use Usd constraint instead of T.
9533 (extendhisi2_i): Add q constraint.
9534 (arc_clzsi2): Add type and length attributes.
9535 (arc_ctzsi2): Likewise.
9536 * config/arc/constraints.md (Usc): Update constraint, the
9537 assembler can parse two relocations for a single instruction.
9539 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9541 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
9542 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
9544 2017-08-31 Olivier Hainque <hainque@adacore.com>
9546 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
9547 match as powerpc-wrs-vxworks*.
9549 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
9551 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
9552 register constraint for by-element operand.
9553 (aarch64_mls_elt_merge<mode>): Likewise.
9555 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9557 * config/arc/arc.c (arc_can_follow_jump): Check for short
9560 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9562 * config.gcc: Use g.opt for arc.
9563 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
9564 functionality moved to ...
9565 (legitimate_scaled_address_p): New function, ...here.
9566 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
9567 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
9568 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
9570 (arc_override_options): Handle G option.
9571 (arc_output_pic_addr_const): Correct function definition.
9572 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
9573 (arc_decl_anon_ns_mem_p): Delete.
9574 (arc_in_small_data_p): Overhaul this function to take into
9575 consideration the value given via G option.
9576 (arc_rewrite_small_data_1): Renamed and corrected old
9577 arc_rewrite_small_data function.
9578 (arc_rewrite_small_data): New function.
9579 (small_data_pattern): Don't use pic_offset_table_rtx.
9580 * config/arc/arc.h (CC1_SPEC): Recognize G option.
9581 * config/arc/simdext.md (movmisalignv2hi): Use
9582 prepare_move_operands function.
9584 (movmisalign*): Likewise.
9585 * doc/invoke.texi (ARC options): Document -G option.
9587 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9589 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
9591 * config/arc/arc.c (arc_print_operand): Output scalled address for
9592 sdata whenever is possible.
9593 (arc_in_small_data_p): Allow sdata for 64bit datum when double
9594 load/stores are available.
9595 (compact_sda_memory_operand): Check for the alignment required by
9596 code density instructions.
9597 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
9599 * config/arc/constraints.md (Usd): Update constraint.
9600 (Us0): New constraint.
9601 (Usc): Update constraint.
9603 2017-08-31 Richard Biener <rguenther@suse.de>
9606 * dwarf2out.c (dwarf2out_early_global_decl): Process each
9609 2017-08-31 Tamar Christina <tamar.christina@arm.com>
9611 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
9612 Resize type_signature.
9614 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9615 Alan Hayward <alan.hayward@arm.com>
9616 David Sherwood <david.sherwood@arm.com>
9618 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
9619 subregs whose inner modes can be stored in GPRs.
9620 (aarch64_classify_index): Likewise.
9622 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9623 Alan Hayward <alan.hayward@arm.com>
9624 David Sherwood <david.sherwood@arm.com>
9626 * config/aarch64/iterators.md (V_cmp_result): Rename to...
9627 (V_INT_EQUIV): ...this.
9628 (v_cmp_result): Rename to...
9629 (v_int_equiv): ...this.
9630 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
9631 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
9632 (copysign<mode>3): Likewise.
9633 (aarch64_simd_bsl<mode>_internal): Likewise.
9634 (aarch64_simd_bsl<mode>): Likewise.
9635 (vec_cmp<mode><mode>): Likewise.
9636 (vcond<mode><mode>): Likewise.
9637 (vcond<v_cmp_mixed><mode>): Likewise.
9638 (vcondu<mode><v_cmp_mixed>): Likewise.
9639 (aarch64_cm<optab><mode>): Likewise.
9640 (aarch64_cmtst<mode>): Likewise.
9641 (aarch64_fac<optab><mode>): Likewise.
9642 (vec_perm_const<mode>): Likewise.
9643 (vcond_mask_<mode><v_cmp_result>): Rename to...
9644 (vcond_mask_<mode><v_int_equiv>): ...this.
9645 (vec_cmp<mode><v_cmp_result>): Rename to...
9646 (vec_cmp<mode><v_int_equiv>): ...this.
9648 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9649 Alan Hayward <alan.hayward@arm.com>
9650 David Sherwood <david.sherwood@arm.com>
9652 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
9654 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
9655 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
9656 (UNSPEC_LD4_DREG): New unspecs.
9657 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
9658 (aarch64_ld2<mode>_dreg_be): Replace with...
9659 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
9661 (aarch64_ld3<mode>_dreg_le)
9662 (aarch64_ld3<mode>_dreg_be): Replace with...
9663 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
9665 (aarch64_ld4<mode>_dreg_le)
9666 (aarch64_ld4<mode>_dreg_be): Replace with...
9667 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
9670 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9672 PR tree-optimization/81987
9673 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
9674 insert an initializer in a location not dominated by the stride
9677 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
9679 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
9680 on the entire header of the finally block in the fallthru case.
9682 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
9684 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
9686 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
9688 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
9689 rs6000_emit_move_from_cr and call renamed function.
9690 (rs6000_emit_prologue): Call renamed functions.
9691 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
9692 movesi_from_cr, remove volatile CRs.
9694 2017-08-30 Jon Beniston <jon@beniston.com>
9695 Richard Biener <rguenther@suse.de>
9697 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
9698 of VECTOR_MODE_P check.
9699 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
9700 element vector types.
9702 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9704 * df.h (df_read_modify_subreg_p): Remove in favor of...
9705 * rtl.h (read_modify_subreg_p): ...this new function. Take a
9706 const_rtx instead of an rtx.
9707 * cprop.c (local_cprop_find_used_regs): Update accordingly.
9708 * df-problems.c (df_word_lr_mark_ref): Likewise.
9709 * ira-lives.c (mark_pseudo_reg_live): Likewise.
9710 (mark_pseudo_reg_dead): Likewise.
9711 (mark_ref_dead): Likewise.
9712 * reginfo.c (init_subregs_of_mode): Likewise.
9713 * sched-deps.c (sched_analyze_1): Likewise.
9714 * df-scan.c (df_def_record_1): Likewise.
9715 (df_uses_record): Likewise.
9716 (df_read_modify_subreg_p): Remove in favor of...
9717 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
9718 const_rtx instead of an rtx.
9720 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9721 Alan Hayward <alan.hayward@arm.com>
9722 David Sherwood <david.sherwood@arm.com>
9724 * rtl.h (partial_subreg_p): New function.
9725 * caller-save.c (save_call_clobbered_regs): Use it.
9726 * calls.c (expand_call): Likewise.
9727 * combine.c (combinable_i3pat): Likewise.
9728 (simplify_set): Likewise.
9729 (make_extraction): Likewise.
9730 (make_compound_operation_int): Likewise.
9731 (gen_lowpart_or_truncate): Likewise.
9732 (force_to_mode): Likewise.
9733 (make_field_assignment): Likewise.
9734 (reg_truncated_to_mode): Likewise.
9735 (record_truncated_value): Likewise.
9736 (move_deaths): Likewise.
9737 * cse.c (record_jump_cond): Likewise.
9738 (cse_insn): Likewise.
9739 * cselib.c (cselib_lookup_1): Likewise.
9740 * expmed.c (extract_bit_field_using_extv): Likewise.
9741 * function.c (assign_parm_setup_reg): Likewise.
9742 * ifcvt.c (noce_convert_multiple_sets): Likewise.
9743 * ira-build.c (create_insn_allocnos): Likewise.
9744 * lra-coalesce.c (merge_pseudos): Likewise.
9745 * lra-constraints.c (match_reload): Likewise.
9746 (simplify_operand_subreg): Likewise.
9747 (curr_insn_transform): Likewise.
9748 * lra-lives.c (process_bb_lives): Likewise.
9749 * lra.c (new_insn_reg): Likewise.
9750 (lra_substitute_pseudo): Likewise.
9751 * regcprop.c (mode_change_ok): Likewise.
9752 (maybe_mode_change): Likewise.
9753 (copyprop_hardreg_forward_1): Likewise.
9754 * reload.c (push_reload): Likewise.
9755 (find_reloads): Likewise.
9756 (find_reloads_subreg_address): Likewise.
9757 * reload1.c (alter_reg): Likewise.
9758 (eliminate_regs_1): Likewise.
9759 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9761 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
9763 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
9764 back to if statements, including unpack.
9766 2017-08-30 Martin Liska <mliska@suse.cz>
9769 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
9771 (func_checker::compare_asm_inputs_outputs): ... this function.
9772 (func_checker::compare_gimple_asm): Use the function to compare
9773 also ASM constrains.
9774 * ipa-icf-gimple.h: Rename the function.
9776 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9777 Alan Hayward <alan.hayward@arm.com>
9778 David Sherwood <david.sherwood@arm.com>
9780 * coretypes.h (complex_mode): New type.
9781 * gdbhooks.py (build_pretty_printer): Handle it.
9782 * machmode.h (complex_mode): New class.
9783 (complex_mode::includes_p): New function.
9784 (is_complex_int_mode): Likewise.
9785 (is_complex_float_mode): Likewise.
9786 * genmodes.c (get_mode_class): Handle complex mode classes.
9787 * function.c (expand_function_end): Use is_complex_int_mode.
9789 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9790 Alan Hayward <alan.hayward@arm.com>
9791 David Sherwood <david.sherwood@arm.com>
9793 * coretypes.h (scalar_mode_pod): New typedef.
9794 * gdbhooks.py (build_pretty_printer): Handle it.
9795 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
9796 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
9797 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
9798 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
9801 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9802 Alan Hayward <alan.hayward@arm.com>
9803 David Sherwood <david.sherwood@arm.com>
9805 * machmode.h (mode_for_vector): Take a scalar_mode instead
9807 * stor-layout.c (mode_for_vector): Likewise.
9808 * explow.c (promote_mode): Use as_a <scalar_mode>.
9809 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
9811 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9812 Alan Hayward <alan.hayward@arm.com>
9813 David Sherwood <david.sherwood@arm.com>
9815 * target.def (preferred_simd_mode): Take a scalar_mode
9816 instead of a machine_mode.
9817 * targhooks.h (default_preferred_simd_mode): Likewise.
9818 * targhooks.c (default_preferred_simd_mode): Likewise.
9819 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
9820 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
9821 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
9822 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
9823 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
9824 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
9825 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
9826 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
9828 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
9829 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
9830 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
9831 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
9832 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
9833 * doc/tm.texi: Regenerate.
9834 * optabs-query.c (can_vec_mask_load_store_p): Return false for
9837 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9838 Alan Hayward <alan.hayward@arm.com>
9839 David Sherwood <david.sherwood@arm.com>
9841 * target.def (scalar_mode_supported_p): Take a scalar_mode
9842 instead of a machine_mode.
9843 * targhooks.h (default_scalar_mode_supported_p): Likewise.
9844 * targhooks.c (default_scalar_mode_supported_p): Likewise.
9845 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
9846 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
9847 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
9848 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
9849 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
9850 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
9851 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
9852 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
9853 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
9854 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
9855 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
9856 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
9858 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
9859 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
9860 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
9861 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
9862 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
9864 * doc/tm.texi: Regenerate.
9866 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9867 Alan Hayward <alan.hayward@arm.com>
9868 David Sherwood <david.sherwood@arm.com>
9870 * coretypes.h (opt_scalar_mode): New typedef.
9871 * gdbhooks.py (build_pretty_printers): Handle it.
9872 * machmode.h (mode_iterator::get_2xwider): Add overload for
9874 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
9876 * expr.c (convert_mode_scalar): Likewise.
9877 * omp-low.c (omp_clause_aligned_alignment): Likewise.
9878 * optabs.c (expand_float): Likewise.
9879 (expand_fix): Likewise.
9880 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
9882 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9883 Alan Hayward <alan.hayward@arm.com>
9884 David Sherwood <david.sherwood@arm.com>
9886 * optabs.c (expand_float): Explicitly check for scalars before
9887 using a branching expansion.
9888 (expand_fix): Likewise.
9890 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9891 Alan Hayward <alan.hayward@arm.com>
9892 David Sherwood <david.sherwood@arm.com>
9894 * expr.c (convert_mode): Split scalar handling out into...
9895 (convert_mode_scalar): ...this new function. Treat the modes
9898 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9899 Alan Hayward <alan.hayward@arm.com>
9900 David Sherwood <david.sherwood@arm.com>
9902 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
9904 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
9906 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9907 Alan Hayward <alan.hayward@arm.com>
9908 David Sherwood <david.sherwood@arm.com>
9910 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
9911 rather than a machine_mode.
9912 (fixed_from_string): Likewise.
9913 (fixed_convert): Likewise.
9914 (fixed_convert_from_int): Likewise.
9915 (fixed_convert_from_real): Likewise.
9916 (real_convert_from_fixed): Likewise.
9917 * fixed-value.c (fixed_from_double_int): Likewise.
9918 (fixed_from_string): Likewise.
9919 (fixed_convert): Likewise.
9920 (fixed_convert_from_int): Likewise.
9921 (fixed_convert_from_real): Likewise.
9922 (real_convert_from_fixed): Likewise.
9923 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
9925 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9926 Alan Hayward <alan.hayward@arm.com>
9927 David Sherwood <david.sherwood@arm.com>
9929 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
9930 of separate mode class checks. Do not allow vector modes here.
9931 (immed_wide_int_const): Use as_a <scalar_mode>.
9932 * explow.c (trunc_int_for_mode): Likewise.
9933 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
9934 (wi::shwi): Likewise.
9935 (wi::min_value): Likewise.
9936 (wi::max_value): Likewise.
9937 * dwarf2out.c (loc_descriptor): Likewise.
9938 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
9941 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9942 Alan Hayward <alan.hayward@arm.com>
9943 David Sherwood <david.sherwood@arm.com>
9945 * tree.h (SCALAR_TYPE_MODE): New macro.
9946 * expr.c (expand_expr_addr_expr_1): Use it.
9947 (expand_expr_real_2): Likewise.
9948 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
9949 (fold_convert_const_fixed_from_int): Likewise.
9950 (fold_convert_const_fixed_from_real): Likewise.
9951 (native_encode_fixed): Likewise
9952 (native_encode_complex): Likewise
9953 (native_encode_vector): Likewise.
9954 (native_interpret_fixed): Likewise.
9955 (native_interpret_real): Likewise.
9956 (native_interpret_complex): Likewise.
9957 (native_interpret_vector): Likewise.
9958 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
9959 (simd_clone_adjust_argument_types): Likewise.
9960 (simd_clone_init_simd_arrays): Likewise.
9961 (simd_clone_adjust): Likewise.
9962 * stor-layout.c (layout_type): Likewise.
9963 * tree.c (build_minus_one_cst): Likewise.
9964 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
9965 * tree-inline.c (estimate_move_cost): Likewise.
9966 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
9967 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
9968 (vectorizable_reduction): Likewise.
9969 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
9970 (vect_recog_mixed_size_cond_pattern): Likewise.
9971 (check_bool_pattern): Likewise.
9972 (adjust_bool_pattern): Likewise.
9973 (search_type_for_mask_1): Likewise.
9974 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
9975 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
9976 (vectorizable_load): Likewise.
9977 (vectorizable_store): Likewise.
9978 * ubsan.c (ubsan_encode_value): Likewise.
9979 * varasm.c (output_constant): Likewise.
9981 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9982 Alan Hayward <alan.hayward@arm.com>
9983 David Sherwood <david.sherwood@arm.com>
9985 * coretypes.h (scalar_mode): New class.
9986 * machmode.h (scalar_mode): Likewise.
9987 (scalar_mode::includes_p): New function.
9988 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
9989 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
9990 * genmodes.c (get_mode_class): Handle remaining scalar modes.
9991 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
9992 * expmed.c (store_bit_field_1): Likewise.
9993 (extract_bit_field_1): Likewise.
9994 * expr.c (write_complex_part): Likewise.
9995 (read_complex_part): Likewise.
9996 (emit_move_complex_push): Likewise.
9997 (expand_expr_real_2): Likewise.
9998 * function.c (assign_parm_setup_reg): Likewise.
9999 (assign_parms_unsplit_complex): Likewise.
10000 * optabs.c (expand_binop): Likewise.
10001 * rtlanal.c (subreg_get_info): Likewise.
10002 * simplify-rtx.c (simplify_immed_subreg): Likewise.
10003 * varasm.c (output_constant_pool_2): Likewise.
10005 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10006 Alan Hayward <alan.hayward@arm.com>
10007 David Sherwood <david.sherwood@arm.com>
10009 * expmed.c (extract_high_half): Use scalar_int_mode and remove
10011 (expmed_mult_highpart_optab): Likewise.
10012 (expmed_mult_highpart): Likewise.
10014 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10015 Alan Hayward <alan.hayward@arm.com>
10016 David Sherwood <david.sherwood@arm.com>
10018 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
10019 instead of a machine_mode.
10020 (builtin_memset_read_str): Likewise.
10021 * builtins.c (c_readstr): Likewise.
10022 (builtin_memcpy_read_str): Likewise.
10023 (builtin_strncpy_read_str): Likewise.
10024 (builtin_memset_read_str): Likewise.
10025 (builtin_memset_gen_str): Likewise.
10026 (expand_builtin_signbit): Use scalar_int_mode for local variables.
10027 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
10028 instead of a machine_mode.
10029 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
10031 (make_extraction): Likewise.
10032 (try_widen_shift_mode): Take and return scalar_int_modes instead
10034 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
10035 a scalar_int_mode instead of a machine_mode.
10036 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
10037 (avr_addr_space_pointer_mode): Likewise.
10038 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
10039 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
10040 (msp430_unwind_word_mode): Likewise.
10041 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
10042 (spu_addr_space_pointer_mode): Likewise.
10043 (spu_addr_space_address_mode): Likewise.
10044 (spu_libgcc_cmp_return_mode): Likewise.
10045 (spu_libgcc_shift_count_mode): Likewise.
10046 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
10047 (rl78_addr_space_pointer_mode): Likewise.
10048 (fl78_unwind_word_mode): Likewise.
10049 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
10051 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
10052 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
10053 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
10054 (mips_valid_pointer_mode): Likewise.
10055 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
10056 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
10057 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
10059 (ft32_addr_space_address_mode): Likewise.
10060 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
10061 scalar_int_mode instead of a machine_mode.
10062 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
10064 (m32c_addr_space_address_mode): Likewise.
10065 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
10066 (rs6000_eh_return_filter_mode): Likewise.
10067 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
10068 (rs6000_eh_return_filter_mode): Likewise.
10069 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
10070 (s390_libgcc_shift_count_mode): Likewise.
10071 (s390_unwind_word_mode): Likewise.
10072 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
10074 * target.def (mode_rep_extended): Likewise.
10075 (valid_pointer_mode): Likewise.
10076 (addr_space.valid_pointer_mode): Likewise.
10077 (eh_return_filter_mode): Return a scalar_int_mode rather than
10079 (libgcc_cmp_return_mode): Likewise.
10080 (libgcc_shift_count_mode): Likewise.
10081 (unwind_word_mode): Likewise.
10082 (addr_space.pointer_mode): Likewise.
10083 (addr_space.address_mode): Likewise.
10084 * doc/tm.texi: Regenerate.
10085 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
10087 (do_jump): Use scalar_int_mode for local variables.
10088 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
10089 rather than a machine_mode.
10090 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
10091 (scompare_loc_descriptor_wide): Likewise.
10092 (scompare_loc_descriptor_narrow): Likewise.
10093 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
10095 * except.c (sjlj_emit_dispatch_table): Likewise.
10096 (expand_builtin_eh_copy_values): Likewise.
10097 * explow.c (convert_memory_address_addr_space_1): Likewise.
10098 Take a scalar_int_mode rather than a machine_mode.
10099 (convert_memory_address_addr_space): Take a scalar_int_mode rather
10100 than a machine_mode.
10101 (memory_address_addr_space): Use scalar_int_mode for local variables.
10102 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
10103 rather than a machine_mode.
10104 * expmed.c (mask_rtx): Likewise.
10105 (init_expmed_one_conv): Likewise.
10106 (expand_mult_highpart_adjust): Likewise.
10107 (extract_high_half): Likewise.
10108 (expmed_mult_highpart_optab): Likewise.
10109 (expmed_mult_highpart): Likewise.
10110 (expand_smod_pow2): Likewise.
10111 (expand_sdiv_pow2): Likewise.
10112 (emit_store_flag_int): Likewise.
10113 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
10115 (extract_low_bits): Likewise.
10116 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
10118 * expr.c (pieces_addr::adjust): Likewise.
10119 (can_store_by_pieces): Likewise.
10120 (store_by_pieces): Likewise.
10121 (clear_by_pieces_1): Likewise.
10122 (expand_expr_addr_expr_1): Likewise.
10123 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
10124 (expand_expr_real_1): Likewise.
10125 (try_casesi): Likewise.
10126 * final.c (shorten_branches): Likewise.
10127 * fold-const.c (fold_convert_const_int_from_fixed): Change the
10128 type of "mode" to machine_mode.
10129 * internal-fn.c (expand_arith_overflow_result_store): Take a
10130 scalar_int_mode rather than a machine_mode.
10131 (expand_mul_overflow): Use scalar_int_mode for local variables.
10132 * loop-doloop.c (doloop_modify): Likewise.
10133 (doloop_optimize): Likewise.
10134 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
10135 than a machine_mode.
10136 (expand_doubleword_shift_condmove): Likewise.
10137 (expand_doubleword_shift): Likewise.
10138 (expand_doubleword_clz): Likewise.
10139 (expand_doubleword_popcount): Likewise.
10140 (expand_doubleword_parity): Likewise.
10141 (expand_absneg_bit): Use scalar_int_mode for local variables.
10142 (prepare_float_lib_cmp): Likewise.
10143 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
10144 rather than a machine_mode.
10145 (convert_memory_address_addr_space): Likewise.
10146 (get_mode_bounds): Likewise.
10147 (get_address_mode): Return a scalar_int_mode rather than a
10149 * rtlanal.c (get_address_mode): Likewise.
10150 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
10151 than a machine_mode.
10152 * targhooks.c (default_mode_rep_extended): Likewise.
10153 (default_valid_pointer_mode): Likewise.
10154 (default_addr_space_valid_pointer_mode): Likewise.
10155 (default_eh_return_filter_mode): Return a scalar_int_mode rather
10156 than a machine_mode.
10157 (default_libgcc_cmp_return_mode): Likewise.
10158 (default_libgcc_shift_count_mode): Likewise.
10159 (default_unwind_word_mode): Likewise.
10160 (default_addr_space_pointer_mode): Likewise.
10161 (default_addr_space_address_mode): Likewise.
10162 * targhooks.h (default_eh_return_filter_mode): Likewise.
10163 (default_libgcc_cmp_return_mode): Likewise.
10164 (default_libgcc_shift_count_mode): Likewise.
10165 (default_unwind_word_mode): Likewise.
10166 (default_addr_space_pointer_mode): Likewise.
10167 (default_addr_space_address_mode): Likewise.
10168 (default_mode_rep_extended): Take a scalar_int_mode rather than
10170 (default_valid_pointer_mode): Likewise.
10171 (default_addr_space_valid_pointer_mode): Likewise.
10172 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
10174 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
10175 rather than a machine_mode.
10176 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
10177 for local variables.
10178 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10179 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
10180 than a machine_mode.
10182 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10183 Alan Hayward <alan.hayward@arm.com>
10184 David Sherwood <david.sherwood@arm.com>
10186 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
10187 the mode argument to scalar_int_mode.
10188 (do_jump_by_parts_zero_rtx): Likewise.
10189 (do_jump_by_parts_equality_rtx): Likewise.
10190 (do_jump_by_parts_greater): Take a mode argument.
10191 (do_jump_by_parts_equality): Likewise.
10192 (do_jump_1): Update calls accordingly.
10194 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10195 Alan Hayward <alan.hayward@arm.com>
10196 David Sherwood <david.sherwood@arm.com>
10198 * is-a.h (safe_dyn_cast): New function.
10199 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
10200 (jump_table_for_label): Likewise.
10201 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
10202 instead of an rtx_insn *.
10203 (shorten_branches): Use dyn_cast instead of LABEL_P and
10204 JUMP_TABLE_DATA_P. Use jump_table_for_label and
10205 rtx_jump_table_data::get_data_mode.
10206 (final_scan_insn): Likewise.
10208 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10209 Alan Hayward <alan.hayward@arm.com>
10210 David Sherwood <david.sherwood@arm.com>
10212 * combine.c (try_combine): Use is_a <scalar_int_mode> when
10213 trying to combine a full-register integer set with a subreg
10216 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10217 Alan Hayward <alan.hayward@arm.com>
10218 David Sherwood <david.sherwood@arm.com>
10220 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
10221 that is always either address_mode or pointer_mode.
10223 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10224 Alan Hayward <alan.hayward@arm.com>
10225 David Sherwood <david.sherwood@arm.com>
10227 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
10228 when the two are known to be equal.
10230 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10231 Alan Hayward <alan.hayward@arm.com>
10232 David Sherwood <david.sherwood@arm.com>
10234 * simplify-rtx.c (simplify_const_unary_operation): Use
10235 is_a <scalar_int_mode> instead of checking for a nonzero
10236 precision. Forcibly convert op_mode to a scalar_int_mode
10237 in that case. More clearly differentiate the operand and
10238 result modes and use the former when deciding what the value
10239 of a count-bits operation should be. Use is_int_mode instead
10240 of checking for a MODE_INT. Remove redundant check for whether
10241 this mode has a zero precision.
10243 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10244 Alan Hayward <alan.hayward@arm.com>
10245 David Sherwood <david.sherwood@arm.com>
10247 * optabs.c (widen_leading): Change the type of the mode argument
10248 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
10249 (widen_bswap): Likewise.
10250 (expand_parity): Likewise.
10251 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
10252 (expand_ffs): Likewise.
10253 (epand_unop): Check for scalar integer modes before calling the
10256 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10257 Alan Hayward <alan.hayward@arm.com>
10258 David Sherwood <david.sherwood@arm.com>
10260 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
10262 (expand_expr_real_1): Update call accordingly.
10264 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10265 Alan Hayward <alan.hayward@arm.com>
10266 David Sherwood <david.sherwood@arm.com>
10268 * expmed.c (store_bit_field_using_insv): Add op0_mode and
10269 value_mode arguments. Use scalar_int_mode internally.
10270 (store_bit_field_1): Rename the new integer mode from imode
10271 to op0_mode and use it instead of GET_MODE (op0). Update calls
10272 to store_split_bit_field, store_bit_field_using_insv and
10273 store_fixed_bit_field.
10274 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
10275 Use scalar_int_mode internally. Use a bit count rather than a mode
10276 when calculating the largest bit size for get_best_mode.
10277 Update calls to store_split_bit_field and store_fixed_bit_field_1.
10278 (store_fixed_bit_field_1): Add mode and value_mode arguments.
10279 Remove assertion that OP0 has a scalar integer mode.
10280 (store_split_bit_field): Add op0_mode and value_mode arguments.
10281 Update calls to extract_fixed_bit_field.
10282 (extract_bit_field_using_extv): Add an op0_mode argument.
10283 Use scalar_int_mode internally.
10284 (extract_bit_field_1): Rename the new integer mode from imode to
10285 op0_mode and use it instead of GET_MODE (op0). Update calls to
10286 extract_split_bit_field, extract_bit_field_using_extv and
10287 extract_fixed_bit_field.
10288 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
10289 to extract_split_bit_field and extract_fixed_bit_field_1.
10290 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
10291 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
10292 on the target mode.
10293 (extract_split_bit_field): Add an op0_mode argument. Update call
10294 to extract_fixed_bit_field.
10296 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10297 Alan Hayward <alan.hayward@arm.com>
10298 David Sherwood <david.sherwood@arm.com>
10300 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
10301 * explow.c (hard_function_value): Likewise.
10302 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
10303 convert_to_mode call outside the loop.
10304 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
10305 for the mode iterator. Require the mode specified by max_pieces
10307 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
10309 (copy_blkmode_to_reg): Likewise.
10310 (set_storage_via_setmem): Likewise.
10311 * optabs.c (prepare_cmp_insn): Likewise.
10312 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
10313 * stor-layout.c (finish_bitfield_representative): Likewise.
10315 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10316 Alan Hayward <alan.hayward@arm.com>
10317 David Sherwood <david.sherwood@arm.com>
10319 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
10320 * expr.c (convert_move): Use them.
10321 (convert_modes): Likewise.
10322 (store_expr_with_bounds): Likewise.
10324 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10325 Alan Hayward <alan.hayward@arm.com>
10326 David Sherwood <david.sherwood@arm.com>
10328 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
10329 parameter for the mode of "x". Remove the "known_x", "known_mode"
10330 and "known_ret" arguments. Change the type of the mode argument
10331 to scalar_int_mode.
10332 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
10333 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
10334 (reg_num_sign_bit_copies_for_combine): Likewise.
10335 * rtlanal.c (nonzero_bits1): Likewise.
10336 (num_sign_bit_copies1): Likewise.
10337 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
10338 (reg_num_sign_bit_copies_general): Likewise.
10339 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
10340 (reg_nonzero_bits_general): Likewise.
10342 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10343 Alan Hayward <alan.hayward@arm.com>
10344 David Sherwood <david.sherwood@arm.com>
10346 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
10347 than in subroutines. Return 1 for non-integer modes.
10348 (cached_num_sign_bit_copies): Change the type of the mode parameter
10349 to scalar_int_mode.
10350 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
10351 classes. Handle CONST_INT_P first and then check whether X also
10352 has a scalar integer mode. Check the same thing for inner registers
10353 of a SUBREG and for values that are being extended or truncated.
10355 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10356 Alan Hayward <alan.hayward@arm.com>
10357 David Sherwood <david.sherwood@arm.com>
10359 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
10360 in subroutines. Return the mode mask for non-integer modes.
10361 (cached_nonzero_bits): Change the type of the mode parameter
10362 to scalar_int_mode.
10363 (nonzero_bits1): Likewise. Remove early exit for other mode
10364 classes. Handle CONST_INT_P first and then check whether X
10365 also has a scalar integer mode.
10367 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10368 Alan Hayward <alan.hayward@arm.com>
10369 David Sherwood <david.sherwood@arm.com>
10371 * expr.c (widest_int_mode_for_size): Make the comment match the code.
10372 Return a scalar_int_mode and assert that the size is greater than
10374 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
10375 (op_by_pieces_d::op_by_pieces_d): Likewise.
10376 (op_by_pieces_d::run): Likewise.
10377 (can_store_by_pieces): Likewise.
10379 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10380 Alan Hayward <alan.hayward@arm.com>
10381 David Sherwood <david.sherwood@arm.com>
10383 * combine.c (extract_left_shift): Add a mode argument and update
10385 (make_compound_operation_int): Change the type of the mode parameter
10386 to scalar_int_mode and update the call to extract_left_shift.
10388 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10389 Alan Hayward <alan.hayward@arm.com>
10390 David Sherwood <david.sherwood@arm.com>
10392 * combine.c (simplify_and_const_int): Change the type of the mode
10393 parameter to scalar_int_mode.
10394 (simplify_and_const_int_1): Likewise. Update recursive call.
10396 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10397 Alan Hayward <alan.hayward@arm.com>
10398 David Sherwood <david.sherwood@arm.com>
10400 * combine.c (simplify_compare_const): Check that the mode is a
10401 scalar_int_mode (rather than VOIDmode) before testing its
10403 (simplify_comparison): Move COMPARISON_P handling out of the
10404 loop and restrict the latter part of the loop to scalar_int_modes.
10405 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
10406 and when considering SUBREG_REGs. Use is_int_mode instead of
10407 checking GET_MODE_CLASS against MODE_INT.
10409 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10410 Alan Hayward <alan.hayward@arm.com>
10411 David Sherwood <david.sherwood@arm.com>
10413 * combine.c (try_widen_shift_mode): Move check for equal modes to...
10414 (simplify_shift_const_1): ...here. Use scalar_int_mode for
10415 shift_unit_mode and for modes involved in scalar shifts.
10417 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10418 Alan Hayward <alan.hayward@arm.com>
10419 David Sherwood <david.sherwood@arm.com>
10421 * combine.c (force_int_to_mode): New function, split out from...
10422 (force_to_mode): ...here. Keep xmode up-to-date and use it
10423 instead of GET_MODE (x).
10425 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10426 Alan Hayward <alan.hayward@arm.com>
10427 David Sherwood <david.sherwood@arm.com>
10429 * optabs-query.h (extraction_insn::struct_mode): Change type to
10430 opt_scalar_int_mode and update comment.
10431 (extraction_insn::field_mode): Change type to scalar_int_mode.
10432 (extraction_insn::pos_mode): Likewise.
10433 * combine.c (make_extraction): Update accordingly.
10434 * optabs-query.c (get_traditional_extraction_insn): Likewise.
10435 (get_optab_extraction_insn): Likewise.
10436 * recog.c (simplify_while_replacing): Likewise.
10437 * expmed.c (narrow_bit_field_mem): Change the type of the mode
10438 parameter to opt_scalar_int_mode.
10440 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10441 Alan Hayward <alan.hayward@arm.com>
10442 David Sherwood <david.sherwood@arm.com>
10444 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
10445 to a scalar_int_mode instead of a machine_mode.
10446 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
10447 (get_best_mode): Return a boolean and use a pointer argument to store
10448 the selected mode. Replace the limit mode parameter with a bit limit.
10449 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
10450 for the values returned by bit_field_mode_iterator::next_mode.
10451 (store_bit_field): Update call to get_best_mode.
10452 (store_fixed_bit_field): Likewise.
10453 (extract_fixed_bit_field): Likewise.
10454 * expr.c (optimize_bitfield_assignment_op): Likewise.
10455 * fold-const.c (optimize_bit_field_compare): Likewise.
10456 (fold_truth_andor_1): Likewise.
10457 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
10458 Update for new type of m_mode.
10459 (get_best_mode): As above.
10461 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10462 Alan Hayward <alan.hayward@arm.com>
10463 David Sherwood <david.sherwood@arm.com>
10465 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
10466 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
10467 (store_bit_field): Check is_a <scalar_int_mode> before calling
10468 strict_volatile_bitfield_p.
10469 (extract_bit_field): Likewise.
10471 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10472 Alan Hayward <alan.hayward@arm.com>
10473 David Sherwood <david.sherwood@arm.com>
10475 * target.def (cstore_mode): Return a scalar_int_mode.
10476 * doc/tm.texi: Regenerate.
10477 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
10478 * targhooks.h (default_cstore_mode): Likewise.
10479 * targhooks.c (default_cstore_mode): Likewise, using a forced
10481 * expmed.c (emit_cstore): Expect the target of the cstore to be
10484 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10485 Alan Hayward <alan.hayward@arm.com>
10486 David Sherwood <david.sherwood@arm.com>
10488 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
10490 (niter_desc): Likewise mode.
10491 (iv_analyze): Add a mode parameter.
10493 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
10494 and change its type to scalar_int_mode.
10495 * loop-iv.c: Update commentary at head of file.
10496 (iv_constant): Pass the mode paraeter before the rtx it describes
10497 and change its type to scalar_int_mode. Remove VOIDmode handling.
10498 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
10499 (iv_extend): Likewise.
10500 (shorten_into_mode): Likewise.
10501 (iv_add): Use scalar_int_mode.
10502 (iv_mult): Likewise.
10503 (iv_shift): Likewise.
10504 (canonicalize_iv_subregs): Likewise.
10505 (get_biv_step_1): Pass the outer_mode parameter before the rtx
10506 it describes and change its mode to scalar_int_mode. Also change
10507 the type of the returned inner_mode to scalar_int_mode.
10508 (get_biv_step): Likewise, turning outer_mode from a pointer
10509 into a direct parameter. Update call to get_biv_step_1.
10510 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
10511 iv_constant and get_biv_step.
10512 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
10513 and change its type to scalar_int_mode. Don't initialise iv->mode
10514 to VOIDmode and remove later checks for its still being VOIDmode.
10515 Update calls to iv_analyze_op and iv_analyze_expr. Check
10516 is_a <scalar_int_mode> when changing the mode under consideration.
10517 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
10518 Update call to iv_analyze_expr.
10519 (iv_analyze_op): Add a mode parameter. Reject subregs whose
10520 inner register is not also a scalar_int_mode. Update call to
10522 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
10523 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
10524 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
10525 separate mode class checks. Update calls to iv_analyze. Remove
10526 fix-up of VOIDmodes after iv_analyze_biv.
10527 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
10528 don't have a scalar_int_mode. Update call to biv_p.
10530 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10531 Alan Hayward <alan.hayward@arm.com>
10532 David Sherwood <david.sherwood@arm.com>
10534 * cfgexpand.c (convert_debug_memory_address): Use
10535 as_a <scalar_int_mode>.
10536 * combine.c (expand_compound_operation): Likewise.
10537 (make_extraction): Likewise.
10538 (change_zero_ext): Likewise.
10539 (simplify_comparison): Likewise.
10540 * cse.c (cse_insn): Likewise.
10541 * dwarf2out.c (minmax_loc_descriptor): Likewise.
10542 (mem_loc_descriptor): Likewise.
10543 (loc_descriptor): Likewise.
10544 * expmed.c (init_expmed_one_mode): Likewise.
10545 (synth_mult): Likewise.
10546 (emit_store_flag_1): Likewise.
10547 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
10548 of a comparison with size.
10549 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
10550 (reduce_to_bit_field_precision): Likewise.
10551 * function.c (expand_function_end): Likewise.
10552 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
10553 * loop-doloop.c (doloop_modify): Likewise.
10554 * optabs.c (expand_binop): Likewise.
10555 (expand_unop): Likewise.
10556 (expand_copysign_absneg): Likewise.
10557 (prepare_cmp_insn): Likewise.
10558 (maybe_legitimize_operand): Likewise.
10559 * recog.c (const_scalar_int_operand): Likewise.
10560 * rtlanal.c (get_address_mode): Likewise.
10561 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10562 (simplify_cond_clz_ctz): Likewise.
10563 * tree-nested.c (get_nl_goto_field): Likewise.
10564 * tree.c (build_vector_type_for_mode): Likewise.
10565 * var-tracking.c (use_narrower_mode): Likewise.
10567 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10568 Alan Hayward <alan.hayward@arm.com>
10569 David Sherwood <david.sherwood@arm.com>
10571 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
10572 * builtins.c (expand_builtin_signbit): Use it.
10573 * cfgexpand.c (expand_debug_expr): Likewise.
10574 * dojump.c (do_jump): Likewise.
10575 (do_compare_and_jump): Likewise.
10576 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
10577 * expmed.c (make_tree): Likewise.
10578 * expr.c (expand_expr_real_2): Likewise.
10579 (expand_expr_real_1): Likewise.
10580 (try_casesi): Likewise.
10581 * fold-const-call.c (fold_const_call_ss): Likewise.
10582 * fold-const.c (unextend): Likewise.
10583 (extract_muldiv_1): Likewise.
10584 (fold_single_bit_test): Likewise.
10585 (native_encode_int): Likewise.
10586 (native_encode_string): Likewise.
10587 (native_interpret_int): Likewise.
10588 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
10589 * internal-fn.c (expand_addsub_overflow): Likewise.
10590 (expand_neg_overflow): Likewise.
10591 (expand_mul_overflow): Likewise.
10592 (expand_arith_overflow): Likewise.
10593 * match.pd: Likewise.
10594 * stor-layout.c (layout_type): Likewise.
10595 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
10596 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
10597 * tree-ssanames.c (get_range_info): Likewise.
10598 * tree-switch-conversion.c (array_value_type) Likewise.
10599 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
10600 (vect_recog_divmod_pattern): Likewise.
10601 (vect_recog_mixed_size_cond_pattern): Likewise.
10602 * tree-vrp.c (extract_range_basic): Likewise.
10603 (simplify_float_conversion_using_ranges): Likewise.
10604 * tree.c (int_fits_type_p): Likewise.
10605 * ubsan.c (instrument_bool_enum_load): Likewise.
10606 * varasm.c (mergeable_string_section): Likewise.
10607 (narrowing_initializer_constant_valid_p): Likewise.
10608 (output_constant): Likewise.
10610 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10611 Alan Hayward <alan.hayward@arm.com>
10612 David Sherwood <david.sherwood@arm.com>
10614 * machmode.h (NARROWEST_INT_MODE): New macro.
10615 * expr.c (alignment_for_piecewise_move): Use it instead of
10616 GET_CLASS_NARROWEST_MODE (MODE_INT).
10617 (push_block): Likewise.
10618 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
10620 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10622 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10623 Alan Hayward <alan.hayward@arm.com>
10624 David Sherwood <david.sherwood@arm.com>
10626 * postreload.c (move2add_valid_value_p): Change the type of the
10627 mode parameter to scalar_int_mode.
10628 (move2add_use_add2_insn): Add a mode parameter and use it instead
10630 (move2add_use_add3_insn): Likewise.
10631 (reload_cse_move2add): Update accordingly.
10633 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10634 Alan Hayward <alan.hayward@arm.com>
10635 David Sherwood <david.sherwood@arm.com>
10637 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
10639 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
10640 * optabs.c (expand_unop): Likewise.
10642 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10643 Alan Hayward <alan.hayward@arm.com>
10644 David Sherwood <david.sherwood@arm.com>
10646 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
10647 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
10648 (popcount_loc_descriptor): Likewise.
10649 (bswap_loc_descriptor): Likewise.
10650 (rotate_loc_descriptor): Likewise.
10651 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
10652 calling the functions above.
10654 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10655 Alan Hayward <alan.hayward@arm.com>
10656 David Sherwood <david.sherwood@arm.com>
10658 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
10660 (try_combine): Likewise.
10661 (simplify_if_then_else): Likewise.
10662 * cse.c (cse_insn): Likewise.
10663 * dwarf2out.c (mem_loc_descriptor): Likewise.
10664 * emit-rtl.c (gen_lowpart_common): Likewise.
10665 * simplify-rtx.c (simplify_truncation): Likewise.
10666 (simplify_binary_operation_1): Likewise.
10667 (simplify_const_relational_operation): Likewise.
10668 (simplify_ternary_operation): Likewise.
10669 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
10671 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10672 Alan Hayward <alan.hayward@arm.com>
10673 David Sherwood <david.sherwood@arm.com>
10675 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
10676 * reload.c (push_reload): Likewise.
10677 (find_reloads): Likewise.
10679 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10680 Alan Hayward <alan.hayward@arm.com>
10681 David Sherwood <david.sherwood@arm.com>
10683 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
10684 (make_compound_operation_int): Likewise.
10685 (change_zero_ext): Likewise.
10686 * expr.c (convert_move): Likewise.
10687 (convert_modes): Likewise.
10688 * fwprop.c (forward_propagate_subreg): Likewise.
10689 * loop-iv.c (get_biv_step_1): Likewise.
10690 * optabs.c (widen_operand): Likewise.
10691 * postreload.c (move2add_valid_value_p): Likewise.
10692 * recog.c (simplify_while_replacing): Likewise.
10693 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10694 (simplify_binary_operation_1): Likewise. Remove redundant
10695 mode equality check.
10697 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10698 Alan Hayward <alan.hayward@arm.com>
10699 David Sherwood <david.sherwood@arm.com>
10701 * combine.c (combine_simplify_rtx): Add checks for
10702 is_a <scalar_int_mode>.
10703 (simplify_if_then_else): Likewise.
10704 (make_field_assignment): Likewise.
10705 (simplify_comparison): Likewise.
10706 * ifcvt.c (noce_try_bitop): Likewise.
10707 * loop-invariant.c (canonicalize_address_mult): Likewise.
10708 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10710 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10711 Alan Hayward <alan.hayward@arm.com>
10712 David Sherwood <david.sherwood@arm.com>
10714 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
10715 is_a <scalar_int_mode> instead of != BLKmode.
10717 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10718 Alan Hayward <alan.hayward@arm.com>
10719 David Sherwood <david.sherwood@arm.com>
10721 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
10722 instead of != VOIDmode.
10723 * combine.c (if_then_else_cond): Likewise.
10724 (change_zero_ext): Likewise.
10725 * dwarf2out.c (mem_loc_descriptor): Likewise.
10726 (loc_descriptor): Likewise.
10727 * rtlanal.c (canonicalize_condition): Likewise.
10728 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
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 * simplify-rtx.c (simplify_binary_operation_1): Use
10735 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
10737 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10738 Alan Hayward <alan.hayward@arm.com>
10739 David Sherwood <david.sherwood@arm.com>
10741 * wide-int.h (int_traits<unsigned char>) New class.
10742 (int_traits<unsigned short>) Likewise.
10743 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
10744 Use GET_MODE_UNIT_PRECISION and remove redundant test for
10746 * combine.c (set_nonzero_bits_and_sign_copies): Use
10747 is_a <scalar_int_mode>.
10748 (find_split_point): Likewise.
10749 (combine_simplify_rtx): Likewise.
10750 (simplify_logical): Likewise.
10751 (expand_compound_operation): Likewise.
10752 (expand_field_assignment): Likewise.
10753 (make_compound_operation): Likewise.
10754 (extended_count): Likewise.
10755 (change_zero_ext): Likewise.
10756 (simplify_comparison): Likewise.
10757 * dwarf2out.c (scompare_loc_descriptor): Likewise.
10758 (ucompare_loc_descriptor): Likewise.
10759 (minmax_loc_descriptor): Likewise.
10760 (mem_loc_descriptor): Likewise.
10761 (loc_descriptor): Likewise.
10762 * expmed.c (init_expmed_one_mode): Likewise.
10763 * lra-constraints.c (lra_constraint_offset): Likewise.
10764 * optabs.c (prepare_libcall_arg): Likewise.
10765 * postreload.c (move2add_note_store): Likewise.
10766 * reload.c (operands_match_p): Likewise.
10767 * rtl.h (load_extend_op): Likewise.
10768 * rtlhooks.c (gen_lowpart_general): Likewise.
10769 * simplify-rtx.c (simplify_truncation): Likewise.
10770 (simplify_unary_operation_1): Likewise.
10771 (simplify_binary_operation_1): Likewise.
10772 (simplify_const_binary_operation): Likewise.
10773 (simplify_const_relational_operation): Likewise.
10774 (simplify_subreg): Likewise.
10775 * stor-layout.c (bitwise_mode_for_mode): Likewise.
10776 * var-tracking.c (adjust_mems): Likewise.
10777 (prepare_call_arguments): Likewise.
10779 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10780 Alan Hayward <alan.hayward@arm.com>
10781 David Sherwood <david.sherwood@arm.com>
10783 * machmode.h (is_int_mode): New fuction.
10784 * combine.c (find_split_point): Use it.
10785 (combine_simplify_rtx): Likewise.
10786 (simplify_if_then_else): Likewise.
10787 (simplify_set): Likewise.
10788 (simplify_shift_const_1): Likewise.
10789 (simplify_comparison): Likewise.
10790 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
10791 * cse.c (notreg_cost): Likewise.
10792 (cse_insn): Likewise.
10793 * cselib.c (cselib_lookup_1): Likewise.
10794 * dojump.c (do_jump_1): Likewise.
10795 (do_compare_rtx_and_jump): Likewise.
10796 * dse.c (get_call_args): Likewise.
10797 * dwarf2out.c (rtl_for_decl_init): Likewise.
10798 (native_encode_initializer): Likewise.
10799 * expmed.c (emit_store_flag_1): Likewise.
10800 (emit_store_flag): Likewise.
10801 * expr.c (convert_modes): Likewise.
10802 (store_field): Likewise.
10803 (expand_expr_real_1): Likewise.
10804 * fold-const.c (fold_read_from_constant_string): Likewise.
10805 * gimple-ssa-sprintf.c (get_format_string): Likewise.
10806 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
10807 * optabs.c (expand_binop): Likewise.
10808 (expand_unop): Likewise.
10809 (expand_abs_nojump): Likewise.
10810 (expand_one_cmpl_abs_nojump): Likewise.
10811 * simplify-rtx.c (mode_signbit_p): Likewise.
10812 (val_signbit_p): Likewise.
10813 (val_signbit_known_set_p): Likewise.
10814 (val_signbit_known_clear_p): Likewise.
10815 (simplify_relational_operation_1): Likewise.
10816 * tree.c (vector_type_mode): Likewise.
10818 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10819 Alan Hayward <alan.hayward@arm.com>
10820 David Sherwood <david.sherwood@arm.com>
10822 * machmode.h (smallest_mode_for_size): Fix formatting.
10823 (smallest_int_mode_for_size): New function.
10824 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
10825 instead of smallest_mode_for_size.
10826 * combine.c (make_extraction): Likewise.
10827 * config/arc/arc.c (arc_expand_movmem): Likewise.
10828 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
10829 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
10830 * config/s390/s390.c (s390_expand_insv): Likewise.
10831 * config/sparc/sparc.c (assign_int_registers): Likewise.
10832 * config/spu/spu.c (spu_function_value): Likewise.
10833 (spu_function_arg): Likewise.
10834 * coverage.c (get_gcov_type): Likewise.
10835 (get_gcov_unsigned_t): Likewise.
10836 * dse.c (find_shift_sequence): Likewise.
10837 * expmed.c (store_bit_field_1): Likewise.
10838 * expr.c (convert_move): Likewise.
10839 (store_field): Likewise.
10840 * internal-fn.c (expand_arith_overflow): Likewise.
10841 * optabs-query.c (get_best_extraction_insn): Likewise.
10842 * optabs.c (expand_twoval_binop_libfunc): Likewise.
10843 * stor-layout.c (layout_type): Likewise.
10844 (initialize_sizetypes): Likewise.
10845 * targhooks.c (default_get_mask_mode): Likewise.
10846 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
10848 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10849 Alan Hayward <alan.hayward@arm.com>
10850 David Sherwood <david.sherwood@arm.com>
10852 * machmode.h (opt_mode::else_blk): New function.
10853 (int_mode_for_mode): Declare.
10854 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
10855 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
10857 * cfgexpand.c (expand_debug_expr): Likewise.
10858 * combine.c (gen_lowpart_or_truncate): Likewise.
10859 (gen_lowpart_for_combine): Likewise.
10860 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
10861 * config/avr/avr.c (avr_to_int_mode): Likewise.
10862 (avr_out_plus_1): Likewise.
10863 (avr_out_plus): Likewise.
10864 (avr_out_round): Likewise.
10865 * config/i386/i386.c (ix86_split_to_parts): Likewise.
10866 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
10867 (s390_expand_vcond): Likewise.
10868 * config/spu/spu.c (spu_split_immediate): Likewise.
10869 (spu_expand_mov): Likewise.
10870 * dse.c (get_stored_val): Likewise.
10871 * expmed.c (store_bit_field_1): Likewise.
10872 (convert_extracted_bit_field): Use int_mode_for_mode instead of
10874 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
10875 (extract_low_bits): Likewise.
10876 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
10877 handling rather than repeating the check.
10878 (emit_group_store): Likewise.
10879 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
10880 * optabs.c (expand_absneg_bit): Likewise.
10881 (expand_copysign_absneg): Likewise.
10882 (expand_copysign_bit): Likewise.
10883 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
10884 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
10885 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
10886 * var-tracking.c (prepare_call_arguments): Likewise.
10887 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
10888 int_mode_for_mode instead of mode_for_size.
10889 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
10891 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10892 Alan Hayward <alan.hayward@arm.com>
10893 David Sherwood <david.sherwood@arm.com>
10895 * machmode.h (int_mode_for_size): New function.
10896 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
10897 instead of mode_for_size.
10898 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
10900 * combine.c (expand_field_assignment): Use int_mode_for_size
10901 instead of mode_for_size.
10902 (make_extraction): Likewise.
10903 (simplify_shift_const_1): Likewise.
10904 (simplify_comparison): Likewise.
10905 * dojump.c (do_jump): Likewise.
10906 * dwarf2out.c (mem_loc_descriptor): Likewise.
10907 * emit-rtl.c (init_derived_machine_modes): Likewise.
10908 * expmed.c (flip_storage_order): Likewise.
10909 (convert_extracted_bit_field): Likewise.
10910 * expr.c (copy_blkmode_from_reg): Likewise.
10911 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
10912 * internal-fn.c (expand_mul_overflow): Likewise.
10913 * lower-subreg.c (simple_move): Likewise.
10914 * optabs-libfuncs.c (init_optabs): Likewise.
10915 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10916 * tree.c (vector_type_mode): Likewise.
10917 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
10918 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
10919 * tree-vect-generic.c (expand_vector_parallel): Likewise.
10920 * tree-vect-stmts.c (vectorizable_load): Likewise.
10921 (vectorizable_store): Likewise.
10923 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10924 Alan Hayward <alan.hayward@arm.com>
10925 David Sherwood <david.sherwood@arm.com>
10927 * coretypes.h (pod_mode): New type.
10928 (scalar_int_mode_pod): New typedef.
10929 * machmode.h (pod_mode): New class.
10930 (int_n_data_t::m): Change type to scalar_int_mode_pod.
10931 * genmodes.c (emit_mode_int_n): Update accordingly.
10932 * lower-subreg.h (target_lower_subreg): Change type to
10933 scalar_int_mode_pod.
10934 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
10935 scalar_int_mode_pod.
10937 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10938 Alan Hayward <alan.hayward@arm.com>
10939 David Sherwood <david.sherwood@arm.com>
10941 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
10942 machine_mode to scalar_int_mode.
10943 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
10944 (rs6000_option_override_internal): Remove cast to int.
10945 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
10946 machine_mode to scalar_int_mode.
10947 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
10948 (rs6000_option_override_internal): Remove cast to int.
10949 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
10950 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
10952 * config/s390/s390.c (s390_expand_builtin): Likewise.
10953 * coretypes.h (scalar_int_mode): New type.
10954 (opt_scalar_int_mode): New typedef.
10955 * machmode.h (scalar_int_mode): New class.
10956 (scalar_int_mode::includes_p): New function.
10957 (byte_mode): Change type to scalar_int_mode.
10958 (word_mode): Likewise.
10959 (ptr_mode): Likewise.
10960 * emit-rtl.c (byte_mode): Likewise.
10961 (word_mode): Likewise.
10962 (ptr_mode): Likewise.
10963 (init_derived_machine_modes): Update accordingly.
10964 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
10965 and MODE_PARTIAL_INT.
10966 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
10967 opt_scalar_int_mode.
10969 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10970 Alan Hayward <alan.hayward@arm.com>
10971 David Sherwood <david.sherwood@arm.com>
10973 * target.def (libgcc_floating_mode_supported_p): Take a
10975 * doc/tm.texi: Regenerate.
10976 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
10978 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
10979 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
10982 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10983 Alan Hayward <alan.hayward@arm.com>
10984 David Sherwood <david.sherwood@arm.com>
10986 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
10987 * doc/tm.texi: Regenerate.
10988 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
10989 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
10990 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
10991 * targhooks.h (default_floatn_mode): Likewise.
10992 * targhooks.c (default_floatn_mode): Likewise.
10993 * tree.c (build_common_tree_nodes): Update accordingly.
10995 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10996 Alan Hayward <alan.hayward@arm.com>
10997 David Sherwood <david.sherwood@arm.com>
10999 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
11000 (mode_iterator::iterate_p): Likewise.
11001 (mode_iterator::get_wider): Likewise.
11002 * expr.c (init_expr_target): Use opt_scalar_float_mode.
11004 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11005 Alan Hayward <alan.hayward@arm.com>
11006 David Sherwood <david.sherwood@arm.com>
11008 * coretypes.h (opt_scalar_float_mode): New typedef.
11009 * machmode.h (float_mode_for_size): New function.
11010 * emit-rtl.c (double_mode): Delete.
11011 (init_emit_once): Use float_mode_for_size.
11012 * stor-layout.c (layout_type): Likewise.
11013 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
11015 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11016 Alan Hayward <alan.hayward@arm.com>
11017 David Sherwood <david.sherwood@arm.com>
11019 * output.h (assemble_real): Take a scalar_float_mode.
11020 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
11021 * config/arm/arm.md (consttable_4): Likewise.
11022 (consttable_8): Likewise.
11023 (consttable_16): Likewise.
11024 * config/mips/mips.md (consttable_float): Likewise.
11025 * config/s390/s390.c (s390_output_pool_entry): Likewise.
11026 * varasm.c (assemble_real): Take a scalar_float_mode.
11027 (output_constant_pool_2): Update accordingly.
11028 (output_constant): Likewise.
11030 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11031 Alan Hayward <alan.hayward@arm.com>
11032 David Sherwood <david.sherwood@arm.com>
11034 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
11035 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
11036 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
11037 (native_encode_real): Likewise.
11038 (native_interpret_real): Likewise.
11039 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
11040 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11042 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11043 Alan Hayward <alan.hayward@arm.com>
11044 David Sherwood <david.sherwood@arm.com>
11046 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
11047 <scalar_float_mode>. Simplify.
11048 (gen_extend_conv_libfunc): Likewise.
11050 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11051 Alan Hayward <alan.hayward@arm.com>
11052 David Sherwood <david.sherwood@arm.com>
11054 * coretypes.h (scalar_float_mode): New type.
11055 * machmode.h (mode_traits::from_int): Use machine_mode if
11056 USE_ENUM_MODES is defined.
11057 (is_a): New function.
11059 (dyn_cast): Likewise.
11060 (scalar_float_mode): New class.
11061 (scalar_float_mode::includes_p): New function.
11062 (is_float_mode): Likewise.
11063 * gdbhooks.py (MachineModePrinter): New class.
11064 (build_pretty_printer): Use it for scalar_float_mode.
11065 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
11066 (format_helper::format_helper): Turn into a template.
11067 * genmodes.c (get_mode_class): New function.
11068 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
11069 or machine_mode if none.
11070 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
11071 as_a <scalar_float_mode>.
11072 * dwarf2out.c (mem_loc_descriptor): Likewise.
11073 (insert_float): Likewise.
11074 (add_const_value_attribute): Likewise.
11075 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11076 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
11077 (expand_unop): Update accordingly.
11078 (expand_abs_nojump): Likewise.
11079 (expand_copysign_absneg): Take a scalar_float_mode.
11080 (expand_copysign_bit): Likewise.
11081 (expand_copysign): Update accordingly.
11083 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11084 Alan Hayward <alan.hayward@arm.com>
11085 David Sherwood <david.sherwood@arm.com>
11087 * coretypes.h (opt_mode): New class.
11088 * machmode.h (opt_mode): Likewise.
11089 (opt_mode::else_void): New function.
11090 (opt_mode::require): Likewise.
11091 (opt_mode::exists): Likewise.
11092 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
11093 (GET_MODE_2XWIDER_MODE): Likewise.
11094 (mode_iterator::get_wider): Update accordingly.
11095 (mode_iterator::get_2xwider): Likewise.
11096 (mode_iterator::get_known_wider): Likewise, turning into a template.
11097 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
11098 forcing a wider mode to exist.
11099 * config/cr16/cr16.h (LONG_REG_P): Likewise.
11100 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11101 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
11102 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
11103 * lower-subreg.c (init_lower_subreg): Likewise.
11104 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
11105 on the final iteration.
11106 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
11107 a wider mode exists before asking for a move pattern.
11108 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
11109 forcing a wider mode to exist.
11110 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
11111 returning false if no such mode exists.
11112 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
11113 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
11114 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
11115 Avoid checking for a MODE_INT if we already know the mode is not a
11117 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
11118 forcing a wider mode to exist.
11119 (expmed_mult_highpart_optab): Likewise.
11120 (expmed_mult_highpart): Likewise.
11121 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
11123 * lto-streamer-in.c (lto_input_mode_table): Likewise.
11124 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
11125 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
11126 * internal-fn.c (expand_mul_overflow): Update use of
11127 GET_MODE_2XWIDER_MODE.
11128 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11129 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
11130 GET_MODE_WIDER_MODE.
11131 (convert_plusminus_to_widen): Likewise.
11132 * tree-switch-conversion.c (array_value_type): Likewise.
11133 * var-tracking.c (emit_note_insn_var_location): Likewise.
11134 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11135 Return false inside rather than outside the loop if no wider mode
11137 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
11138 and GET_MODE_2XWIDER_MODE
11139 (can_compare_p): Use else_void.
11140 * gdbhooks.py (OptMachineModePrinter): New class.
11141 (build_pretty_printer): Use it for opt_mode.
11143 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11144 Alan Hayward <alan.hayward@arm.com>
11145 David Sherwood <david.sherwood@arm.com>
11147 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
11148 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
11150 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11151 Alan Hayward <alan.hayward@arm.com>
11152 David Sherwood <david.sherwood@arm.com>
11154 * machmode.h (mode_traits): New structure.
11155 (get_narrowest_mode): New function.
11156 (mode_iterator::start): Likewise.
11157 (mode_iterator::iterate_p): Likewise.
11158 (mode_iterator::get_wider): Likewise.
11159 (mode_iterator::get_known_wider): Likewise.
11160 (mode_iterator::get_2xwider): Likewise.
11161 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
11162 (FOR_EACH_MODE): Likewise.
11163 (FOR_EACH_MODE_FROM): Likewise.
11164 (FOR_EACH_MODE_UNTIL): Likewise.
11165 (FOR_EACH_WIDER_MODE): Likewise.
11166 (FOR_EACH_2XWIDER_MODE): Likewise.
11167 * builtins.c (expand_builtin_strlen): Use new mode iterators.
11168 * combine.c (simplify_comparison): Likewise
11169 * config/i386/i386.c (type_natural_mode): Likewise.
11170 * cse.c (cse_insn): Likewise.
11171 * dse.c (find_shift_sequence): Likewise.
11172 * emit-rtl.c (init_derived_machine_modes): Likewise.
11173 (init_emit_once): Likewise.
11174 * explow.c (hard_function_value): Likewise.
11175 * expmed.c (extract_fixed_bit_field_1): Likewise.
11176 (extract_bit_field_1): Likewise.
11177 (expand_divmod): Likewise.
11178 (emit_store_flag_1): Likewise.
11179 * expr.c (init_expr_target): Likewise.
11180 (convert_move): Likewise.
11181 (alignment_for_piecewise_move): Likewise.
11182 (widest_int_mode_for_size): Likewise.
11183 (emit_block_move_via_movmem): Likewise.
11184 (copy_blkmode_to_reg): Likewise.
11185 (set_storage_via_setmem): Likewise.
11186 (compress_float_constant): Likewise.
11187 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11188 * optabs-query.c (get_best_extraction_insn): Likewise.
11189 * optabs.c (expand_binop): Likewise.
11190 (expand_twoval_unop): Likewise.
11191 (expand_twoval_binop): Likewise.
11192 (widen_leading): Likewise.
11193 (widen_bswap): Likewise.
11194 (expand_parity): Likewise.
11195 (expand_unop): Likewise.
11196 (prepare_cmp_insn): Likewise.
11197 (prepare_float_lib_cmp): Likewise.
11198 (expand_float): Likewise.
11199 (expand_fix): Likewise.
11200 (expand_sfix_optab): Likewise.
11201 * postreload.c (move2add_use_add2_insn): Likewise.
11202 * reg-stack.c (reg_to_stack): Likewise.
11203 * reginfo.c (choose_hard_reg_mode): Likewise.
11204 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11205 * stor-layout.c (mode_for_size): Likewise.
11206 (smallest_mode_for_size): Likewise.
11207 (mode_for_vector): Likewise.
11208 (finish_bitfield_representative): Likewise.
11209 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
11210 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
11211 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11212 * var-tracking.c (prepare_call_arguments): Likewise.
11214 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11215 Alan Hayward <alan.hayward@arm.com>
11216 David Sherwood <david.sherwood@arm.com>
11218 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
11219 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
11220 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
11221 * machmode.h (mode_size): Move earlier in file.
11222 (mode_precision): Likewise.
11223 (mode_inner): Likewise.
11224 (mode_nunits): Likewise.
11225 (mode_unit_size): Likewise.
11226 (unit_unit_precision): Likewise.
11227 (mode_wider): Likewise.
11228 (mode_2xwider): Likewise.
11229 (machine_mode): New class.
11230 (mode_to_bytes): New function.
11231 (mode_to_bits): Likewise.
11232 (mode_to_precision): Likewise.
11233 (mode_to_inner): Likewise.
11234 (mode_to_unit_size): Likewise.
11235 (mode_to_unit_precision): Likewise.
11236 (mode_to_nunits): Likewise.
11237 (GET_MODE_SIZE): Use mode_to_bytes.
11238 (GET_MODE_BITSIZE): Use mode_to_bits.
11239 (GET_MODE_PRECISION): Use mode_to_precision.
11240 (GET_MODE_INNER): Use mode_to_inner.
11241 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
11242 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
11243 (GET_MODE_NUNITS): Use mode_to_nunits.
11244 * system.h (ALWAYS_INLINE): New macro.
11245 * config/powerpcspe/powerpcspe-c.c
11246 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
11247 int for arg1_mode and arg2_mode.
11249 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11250 Alan Hayward <alan.hayward@arm.com>
11251 David Sherwood <david.sherwood@arm.com>
11253 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
11254 Prefix mode names with E_ in case statements.
11255 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11256 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
11257 (aarch64_split_simd_move): Likewise.
11258 (aarch64_gen_storewb_pair): Likewise.
11259 (aarch64_gen_loadwb_pair): Likewise.
11260 (aarch64_gen_store_pair): Likewise.
11261 (aarch64_gen_load_pair): Likewise.
11262 (aarch64_get_condition_code_1): Likewise.
11263 (aarch64_constant_pool_reload_icode): Likewise.
11264 (get_rsqrte_type): Likewise.
11265 (get_rsqrts_type): Likewise.
11266 (get_recpe_type): Likewise.
11267 (get_recps_type): Likewise.
11268 (aarch64_gimplify_va_arg_expr): Likewise.
11269 (aarch64_simd_container_mode): Likewise.
11270 (aarch64_emit_load_exclusive): Likewise.
11271 (aarch64_emit_store_exclusive): Likewise.
11272 (aarch64_expand_compare_and_swap): Likewise.
11273 (aarch64_gen_atomic_cas): Likewise.
11274 (aarch64_emit_bic): Likewise.
11275 (aarch64_emit_atomic_swap): Likewise.
11276 (aarch64_emit_atomic_load_op): Likewise.
11277 (aarch64_evpc_trn): Likewise.
11278 (aarch64_evpc_uzp): Likewise.
11279 (aarch64_evpc_zip): Likewise.
11280 (aarch64_evpc_ext): Likewise.
11281 (aarch64_evpc_rev): Likewise.
11282 (aarch64_evpc_dup): Likewise.
11283 (aarch64_gen_ccmp_first): Likewise.
11284 (aarch64_gen_ccmp_next): Likewise.
11285 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11286 (alpha_emit_xfloating_libcall): Likewise.
11287 (emit_insxl): Likewise.
11288 (alpha_arg_type): Likewise.
11289 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
11290 (arc_preferred_simd_mode): Likewise.
11291 (arc_secondary_reload): Likewise.
11292 (get_arc_condition_code): Likewise.
11293 (arc_print_operand): Likewise.
11294 (arc_legitimate_constant_p): Likewise.
11295 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11296 * config/arc/arc.md (casesi_load): Likewise.
11297 (casesi_compact_jump): Likewise.
11298 * config/arc/predicates.md (proper_comparison_operator): Likewise.
11299 (cc_use_register): Likewise.
11300 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11301 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
11302 (arm_init_iwmmxt_builtins): Likewise.
11303 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
11304 (neon_expand_vector_init): Likewise.
11305 (arm_attr_length_move_neon): Likewise.
11306 (maybe_get_arm_condition_code): Likewise.
11307 (arm_emit_vector_const): Likewise.
11308 (arm_preferred_simd_mode): Likewise.
11309 (arm_output_iwmmxt_tinsr): Likewise.
11310 (thumb1_output_casesi): Likewise.
11311 (thumb2_output_casesi): Likewise.
11312 (arm_emit_load_exclusive): Likewise.
11313 (arm_emit_store_exclusive): Likewise.
11314 (arm_expand_compare_and_swap): Likewise.
11315 (arm_evpc_neon_vuzp): Likewise.
11316 (arm_evpc_neon_vzip): Likewise.
11317 (arm_evpc_neon_vrev): Likewise.
11318 (arm_evpc_neon_vtrn): Likewise.
11319 (arm_evpc_neon_vext): Likewise.
11320 (arm_validize_comparison): Likewise.
11321 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
11322 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
11323 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
11324 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
11325 (c6x_preferred_simd_mode): Likewise.
11326 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
11327 (epiphany_rtx_costs): Likewise.
11328 * config/epiphany/predicates.md (proper_comparison_operator):
11330 * config/frv/frv.c (condexec_memory_operand): Likewise.
11331 (frv_emit_move): Likewise.
11332 (output_move_single): Likewise.
11333 (output_condmove_single): Likewise.
11334 (frv_hard_regno_mode_ok): Likewise.
11335 (frv_matching_accg_mode): Likewise.
11336 * config/h8300/h8300.c (split_adds_subs): Likewise.
11337 (h8300_rtx_costs): Likewise.
11338 (h8300_print_operand): Likewise.
11339 (compute_mov_length): Likewise.
11340 (output_logical_op): Likewise.
11341 (compute_logical_op_length): Likewise.
11342 (compute_logical_op_cc): Likewise.
11343 (h8300_shift_needs_scratch_p): Likewise.
11344 (output_a_shift): Likewise.
11345 (compute_a_shift_length): Likewise.
11346 (compute_a_shift_cc): Likewise.
11347 (expand_a_rotate): Likewise.
11348 (output_a_rotate): Likewise.
11349 * config/i386/i386.c (classify_argument): Likewise.
11350 (function_arg_advance_32): Likewise.
11351 (function_arg_32): Likewise.
11352 (function_arg_64): Likewise.
11353 (function_value_64): Likewise.
11354 (ix86_gimplify_va_arg): Likewise.
11355 (ix86_legitimate_constant_p): Likewise.
11356 (put_condition_code): Likewise.
11357 (split_double_mode): Likewise.
11358 (ix86_avx256_split_vector_move_misalign): Likewise.
11359 (ix86_expand_vector_logical_operator): Likewise.
11360 (ix86_split_idivmod): Likewise.
11361 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
11362 (ix86_build_const_vector): Likewise.
11363 (ix86_build_signbit_mask): Likewise.
11364 (ix86_match_ccmode): Likewise.
11365 (ix86_cc_modes_compatible): Likewise.
11366 (ix86_expand_branch): Likewise.
11367 (ix86_expand_sse_cmp): Likewise.
11368 (ix86_expand_sse_movcc): Likewise.
11369 (ix86_expand_int_sse_cmp): Likewise.
11370 (ix86_expand_vec_perm_vpermi2): Likewise.
11371 (ix86_expand_vec_perm): Likewise.
11372 (ix86_expand_sse_unpack): Likewise.
11373 (ix86_expand_int_addcc): Likewise.
11374 (ix86_split_to_parts): Likewise.
11375 (ix86_vectorize_builtin_gather): Likewise.
11376 (ix86_vectorize_builtin_scatter): Likewise.
11377 (avx_vpermilp_parallel): Likewise.
11378 (inline_memory_move_cost): Likewise.
11379 (ix86_tieable_integer_mode_p): Likewise.
11380 (x86_maybe_negate_const_int): Likewise.
11381 (ix86_expand_vector_init_duplicate): Likewise.
11382 (ix86_expand_vector_init_one_nonzero): Likewise.
11383 (ix86_expand_vector_init_one_var): Likewise.
11384 (ix86_expand_vector_init_concat): Likewise.
11385 (ix86_expand_vector_init_interleave): Likewise.
11386 (ix86_expand_vector_init_general): Likewise.
11387 (ix86_expand_vector_set): Likewise.
11388 (ix86_expand_vector_extract): Likewise.
11389 (emit_reduc_half): Likewise.
11390 (ix86_emit_i387_round): Likewise.
11391 (ix86_mangle_type): Likewise.
11392 (ix86_expand_round_sse4): Likewise.
11393 (expand_vec_perm_blend): Likewise.
11394 (canonicalize_vector_int_perm): Likewise.
11395 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
11396 (expand_vec_perm_1): Likewise.
11397 (expand_vec_perm_interleave3): Likewise.
11398 (expand_vec_perm_even_odd_pack): Likewise.
11399 (expand_vec_perm_even_odd_1): Likewise.
11400 (expand_vec_perm_broadcast_1): Likewise.
11401 (ix86_vectorize_vec_perm_const_ok): Likewise.
11402 (ix86_expand_vecop_qihi): Likewise.
11403 (ix86_expand_mul_widen_hilo): Likewise.
11404 (ix86_expand_sse2_abs): Likewise.
11405 (ix86_expand_pextr): Likewise.
11406 (ix86_expand_pinsr): Likewise.
11407 (ix86_preferred_simd_mode): Likewise.
11408 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
11409 * config/i386/sse.md (*andnot<mode>3): Likewise.
11410 (<mask_codefor><code><mode>3<mask_name>): Likewise.
11411 (*<code><mode>3): Likewise.
11412 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
11413 (ia64_expand_atomic_op): Likewise.
11414 (ia64_arg_type): Likewise.
11415 (ia64_mode_to_int): Likewise.
11416 (ia64_scalar_mode_supported_p): Likewise.
11417 (ia64_vector_mode_supported_p): Likewise.
11418 (expand_vec_perm_broadcast): Likewise.
11419 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
11420 (iq2000_function_arg_advance): Likewise.
11421 (iq2000_function_arg): Likewise.
11422 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
11423 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
11424 (m68k_libcall_value): Likewise.
11425 (m68k_function_value): Likewise.
11426 (sched_attr_op_type): Likewise.
11427 * config/mcore/mcore.c (mcore_output_move): Likewise.
11428 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
11430 (microblaze_function_arg): Likewise.
11431 * config/mips/mips.c (mips16_build_call_stub): Likewise.
11432 (mips_print_operand): Likewise.
11433 (mips_mode_ok_for_mov_fmt_p): Likewise.
11434 (mips_vector_mode_supported_p): Likewise.
11435 (mips_preferred_simd_mode): Likewise.
11436 (mips_expand_vpc_loongson_even_odd): Likewise.
11437 (mips_expand_vec_unpack): Likewise.
11438 (mips_expand_vi_broadcast): Likewise.
11439 (mips_expand_vector_init): Likewise.
11440 (mips_expand_vec_reduc): Likewise.
11441 (mips_expand_msa_cmp): Likewise.
11442 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
11443 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
11444 (cc_flags_for_mode): Likewise.
11445 * config/msp430/msp430.c (msp430_print_operand): Likewise.
11446 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
11447 (nds32_output_casesi_pc_relative): Likewise.
11448 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11449 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
11450 (nvptx_gen_unpack): Likewise.
11451 (nvptx_gen_pack): Likewise.
11452 (nvptx_gen_shuffle): Likewise.
11453 (nvptx_gen_wcast): Likewise.
11454 (nvptx_preferred_simd_mode): Likewise.
11455 * config/pa/pa.c (pa_secondary_reload): Likewise.
11456 * config/pa/predicates.md (base14_operand): Likewise.
11457 * config/powerpcspe/powerpcspe-c.c
11458 (altivec_resolve_overloaded_builtin): Likewise.
11459 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
11461 (rs6000_preferred_simd_mode): Likewise.
11462 (output_vec_const_move): Likewise.
11463 (rs6000_expand_vector_extract): Likewise.
11464 (rs6000_split_vec_extract_var): Likewise.
11465 (reg_offset_addressing_ok_p): Likewise.
11466 (rs6000_legitimate_offset_address_p): Likewise.
11467 (rs6000_legitimize_address): Likewise.
11468 (rs6000_emit_set_const): Likewise.
11469 (rs6000_const_vec): Likewise.
11470 (rs6000_emit_move): Likewise.
11471 (spe_build_register_parallel): Likewise.
11472 (rs6000_darwin64_record_arg_recurse): Likewise.
11473 (swap_selector_for_mode): Likewise.
11474 (spe_init_builtins): Likewise.
11475 (paired_init_builtins): Likewise.
11476 (altivec_init_builtins): Likewise.
11477 (do_load_for_compare): Likewise.
11478 (rs6000_generate_compare): Likewise.
11479 (rs6000_expand_float128_convert): Likewise.
11480 (emit_load_locked): Likewise.
11481 (emit_store_conditional): Likewise.
11482 (rs6000_output_function_epilogue): Likewise.
11483 (rs6000_handle_altivec_attribute): Likewise.
11484 (rs6000_function_value): Likewise.
11485 (emit_fusion_gpr_load): Likewise.
11486 (emit_fusion_p9_load): Likewise.
11487 (emit_fusion_p9_store): Likewise.
11488 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
11489 (fusion_gpr_mem_load): Likewise.
11490 (fusion_addis_mem_combo_load): Likewise.
11491 (fusion_addis_mem_combo_store): Likewise.
11492 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
11493 (fusion_gpr_mem_load): Likewise.
11494 (fusion_addis_mem_combo_load): Likewise.
11495 (fusion_addis_mem_combo_store): Likewise.
11496 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11498 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
11499 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
11500 (rs6000_preferred_simd_mode): Likewise.
11501 (output_vec_const_move): Likewise.
11502 (rs6000_expand_vector_extract): Likewise.
11503 (rs6000_split_vec_extract_var): Likewise.
11504 (reg_offset_addressing_ok_p): Likewise.
11505 (rs6000_legitimate_offset_address_p): Likewise.
11506 (rs6000_legitimize_address): Likewise.
11507 (rs6000_emit_set_const): Likewise.
11508 (rs6000_const_vec): Likewise.
11509 (rs6000_emit_move): Likewise.
11510 (rs6000_darwin64_record_arg_recurse): Likewise.
11511 (swap_selector_for_mode): Likewise.
11512 (paired_init_builtins): Likewise.
11513 (altivec_init_builtins): Likewise.
11514 (rs6000_expand_float128_convert): Likewise.
11515 (emit_load_locked): Likewise.
11516 (emit_store_conditional): Likewise.
11517 (rs6000_output_function_epilogue): Likewise.
11518 (rs6000_handle_altivec_attribute): Likewise.
11519 (rs6000_function_value): Likewise.
11520 (emit_fusion_gpr_load): Likewise.
11521 (emit_fusion_p9_load): Likewise.
11522 (emit_fusion_p9_store): Likewise.
11523 * config/rx/rx.c (rx_gen_move_template): Likewise.
11524 (flags_from_mode): Likewise.
11525 * config/s390/predicates.md (s390_alc_comparison): Likewise.
11526 (s390_slb_comparison): Likewise.
11527 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
11528 (s390_vector_mode_supported_p): Likewise.
11529 (s390_cc_modes_compatible): Likewise.
11530 (s390_match_ccmode_set): Likewise.
11531 (s390_canonicalize_comparison): Likewise.
11532 (s390_emit_compare_and_swap): Likewise.
11533 (s390_branch_condition_mask): Likewise.
11534 (s390_rtx_costs): Likewise.
11535 (s390_secondary_reload): Likewise.
11536 (__SECONDARY_RELOAD_CASE): Likewise.
11537 (s390_expand_cs): Likewise.
11538 (s390_preferred_simd_mode): Likewise.
11539 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
11540 * config/sh/sh.c (sh_print_operand): Likewise.
11541 (dump_table): Likewise.
11542 (sh_secondary_reload): Likewise.
11543 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11544 * config/sh/sh.md (casesi_worker_1): Likewise.
11545 (casesi_worker_2): Likewise.
11546 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
11547 (fcc_comparison_operator): Likewise.
11548 * config/sparc/sparc.c (sparc_expand_move): Likewise.
11549 (emit_soft_tfmode_cvt): Likewise.
11550 (sparc_preferred_simd_mode): Likewise.
11551 (output_cbranch): Likewise.
11552 (sparc_print_operand): Likewise.
11553 (sparc_expand_vec_perm_bmask): Likewise.
11554 (vector_init_bshuffle): Likewise.
11555 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11556 (spu_vector_mode_supported_p): Likewise.
11557 (spu_expand_insv): Likewise.
11558 (spu_emit_branch_or_set): Likewise.
11559 (spu_handle_vector_attribute): Likewise.
11560 (spu_builtin_splats): Likewise.
11561 (spu_builtin_extract): Likewise.
11562 (spu_builtin_promote): Likewise.
11563 (spu_expand_sign_extend): Likewise.
11564 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11565 (tilegx_simd_int): Likewise.
11566 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
11567 (tilepro_simd_int): Likewise.
11568 * config/v850/v850.c (const_double_split): Likewise.
11569 (v850_print_operand): Likewise.
11570 (ep_memory_offset): Likewise.
11571 * config/vax/vax.c (vax_rtx_costs): Likewise.
11572 (vax_output_int_move): Likewise.
11573 (vax_output_int_add): Likewise.
11574 (vax_output_int_subtract): Likewise.
11575 * config/visium/predicates.md (visium_branch_operator): Likewise.
11576 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
11577 (visium_print_operand_address): Likewise.
11578 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11579 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
11580 (xtensa_expand_conditional_branch): Likewise.
11581 (xtensa_copy_incoming_a7): Likewise.
11582 (xtensa_output_literal): Likewise.
11583 * dfp.c (decimal_real_maxval): Likewise.
11584 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
11586 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11587 Alan Hayward <alan.hayward@arm.com>
11588 David Sherwood <david.sherwood@arm.com>
11590 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
11591 (mode_nunits_inline): Likewise.
11592 (mode_inner_inline): Likewise.
11593 (mode_unit_size_inline): Likewise.
11594 (mode_unit_precision_inline): Likewise.
11595 (emit_insn_modes_h): Likewise. Also emit a #define of the
11597 (emit_mode_wider): Add an E_ prefix to mode names.
11598 (emit_mode_complex): Likewise.
11599 (emit_mode_inner): Likewise.
11600 (emit_mode_adjustments): Likewise.
11601 (emit_mode_int_n): Likewise.
11602 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
11603 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
11604 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
11605 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
11606 (CRC32_BUILTIN, ENTRY): Likewise.
11607 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
11608 (aarch64_pop_regs): Likewise.
11609 (aarch64_process_components): Likewise.
11610 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
11611 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
11612 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
11613 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
11614 * config/arm/arm.c (arm_init_libfuncs): Likewise.
11615 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
11617 * config/i386/i386-builtin.def (pcmpestr): Likewise.
11618 (pcmpistr): Likewise.
11619 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
11620 * config/mmix/mmix.c (mmix_output_condition): Likewise.
11621 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
11623 * config/rl78/rl78.c (mduc_regs): Likewise.
11624 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
11625 (htm_expand_builtin): Likewise.
11626 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
11627 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
11628 * config/xtensa/xtensa.c (print_operand): Likewise.
11629 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
11630 (NUM_MODE_VECTOR_INT): Likewise.
11631 * genoutput.c (null_operand): Likewise.
11632 (output_operand_data): Likewise.
11633 * genrecog.c (print_parameter_value): Likewise.
11634 * lra.c (debug_operand_data): Likewise.
11636 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11638 * dwarf2out.c (scompare_loc_descriptor_wide)
11639 (scompare_loc_descriptor_narrow): New functions, split out from...
11640 (scompare_loc_descriptor): ...here.
11641 * expmed.c (emit_store_flag_int): New function, split out from...
11642 (emit_store_flag): ...here.
11644 2017-08-30 Richard Biener <rguenther@suse.de>
11646 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
11647 (dwarf2out_early_finish): Move setting of AT_pubnames from
11648 early debug output to early finish.
11650 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11652 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
11653 and -mdata-region to the assembler.
11655 2017-08-30 Richard Biener <rguenther@suse.de>
11657 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
11659 (gen_subprogram_die): Add DW_AT_object_pointer only early.
11660 (dwarf2out_early_global_decl): Only generate a DIE for the
11661 abstract origin if it doesn't already exist or is a declaration DIE.
11662 (resolve_addr): Do not add the linkage name twice when
11663 generating a stub DIE for the DW_TAG_GNU_call_site target.
11665 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11667 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11668 Use machine_mode rather than int for arg1_mode.
11670 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
11673 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
11674 that the second argument of the built-in functions to unpack
11675 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
11676 switch statement instead a lot of if statements.
11677 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
11678 Allow 64-bit values to be in Altivec registers as well as
11679 traditional floating point registers.
11680 (pack<mode>, FMOVE128_VSX iterator): Likewise.
11682 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
11684 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
11685 MAX_REGS_PER_ADDRESS == 1.
11687 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
11689 * config/i386/i386.opt (flag_fentry): Do not init to -1.
11690 * config/i386/i386.c (ix86_option_override_internal): Simplify
11691 setting of opts->x_flag_entry.
11693 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11694 Jakub Jelinek <jakub@redhat.com>
11695 Richard Biener <rguenther@suse.de>
11697 PR tree-optimization/81503
11698 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
11699 folded constant fits in the target type; reorder tests for clarity.
11701 2017-08-29 Martin Liska <mliska@suse.cz>
11703 * passes.def: Include pass_lower_switch.
11704 * stmt.c (dump_case_nodes): Remove and move to
11705 tree-switch-conversion.
11706 (case_values_threshold): Likewise.
11707 (expand_switch_as_decision_tree_p): Likewise.
11708 (emit_case_decision_tree): Likewise.
11709 (expand_case): Likewise.
11710 (balance_case_nodes): Likewise.
11711 (node_has_low_bound): Likewise.
11712 (node_has_high_bound): Likewise.
11713 (node_is_bounded): Likewise.
11714 (emit_case_nodes): Likewise.
11715 (struct simple_case_node): New struct.
11716 (add_case_node): Remove.
11717 (emit_case_dispatch_table): Use vector instead of case_list.
11718 (reset_out_edges_aux): Remove.
11719 (compute_cases_per_edge): Likewise.
11720 (expand_case): Build list of simple_case_node.
11721 (expand_sjlj_dispatch_table): Use it.
11722 * tree-switch-conversion.c (struct case_node): Moved from
11723 stmt.c and adjusted.
11724 (emit_case_nodes): Likewise.
11725 (node_has_low_bound): Likewise.
11726 (node_has_high_bound): Likewise.
11727 (node_is_bounded): Likewise.
11728 (case_values_threshold): Likewise.
11729 (reset_out_edges_aux): Likewise.
11730 (compute_cases_per_edge): Likewise.
11731 (add_case_node): Likewise.
11732 (dump_case_nodes): Likewise.
11733 (balance_case_nodes): Likewise.
11734 (expand_switch_as_decision_tree_p): Likewise.
11735 (emit_jump): Likewise.
11736 (emit_case_decision_tree): Likewise.
11737 (try_switch_expansion): Likewise.
11738 (do_jump_if_equal): Likewise.
11739 (emit_cmp_and_jump_insns): Likewise.
11740 (fix_phi_operands_for_edge): New function.
11741 (record_phi_operand_mapping): Likewise.
11742 (class pass_lower_switch): New pass.
11743 (pass_lower_switch::execute): New function.
11744 (make_pass_lower_switch): Likewise.
11745 (conditional_probability):
11746 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
11747 * tree-pass.h: Add make_pass_lower_switch.
11749 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11752 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
11755 2017-08-29 Richard Biener <rguenther@suse.de>
11757 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
11758 we do not add a DW_AT_inline attribute twice.
11759 (gen_subprogram_die): Remove code setting DW_AT_inline on
11760 DECL_ABSTRACT_P nodes.
11762 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
11764 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
11765 calls to internal functions.
11766 (gimplify_modify_expr): Likewise.
11767 * tree-call-cdce.c (use_internal_fn): Likewise.
11768 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
11769 (convert_to_divmod): Set the nothrow flag.
11770 * tree-if-conv.c (predicate_mem_writes): Likewise.
11771 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
11772 (vectorizable_call): Likewise.
11773 (vectorizable_store): Likewise.
11774 (vectorizable_load): Likewise.
11775 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
11776 (vect_recog_mask_conversion_pattern): Likewise.
11778 2017-08-29 Martin Liska <mliska@suse.cz>
11781 * gcc.c (driver_handle_option): Add new argument.
11782 * opts-common.c (handle_option): Pass
11783 target_option_override_hook.
11784 * opts-global.c (lang_handle_option): Add new option.
11785 (set_default_handlers): Add new argument.
11786 (decode_options): Likewise.
11787 * opts.c (target_handle_option): Likewise.
11788 (common_handle_option): Call target_option_override_hook.
11789 * opts.h (struct cl_option_handler_func): Add hook for
11790 target option override.
11791 (struct cl_option_handlers): Likewise.
11792 (set_default_handlers): Add new argument.
11793 (decode_options): Likewise.
11794 (common_handle_option): Likewise.
11795 (target_handle_option): Likewise.
11796 * toplev.c (toplev::main): Pass targetm.target_option.override
11799 2017-08-29 Richard Biener <rguenther@suse.de>
11800 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
11802 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
11803 life to the active subtree.
11805 2017-08-28 Jeff Law <law@redhat.com>
11807 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
11808 derive_equivalences.
11809 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
11810 Code moved into....
11811 (edge_info::derive_equivalences): New private member function
11813 * tree-ssa-dom.c (class edge_info): Changed from a struct
11814 to a class. Add ctor/dtor, methods and data members.
11815 (edge_info::edge_info): Renamed from allocate_edge_info.
11816 Initialize additional members.
11817 (edge_info::~edge_info): New.
11818 (free_dom_edge_info): Delete the edge info.
11819 (record_edge_info): Use new class & associated member functions.
11820 Tighten forms for testing for edge equivalences.
11821 (record_temporary_equivalences): Iterate over the simple
11822 equivalences rather than assuming there's only one per edge.
11823 (cprop_into_successor_phis): Iterate over the simple
11824 equivalences rather than assuming there's only one per edge.
11825 (optimize_stmt): Use operand_equal_p rather than pointer
11826 equality for mini-DSE code.
11828 2017-08-28 Nathan Sidwell <nathan@acm.org>
11830 * gcc.c (execute): Fold SIGPIPE handling into switch
11831 statement. Adjust internal error message.
11833 2017-08-28 Richard Biener <rguenther@suse.de>
11836 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
11837 Do nothing for removed DIEs.
11839 2017-08-28 Richard Biener <rguenther@suse.de>
11841 PR tree-optimization/81977
11842 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
11845 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
11848 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
11849 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
11850 using targetm.gen_mem_thread_fence.
11852 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
11855 * config/i386/i386.md (*<btsc><mode>): Change operand 2
11856 predicate to register_operand. Reorder operands.
11857 (*btr<mode>): Ditto.
11858 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
11859 (*btr<mode>_mask): Ditto.
11861 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
11863 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
11864 * config/rs6000/xmmintrin.h: New file.
11865 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
11867 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11870 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
11871 parameter and_insn and return it.
11872 (recombine_lvx_pattern): Insert a copy to ensure availability of
11873 the base register of the copied masking operation at the point of
11874 the instruction replacement.
11875 (recombine_stvx_pattern): Likewise.
11877 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
11879 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
11880 undocumented switches.
11881 (-mpower9-dform-vector): Likewise.
11882 (-mpower9-dform): Likewise.
11883 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
11884 comments to delete references to -mpower9-dform* switches.
11885 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
11886 Delete reference to -mpower9-dform* switches, test for
11887 -mpower9-vector instead.
11888 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
11889 (OTHER_P9_VECTOR_MASKS): Likewise.
11890 (POWERPC_MASKS): Likewise.
11891 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
11892 tests against -mpower9-dform* to -mpower9-vector. Delete code
11893 that checked for -mpower9-dform* consistancy with other options.
11894 Add test for -mpower9-misc to enable other power9 switches.
11895 (rs6000_init_hard_regno_mode_ok): Likewise.
11896 (rs6000_option_override_internal): Likewise.
11897 (rs6000_emit_prologue): Likewise.
11898 (rs6000_emit_epilogue): Likewise.
11899 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
11900 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
11901 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
11903 (emit_fusion_p9_store): Likewise.
11904 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
11905 resetting these macros if the assembler does not support ISA 3.0
11907 (TARGET_P9_DFORM_VECTOR): Likewise.
11908 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
11909 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
11911 2017-08-25 Alan Modra <amodra@gmail.com>
11913 PR rtl-optimization/81747
11914 * cse.c (cse_extended_basic_block): Don't attempt to record
11915 equivalences for degenerate conditional branches that branch
11916 to their fall-through.
11918 2017-08-24 Martin Sebor <msebor@redhat.com>
11920 PR middle-end/81908
11921 * gimple-fold.c (size_must_be_zero_p): New function.
11922 (gimple_fold_builtin_memory_op): Call it.
11924 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
11926 * config/rs6000/mm_malloc.h: New file.
11928 2017-08-24 Bin Cheng <bin.cheng@arm.com>
11930 PR tree-optimization/81913
11931 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
11932 analysis when either IVs in condition can wrap.
11934 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
11936 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
11937 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
11939 2017-08-24 Richard Biener <rguenther@suse.de>
11942 * targhooks.c (default_target_can_inline_p): Properly
11943 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
11944 is present and always compare.
11945 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
11946 infer -mfpmath=sse from TARGET_SSE_P.
11947 (ix86_can_inline_p): Properly use target_option_default_node when
11948 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
11950 2017-08-24 Richard Biener <rguenther@suse.de>
11953 * dwarf2out.c (output_die): Handle flag_generate_offload like
11955 (output_comp_unit): Likewise.
11956 (gen_array_type_die): Likewise.
11957 (dwarf2out_early_finish): Likewise.
11958 (note_variable_value_in_expr): Likewise.
11959 (dwarf2out_finish): Likewise. Adjust assert.
11960 * cgraphunit.c (symbol_table::compile): Move setting of
11961 flag_generate_offload earlier ...
11962 (symbol_table::finalize_compilation_unit): ... here, before
11963 early debug finalization.
11965 2017-08-24 Richard Biener <rguenther@suse.de>
11967 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
11968 and ipa-fnsummary.h.
11969 (ix86_can_inline_p): When ix86_fpmath flags do not match
11970 check whether the callee uses FP math at all.
11972 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
11974 PR middle-end/81931
11975 * tree-ssanames.c (get_nonzero_bits): Use element_precision
11976 instead of TYPE_PRECISION.
11978 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
11979 Alan Hayward <alan.hayward@arm.com>
11980 David Sherwood <david.sherwood@arm.com>
11982 * combine.c (make_extraction): Use subreg_offset_from_lsb.
11984 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
11986 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
11988 (ix86_frame::stack_realign_allocate): New field.
11989 (struct machine_frame_state): Modify comments.
11990 (machine_frame_state::sp_realigned_fp_end): New field.
11991 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
11992 layout calculation.
11993 (sp_valid_at): Add assertion to assure no attempt to access invalid
11994 offset of a realigned stack.
11995 (fp_valid_at): Likewise.
11996 (choose_baseaddr): Modify comments.
11997 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
11998 ix86_expand_prologue.
11999 (ix86_expand_prologue): Modify stack realignment and allocation.
12000 (ix86_expand_epilogue): Modify comments.
12001 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
12002 avx2_runtime, avx512f, and avx512f_runtime.
12004 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
12006 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
12007 (mstackrealign): Do not init to -1.
12008 * config/i386/i386.c (ix86_option_override_internal):
12009 Check opts_set, not opts when setting default value of
12010 opts->x_ix86_force_align_arg_pointer.
12012 2017-08-23 Richard Biener <rguenther@suse.de>
12014 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
12015 lang_hooks.decl_printable_name.
12016 * print-rtl-function.c (print_rtx_function): Likewise.
12017 * tree-pretty-print.c (dump_function_header): Likewise.
12019 2017-08-23 Richard Biener <rguenther@suse.de>
12022 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
12023 -g0 at compile-time.
12025 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12027 PR middle-end/19706
12028 * doc/sourcebuild.texi (Other hardware attributes):
12031 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12033 PR middle-end/19706
12034 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
12035 Fix single-use check.
12037 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
12039 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
12041 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12043 * doc/install.texi: Modify to add more details on running selected
12046 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12048 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
12049 is combined with -mabi=ms.
12050 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
12053 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12055 PR tree-optimization/81488
12056 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
12057 and cached_basis fields.
12058 (MAX_SPREAD): New constant.
12059 (alloc_cand_and_find_basis): Initialize new fields.
12060 (clear_visited): New function.
12061 (create_phi_basis_1): Rename from create_phi_basis, set visited
12062 and cached_basis fields.
12063 (create_phi_basis): New wrapper function.
12064 (phi_add_costs_1): Rename from phi_add_costs, add spread
12065 parameter, set visited field, short-circuit when limits reached.
12066 (phi_add_costs): New wrapper function.
12067 (record_phi_increments_1): Rename from record_phi_increments, set
12069 (record_phi_increments): New wrapper function.
12070 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
12071 (phi_incr_cost): New wrapper function.
12072 (all_phi_incrs_profitable_1): Rename from
12073 all_phi_incrs_profitable, set visited field.
12074 (all_phi_incrs_profitable): New wrapper function.
12076 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12077 Alan Hayward <alan.hayward@arm.com>
12078 David Sherwood <david.sherwood@arm.com>
12080 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
12081 that takes the outer and inner modes.
12082 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
12083 comparison as the canonical test for a paradoxical subreg.
12084 * combine.c (simplify_set): Use paradoxical_subreg_p.
12085 (make_extraction): Likewise.
12086 (force_to_mode): Likewise.
12087 (rtx_equal_for_field_assignment_p): Likewise.
12088 (gen_lowpart_for_combine): Likewise.
12089 (simplify_comparison): Likewise.
12090 * cse.c (equiv_constant): Likewise.
12091 * expmed.c (store_bit_field_1): Likewise.
12092 * final.c (alter_subreg): Likewise.
12093 * fwprop.c (propagate_rtx): Likewise.
12094 (forward_propagate_subreg): Likewise.
12095 * ira-conflicts.c (ira_build_conflicts): Likewise.
12096 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
12097 * lra-constraints.c (curr_insn_transform): Likewise.
12098 (split_reg): Likewise.
12099 * lra-eliminations.c (move_plus_up): Likewise.
12100 (lra_eliminate_regs_1): Likewise.
12101 * recog.c (general_operand): Likewise.
12102 * ree.c (combine_reaching_defs): Likewise.
12103 * reload.c (push_reload): Likewise.
12104 (find_reloads): Likewise.
12105 * reload1.c (elimination_effects): Likewise.
12106 (compute_reload_subreg_offset): Likewise.
12107 (choose_reload_regs): Likewise.
12108 * rtlanal.c (subreg_lsb_1): Likewise.
12109 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12110 (simplify_subreg): Likewise.
12111 * var-tracking.c (track_loc_p): Likewise.
12112 * emit-rtl.c (byte_lowpart_offset): Likewise.
12113 (paradoxical_subreg_p): Delete out-of-line definition.
12115 2017-08-22 Jeff Law <law@redhat.com>
12117 PR tree-optimization/81741
12118 PR tree-optimization/71947
12119 * tree-ssa-dom.c: Include tree-inline.h.
12120 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
12121 equivalences if one is more expensive to compute than the other.
12122 * tree-ssa-scopedtables.h (class const_or_copies): Make
12123 record_const_or_copy_raw method private.
12124 (class avail_exprs_stack): New method simplify_binary_operation.
12125 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
12126 avail_exprs_stack::simplify_binary_operation as needed.
12127 (avail_exprs_stack::simplify_binary_operation): New function.
12129 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
12131 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
12132 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
12133 (DOT_SYMBOLS): Likewise.
12134 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
12135 (RELOCATABLE_NEEDS_FIXUP): Likewise.
12136 (RS6000_ABI_NAME): Likewise.
12137 (TARGET_CMODEL): Likewise.
12138 (TOC_SECTION_ASM_OP): Likewise.
12139 (SET_CMODEL): New macro.
12140 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
12142 2017-08-22 Richard Biener <rguenther@suse.de>
12144 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
12145 to allow for free-lang-data replacements similar to verify_type_variant.
12147 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12148 Alan Hayward <alan.hayward@arm.com>
12149 David Sherwood <david.sherwood@arm.com>
12151 * config/aarch64/aarch64.md (casesi): Use DImode rather than
12152 VOIDmode for the LABEL_REF.
12154 2017-08-22 Richard Biener <rguenther@suse.de>
12156 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
12158 2017-08-22 Richard Biener <rguenther@suse.de>
12160 * common.opt (feliminate-dwarf2-dups): Ignore.
12161 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
12162 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
12163 same_die_p_wrap, compute_section_prefix,
12164 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
12165 (comdat_symbol_id, comdat_symbol_number): Likewise.
12166 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
12168 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
12169 (output_die): Mark unreachable path unreachable.
12170 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
12171 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
12172 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
12173 (dwarf2out_early_finish): Likewise.
12175 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
12177 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
12179 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
12182 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
12183 not VAR_P. Filter attribute warnings with OPT_Wattributes.
12184 (avr_attribute_table) <io, io_low, address>: Initialize
12185 .decl_required with true.
12187 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
12189 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
12190 undocumented debugging options.
12191 (-mvsx-scalar-double): Likewise.
12192 (-mallow-df-permute): Likewise.
12193 (-mvectorize-builtins): Likewise.
12194 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12195 (rs6000_builtin_vectorized_function): Likewise.
12196 (rs6000_builtin_md_vectorized_function): Likewise.
12197 (rs6000_opt_vars): Likewise.
12199 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
12202 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
12203 (*btrq_imm): Rename from *btrq.
12204 (*btcq_imm): Rename from *btcq.
12205 (btsc): New code attribute.
12206 (*<btsc><mode>): New insn pattern.
12207 (*btr<mode>): Ditto.
12208 (*<btsc><mode>_mask): New insn_and_split pattern.
12209 (*btr<mode>_mask): Ditto.
12211 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12212 Alan Hayward <alan.hayward@arm.com>
12213 David Sherwood <david.sherwood@arm.com>
12215 * function.c (pad_below): Simplify padding calculation.
12217 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12218 Alan Hayward <alan.hayward@arm.com>
12219 David Sherwood <david.sherwood@arm.com>
12221 * target.def (function_prologue): Remove frame size argument.
12222 (function_epilogue): Likewise.
12223 * doc/tm.texi: Regenerate.
12224 * final.c (final_start_function): Update call to function_prologue.
12225 (final_end_function): Update call to function_epilogue.
12226 (default_function_pro_epilogue): Remove frame size argument.
12227 * output.h (default_function_pro_epilogue): Likewise.
12228 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
12229 (arm_output_function_prologue): Likewise.
12230 * config/frv/frv.c (frv_function_prologue): Likewise.
12231 (frv_function_epilogue): Likewise.
12232 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
12233 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
12234 (ia64_output_function_epilogue): Likewise.
12235 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
12236 (m32r_output_function_epilogue): Likewise.
12237 * config/microblaze/microblaze.c (microblaze_function_prologue)
12238 (microblaze_function_epilogue): Likewise.
12239 * config/mips/mips.c (mips_output_function_prologue): Likewise.
12240 (mips_output_function_epilogue): Likewise.
12241 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
12242 (mmix_target_asm_function_epilogue): Likewise.
12243 * config/msp430/msp430.c (msp430_start_function): Likewise.
12244 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
12245 (nds32_asm_function_epilogue): Likewise.
12246 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
12247 * config/pa/pa.c (pa_output_function_prologue): Likewise.
12248 (pa_output_function_epilogue): Likewise.
12249 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
12250 (rs6000_output_function_epilogue): Likewise.
12251 * config/rl78/rl78.c (rl78_start_function): Likewise.
12252 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
12253 (rs6000_output_function_epilogue): Likewise.
12254 * config/rx/rx.c (rx_output_function_prologue): Likewise.
12255 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
12256 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
12257 (sparc_asm_function_epilogue): Likewise.
12259 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12261 * tree.h (type_has_mode_precision_p): New function.
12262 * convert.c (convert_to_integer_1): Use it.
12263 * expr.c (expand_expr_real_2): Likewise.
12264 (expand_expr_real_1): Likewise.
12265 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
12266 * match.pd: Likewise.
12267 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
12268 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
12269 * tree-tailcall.c (process_assignment): Likewise.
12270 * tree-vect-loop.c (vectorizable_reduction): Likewise.
12271 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
12272 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
12273 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12274 (vectorizable_assignment): Likewise.
12275 (vectorizable_shift): Likewise.
12276 (vectorizable_operation): Likewise.
12277 * tree-vrp.c (register_edge_assert_for_2): Likewise.
12279 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
12281 * match.pd: Add pow (C, x) simplification.
12283 2017-08-21 Richard Biener <rguenther@suse.de>
12285 PR tree-optimization/81900
12286 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
12287 for blocks with abnormal predecessors.
12288 (compute_antic): Do not set visited flag prematurely.
12290 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
12293 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
12295 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12297 * stor-layout.h (vector_type_mode): Move to...
12298 * tree.h (vector_type_mode): ...here.
12299 * stor-layout.c (vector_type_mode): Move to...
12300 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
12302 2017-08-21 Richard Biener <rguenther@suse.de>
12304 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
12305 register_external_die hooks.
12306 (debug_false_tree_charstarstar_uhwistar): Declare.
12307 (debug_nothing_tree_charstar_uhwi): Likewise.
12308 * debug.c (do_nothing_debug_hooks): Adjust.
12309 (debug_false_tree_charstarstar_uhwistar): New do nothing.
12310 (debug_nothing_tree_charstar_uhwi): Likewise.
12311 * dbxout.c (dbx_debug_hooks): Adjust.
12312 (xcoff_debug_hooks): Likewise.
12313 * sdbout.c (sdb_debug_hooks): Likewise.
12314 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
12315 * dwarf2out.c (macinfo_label_base): New global.
12316 (dwarf2out_register_external_die): New function for the
12317 register_external_die hook.
12318 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
12319 (dwarf2_debug_hooks): Use them.
12320 (dwarf2_lineno_debug_hooks): Adjust.
12321 (struct die_struct): Add with_offset flag.
12322 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
12323 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
12324 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
12325 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
12326 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
12327 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
12328 defining section names for the early LTO debug variants.
12329 (reset_indirect_string): New helper.
12330 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
12331 (print_dw_val): Add support for offsetted symbol references.
12332 (get_ultimate_context): Split out from is_cxx.
12333 (is_cxx): Use get_ultimate_context.
12334 (is_fortran): Add decl overload.
12335 (compute_comp_unit_symbol): Split out worker from
12336 compute_section_prefix.
12337 (compute_section_prefix): Call compute_comp_unit_symbol and
12338 set comdat_type_p here.
12339 (output_die): Skip DIE symbol output for the LTO added one.
12340 Handle DIE symbol references with offset.
12341 (output_comp_unit): Guard section name mangling properly.
12342 For LTO debug sections emit a symbol at the section beginning
12343 which we use to refer to its DIEs.
12344 (add_abstract_origin_attribute): For DIEs registered via
12345 dwarf2out_register_external_die directly refer to the early
12346 DIE rather than indirectly through the shadow one we created.
12347 Remove obsolete call to dwarf2out_abstract_function for
12348 non-function/block origins.
12349 (gen_array_type_die): When generating early LTO debug do
12350 not emit DW_AT_string_length.
12351 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
12352 late when in LTO. As suggested place a gcc_unreachable for
12353 the DECL_ABSTRACT_P case.
12354 (gen_subprogram_die): Avoid another specification DIE
12355 for early built declarations/definitions for the late LTO case.
12356 (gen_variable_die): Add type references for late duplicated VLA dies
12358 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
12359 we have the abstract instance already.
12360 (process_scope_var): Adjust decl DIE contexts in LTO which
12361 first puts them in limbo.
12362 (gen_decl_die): Do not generate type DIEs late apart from
12363 types for VLAs or for decls we do not yet have a DIE. Do not
12364 call dwarf2out_abstract_function late.
12365 (dwarf2out_early_global_decl): Make sure to create DIEs
12366 for abstract instances of a decl first.
12367 (dwarf2out_late_global_decl): Adjust comment.
12368 (output_macinfo_op): With multiple macro sections use
12369 macinfo_label_base to distinguish labels.
12370 (output_macinfo): Likewise. Update macinfo_label_base.
12371 Pass in the line info label.
12372 (note_variable_value_in_expr): When generating LTO resolve
12373 all variable values here by generating DIEs as needed.
12374 (init_sections_and_labels): Add early LTO debug flag parameter
12375 and generate different sections and names if set. Add generation
12376 counter for the labels so we can have multiple of them.
12377 (reset_dies): Helper to allow DIEs to be output multiple times.
12378 (dwarf2out_finish): When outputting DIEs to the fat part of an
12379 LTO object first reset DIEs.
12380 (dwarf2out_early_finish): Output early DIEs when generating LTO.
12381 (modified_type_die): Check for decl_ultimate_origin being self
12383 (gen_type_die_with_usage): Likewise.
12384 (gen_typedef_die): Allow decl_ultimate_origin being self.
12385 (set_decl_abstract_flags): Remove.
12386 (set_block_abstract_flags): Likewise.
12387 (dwarf2out_abstract_function): Treat the early generated DIEs
12388 as the abstract copy and only add DW_AT_inline and
12389 DW_AT_artificial here and call set_decl_origin_self.
12390 If the DIE has an abstract origin don't do anything.
12391 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
12392 if we have none yet (Go fails to build one, PR78628).
12393 (variably_modified_type_p): Prevent endless recursion for Ada
12394 cyclic pointer types.
12395 * lto-streamer-in.c: Include debug.h.
12396 (dref_queue): New global.
12397 (lto_read_tree_1): Stream in DIE references.
12398 (lto_input_tree): Register DIE references.
12399 (input_function): Stream DECL_DEBUG_ARGS.
12400 * lto-streamer-out.c: Include debug.h.
12401 (lto_write_tree_1): Output DIE references.
12402 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
12403 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
12404 (output_function): Stream DECL_DEBUG_ARGS.
12405 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
12406 Stream DECL_ABSTRACT_ORIGIN.
12407 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
12408 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
12409 DECL_CONTEXT for file-scope decls.
12410 * lto-streamer.h (struct dref_entry): Declare.
12411 (dref_queue): Likewise.
12412 * cfgexpand.c (pass_expand::execute): Do not call the
12413 outlining_inline_function hook here.
12414 * lto-wrapper.c (debug_obj): New global.
12415 (tool_cleanup): Unlink it if required.
12416 (debug_objcopy): New function.
12417 (run_gcc): Handle early debug sections in the IL files by
12418 extracting them to separate files, partially linkin them and
12419 feeding the result back as result to the linker.
12420 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
12421 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
12422 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
12423 sections into a separate segment.
12424 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
12426 (darwin_asm_dwarf_section): Likewise.
12427 (darwin_asm_output_dwarf_offset): Likewise.
12428 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
12430 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12431 Alan Hayward <alan.hayward@arm.com>
12432 David Sherwood <david.sherwood@arm.com>
12434 * read-md.h (md_reader::record_potential_iterator_use): Replace
12435 pointer argument with an rtx and an index.
12436 * read-rtl.c (iterator_group::apply_iterator): Likewise.
12437 (apply_mode_iterator): Likewise.
12438 (apply_code_iterator): Likewise.
12439 (apply_int_iterator): Likewise.
12440 (apply_subst_iterator): Likewise.
12441 (record_iterator_use): Likewise.
12442 (record_attribute_use): Likewise.
12443 (md_reader::record_potential_iterator_use): Likewise. Update calls
12444 to record_iterator_use and apply_iterator.
12445 (iterator_use): Replace ptr with x and index.
12446 (attribute_use): Likewise.
12447 (apply_attribute_uses): Update calls to apply_iterator.
12448 (apply_iterators): Likewise. Update initialization of iterator_use.
12449 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
12450 and record_potential_iterator_use.
12451 (rtx_reader::read_rtx_operand): Likewise.
12453 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12454 Alan Hayward <alan.hayward@arm.com>
12455 David Sherwood <david.sherwood@arm.com>
12457 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
12460 2017-08-21 Richard Biener <rguenther@suse.de>
12462 PR middle-end/81884
12463 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
12464 at struct end conservatively when comparing common bases.
12466 2017-08-21 Richard Biener <rguenther@suse.de>
12468 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
12469 (mem_ref_in_stmt): Remove.
12470 (determine_max_movement): Use ref index to get at the reference.
12471 (invariantness_dom_walker::before_dom_children): Deal with
12472 lim data already initialized.
12473 (gather_mem_refs_stmt): Initialize lim data and record ref index.
12475 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12477 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
12478 (TARGET_ISA_ROUND): Ditto.
12479 (TARGET_ROUND): Ditto.
12480 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
12481 * config/i386/i386.md: Ditto.
12482 * config/i386/sse.md: Ditto.
12483 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
12484 with OPTION_MASK_ISA_SSE4_1.
12486 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12489 * doc/extend.texi (x86 Built-in Functions): Correct the name of
12490 __builtin_ia32_lzcnt_u16.
12492 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
12495 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
12496 (rs6000_set_current_function): Rewrite function to use it.
12498 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
12501 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
12502 and TYPE_WARN_IF_NOT_ALIGN.
12503 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
12504 (handle_warn_if_not_align): New.
12505 (place_union_field): Call handle_warn_if_not_align.
12506 (place_field): Call handle_warn_if_not_align.
12507 Copy TYPE_WARN_IF_NOT_ALIGN.
12508 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
12509 (layout_type): Likewise.
12510 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
12512 (tree_decl_common): Add warn_if_not_align.
12513 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
12514 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
12515 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
12516 (DECL_WARN_IF_NOT_ALIGN): Likewise.
12517 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
12518 * doc/extend.texi: Document warn_if_not_aligned attribute.
12519 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
12521 2017-08-17 Martin Liska <mliska@suse.cz>
12524 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
12525 (get_data_dependence): Use it as pointer type.
12526 (distribute_loop): Likewise.
12528 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12530 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
12531 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
12532 (p8_vmrgow_<mode>_direct): New define_insn.
12533 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
12534 handle endianness for vmrgew and vmrgow permute patterns.
12536 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
12538 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
12539 * config/rs6000/rs6000-cpus.def: Remove comment.
12540 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
12541 (POWERPC_MASKS): Likewise.
12542 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
12543 use of TARGET_VSX_TIMODE.
12544 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
12545 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
12546 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
12547 (rs6000_option_override_internal): Remove dead code.
12548 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
12549 (rs6000_legitimize_reload_address): Likewise.
12550 (rs6000_legitimate_address_p): Likewise.
12551 (rs6000_opt_masks): Delete "vsx-timode".
12552 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
12553 from function comment.
12554 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
12555 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
12556 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
12558 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
12559 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
12560 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
12561 (VSX_TI): Likewise.
12563 (define_peephole2): Likewise.
12565 2017-08-17 Martin Sebor <msebor@redhat.com>
12568 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
12569 past the end of an array.
12570 (test_pp_format): Add test cases.
12572 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
12574 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
12575 missing ECF_NOTHROW flags.
12577 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
12580 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
12581 operands residing in integer registers.
12582 (*vsx_le_perm_load_<mode>): Likewise.
12583 (*vsx_le_perm_store_<mode>): Likewise.
12584 (define_peephole2): Add peepholes to optimize the above.
12586 2017-08-17 Marek Polacek <polacek@redhat.com>
12588 PR middle-end/81814
12589 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
12590 to mimic what shorten_compare did. Change the return type to bool.
12591 (fold_cond_expr_with_comparison): Update call to
12592 operand_equal_for_comparison_p.
12593 (fold_ternary_loc): Likewise.
12595 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
12597 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
12599 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
12601 2017-08-17 Richard Biener <rguenther@suse.de>
12603 * tree-ssa-structalias.c (solve_graph): When propagating
12604 to successors update the graphs succ edges and avoid duplicate work.
12606 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
12609 * config/i386/i386.c (ix86_option_override_internal): Save target
12610 specific options after ix86_stack_protector_guard_reg was changed.
12612 2017-08-17 Richard Biener <rguenther@suse.de>
12614 PR tree-optimization/81827
12615 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
12616 (new_var_info): Initialize it conservatively.
12617 (get_call_vi): Mark register vars.
12618 (new_scalar_tmp_constraint_exp): Likewise.
12619 (handle_rhs_call): Likewise.
12620 (handle_const_call): Likewise.
12621 (create_function_info_for): Likewise.
12622 (solve_constraints): Sort varinfos to separate register from
12623 non-register vars to pack points-to solution bitmaps during
12626 2017-08-17 Marek Polacek <polacek@redhat.com>
12628 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
12630 2017-08-17 Richard Biener <rguenther@suse.de>
12632 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
12633 to true when overflow is undefined and we saturated the result.
12635 2017-08-17 Alan Modra <amodra@gmail.com>
12638 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
12639 Don't use store multiple if only one reg needs saving.
12640 (interesting_frame_related_regno): New function.
12641 (rs6000_frame_related): Don't emit frame info for regs that
12643 (rs6000_emit_epilogue): Likewise.
12645 2017-08-16 Nathan Sidwell <nathan@acm.org>
12647 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
12648 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
12649 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
12650 * tree.c (free_lang_data_in_type): Use else-if chain. Always
12651 clear TYPE_LANG_1. Remove obsolete member-function stripping.
12652 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
12653 (verify_type): Adjust for TYPE_BINFO move.
12654 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
12655 process TYPE_BINFO directly.
12656 (hash_tree): Likewise.
12657 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
12659 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
12662 2017-08-16 David Malcolm <dmalcolm@redhat.com>
12664 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
12666 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
12669 * config/i386/i386.md (*anddi_1_btr): Change predicates of
12670 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
12671 Add ix86_binary_operator_ok to insn constraint.
12672 (*iordi_1_bts): Ditto.
12673 (*xordi_1_btc): Ditto.
12674 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
12675 Update corresponding peephole2 pattern.
12679 2017-08-16 Bin Cheng <bin.cheng@arm.com>
12681 PR tree-optimization/81832
12682 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
12683 copy loop header which has IFN_LOOP_DIST_ALIAS call.
12685 2017-08-16 Marek Polacek <polacek@redhat.com>
12688 * fold-const.c (fold_indirect_ref_1): Restore original behavior
12689 regarding size_zero_node.
12691 2017-08-16 Martin Liska <mliska@suse.cz>
12694 * config.gcc: Respect previously set extra_objs in case
12697 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
12699 PR tree-optimization/81835
12700 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
12701 the phi SSA_NAME. Check that the condition in a COND_EXPR does
12702 not depend on the phi.
12704 2017-08-16 Alan Modra <amodra@gmail.com>
12706 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
12709 2017-08-16 Alan Modra <amodra@gmail.com>
12711 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
12712 (save_reg_p): ..into this. Update all callers.
12713 (first_reg_to_save): Simplify.
12715 2017-08-16 Alan Modra <amodra@gmail.com>
12717 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
12720 2017-08-15 Joseph Myers <joseph@codesourcery.com>
12724 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
12725 constant count if that count is less than 32.
12727 2017-08-15 Nathan Sidwell <nathan@acm.org>
12729 * gcc.c (execute): Emit friendlier message if inferior is killed
12730 by an external cause.
12732 2017-08-15 Richard Biener <rguenther@suse.de>
12734 PR tree-optimization/81790
12735 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
12736 CONSTRUCTORs from simplifying and VN.
12738 2017-08-14 Martin Sebor <msebor@redhat.com>
12740 * builtin-attrs.def: Add comments.
12742 2017-08-14 Martin Sebor <msebor@redhat.com>
12745 * doc/extend.texi (attribute nonstring): Document new attribute.
12747 2017-08-14 Martin Sebor <msebor@redhat.com>
12750 * tree-diagnostic.c (default_tree_printer): Handle %G.
12751 * gimple-pretty-print.h (percent_G_format): Declare new function.
12752 * gimple-pretty-print.c (percent_G_format): Define.
12753 * tree-pretty-print.c (percent_K_format): Add argument.
12755 2017-08-14 Martin Sebor <msebor@redhat.com>
12757 PR translation/79998
12758 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
12759 Remove a stray space.
12761 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
12764 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
12765 (*iordi_1_bts): Ditto.
12766 (*xordi_1_btc): Ditto.
12768 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12771 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
12772 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12774 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
12775 quoted strings, and make more translator-friendly.
12776 (darwin_rs6000_override_options): Likewise.
12777 (rs6000_option_override_internal): Likewise.
12778 (rs6000_return_in_memory): Fix overlong line.
12779 (init_cmulative_args): Use quoted strings, and make more
12780 translator-friendly.
12781 (rs6000_pass_by_reference): Fix overlong line.
12782 (def_builtin): Use quoted strings.
12783 (altivec_expand_predicate_builtin): Use quoted strings, and make
12784 more translator-friendly.
12785 (htm_expand_builtin): Use quoted strings.
12786 (cpu_expand_builtin): Use quoted strings, and make more
12787 translator-friendly.
12788 (altivec_expand_builtin): Likewise.
12789 (paired_expand_predicate_builtin): Likewise.
12790 (rs6000_invalid_builtin): Likewise.
12791 (builtin_function_type): Use quoted strings.
12792 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
12793 more translator-friendly.
12794 (rs6000_trampoline_init): Likewise.
12795 (rs6000_handle_altivec_attribute): Likewise.
12796 (rs6000_inner_target_options): Use quoted strings.
12797 (rs6000_disable_incompatible_switches): Likewise.
12798 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
12799 strings, and make more translator-friendly.
12800 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
12802 2017-08-14 Bin Cheng <bin.cheng@arm.com>
12804 PR tree-optimization/81799
12805 * tree-loop-distribution.c (version_loop_by_alias_check): Force
12806 cond_expr to simple gimple operand.
12808 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
12810 PR middle-end/46932
12811 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
12813 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
12817 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
12818 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
12819 TARGET_GASISR_PROLOGUES.
12820 * config/avr/avr.c (avr_option_override): Same.
12821 (avr_pass_pre_proep::execute): Same.
12823 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
12826 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
12827 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
12829 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
12831 * config/i386/i386.md (*load_tp_<mode>): Redefine as
12832 define_insn_and_split. Split to a memory load from 0 in
12833 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
12834 using PTR mode iterator.
12835 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
12836 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
12837 (*add_tp_<mode>): Redefine as define_insn_and_split.
12838 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
12839 address space. Merge with *add_tp_x32 using PTR mode iterator.
12840 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
12841 Split to an add with a memory load from 0 in
12842 DEFAULT_TLS_SEG_REG address space.
12844 2017-08-12 Andrew Pinski <apinski@cavium.com>
12846 * config/aarch64/aarch64-option-extensions.def (rdma):
12847 Fix feature string to what Linux prints out in /proc/cpuinfo.
12849 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
12852 * dwarf2out.c (modified_type_die): For C typedef types that have
12853 an ultimate origin, process the ultimate origin instead of the
12855 (gen_typedef_die): Assert that input DECLs have no ultimate
12857 (gen_type_die_with_usage): For typedef variants that have an
12858 ultimate origin, just call gen_decl_die on the original DECL.
12859 (process_scope_var): Avoid creating DIEs for local typedefs and
12860 concrete static variables.
12862 2017-08-12 Alan Modra <amodra@gmail.com>
12866 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
12867 match gnu-user.h startfile.
12868 (ENDFILE_LINUX_SPEC): Similarly.
12870 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
12873 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
12875 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
12877 2017-08-11 Tamar Christina <tamar.christina@arm.com>
12878 * config/aarch64/aarch64.md (mov<mode>): Change.
12879 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
12880 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
12881 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
12883 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
12885 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
12886 for storage order barriers.
12888 2017-08-11 Martin Liska <mliska@suse.cz>
12891 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
12892 variables of void type.
12894 2017-08-11 Martin Liska <mliska@suse.cz>
12896 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
12897 TARGET_SUPPORTS_ALIASES.
12898 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
12899 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
12900 (optimize_weakref): Likewise.
12901 * symtab.c (symtab_node::noninterposable_alias): Likewise.
12902 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
12903 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
12905 2017-08-11 Martin Liska <mliska@suse.cz>
12908 * config/i386/i386.c (make_resolver_func): Do complete
12909 refactoring of the function.
12911 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
12914 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
12915 * config/i386/i386.c (ix86_stack_protect_guard): Use
12916 ix86_stack_protect_guard_symbol_str to generate varible declaration.
12917 * doc/invoke.texi (x86 Options): Document
12918 -mstack-protector-guard-symbol= option.
12920 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
12922 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
12923 * config/i386/i386.c (ix86_split_stack_guard): New function.
12924 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
12925 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
12926 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
12927 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
12928 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
12929 (split_stack_space_check): Call ix86_split_stack_guard.
12931 2017-08-10 Martin Sebor <msebor@redhat.com>
12933 * print-tree.c (print_node): Print location using the established
12935 Replace spaces with colons.
12936 (debug_raw, debug): Ditto.
12938 2017-08-10 Martin Sebor <msebor@redhat.com>
12941 * pretty-print.c (pp_format): Correct the handling of %s precision.
12943 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
12946 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
12948 (ix86_finalize_stack_frame_flags): This. Also clear
12949 frame_pointer_needed if -fno-omit-frame-pointer is used without
12951 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
12952 with ix86_finalize_stack_frame_flags.
12953 (ix86_expand_epilogue): Likewise.
12954 (ix86_expand_split_stack_prologue): Likewise.
12955 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
12957 2017-08-10 Martin Liska <mliska@suse.cz>
12960 * c-attribs.c (handle_target_attribute):
12961 Report warning for an empty string argument of target attribute.
12963 2017-08-09 Jakub Jelinek <jakub@redhat.com>
12966 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
12968 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
12969 or DECL_NONLOCAL labels.
12970 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
12971 or DECL_NONLOCAL labels here.
12973 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
12975 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
12976 to indicate when early gimple folding has been disabled.
12977 (rs6000_gimple_fold_builtin): Add debug content.
12978 (rs6000_invalid_builtin): Fix whitespace.
12979 (rs6000_expand_builtin): Fix whitespace.
12980 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
12982 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
12985 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
12986 SAVE_MULTIPLE if not all the registers that saves, should be saved.
12988 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
12990 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
12991 (qdf24xx): Likewise.
12992 * config/aarch64/aarch64.md: Include falkor.md.
12993 * config/aarch64/falkor.md: New.
12995 2017-08-09 Marek Polacek <polacek@redhat.com>
12998 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
12999 * diagnostic.c (emit_diagnostic): Add a comment.
13000 (emit_diagnostic_valist): New function.
13002 2017-08-09 Marek Polacek <polacek@redhat.com>
13005 * input.c (make_location): New overload.
13006 * input.h (make_location): Declare.
13008 2017-08-08 Alan Modra <amodra@gmail.com>
13009 H.J. Lu <hongjiu.lu@intel.com>
13012 * gcc.c (NO_PIE_SPEC): Delete.
13013 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
13015 (LINK_PIE_SPEC): ..to here.
13016 (LINK_COMMAND_SPEC): Support -no-pie.
13017 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
13018 chain of crtbegin*.o selection, update for PIE_SPEC changes and
13020 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
13021 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
13022 (ENDFILE_CRTEND_SPEC): Similarly.
13024 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
13027 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
13028 (mstack-protector-guard-offset=): Ditto.
13029 * config/i386/i386.c (ix86_option_override): Handle
13030 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
13032 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
13033 ix86_stack_protect_guard_offset variables.
13034 (TARGET_STACK_PROTECT_GUARD): Always define.
13035 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
13036 and -mstack-protector-guard-offset= options.
13038 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13040 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
13041 boundary case for the last candidate.
13043 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13045 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
13046 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
13048 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13050 PR middle-end/19706
13051 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
13052 * config/aarch64/aarch64-builtins.c
13053 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
13054 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
13055 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
13057 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13058 Andrew Pinski <pinskia@gmail.com>
13060 PR middle-end/19706
13061 * internal-fn.def (XORSIGN): New.
13062 * optabs.def (xorsign_optab): New.
13063 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
13064 (convert_expand_mult_copysign): New.
13065 (pass_optimize_widening_mul::execute): Call
13066 convert_expand_mult_copysign.
13068 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13070 PR tree-optimization/81354
13071 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
13072 Insert on edges rather than explicitly creating landing pads.
13073 (analyze_candidates_and_replace): Commit edge inserts.
13075 2017-08-08 Richard Biener <rguenther@suse.de>
13077 PR middle-end/81719
13078 * tree-ssa-loop-niter.c: Include tree-dfa.h.
13079 (expand_simple_operations): Also look through ADDR_EXPRs with
13080 MEM_REF bases treating them as POINTER_PLUS_EXPR.
13082 2017-08-08 Richard Biener <rguenther@suse.de>
13084 PR tree-optimization/81723
13085 * tree-vect-slp.c (struct bst_traits): New hash traits.
13086 (bst_fail): New global.
13087 (vect_build_slp_tree_2): New worker, split out from ...
13088 (vect_build_slp_tree): ... this now wrapping it with using
13089 bst_fail set to cache SLP tree build fails. Properly handle
13091 (vect_analyze_slp_instance): Allocate and free bst_fail.
13093 2017-08-08 Martin Liska <mliska@suse.cz>
13096 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
13097 LABEL_DECLs that can be from a different function.
13099 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13101 PR tree-optimization/81744
13102 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
13103 loop's number of iterations.
13105 2017-08-08 Martin Liska <mliska@suse.cz>
13107 * asan.c: Include header files.
13108 * attribs.c (build_decl_attribute_variant): New function moved
13110 (build_type_attribute_qual_variant): Likewise.
13111 (cmp_attrib_identifiers): Likewise.
13112 (simple_cst_list_equal): Likewise.
13113 (omp_declare_simd_clauses_equal): Likewise.
13114 (attribute_value_equal): Likewise.
13115 (comp_type_attributes): Likewise.
13116 (build_type_attribute_variant): Likewise.
13117 (lookup_ident_attribute): Likewise.
13118 (remove_attribute): Likewise.
13119 (merge_attributes): Likewise.
13120 (merge_type_attributes): Likewise.
13121 (merge_decl_attributes): Likewise.
13122 (merge_dllimport_decl_attributes): Likewise.
13123 (handle_dll_attribute): Likewise.
13124 (attribute_list_equal): Likewise.
13125 (attribute_list_contained): Likewise.
13126 * attribs.h (lookup_attribute): New function moved from tree.[ch].
13127 (lookup_attribute_by_prefix): Likewise.
13128 * bb-reorder.c: Include header files.
13129 * builtins.c: Likewise.
13130 * calls.c: Likewise.
13131 * cfgexpand.c: Likewise.
13132 * cgraph.c: Likewise.
13133 * cgraphunit.c: Likewise.
13134 * convert.c: Likewise.
13135 * dwarf2out.c: Likewise.
13136 * final.c: Likewise.
13137 * fold-const.c: Likewise.
13138 * function.c: Likewise.
13139 * gimple-expr.c: Likewise.
13140 * gimple-fold.c: Likewise.
13141 * gimple-pretty-print.c: Likewise.
13142 * gimple.c: Likewise.
13143 * gimplify.c: Likewise.
13144 * hsa-common.c: Likewise.
13145 * hsa-gen.c: Likewise.
13146 * internal-fn.c: Likewise.
13147 * ipa-chkp.c: Likewise.
13148 * ipa-cp.c: Likewise.
13149 * ipa-devirt.c: Likewise.
13150 * ipa-fnsummary.c: Likewise.
13151 * ipa-inline.c: Likewise.
13152 * ipa-visibility.c: Likewise.
13154 * lto-cgraph.c: Likewise.
13155 * omp-expand.c: Likewise.
13156 * omp-general.c: Likewise.
13157 * omp-low.c: Likewise.
13158 * omp-offload.c: Likewise.
13159 * omp-simd-clone.c: Likewise.
13160 * opts-global.c: Likewise.
13161 * passes.c: Likewise.
13162 * predict.c: Likewise.
13163 * sancov.c: Likewise.
13164 * sanopt.c: Likewise.
13165 * symtab.c: Likewise.
13166 * toplev.c: Likewise.
13167 * trans-mem.c: Likewise.
13168 * tree-chkp.c: Likewise.
13169 * tree-eh.c: Likewise.
13170 * tree-into-ssa.c: Likewise.
13171 * tree-object-size.c: Likewise.
13172 * tree-parloops.c: Likewise.
13173 * tree-profile.c: Likewise.
13174 * tree-ssa-ccp.c: Likewise.
13175 * tree-ssa-live.c: Likewise.
13176 * tree-ssa-loop.c: Likewise.
13177 * tree-ssa-sccvn.c: Likewise.
13178 * tree-ssa-structalias.c: Likewise.
13179 * tree-ssa.c: Likewise.
13180 * tree-streamer-in.c: Likewise.
13181 * tree-vectorizer.c: Likewise.
13182 * tree-vrp.c: Likewise.
13183 * tsan.c: Likewise.
13184 * ubsan.c: Likewise.
13185 * varasm.c: Likewise.
13186 * varpool.c: Likewise.
13187 * tree.c: Remove functions moved to attribs.[ch].
13188 * tree.h: Likewise.
13189 * config/aarch64/aarch64.c: Add attrs.h header file.
13190 * config/alpha/alpha.c: Likewise.
13191 * config/arc/arc.c: Likewise.
13192 * config/arm/arm.c: Likewise.
13193 * config/avr/avr.c: Likewise.
13194 * config/bfin/bfin.c: Likewise.
13195 * config/c6x/c6x.c: Likewise.
13196 * config/cr16/cr16.c: Likewise.
13197 * config/cris/cris.c: Likewise.
13198 * config/darwin.c: Likewise.
13199 * config/epiphany/epiphany.c: Likewise.
13200 * config/fr30/fr30.c: Likewise.
13201 * config/frv/frv.c: Likewise.
13202 * config/ft32/ft32.c: Likewise.
13203 * config/h8300/h8300.c: Likewise.
13204 * config/i386/winnt.c: Likewise.
13205 * config/ia64/ia64.c: Likewise.
13206 * config/iq2000/iq2000.c: Likewise.
13207 * config/lm32/lm32.c: Likewise.
13208 * config/m32c/m32c.c: Likewise.
13209 * config/m32r/m32r.c: Likewise.
13210 * config/m68k/m68k.c: Likewise.
13211 * config/mcore/mcore.c: Likewise.
13212 * config/microblaze/microblaze.c: Likewise.
13213 * config/mips/mips.c: Likewise.
13214 * config/mmix/mmix.c: Likewise.
13215 * config/mn10300/mn10300.c: Likewise.
13216 * config/moxie/moxie.c: Likewise.
13217 * config/msp430/msp430.c: Likewise.
13218 * config/nds32/nds32-isr.c: Likewise.
13219 * config/nds32/nds32.c: Likewise.
13220 * config/nios2/nios2.c: Likewise.
13221 * config/nvptx/nvptx.c: Likewise.
13222 * config/pa/pa.c: Likewise.
13223 * config/pdp11/pdp11.c: Likewise.
13224 * config/powerpcspe/powerpcspe.c: Likewise.
13225 * config/riscv/riscv.c: Likewise.
13226 * config/rl78/rl78.c: Likewise.
13227 * config/rx/rx.c: Likewise.
13228 * config/s390/s390.c: Likewise.
13229 * config/sh/sh.c: Likewise.
13230 * config/sol2.c: Likewise.
13231 * config/sparc/sparc.c: Likewise.
13232 * config/spu/spu.c: Likewise.
13233 * config/stormy16/stormy16.c: Likewise.
13234 * config/tilegx/tilegx.c: Likewise.
13235 * config/tilepro/tilepro.c: Likewise.
13236 * config/v850/v850.c: Likewise.
13237 * config/vax/vax.c: Likewise.
13238 * config/visium/visium.c: Likewise.
13239 * config/xtensa/xtensa.c: Likewise.
13241 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13244 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
13245 constraints since the -mupper-regs-* switches have been
13247 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
13248 into a vector from a double word element that was extracted from
13249 another vector, and eliminate extra XXPERMDI instructions.
13250 (vsx_concat_<mode>_2): Likewise.
13251 (vsx_concat_<mode>_3): Likewise.
13252 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
13253 concat to allow optimizing inserts from previous extracts.
13255 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
13257 * config/i386/i386.c (ix86_stack_protect_guard): Generate
13258 memory reference to a SSP offset in TLS address space.
13259 (ix86_print_operand) <case '@'>: Remove.
13260 (ix86_print_operand_punct_valid_p): Remove '@' code.
13261 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
13262 UNSPEC_SP_TLS_TEST.
13263 (stack_tls_protect_set_<mode>): Remove.
13264 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
13265 (stack_tls_protect_test_<mode>): Remove.
13266 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
13268 2017-08-07 Olivier Hainque <hainque@adacore.com>
13271 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
13273 2017-08-07 Douglas Rupp <rupp@adacore.com>
13275 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
13276 variable was referenced as multidir in command.
13278 2017-08-07 Jakub Jelinek <jakub@redhat.com>
13281 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
13284 2017-08-07 Martin Liska <mliska@suse.cz>
13286 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
13287 * config/rl78/rl78.c: Add include of attribs.h.
13288 * config/sh/sh.c: Likewise.
13289 * config/v850/v850.c: Likewise.
13291 2017-08-07 Tom de Vries <tom@codesourcery.com>
13293 PR middle-end/78266
13294 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
13296 2017-08-07 Martin Liska <mliska@suse.cz>
13298 * config/mips/mips.c: Include attribs.h.
13300 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
13303 * doc/invoke.texi: Document change in behvaior for -Ofast for
13306 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
13308 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
13310 (aarch64_pop_regs): Likewise.
13311 (aarch64_gen_load_pair): Likewise.
13312 (aarch64_save_callee_saves): Likewise.
13313 (aarch64_restore_callee_saves): Likewise.
13315 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13317 * config/i386/i386.c: Revert the last change.
13319 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13322 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13324 (ix86_finalize_stack_frame_flags): This. Also clear
13325 frame_pointer_needed if -fno-omit-frame-pointer is used without
13327 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13328 with ix86_finalize_stack_frame_flags.
13329 (ix86_expand_epilogue): Likewise.
13330 (ix86_expand_split_stack_prologue): Likewise.
13332 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13335 * config/i386/i386.c (get_builtin_code_for_version): Set priority
13336 to P_AES for Westmere.
13338 2017-08-07 Jonathan Yong <10walls@gmail.com>
13340 * config/i386/mingw.opt (fset-stack-executable): Removed.
13341 * config/i386/cygming.opt (fset-stack-executable): Moved
13343 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
13345 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
13347 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
13349 2017-08-07 Marek Polacek <polacek@redhat.com>
13351 PR middle-end/81737
13352 * fold-const.c (fold_indirect_ref_1): Check type_domain.
13354 2017-08-07 Martin Liska <mliska@suse.cz>
13356 * attribs.h (canonicalize_attr_name): New function.
13357 (cmp_attribs): Move from c-format.c and adjusted.
13358 (is_attribute_p): Moved from tree.h.
13359 * tree-inline.c: Add new includes.
13360 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
13361 (private_is_attribute_p): Remove.
13362 (private_lookup_attribute): Likewise.
13363 (private_lookup_attribute_by_prefix): Simplify.
13364 (remove_attribute): Use is_attribute_p.
13365 * tree.h: Remove removed declarations.
13367 2017-08-07 Jakub Jelinek <jakub@redhat.com>
13369 PR middle-end/81698
13370 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
13371 instead of computing it in the function. Formatting fix.
13372 (expand_case): Don't rely on default_edge being the first edge,
13373 clear it if removing it, pass default_edge to
13374 emit_case_dispatch_table.
13375 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
13378 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
13380 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
13381 insn in the function, emit NOP after the insn.
13383 2017-08-06 Tom de Vries <tom@codesourcery.com>
13385 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
13388 2017-08-06 Tom de Vries <tom@codesourcery.com>
13390 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
13393 2017-08-04 Yury Gribov <tetra2005@gmail.com>
13395 PR tree-optimization/57371
13396 * match.pd: New pattern.
13398 2017-08-04 Marek Polacek <polacek@redhat.com>
13400 PR middle-end/81695
13401 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
13402 perform the computation in offset_int.
13404 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13406 PR tree-optimization/81136
13407 * tree-vectorizer.h: Include tree-hash-traits.h.
13408 (vec_base_alignments): New typedef.
13409 (vec_info): Add a base_alignments field.
13410 (vect_record_base_alignments): Declare.
13411 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
13413 (DR_IS_CONDITIONAL_IN_STMT): New macro.
13414 (create_data_ref): Add an is_conditional_in_stmt argument.
13415 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
13416 the is_conditional_in_stmt field.
13417 (data_ref_loc): Add an is_conditional_in_stmt field.
13418 (get_references_in_stmt): Set the is_conditional_in_stmt field.
13419 (find_data_references_in_stmt): Update call to create_data_ref.
13420 (graphite_find_data_references_in_stmt): Likewise.
13421 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
13422 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
13423 (vect_record_base_alignment): New function.
13424 (vect_record_base_alignments): Likewise.
13425 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
13426 for nested statements even if we fail to compute a misalignment.
13427 Use pooled base alignments for unconditional references.
13428 (vect_find_same_alignment_drs): Compare base addresses instead
13430 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
13431 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
13433 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13435 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
13436 Add an explicit name for the enum. Use auto_vec for slp_instances
13437 and grouped_stores.
13438 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
13440 (_bb_vec_info): Add a constructor and destructor.
13441 (vinfo_for_stmt): Return NULL for uids of -1 as well.
13442 (destroy_loop_vec_info): Delete.
13443 (vect_destroy_datarefs): Likewise.
13444 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
13445 (vec_info::vec_info): New function.
13446 (vec_info::~vec_info): Likewise.
13447 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
13448 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
13449 destroy_loop_vec_info.
13450 * tree-vect-loop.c (new_loop_vec_info): Replace with...
13451 (_loop_vec_info::_loop_vec_info): ...this.
13452 (destroy_loop_vec_info): Replace with...
13453 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
13454 the stmt_vec_infos. Leave handling of vec_info information to its
13455 destructor. Remove explicit vector releases.
13456 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
13457 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
13458 * tree-vect-slp.c (new_bb_vec_info): Replace with...
13459 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
13460 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
13461 (destroy_bb_vec_info): Replace with...
13462 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
13463 information to its destructor.
13464 (vect_slp_analyze_bb_1): Use new and delete instead of
13465 new_bb_vec_info and destroy_bb_vec_info.
13466 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
13469 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13471 * tree-data-ref.h (subscript): Add access_fn field.
13472 (data_dependence_relation): Add could_be_independent_p.
13473 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
13474 (same_access_functions): Move to tree-data-ref.c.
13475 * tree-data-ref.c (ref_contains_union_access_p): New function.
13476 (access_fn_component_p): Likewise.
13477 (access_fn_components_comparable_p): Likewise.
13478 (dr_analyze_indices): Add a reference to access_fn_component_p.
13479 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
13481 (constant_access_functions): Likewise.
13482 (add_other_self_distances): Likewise.
13483 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
13484 (initialize_data_dependence_relation): Use XCNEW and remove
13485 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
13486 of access functions that have the same type. Allow the
13487 subsequence to end with different bases in some circumstances.
13488 Record the chosen access functions in SUB_ACCESS_FN.
13489 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
13490 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
13491 (subscript_dependence_tester_1): Likewise dra and drb.
13492 (build_classic_dist_vector): Update calls accordingly.
13493 (subscript_dependence_tester): Likewise.
13494 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
13495 DDR_COULD_BE_INDEPENDENT_P.
13496 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
13497 comp_alias_ddrs instead of may_alias_ddrs.
13498 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
13500 (vect_analyze_data_ref_dependence): Use it if
13501 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
13502 distance vectors if that fails.
13503 (dependence_distance_ge_vf): New function.
13504 (vect_prune_runtime_alias_test_list): Use it. Don't clear
13505 LOOP_VINFO_MAY_ALIAS_DDRS.
13507 2017-08-04 Richard Biener <rguenther@suse.de>
13509 PR middle-end/81705
13510 * fold-const.c (fold_binary_loc): Properly restrict
13511 minus_var0 && minus_var1 case when associating undefined overflow
13514 2017-08-04 Tom de Vries <tom@codesourcery.com>
13516 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
13518 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13520 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13521 Don't start diagnostic messages with a capital letter.
13522 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13524 (rs6000_invalid_builtin): Likewise.
13525 (rs6000_trampoline_init): Likewise.
13527 2017-08-03 Jakub Jelinek <jakub@redhat.com>
13530 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
13531 after setting changeable df flags.
13533 2017-08-03 Richard Biener <rguenther@suse.de>
13535 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
13536 up if the use is in USE - X.
13538 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
13540 * toplev.c (dumpfile.h): New include.
13541 (internal_error_reentered): New static function. Use it...
13542 (internal_error_function): ...here to handle reentered internal_error.
13544 2017-08-03 Richard Biener <rguenther@suse.de>
13546 PR middle-end/81148
13547 * fold-const.c (split_tree): Add minus_var and minus_con
13548 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
13549 here but always use minus_*.
13550 (associate_trees): Assert we never associate with MINUS_EXPR
13551 and NULL first operand. Do not recurse for PLUS_EXPR operands
13552 when associating as MINUS_EXPR either.
13553 (fold_binary_loc): Track minus_var and minus_con.
13555 2017-08-03 Tom de Vries <tom@codesourcery.com>
13558 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
13559 ACCEL_COMPILER, apply finish_options on
13560 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
13562 2017-08-03 Tom de Vries <tom@codesourcery.com>
13565 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
13566 function_entry_patch_area_size > 0.
13568 2017-08-03 Jakub Jelinek <jakub@redhat.com>
13571 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
13572 instead of 10??LU, perform unit multiplication in wide_int,
13573 don't change alloc_object_size_limit if the limit is larger
13576 PR tree-optimization/81655
13577 PR tree-optimization/81588
13578 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
13579 the case when ranges[i].low and high are 1 for unsigned type with
13582 PR middle-end/81052
13583 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
13584 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
13586 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13588 * tree-vrp.h: Add include guard.
13590 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
13593 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
13594 (ud2): New insn pattern.
13595 * config/i386/i386.c (ix86_expand_epilogue):
13596 For naked functions, generate ud2 instead of trap insn.
13598 2017-08-02 Marek Polacek <polacek@redhat.com>
13601 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
13603 2017-08-02 Tom de Vries <tom@codesourcery.com>
13604 Cesar Philippidis <cesar@codesourcery.com>
13606 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
13607 Add missing edge probabilities.
13609 2017-08-02 Tamar Christina <tamar.christina@arm.com>
13611 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
13612 Correct endianness.
13614 2017-08-02 Jakub Jelinek <jakub@redhat.com>
13616 PR middle-end/79499
13617 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
13618 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
13619 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
13620 prologue_seq sequences - if any.
13622 2017-08-02 Richard Biener <rguenther@suse.de>
13624 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
13625 via vectors if supported, integer extracts via punning if supported
13626 or otherwise vector extracts.
13628 2017-08-02 Richard Biener <rguenther@suse.de>
13630 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
13632 (bitmap_insert_into_set): ... this.
13634 2017-08-02 Richard Biener <rguenther@suse.de>
13636 PR tree-optimization/81633
13638 2015-08-17 Alan Hayward <alan.hayward@arm.com>
13640 PR tree-optimization/71752
13641 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
13643 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
13645 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
13646 (machine_function::call_ms2sysv_pad_out): Remove field.
13647 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
13648 (ix86_compute_frame_layout): Likewise.
13650 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
13653 * config/i386/i386.c (ix86_set_func_type): Disallow naked
13654 attribute with interrupt attribute.
13656 2017-08-01 Andrew Pinski <apinski@cavium.com>
13658 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
13659 BIT_INSERT_EXPR's operand 1
13660 to see if the types precision matches.
13662 2017-08-01 Martin Liska <mliska@suse.cz>
13664 PR middle-end/70140
13665 * builtins.c (expand_builtin_memcpy_args): Remove.
13666 (expand_builtin_memcpy): Call newly added function
13667 expand_builtin_memory_copy_args.
13668 (expand_builtin_memcpy_with_bounds): Likewise.
13669 (expand_builtin_mempcpy): Remove last argument.
13670 (expand_builtin_mempcpy_with_bounds): Likewise.
13671 (expand_builtin_memory_copy_args): New function created from
13672 expand_builtin_mempcpy_args with small modifications.
13673 (expand_builtin_mempcpy_args): Remove.
13674 (expand_builtin_stpcpy): Remove unused argument.
13675 (expand_builtin): Likewise.
13676 (expand_builtin_with_bounds): Likewise.
13678 2017-08-01 Martin Liska <mliska@suse.cz>
13681 Make mempcpy more optimal (PR middle-end/70140).
13683 2017-08-01 Jakub Jelinek <jakub@redhat.com>
13686 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
13687 __builtin_vec_cmpne verify both arguments are compatible vectors
13688 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
13689 verify arg1_type is a pointer or array type. For __builtin_vec_st,
13690 move computation of aligned to after checking the argument types.
13694 * config/rs6000/vsx.md (vextract_fp_from_shorth,
13695 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
13698 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
13699 Doug Rupp <rupp@adacore.com>
13700 Olivier Hainque <hainque@adacore.com>
13702 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
13703 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
13705 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
13706 for TARGET_OS_CPP_BUILTIN.
13707 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
13708 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
13710 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
13711 passing required abi options to the assembler for EABI configurations.
13712 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
13713 of .text.hot and .text.unlikely sections for kernel modules when
13714 using ARM style exceptions.
13715 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
13716 options. Add EXTRA_CC1_SPEC.
13717 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
13719 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
13721 (ARM_TARGET2_DWARF_FORMAT): Define.
13722 * config/arm/t-vxworks: Adjust multilib control to removal of the
13723 Diab command line options.
13725 2017-08-01 Martin Liska <mliska@suse.cz>
13727 PR gcov-profile/81561
13728 * gcov.c (unblock): Make unblocking safe as we need to preserve
13729 index correspondence of blocks and block_lists.
13731 2017-08-01 Richard Biener <rguenther@suse.de>
13733 PR tree-optimization/81181
13734 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
13735 (compute_antic): ... end of iteration here.
13737 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
13739 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
13740 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
13741 (ftree-slp-vectorize): Likewise.
13742 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
13743 can no longer be set independent of flag_tree_loop_vectorize.
13744 * omp-general.c (emp_max_vf): Likewise.
13745 * opts.c (enable_fdo_optimizations): Remove references to
13746 flag_tree_vectorize, these are now implicit.
13747 (common_handle_option): Remove handling for OPT_ftree_vectorize,
13748 and leave it for the options machinery.
13750 2017-08-01 Martin Liska <mliska@suse.cz>
13752 PR middle-end/70140
13753 * builtins.c (expand_builtin_memcpy_args): Remove.
13754 (expand_builtin_memcpy): Call newly added function
13755 expand_builtin_memory_copy_args.
13756 (expand_builtin_memcpy_with_bounds): Likewise.
13757 (expand_builtin_mempcpy): Remove last argument.
13758 (expand_builtin_mempcpy_with_bounds): Likewise.
13759 (expand_builtin_memory_copy_args): New function created from
13760 expand_builtin_mempcpy_args with small modifications.
13761 (expand_builtin_mempcpy_args): Remove.
13762 (expand_builtin_stpcpy): Remove unused argument.
13763 (expand_builtin): Likewise.
13764 (expand_builtin_with_bounds): Likewise.
13766 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
13769 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
13770 print "ds:" only for immediates in generic address space.
13772 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
13775 * config/i386/i386.c (ix86_funciton_naked): New prototype.
13776 (ix86_function_ok_for_sibcall): Return false for naked functions.
13778 2017-08-01 Richard Biener <rguenther@suse.de>
13780 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
13781 (compute_antic): Seed worklist with exit block predecessors.
13782 * cfganal.c (dfs_find_deadend): For a cycle return the source
13783 of the edge closing it.
13785 2017-08-01 Tamar Christina <tamar.christina@arm.com>
13787 * config/aarch64/aarch64.c
13788 (aarch64_can_const_movi_rtx_p): Move 0 check.
13790 2017-08-01 Bin Cheng <bin.cheng@arm.com>
13792 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
13793 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
13795 * match.pd: Ditto in address comparison pattern.
13797 2017-08-01 Bin Cheng <bin.cheng@arm.com>
13799 PR tree-optimization/81627
13800 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
13801 closed ssa form for store-store chain.
13803 2017-08-01 Bin Cheng <bin.cheng@arm.com>
13805 PR tree-optimization/81620
13806 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
13807 for store-store chain.
13809 2017-08-01 Jakub Jelinek <jakub@redhat.com>
13811 PR tree-optimization/81588
13812 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
13813 ranges[i].in_p, invert comparison code ccode. For >/>=,
13814 swap rhs1 and rhs2 and comparison code unconditionally,
13815 for </<= don't do that. Don't swap rhs1/rhs2 again if
13816 ranges[i].in_p, instead invert comparison code ccode if
13817 opcode or oe->rank is BIT_IOR_EXPR.
13820 * optabs.def (vec_extract_optab, vec_init_optab): Change from
13821 a direct optab to conversion optab.
13822 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
13823 with GET_MODE_INNER as last argument instead of optab_handler.
13824 * expmed.c (extract_bit_field_1): Likewise. Use vector from
13825 vector extraction if possible and optab is available.
13826 * expr.c (store_constructor): Use convert_optab_handler instead
13827 of optab_handler. Use vector initialization from smaller
13828 vectors if possible and optab is available.
13829 * tree-vect-stmts.c (vectorizable_load): Likewise.
13830 * doc/md.texi (vec_extract, vec_init): Document that the optabs
13831 now have two modes.
13832 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
13833 of vec_init from half-sized vectors with the same element mode.
13834 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
13835 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
13836 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
13837 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
13838 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
13839 after mode in gen_vec_extract* calls.
13840 (vec_extract<mode>): Renamed to ...
13841 (vec_extract<mode><ssescalarmodelower>): ... this.
13842 (vec_extract<mode><ssehalfvecmodelower>): New expander.
13843 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
13844 element mode after mode in gen_vec_init* calls.
13845 (VEC_INIT_HALF_MODE): New mode iterator.
13846 (vec_init<mode>): Renamed to ...
13847 (vec_init<mode><ssescalarmodelower>): ... this.
13848 (vec_init<mode><ssehalfvecmodelower>): New expander.
13849 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
13850 (vec_extractv2sfsf): ... this.
13851 (vec_initv2sf): Renamed to ...
13852 (vec_initv2sfsf): ... this.
13853 (vec_extractv2si): Renamed to ...
13854 (vec_extractv2sisi): ... this.
13855 (vec_initv2si): Renamed to ...
13856 (vec_initv2sisi): ... this.
13857 (vec_extractv4hi): Renamed to ...
13858 (vec_extractv4hihi): ... this.
13859 (vec_initv4hi): Renamed to ...
13860 (vec_initv4hihi): ... this.
13861 (vec_extractv8qi): Renamed to ...
13862 (vec_extractv8qiqi): ... this.
13863 (vec_initv8qi): Renamed to ...
13864 (vec_initv8qiqi): ... this.
13865 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
13866 (vec_init<mode>): Renamed to ...
13867 (vec_init<mode><VEC_base_l>): ... this.
13868 (vec_extract<mode>): Renamed to ...
13869 (vec_extract<mode><VEC_base_l>): ... this.
13870 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
13871 (vec_initv2sfsf): ... this.
13872 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
13873 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
13874 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
13875 element mode after mode in gen_vec_init* calls.
13876 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
13877 (vec_init<mode><Vel>): ... this.
13878 (vec_extract<mode>): Renamed to ...
13879 (vec_extract<mode><Vel>): ... this.
13880 * config/aarch64/iterators.md (Vel): New mode attribute.
13881 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
13882 Add element mode after mode in gen_vec_extract* calls.
13883 * config/s390/vector.md (non_vec_l): New mode attribute.
13884 (vec_extract<mode>): Renamed to ...
13885 (vec_extract<mode><non_vec_l>): ... this.
13886 (vec_init<mode>): Renamed to ...
13887 (vec_init<mode><non_vec_l>): ... this.
13888 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
13889 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
13891 * config/arm/iterators.md (V_elem_l): New mode attribute.
13892 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
13893 (vec_extract<mode><V_elem_l>): ... this.
13894 (vec_extractv2di): Renamed to ...
13895 (vec_extractv2didi): ... this.
13896 (vec_init<mode>): Renamed to ...
13897 (vec_init<mode><V_elem_l>): ... this.
13898 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
13899 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
13900 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
13901 Add element mode after gen_vec_extract* calls.
13902 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
13903 (vec_init<mode><unitmode>): ... this.
13904 (vec_extract<mode>): Renamed to ...
13905 (vec_extract<mode><unitmode>): ... this.
13906 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
13907 (vec_init<mode><unitmode>): ... this.
13908 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
13909 (vec_initv2sfsf): ... this.
13910 (vec_extractv2sf): Renamed to ...
13911 (vec_extractv2sfsf): ... this.
13912 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
13913 Add element mode after gen_vec_extract* calls.
13914 * config/mips/mips.md (unitmode): New mode iterator.
13915 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
13916 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
13917 * config/spu/spu.md (inner_l): New mode attribute.
13918 (vec_init<mode>): Renamed to ...
13919 (vec_init<mode><inner_l>): ... this.
13920 (vec_extract<mode>): Renamed to ...
13921 (vec_extract<mode><inner_l>): ... this.
13922 * config/sparc/sparc.md (veltmode): New mode iterator.
13923 (vec_init<VMALL:mode>): Renamed to ...
13924 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
13925 * config/ia64/vect.md (vec_initv2si): Renamed to ...
13926 (vec_initv2sisi): ... this.
13927 (vec_initv2sf): Renamed to ...
13928 (vec_initv2sfsf): ... this.
13929 (vec_extractv2sf): Renamed to ...
13930 (vec_extractv2sfsf): ... this.
13931 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
13932 (vec_init<mode>): Renamed to ...
13933 (vec_init<mode><VEC_base_l>): ... this.
13934 (vec_extract<mode>): Renamed to ...
13935 (vec_extract<mode><VEC_base_l>): ... this.
13936 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
13937 (vec_initv2sfsf): ... this.
13938 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
13939 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
13940 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
13941 gen_vec_init* calls.
13943 2017-08-01 Richard Biener <rguenther@suse.de>
13945 PR tree-optimization/81297
13946 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
13947 TREE_OVERFLOW from INTEGER_CSTs.
13949 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
13951 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
13953 2017-07-31 Carl Love <cel@us.ibm.com>
13955 * config/rs6000/rs6000-c: Add support for built-in functions
13956 vector signed char vec_xl_be (signed long long, signed char *);
13957 vector unsigned char vec_xl_be (signed long long, unsigned char *);
13958 vector signed int vec_xl_be (signed long long, signed int *);
13959 vector unsigned int vec_xl_be (signed long long, unsigned int *);
13960 vector signed long long vec_xl_be (signed long long, signed long long *);
13961 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
13962 vector signed short vec_xl_be (signed long long, signed short *);
13963 vector unsigned short vec_xl_be (signed long long, unsigned short *);
13964 vector double vec_xl_be (signed long long, double *);
13965 vector float vec_xl_be (signed long long, float *);
13966 * config/rs6000/altivec.h (vec_xl_be): Add #define.
13967 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
13968 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
13970 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
13971 (altivec_expand_builtin): Add switch statement to call
13972 altivec_expand_xl_be for each builtin.
13973 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
13974 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
13975 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
13976 __builtin_vsx_le_be_v16qi.
13977 * doc/extend.texi: Update the built-in documentation file for the
13978 new built-in functions.
13980 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
13983 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
13985 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
13987 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13989 * config.gcc: Add z14.
13990 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
13991 CPU model numbers for z13s and z14.
13992 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
13994 * config/s390/s390-opts.h (enum processor_type): Rename
13995 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
13996 * config/s390/s390.c (processor_table): Add field for CPU name to
13997 be passed to Binutils.
13998 (s390_asm_output_machine_for_arch): Use the new field in
13999 processor_table for Binutils.
14000 (s390_expand_builtin): Replace arch12 with z14.
14001 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14002 (s390_get_sched_attrmask): Likewise.
14003 (s390_get_unit_mask): Likewise.
14004 * config/s390/s390.opt: Add z14 to processor_type enum.
14006 2017-07-31 Martin Jambor <mjambor@suse.cz>
14009 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
14010 regardless of optimization level.
14012 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
14013 Martin Liska <mliska@suse.cz>
14015 * predict.def: Remove old comment and adjust probability.
14016 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
14017 PREDICT statements.
14019 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14022 * config/i386/i386.c (ix86_function_naked): New function.
14023 (ix86_can_use_return_insn_p): Return false for naked functions.
14024 (ix86_expand_prologue): Skip prologue for naked functions.
14025 (ix86_expand_epilogue): Skip epilogue for naked functions
14026 and emit trap instruction.
14027 (ix86_warn_func_return): New function.
14028 (ix86_attribute_table): Add "naked" attribute specification.
14029 (TARGET_WARN_FUNC_RETURN): Define.
14030 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
14032 2017-07-31 Martin Liska <mliska@suse.cz>
14034 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
14035 (dump_gimple_bb_header): Always dump BB info.
14036 (pp_cfg_jump): Do not append info about BB when dumping a jump.
14038 2017-07-31 Martin Liska <mliska@suse.cz>
14041 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
14042 also with current_function_decl non-null equality.
14044 2017-07-31 Jakub Jelinek <jakub@redhat.com>
14047 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
14048 change type to the element type, instead add eltype variable and
14049 use it where we are interested in the element type.
14051 PR tree-optimization/81603
14052 * ipa-polymorphic-call.c
14053 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
14054 offset arithmetic in offset_int, bail out if the resulting bit offset
14055 doesn't fit into shwi.
14057 2017-07-31 Martin Liska <mliska@suse.cz>
14059 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
14060 (gimplify_save_expr): Fix comment.
14062 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
14065 * config/i386/i386.c (ix86_function_arg): Update arguments for
14067 (ix86_compute_frame_layout): Set the initial stack offset to
14068 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
14069 INCOMING_FRAME_SP_OFFSET.
14070 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
14071 stack before exception handler returns.
14072 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
14073 the 'ERROR_CODE' for exception handler.
14075 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
14077 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
14078 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
14079 (ASM_OUTPUT_REG_POP): Ditto.
14080 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
14081 instead of asm_fprintf to output pure string.
14083 2017-07-29 Jakub Jelinek <jakub@redhat.com>
14085 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
14086 to imported_module_or_decl hook.
14087 (debug_nothing_tree_tree_tree_bool): Remove.
14088 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
14089 * debug.c (do_nothing_debug_hooks): Use
14090 debug_nothing_tree_tree_tree_bool_bool instead of
14091 debug_nothing_tree_tree_tree_bool.
14092 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14093 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
14094 * sdbout.c (sdb_debug_hooks): Likewise.
14095 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
14096 (gen_namespace_die): Add DW_AT_export_symbols attribute if
14098 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
14099 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
14100 attribute, don't add anything.
14102 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14104 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
14105 (fold_build2_stat_loc): Likewise.
14106 (fold_build3_stat_loc): Likewise.
14107 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
14108 (fold_build1_loc): Remove macro.
14109 (fold_build2_loc): Likewise.
14110 (fold_build3_loc): Likewise.
14112 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14114 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
14115 (gimple_build_debug_bind_source_stat): Likewise.
14116 * gimple.h (gimple_build_debug_bind): Remove macro.
14117 (gimple_build_debug_bind_source): Likewise.
14119 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14121 * bitmap.c (bitmap_alloc): Adjust.
14122 (bitmap_gc_alloc): Likewise.
14123 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
14125 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14127 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
14128 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
14129 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
14130 (bitmap_gc_alloc_stat): Likewise.
14131 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
14133 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14135 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
14136 * rtl.h (shallow_copy_rtx): Remove macro.
14138 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14140 * emit-rtl.c (gen_raw_REG): Adjust.
14141 * gengenrtl.c (gendef): Likewise.
14142 * rtl.c (rtx_alloc_stat): Remove _stat from name.
14143 * rtl.h (rtx_alloc): Remove macro.
14145 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14147 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
14148 (build_tree_list_stat): Likewise.
14149 * tree.h (build_tree_list): Remove macro.
14150 (build_tree_list_vec): Likewise.
14152 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14154 * tree.c (make_vector_stat): Remove _stat from name.
14155 (build_vector_stat): Likewise.
14156 * tree.h (make_vector_stat): Remove macro.
14157 (build_vector_stat): Likewise.
14159 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14161 * tree.h (build_var_debug_value): Remove prototype.
14163 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14165 * tree.c (tree_cons_stat): Remove _stat from name.
14166 * tree.h (tree_cons): Remove macro.
14168 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14170 * tree.c (build_vl_exp_stat): Remove _stat from name.
14171 * tree.h (build_vl_exp): Remove macro.
14173 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14175 * tree.c (build_decl_stat): Remove _stat from name.
14176 * tree.h (build_decl): Remove macro.
14178 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14180 * gimple.c (gimple_build_with_ops_stat): Adjust.
14181 (gimple_alloc_stat): Remove _stat from name.
14182 * gimple.h (gimple_alloc): Remove macro.
14184 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14186 * tree.c (make_tree_vec_stat): Remove _stat from name.
14187 (grow_tree_vec_stat): Likewise.
14188 * tree.h (make_tree_vec_stat): Adjust prototype.
14189 (grow_tree_vec_stat): Likewise.
14190 (make_tree_vec): Remove macro.
14191 (grow_tree_vec): Likewise.
14193 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14195 * fold-const.c (fold_build1_stat_loc): Adjust.
14196 (fold_build2_stat_loc): Likewise.
14197 (fold_build3_stat_loc): Likewise.
14198 * tree.c (build0_stat): Remove _stat from name.
14199 (build1_stat): Likewise.
14200 (build2_stat): Likewise.
14201 (build3_stat): Likewise.
14202 (build4_stat): Likewise.
14203 (build5_stat): Likewise.
14204 * tree.h (build1_loc): Remove macro, and rename _stat function
14206 (build2_loc): Likewise.
14207 (build3_loc): Likewise.
14208 (build4_loc): Likewise.
14209 (build5_loc): Likewise.
14211 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14213 * tree.c (make_int_cst_stat): Remove _stat from name.
14214 * tree.h (make_int_cst_stat): Adjust prototype.
14215 (make_int_cst): Remove macro.
14217 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14219 * tree.c (make_tre_binfo_stat): Remove _stat from name.
14220 * tree.h (make_tree_binfo_stat): Adjust prototype.
14221 (make_tree_binfo): Remove.
14223 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14225 * tree.c (copy_node_stat): Rename to copy_node.
14226 (build_distinct_type_copy): Adjust.
14227 * tree.h (copy_node_stat): Adjust prototype.
14228 (copy_node): Remove macro.
14230 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14232 * tree.c (make_node_stat): rename to make_node.
14233 (build_tree_list_stat): Adjust.
14234 (build0_stat): Likewise.
14235 (build2_stat): Likewise.
14236 (build3_stat): Likewise.
14237 (build4_stat): Likewise.
14238 (build5_stat): Likewise.
14239 (build_decl_stat): Likewise.
14240 * tree.h (make_node_stat): Adjust prototype.
14241 (make_node): remove macro.
14243 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
14245 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
14246 (PPC_FEATURE2_SCV): Likewise.
14247 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
14249 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14251 * config/aarch64/aarch64.c
14252 (aarch64_internal_mov_immediate): Add new special pattern.
14253 * config/aarch64/aarch64.md (*movdi_aarch64):
14254 Add reg/32bit const mov case.
14256 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14257 Richard Sandiford <richard.sandiford@linaro.org>
14259 * config/aarch64/aarch64.md (mov<mode>): Generalize.
14260 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14261 Add integer and movi cases.
14262 (movi-split-hf-df-sf split, fp16): New.
14263 (enabled): Added TARGET_FP_F16INST.
14264 * config/aarch64/iterators.md (GPF_HF): New.
14265 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
14267 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14269 * config/aarch64/aarch64.c
14270 (aarch64_simd_container_mode): Add prototype.
14271 (aarch64_expand_mov_immediate): Add HI support.
14272 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
14273 (aarch64_can_const_movi_rtx_p): New.
14274 (aarch64_preferred_reload_class):
14275 Remove restrictions of using FP registers for certain SIMD operations.
14276 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
14277 (aarch64_valid_floating_const): Add integer move validation.
14278 (aarch64_simd_imm_scalar_p): Remove.
14279 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
14280 (aarch64_legitimate_constant_p): Expand list of supported cases.
14281 * config/aarch64/aarch64-protos.h
14282 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
14283 (aarch64_reinterpret_float_as_int): New.
14284 (aarch64_simd_imm_scalar_p): Remove.
14285 * config/aarch64/constraints.md (Uvi): New.
14286 (Dd): Split into Ds and new Dd.
14287 * config/aarch64/aarch64.md (*movsi_aarch64):
14289 (*movdi_aarch64): Add SIMD mov case.
14291 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14293 * tree-predcom.c: (struct chain): Handle store-store chain in which
14294 stores for elimination only store loop invariant values.
14295 (execute_pred_commoning_chain): Ditto.
14296 (prepare_initializers_chain_store_elim): Ditto.
14297 (prepare_finalizers): Ditto.
14298 (is_inv_store_elimination_chain): New function.
14299 (initialize_root_vars_store_elim_1): New function.
14301 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14303 * tree-predcom.c: Revise general description of the pass.
14304 (enum chain_type): New enum type for store elimination.
14305 (struct chain): New field supporting store elimination.
14306 (struct component): Ditto.
14307 (dump_chain): Dump store-stores chain.
14308 (release_chain): Release resources.
14309 (split_data_refs_to_components): Compute and create component
14310 contains only stores for elimination.
14311 (get_chain_last_ref_at): New function.
14312 (make_invariant_chain): Initialization.
14313 (make_rooted_chain): Specify chain type in parameter and record it.
14314 (add_looparound_copies): Skip for store-stores chain.
14315 (determine_roots_comp): Compute type of chain and pass it to
14317 (initialize_root_vars_store_elim_2): New function.
14318 (finalize_eliminated_stores): New function.
14319 (remove_stmt): Handle store for elimination.
14320 (execute_pred_commoning_chain): Execute predictive commoning on
14321 store-store chains.
14322 (determine_unroll_factor): Skip unroll for store-stores chain.
14323 (prepare_initializers_chain_store_elim): New function.
14324 (prepare_initializers_chain): Hanlde store-store chain.
14325 (prepare_finalizers_chain, prepare_finalizers): New function.
14326 (tree_predictive_commoning_loop): Return integer value indicating
14327 if loop is unrolled or lcssa form is corrupted.
14328 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
14330 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14332 * tree-predcom.c (initialize_root): Delete.
14333 (execute_pred_commoning_chain): Initialize root vars and replace
14334 reference of non-combined chain directly, rather than call above
14337 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14339 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
14340 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
14342 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14344 * tree-predcom.c (struct chain): New field init_seq.
14345 (release_chain): Release init_seq.
14346 (prepare_initializers_chain): Record intialization stmts in above
14348 (insert_init_seqs): New function.
14349 (tree_predictive_commoning_loop): Call insert_init_seqs.
14351 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14353 * tree-predcom.c (determine_roots_comp): Skip trivial components.
14355 2017-07-28 Richard Biener <rguenther@suse.de>
14357 * match.pd: Remove superfluous :c.
14358 * genmatch.c (simplify::id): Add member.
14359 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
14361 (current_id): New global.
14362 (dt_node::parent): Move from ...
14363 (dt_operand::parent): ... here. Add for_id member.
14364 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
14365 (decision_tree::find_node): Relax order requirement when
14366 merging DT_TRUE nodes to ones inbetween the current simplify
14367 and the one we try to merge with. Add diagnostic whenever
14368 we need to enforce pattern order by not merging.
14369 (decision_tree::insert): Set current_id.
14370 (decision_tree::print_node): Dump parent node and for_id.
14371 (parser::last_id): Add member.
14372 (parser::push_simplify): Assign unique id.
14373 (parser::parser): Initialize last_id.
14375 2017-07-28 Martin Liska <mliska@suse.cz>
14378 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
14379 gimple_build_debug_bind.
14381 2017-07-28 Richard Biener <rguenther@suse.de>
14383 PR tree-optimization/81502
14384 * match.pd: Add pattern combining BIT_INSERT_EXPR with
14386 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
14388 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
14389 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
14390 for BIT_FIELD_REF args.
14391 * fold-const.c (make_bit_field_ref): Likewise.
14392 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
14394 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14397 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
14398 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
14399 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
14400 Or it into SANITIZER_UNDEFINED.
14401 * ubsan.c: Include gimple-fold.h and varasm.h.
14402 (ubsan_expand_ptr_ifn): New function.
14403 (instrument_pointer_overflow): New function.
14404 (maybe_instrument_pointer_overflow): New function.
14405 (instrument_object_size): Formatting fix.
14406 (pass_ubsan::execute): Call instrument_pointer_overflow
14407 and maybe_instrument_pointer_overflow.
14408 * internal-fn.c (expand_UBSAN_PTR): New function.
14409 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
14410 * sanitizer.def (__ubsan_handle_pointer_overflow,
14411 __ubsan_handle_pointer_overflow_abort): New builtins.
14412 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
14413 * internal-fn.def (UBSAN_PTR): New internal function.
14414 * opts.c (sanitizer_opts): Add pointer-overflow.
14415 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
14416 * fold-const.c (build_range_check): Compute pointer range check in
14417 integral type if pointer arithmetics would be needed. Formatting
14420 2017-07-28 Martin Liska <mliska@suse.cz>
14423 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
14424 parameters that are of a variable-length.
14426 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
14428 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
14430 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
14431 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
14432 (CRT_CALL_STATIC_FUNCTION): Likewise.
14433 (ASM_DEFAULT_SPEC): New define.
14434 (ASM_SPEC32): Likewise.
14435 (ASM_SPEC64): Likewise.
14436 (ASM_SPEC_COMMON): Likewise.
14437 (ASM_SPEC): Likewise.
14438 (INVALID_64BIT): Likewise.
14439 (LINK_OS_DEFAULT_SPEC): Likewise.
14440 (LINK_OS_SPEC32): Likewise.
14441 (LINK_OS_SPEC64): Likewise.
14442 (POWERPC_LINUX): Likewise.
14443 (PTRDIFF_TYPE): Likewise.
14444 (RESTORE_FP_PREFIX): Likewise.
14445 (RESTORE_FP_SUFFIX): Likewise.
14446 (SAVE_FP_PREFIX): Likewise.
14447 (SAVE_FP_SUFFIX): Likewise.
14448 (SIZE_TYPE): Likewise.
14449 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
14450 (TARGET_64BIT): Likewise.
14451 (TARGET_64BIT): Likewise.
14452 (TARGET_AIX): Likewise.
14453 (WCHAR_TYPE_SIZE): Likewise.
14454 (WCHAR_TYPE): Undefine.
14455 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
14456 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
14457 (CPP_OS_RTEMS_SPEC): Delete.
14458 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
14459 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
14461 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
14463 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14465 PR tree-optimization/81578
14466 * tree-parloops.c (build_new_reduction): Bail out if
14467 reduction_code isn't one of the standard OpenMP reductions.
14468 Move the details printing after that decision.
14470 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14472 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
14473 related to reload_in_progress.
14474 (splat_input_operand): Likewise.
14475 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
14477 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
14479 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
14480 (TARGET_INSTANTIATE_DECLS): Likewise.
14481 (legitimate_indexed_address_p): Delete reload_in_progress code.
14482 (rs6000_debug_legitimate_address_p): Likewise.
14483 (rs6000_eliminate_indexed_memrefs): Likewise.
14484 (rs6000_emit_le_vsx_store): Likewise.
14485 (rs6000_emit_move_si_sf_subreg): Likewise.
14486 (rs6000_emit_move): Likewise.
14487 (register_to_reg_type): Likewise.
14488 (rs6000_pre_atomic_barrier): Likewise.
14489 (rs6000_machopic_legitimize_pic_address): Likewise.
14490 (rs6000_allocate_stack_temp): Likewise.
14491 (rs6000_address_for_fpconvert): Likewise.
14492 (rs6000_address_for_altivec): Likewise.
14493 (rs6000_secondary_memory_needed_rtx): Delete function.
14494 (rs6000_check_sdmode): Likewise.
14495 (rs6000_alloc_sdmode_stack_slot): Likewise.
14496 (rs6000_instantiate_decls): Likewise.
14497 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
14498 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
14499 Delete reload_in_progress.
14500 (*vec_reload_and_plus_<mptrsize>): Likewise.
14501 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
14502 (vsx_div_v2di): Likewise.
14503 (vsx_udiv_v2di): Likewise.
14505 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14507 * config/rs6000/rs6000.opt (mlra): Replace with stub.
14508 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
14509 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
14510 (rs6000_debug_reg_global): Delete print of LRA status.
14511 (rs6000_option_override_internal): Delete dead LRA related code.
14512 (rs6000_lra_p): Delete function.
14513 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
14515 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
14517 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
14518 * config/riscv/rtems.h: New file.
14520 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14521 Sudakshina Das <sudi.das@arm.com>
14523 * config/aarch64/aarch64.md
14524 (define_split for and<mode>3nr_compare): Move
14525 non aarch64_logical_operand to a register.
14526 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
14527 register immediate operand to a register.
14528 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
14530 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14532 PR middle-end/81564
14533 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
14535 2017-07-27 Richard Biener <rguenther@suse.de>
14537 PR tree-optimization/81573
14538 PR tree-optimization/81494
14539 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
14540 multi defuse cycle case.
14542 2017-07-27 Richard Biener <rguenther@suse.de>
14544 PR tree-optimization/81571
14545 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
14548 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
14550 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
14551 earlier and only if MASK_FPU is set. Adjust formatting.
14553 2017-07-27 Martin Liska <mliska@suse.cz>
14555 * opt-functions.awk: Add validation of value of Init.
14556 * optc-gen.awk: Pass new argument.
14558 2017-07-27 Martin Liska <mliska@suse.cz>
14560 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
14561 Fix wrong condition.
14563 2017-07-27 Martin Liska <mliska@suse.cz>
14565 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
14566 BBs and edges seen by autoFDO.
14568 2017-07-27 Richard Biener <rguenther@suse.de>
14570 PR tree-optimization/81502
14571 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
14572 with incompatible but same sized type.
14573 (execute_update_addresses_taken): Likewise.
14575 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
14577 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
14578 flag_tree_loop_vectorize rather than flag_tree_vectorize.
14580 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14583 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
14584 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
14585 Change s_operand to memory_operand.
14587 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
14589 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
14590 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
14591 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
14592 Emit instructions rather than returning an expression. Handle TFmode
14593 and KFmode by casting to TImode.
14594 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
14595 (rs6000_emit_le_vsx_store): Likewise.
14596 * config/rs6000/vsx.md (VSX_TI): New iterator.
14597 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
14598 (*vsx_le_undo_permute_<mode>): Likewise.
14599 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
14600 emit the split sequence.
14601 (*vsx_le_perm_store_<mode>): Likewise.
14603 2017-07-27 Jakub Jelinek <jakub@redhat.com>
14605 PR tree-optimization/81555
14606 PR tree-optimization/81556
14607 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
14608 if true, force CHANGED for the recursive invocation.
14609 (reassociate_bb): Remember original length of ops array, pass
14610 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
14612 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
14613 attributes for noipa attribute. For naked attribute use
14614 lookup_attribute first before lookup_attribute_spec.
14615 * final.c (rest_of_handle_final): Disable IPA RA for functions with
14617 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
14618 for functions with noipa attribute.
14619 (cgraph_externally_visible_p): Return true for functions with noipa
14621 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
14622 for functions with noipa attribute.
14623 * doc/extend.texi: Document noipa function attribute.
14624 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
14625 also for functions with noipa attribute.
14626 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
14628 2017-07-26 Andrew Pinski <apinski@cavium.com>
14630 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
14631 vec_unalign_load_cost and vec_unalign_store_cost.
14633 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14635 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
14636 -mvsx-small-integer option.
14637 (ISA_3_0_MASKS_IEEE): Likewise.
14638 (OTHER_VSX_VECTOR_MASKS): Likewise.
14639 (POWERPC_MASKS): Likewise.
14640 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
14641 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
14642 code, only testing for DImode being allowed in non-VSX floating
14644 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
14645 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
14647 (rs6000_option_override_internal): Delete -mvsx-small-integer.
14648 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
14649 TARGET_P8_VECTOR test.
14650 (rs6000_secondary_reload_simple_move): Likewise.
14651 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
14652 since TARGET_P9_VECTOR was already tested.
14653 (rs6000_opt_masks): Remove -mvsx-small-integer.
14654 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
14655 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14657 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
14658 test for TARGET_VEXTRACTUB was used, and that uses
14660 (p9 extract splitter): Likewise.
14661 (vsx_extract_<mode>_di_p9): Likewise.
14662 (vsx_extract_<mode>_store_p9): Likewise.
14663 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
14664 for TARGET_P9_VECTOR was used. Delete code that is now dead with
14665 the elimination of TARGET_VSX_SMALL_INTEGER.
14666 (vsx_extract_<mode>_p8): Likewise.
14667 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
14668 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
14669 (vsx_set_<mode>_p9): Likewise.
14670 (vsx_set_v4sf_p9): Likewise.
14671 (vsx_set_v4sf_p9_zero): Likewise.
14672 (vsx_insert_extract_v4sf_p9): Likewise.
14673 (vsx_insert_extract_v4sf_p9_2): Likewise.
14674 * config/rs6000/rs6000.md (sign extend splitter): Change
14675 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
14676 (floatsi<mode>2_lfiwax_mem): Likewise.
14677 (floatunssi<mode>2_lfiwzx_mem): Likewise.
14678 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
14679 since a test for TARGET_P9_VECTOR was used.
14680 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14681 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
14682 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14683 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
14684 TARGET_P8_VECTOR test.
14685 (fix_trunc<mode>si2_stfiwx): Likewise.
14686 (fix_trunc<mode>si2_internal): Likewise.
14687 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
14688 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14690 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
14691 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
14692 TARGET_P8_VECTOR test.
14693 (fixuns_trunc<mode>si2_stfiwx): Likewise.
14694 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
14695 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14697 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
14698 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
14699 since a test for TARGET_P9_VECTOR was used.
14700 (splitter for loading small constants): Likewise.
14702 2017-07-26 Andrew Pinski <apinski@cavium.com>
14704 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
14707 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
14710 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
14711 (fp_valid_at): Likewise.
14713 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
14715 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
14716 (qdf24xx_addrcost_table): Likewise.
14717 (cortexa57_tunings): Update to use generic_branch_cost.
14718 (cortexa72_tunings): Likewise.
14719 (cortexa73_tunings): Likewise.
14720 (qdf24xx_tunings): Likewise.
14722 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
14724 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
14725 (thunderx2t99_branch_cost): Likewise.
14726 (cortexa35_tunings): Update to use generic_branch_cost.
14727 (cortexa53_tunings): Likewise.
14728 (cortexa57_tunings): Likewise.
14729 (cortexa72_tunings): Likewise.
14730 (cortexa73_tunings): Likewise.
14731 (thunderx2t99_tunings): Likewise.
14733 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
14735 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
14736 (sparc_option_override): Honour MASK_FSMULD.
14737 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
14738 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
14739 * config/sparc/sparc.opt (mfsmuld): New option.
14740 * doc/invoke.texi (mfsmuld): Document option.
14742 2017-07-26 Marek Polacek <polacek@redhat.com>
14744 PR middle-end/70992
14745 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
14747 2017-07-26 Richard Biener <rguenther@suse.de>
14749 * gimple-match-head.c (do_valueize): Return OP if valueize
14751 (get_def): New helper to get at the def stmt of a SSA name
14752 if valueize allows.
14753 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
14754 do_valueize to get at the def stmt.
14755 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
14757 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
14759 PR middle-end/46932
14760 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
14762 2017-07-26 Martin Liska <mliska@suse.cz>
14765 * function.c (expand_function_start): Make expansion of
14766 nonlocal_goto_save_area after parm_birth_insn.
14768 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
14770 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
14771 from all CPU target flags enable members.
14773 2017-07-26 Richard Biener <rguenther@suse.de>
14775 * genmatch.c (dt_simplify::gen): Make iterator vars const.
14776 (decision_tree::gen): Make 'type' const.
14777 (write_predicate): Likewise.
14779 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
14781 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
14782 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
14783 (rs6000_option_override_internal): Likewise.
14784 (rs6000_expand_vector_set): Likewise.
14785 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
14786 (TARGET_UPPER_REGS_SF): Likewise.
14787 (TARGET_UPPER_REGS_DI): Likewise.
14788 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
14789 (TARGET_DIRECT_MOVE_64BIT): Likewise.
14790 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
14791 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14792 (Splitters for DI constants in Altivec registers): Likewise.
14793 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
14794 (vsx_set_v4sf_p9): Likewise.
14795 (vsx_set_v4sf_p9_zero): Likewise.
14796 (vsx_insert_extract_v4sf_p9): Likewise.
14797 (vsx_insert_extract_v4sf_p9_2): Likewise.
14799 2017-07-25 Carl Love <cel@us.ibm.com>
14801 * doc/extend.texi: Update the built-in documentation file for the
14802 existing built-in functions
14803 vector signed char vec_cnttz (vector signed char);
14804 vector unsigned char vec_cnttz (vector unsigned char);
14805 vector signed short vec_cnttz (vector signed short);
14806 vector unsigned short vec_cnttz (vector unsigned short);
14807 vector signed int vec_cnttz (vector signed int);
14808 vector unsigned int vec_cnttz (vector unsigned int);
14809 vector signed long long vec_cnttz (vector signed long long);
14810 vector unsigned long long vec_cnttz (vector unsigned long long);
14812 2017-07-25 Andrew Pinski <apinski@cavium.com>
14814 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
14815 accesses where the use is for the first operand of a BIT_INSERT.
14817 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
14820 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
14821 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
14823 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
14825 * config/i386/gstabs.h: Delete.
14826 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
14828 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
14830 * config/i386/i386.c (ix86_decompose_address): Do not check for
14831 register RTX when looking at index_reg or base_reg.
14832 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
14834 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
14836 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
14837 to update EH info here.
14839 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
14841 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
14843 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
14845 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
14847 2017-07-25 Torsten Duwe <duwe@suse.de>
14849 * common.opt: Introduce -fpatchable-function-entry
14850 command line option, and its variables function_entry_patch_area_size
14851 and function_entry_patch_area_start.
14852 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
14853 including a two-value parser.
14854 * target.def (print_patchable_function_entry): New target hook.
14855 * targhooks.h (default_print_patchable_function_entry): New function.
14856 * targhooks.c (default_print_patchable_function_entry): Likewise.
14857 * toplev.c (process_options): Switch off IPA-RA if
14858 patchable function entries are being generated.
14859 * varasm.c (assemble_start_function): Look at the
14860 patchable-function-entry command line switch and current
14861 function attributes and maybe generate NOP instructions by
14862 calling the print_patchable_function_entry hook.
14863 * doc/extend.texi: Document patchable_function_entry attribute.
14864 * doc/invoke.texi: Document -fpatchable_function_entry
14865 command line option.
14866 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
14868 * doc/tm.texi: Re-generate.
14870 2017-07-25 Jakub Jelinek <jakub@redhat.com>
14873 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
14874 TARGET_AVX512DQ rather than TARGET_AVX512BW.
14876 2017-07-25 Tamar Christina <tamar.christina@arm.com>
14878 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
14880 2017-07-25 Richard Biener <rguenther@suse.de>
14882 PR tree-optimization/81455
14883 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
14884 not walk in cycles when looking for guards.
14886 2017-07-25 Richard Biener <rguenther@suse.de>
14888 PR tree-optimization/81529
14889 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
14890 when optimizing backedge uses.
14892 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
14894 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
14896 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
14897 to dl_section_ref. On AIX, append an expression to subtract
14898 the size of the section length to dl_section_ref.
14900 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
14902 * configure.ac: If any of the config.* scripts fail, exit 1.
14903 * configure: Regenerate.
14905 2017-07-25 Richard Biener <rguenther@suse.de>
14907 PR middle-end/81546
14908 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
14909 of immediate uses, be more verbose on errors.
14911 2017-07-25 Richard Biener <rguenther@suse.de>
14913 PR tree-optimization/81510
14914 * tree-vect-loop.c (vect_is_simple_reduction): When the
14915 reduction stmt is not inside the loop bail out.
14917 2017-07-25 Richard Biener <rguenther@suse.de>
14919 PR tree-optimization/81303
14920 * tree-vect-loop-manip.c (vect_loop_versioning): Build
14921 profitability check against LOOP_VINFO_NITERSM1.
14923 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
14925 * domwalk.c (cmp_bb_postorder): Simplify.
14926 (sort_bbs_postorder): New function. Use it...
14927 (dom_walker::walk): ...here to optimize common cases.
14929 2017-07-25 Martin Liska <mliska@suse.cz>
14932 * ipa-visibility.c (function_and_variable_visibility): Make the
14933 redirection just on target that supports aliasing.
14934 Fix GNU coding style.
14936 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
14939 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
14941 * config/arm/rtems.h: Likewise.
14942 * config/bfin/rtems.h: Likewise.
14943 * config/i386/rtemself.h: Likewise.
14944 * config/lm32/rtems.h: Likewise.
14945 * config/m32c/rtems.h: Likewise.
14946 * config/m68k/rtemself.h: Likewise.
14947 * config/microblaze/rtems.h: Likewise.
14948 * config/mips/rtems.h: Likewise.
14949 * config/moxie/rtems.h: Likewise.
14950 * config/nios2/rtems.h: Likewise.
14951 * config/powerpcspe/rtems.h: Likewise.
14952 * config/rs6000/rtems.h: Likewise.
14953 * config/rtems.h: Likewise.
14954 * config/sh/rtems.h: Likewise.
14955 * config/sh/rtemself.h: Likewise.
14956 * config/sparc/rtemself.h: Likewise.
14958 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
14961 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
14962 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
14963 * tree-ssa-structalias.c (alias_get_name): Same.
14965 2017-07-25 Bin Cheng <bin.cheng@arm.com>
14968 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
14969 instructions if no du chain is found.
14971 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
14973 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
14975 2017-07-25 Richard Biener <rguenther@suse.de>
14977 PR middle-end/81505
14978 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
14981 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
14983 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
14984 upper-regs options.
14985 (ISA_2_7_MASKS_SERVER): Likewise.
14986 (ISA_3_0_MASKS_IEEE): Likewise.
14987 (OTHER_P8_VECTOR_MASKS): Likewise.
14988 (OTHER_VSX_VECTOR_MASKS): Likewise.
14989 (POWERPC_MASKS): Likewise.
14990 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
14991 duplicate list of options.
14992 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
14993 explicit -mupper-regs options.
14994 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
14995 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
14996 alias for -mupper-regs-df.
14997 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
14998 (rs6000_init_hard_regno_mode_ok): Likewise.
14999 (rs6000_option_override_internal): Likewise.
15000 (rs6000_opt_masks): Likewise.
15001 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
15002 options in terms of whether -mvsx or -mpower8-vector was used.
15003 (TARGET_UPPER_REGS_DI): Likewise.
15004 (TARGET_UPPER_REGS_SF): Likewise.
15005 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
15006 -mupper-regs-* options.
15008 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15010 * passes.c (emergency_dump_function): Print some empty lines and a
15011 header before the RTL dump.
15013 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15015 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
15017 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
15020 * config/aarch64/aarch64.c (aarch64_classify_symbol):
15021 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
15023 2017-07-24 Carl Love <cel@us.ibm.com>
15025 * config/rs6000/rs6000-c.c: Add support for built-in functions
15026 vector float vec_extract_fp32_from_shorth (vector unsigned short);
15027 vector float vec_extract_fp32_from_shortl (vector unsigned short);
15028 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
15029 vec_extract_fp_from_shortl): Add defines for the two builtins.
15030 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
15031 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
15033 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
15034 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
15035 * doc/extend.texi: Update the built-in documentation file for the
15036 new built-in function.
15038 2017-07-24 Jakub Jelinek <jakub@redhat.com>
15041 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
15043 * doc/generic.texi: Likewise.
15044 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
15045 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
15047 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
15049 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
15050 (aarch64_mls_elt_merge<mode>): Likewise.
15052 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
15054 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
15055 having __cxa_atexit.
15057 2017-07-23 Michael Collison <michael.collison@arm.com>
15059 * config/arm/arm.c (arm_option_override): Deprecate
15060 use of -mstructure-size-boundary.
15061 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
15062 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
15064 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15067 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
15068 Reduce cost estimate for direct moves.
15070 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
15073 * config/i386/i386.c (ix86_option_override_internal): Disable
15074 BMI, BMI2 and TBM instructions for -m16.
15076 2017-07-21 Carl Love <cel@us.ibm.com>
15078 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15079 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15080 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15081 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15082 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15083 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15084 VMULOSW): New enum "unspec" values.
15085 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15086 altivec_vmulosw): New patterns.
15087 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15088 VMULOSW): Add definitions.
15090 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
15092 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
15093 (qdf24xx): Likewise.
15094 * config/aarch64/aarch64-options-extensions.def (rdma); New.
15095 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
15096 (AARCH64_FL_V8_1): Renumber.
15097 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
15098 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
15099 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
15100 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
15101 rdma to feature modifiers list.
15103 2017-07-21 Yury Gribov <tetra2005@gmail.com>
15105 PR middle-end/56727
15106 * ipa-visibility (function_and_variable_visibility): Convert
15107 recursive PLT call to direct call if appropriate.
15109 2017-07-21 Andrew Pinski <apinski@cavium.com>
15111 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
15112 operand 1 to see if the types precision matches.
15113 * fold-const.c (operand_equal_p): Likewise.
15115 2017-07-21 Richard Biener <rguenther@suse.de>
15117 PR tree-optimization/81303
15118 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
15119 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
15120 (vect_peeling_hash_get_lowest_cost): Adjust.
15121 (vect_enhance_data_refs_alignment): Likewise. Use
15122 vect_get_peeling_costs_all_drs to compute the penalty for no
15123 peeling to match up costs.
15125 2017-07-21 Richard Biener <rguenther@suse.de>
15127 PR tree-optimization/81500
15128 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
15129 we didn't identify a reduction path.
15131 2017-07-21 Tom de Vries <tom@codesourcery.com>
15132 Cesar Philippidis <cesar@codesourcery.com>
15134 PR gcov-profile/81442
15135 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
15138 2017-07-21 Tom de Vries <tom@codesourcery.com>
15141 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
15143 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
15144 nvptx_override_options_after_change.
15146 2017-07-21 Ulrich Drepper <drepper@redhat.com>
15148 * dwarf2out.c (output_file_names): Avoid double testing for
15149 dwarf_version >= 5.
15151 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
15153 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
15155 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15157 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
15159 (try_crossjump_to_edge): Do not punt on partitioned functions.
15161 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15163 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
15164 Put all BBs reachable only via paths crossing cold region to cold
15166 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
15168 2016-07-21 Richard Biener <rguenther@suse.de>
15170 PR tree-optimization/81303
15171 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
15172 into account prologue and epilogue iterations when raising
15173 min_profitable_iters to sth at least covering one vector iteration.
15175 2017-07-21 Tamar Christina <tamar.christina@arm.com>
15177 * config/arm/arm.c (arm_test_cpu_arch_dat):
15180 2017-07-20 Nathan Sidwell <nathan@acm.org>
15182 Remove TYPE_METHODS.
15183 * tree.h (TYPE_METHODS): Delete.
15184 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
15185 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
15186 (dbxout_type_methods): Scan TYPE_FIELDS.
15187 (dbxout_type): Don't check TYPE_METHODS here.
15188 * function.c (use_register_for_decl): Always ignore register for
15189 class types when not optimizing.
15190 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
15191 * tree.c (free_lang_data_in_type): Stitch out member functions and
15192 templates from TYPE_FIELDS.
15193 (build_distinct_type_copy, verify_type_variant,
15194 verify_type): Member fns are on TYPE_FIELDS.
15195 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
15196 * tree-pretty-print.c (dump_generic_node): Likewise.
15198 2017-07-20 Jakub Jelinek <jakub@redhat.com>
15201 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
15202 V2TImode and V4TImode.
15203 (ix86_expand_vector_extract): Likewise.
15204 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
15205 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
15206 (ssescalarmode): Handle V4TImode and V2TImode.
15207 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
15208 (*vec_extractv2ti, *vec_extractv4ti): New insns.
15209 (VEXTRACTI128_MODE): New mode iterator.
15210 (splitter for *vec_extractv?ti first element): New.
15211 (VEC_INIT_MODE): New mode iterator.
15212 (vec_init<mode>): Consolidate 3 expanders into one using
15213 VEC_INIT_MODE mode iterator.
15215 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
15217 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
15218 non_spilled_static_chain_regno_p.
15220 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
15222 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
15224 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
15226 * bb-reorder.c (connect_traces): Allow copying of blocks within
15229 2017-07-20 Richard Biener <rguenther@suse.de>
15231 * gimple.h (gimple_phi_result): Add gphi * overload.
15232 (gimple_phi_result_ptr): Likewise.
15233 (gimple_phi_arg): Likewise. Adjust index assert to only
15234 allow actual argument accesses rather than all slots available
15236 (gimple_phi_arg_def): Add gphi * overload.
15237 * tree-phinodes.c (make_phi_node): Initialize only actual
15239 (resize_phi_node): Clear memory not covered by old node,
15240 do not initialize excess argument slots.
15241 (reserve_phi_args_for_new_edge): Initialize new argument slot
15244 2017-07-20 Bin Cheng <bin.cheng@arm.com>
15246 PR tree-optimization/81388
15248 2016-07-21 Bin Cheng <bin.cheng@arm.com>
15250 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
15251 by removing computation of may_be_zero.
15253 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15254 Tom de Vries <tom@codesourcery.com>
15256 PR middle-end/81030
15257 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
15258 when gimple level profile disagrees with what RTL expander did.
15260 2017-07-20 Richard Biener <rguenther@suse.de>
15262 PR tree-optimization/61171
15263 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
15264 (vect_analyze_stmt): Add slp instance parameter.
15265 (vectorizable_reduction): Likewise.
15266 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
15267 (vect_is_simple_reduction): Deal with chains not detected
15268 as SLP reduction chain, specifically not properly associated
15269 chains containing a mix of plus/minus.
15270 (get_reduction_op): Remove.
15271 (get_initial_defs_for_reduction): Simplify, pass in whether
15272 this is a reduction chain, pass in the SLP node for the PHIs.
15273 (vect_create_epilog_for_reduction): Get the SLP instance as
15275 (vectorizable_reduction): Get the SLP instance as arg.
15276 During analysis remember the SLP node with the PHIs in the
15277 instance. Simplify getting at the vectorized reduction PHIs.
15278 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
15279 through SLP instance.
15280 (vect_slp_analyze_operations): Likewise.
15281 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
15282 (vect_transform_stmt): Likewise.
15284 2017-07-20 Tom de Vries <tom@codesourcery.com>
15286 PR tree-optimization/81489
15287 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
15288 read of phi arg location to before loop that modifies phi.
15290 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15292 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
15295 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15297 PR middle-end/81331
15298 * except.c (execute): Fix ordering issue.
15300 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
15302 PR rtl-optimization/81423
15303 * combine.c (make_compound_operation_int): Don't try to optimize
15304 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
15306 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
15308 PR rtl-optimization/81423
15309 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
15310 with a constant that is -1 in the truncated to mode.
15312 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15314 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
15315 (determine_unlikely_bbs): ... here.
15316 * predict.h (propagate_unlikely_bbs_forward): Declare.
15317 * cfgexpand.c (pass_expand::execute): Use it.
15318 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
15320 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
15321 propagate_unlikely_bbs_forward.
15323 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15325 PR middle-end/81331
15326 * except.c (maybe_add_nop_after_section_switch): New function.
15329 2017-07-19 Tom de Vries <tom@codesourcery.com>
15331 * gimple.h (gimple_phi_set_arg): Make assert more strict.
15333 2017-07-19 Tom de Vries <tom@codesourcery.com>
15335 * gimple.h (gimple_phi_arg): Make assert more strict.
15337 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
15339 * config.gcc (powerpc*-*-*): Add mmintrin.h.
15340 * config/rs6000/mmintrin.h: New file.
15341 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
15343 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15345 PR tree-optimization/81346
15346 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
15348 2017-07-19 Tom de Vries <tom@codesourcery.com>
15350 * config/nvptx/nvptx.md (VECIM): Add V2DI.
15352 2017-07-19 Tom de Vries <tom@codesourcery.com>
15354 * config/nvptx/nvptx-modes.def: Add V2DImode.
15355 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
15356 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
15357 (nvptx_output_mov_insn): Handle lack of mov.b128.
15358 (nvptx_print_operand): Handle 'H' and 'L' codes.
15359 (nvptx_vector_mode_supported): Allow V2DImode.
15360 (nvptx_preferred_simd_mode): New function.
15361 (nvptx_data_alignment): New function.
15362 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
15363 nvptx_preferred_simd_mode.
15364 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
15366 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
15368 2017-07-19 Tom de Vries <tom@codesourcery.com>
15370 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
15371 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
15372 (nvptx_vector_mode_supported): New function. Allow V2SImode.
15373 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
15374 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
15375 (mov<VECIM>_insn): New define_insn.
15376 (define_expand "mov<VECIM>): New define_expand.
15378 2017-07-19 Tom de Vries <tom@codesourcery.com>
15380 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
15382 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15384 PR tree-optimization/81346
15385 * fold-const.h (fold_div_compare, range_check_type): Declare.
15386 * fold-const.c (range_check_type): New function.
15387 (build_range_check): Use range_check_type.
15388 (fold_div_compare): No longer static, rewritten into
15389 a match.pd helper function.
15390 (fold_comparison): Don't call fold_div_compare here.
15391 * match.pd (X / C1 op C2): New optimization using fold_div_compare
15392 as helper function.
15394 2017-07-19 Nathan Sidwell <nathan@acm.org>
15396 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
15397 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
15398 * tree.c (find_decls_types_r, verify_type): Use
15399 TYPE_{MIN,MAX}_VALUE_RAW.
15400 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
15401 (hash_tree): Likewise.
15402 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
15404 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
15407 2017-07-18 Tom de Vries <tom@codesourcery.com>
15409 PR middle-end/81464
15410 * omp-expand.c (expand_omp_for_static_chunk): Handle
15411 equal-argument loop exit phi.
15413 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
15416 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
15417 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
15418 operand 2 predicate.
15419 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
15420 operand 2 predicate.
15421 (ror,rol -> rorx splitters): Use const_int_operand as
15422 operand 2 predicate.
15424 2017-06-18 Richard Biener <rguenther@suse.de>
15426 PR tree-optimization/81410
15427 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
15428 the gap in the ! slp_perm SLP case after each group.
15430 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15432 PR middle-end/81463
15433 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
15436 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15438 PR middle-end/81462
15439 * predict.c (set_even_probabilities): Cleanup; do not affect
15440 probabilities that are already known.
15441 (combine_predictions_for_bb): Call even when count is set.
15443 2017-07-18 Nathan Sidwell <nathan@acm.org>
15445 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
15448 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15451 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
15452 optimization for loop niter analysis.
15454 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
15457 * config/avr/avr.c (avr_optimize_casesi): Don't use
15458 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
15460 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
15462 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
15463 body_cost_vec from _vect_peel_extended_info.
15464 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
15465 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
15468 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15470 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
15472 2017-07-18 Richard Biener <rguenther@suse.de>
15474 PR tree-optimization/80620
15475 PR tree-optimization/81403
15476 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
15477 info when re-using a VN table entry.
15479 2017-07-18 Richard Biener <rguenther@suse.de>
15481 PR tree-optimization/81418
15482 * tree-vect-loop.c (vectorizable_reduction): Properly compute
15483 vectype_in. Verify that with lane-reducing reduction operations
15484 we have a single def-use cycle.
15486 2017-07-17 Carl Love <cel@us.ibm.com>
15488 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
15490 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15491 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15492 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15493 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15494 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15495 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15496 VMULOSW): New enum "unspec" values.
15497 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
15498 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
15499 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15500 altivec_vmulosw): New patterns.
15501 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15502 VMULOSW): Add definitions.
15503 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
15505 * config/alpha/alpha.c: Include predict.h.
15507 2017-07-17 Yury Gribov <tetra2005@gmail.com>
15509 * tree-vrp.c (compare_assert_loc): Fix comparison function
15510 to return predictable results.
15512 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15514 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
15516 (subdi3): Likewise.
15517 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
15518 * doc/invoke.texi (mexpand-adddi): Update text.
15520 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15522 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
15523 that also clobbers the CC register. The old expand code is moved
15525 (*arc_clzsi2): ... here.
15526 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
15527 the CC register. The old expand code is moved to ...
15528 (arc_ctzsi2): ... here.
15530 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15532 * config/arc/arc.opt (mindexed-loads): Use initial value
15533 TARGET_INDEXED_LOADS_DEFAULT.
15534 (mauto-modify-reg): Use initial value
15535 TARGET_AUTO_MODIFY_REG_DEFAULT.
15536 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
15537 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
15538 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
15539 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
15541 2017-07-17 Martin Liska <mliska@suse.cz>
15544 * opts.c (finish_options): Do not allow -fgnu-tm
15545 w/ -fsanitize={kernel-,}address. Say sorry.
15547 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15550 * tree-loop-distribution.c (classify_partition): Only assert on
15551 numer of iterations.
15552 (merge_dep_scc_partitions): Delete prameter. Update function call.
15553 (distribute_loop): Remove code handling loop with unknown niters.
15554 (pass_loop_distribution::execute): Skip loop with unknown niters.
15556 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15559 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
15560 function sort_partitions_by_post_order.
15562 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15564 PR tree-optimization/81374
15565 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
15566 the max index of basic blocks, rather than number of basic blocks.
15568 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15570 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
15572 (arc_legitimate_pic_operand_p): Likewise.
15573 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
15575 (arc_needs_pcl_p): Likewise.
15576 (arc_legitimate_pc_offset_p): Likewise.
15577 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
15578 function is also used in constrains.md.
15579 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
15580 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
15581 PLUS. Only return true/false in known cases, otherwise assert.
15582 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
15583 is already called in arc_legitimate_constant_p.
15584 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
15586 (LEGITIMATE_PIC_OPERAND_P): Use
15587 arc_raw_symbolic_reference_mentioned_p function.
15588 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
15593 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15594 Andrew Burgess <andrew.burgess@embecosm.com>
15596 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
15597 (arc_return_address_register): New function.
15598 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
15599 (arc_handle_fndecl_attribute): Add naked attribute.
15600 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
15601 (TARGET_WARN_FUNC_RETURN): Likewise.
15602 (arc_allocate_stack_slots_for_args): New function.
15603 (arc_warn_func_return): Likewise.
15604 (machine_function): Change type fn_type.
15605 (arc_compute_function_type): Consider new naked function type,
15606 change function return type.
15607 (arc_must_save_register): Adapt to handle new
15608 arc_compute_function_type's return type.
15609 (arc_expand_prologue): Likewise.
15610 (arc_expand_epilogue): Likewise.
15611 (arc_return_address_regs): Delete.
15612 (arc_return_address_register): New function.
15613 (arc_epilogue_uses): Use above function.
15614 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
15615 (arc_function_type): Change encoding, add naked type.
15616 (ARC_INTERRUPT_P): Change to handle the new encoding.
15617 (ARC_FAST_INTERRUPT_P): Likewise.
15618 (ARC_NORMAL_P): Define.
15619 (ARC_NAKED_P): Likewise.
15620 (arc_compute_function_type): Delete prototype.
15621 * config/arc/arc.md (in_ret_delay_slot): Use
15622 arc_return_address_register function.
15623 (simple_return): Likewise.
15624 (p_return_i): Likewise.
15626 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15628 PR tree-optimization/81428
15629 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
15630 can't be built for those types.
15632 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
15634 Remove stuff dead since r239246.
15636 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
15637 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
15638 (avr_inform_devices): Remove dead stuff.
15640 2017-07-17 Tamar Christina <tamar.christina@arm.com>
15642 * config/arm/arm_neon.h: Fix softp typo.
15644 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15646 PR tree-optimization/81365
15647 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
15648 aggregate moves onto bb predecessor edges, make sure there are no
15649 loads that could alias the lhs in between the start of bb and the
15652 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
15655 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
15656 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
15657 [LSHIFTRT, outer_code = TRUNCATE]: Same.
15659 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15661 PR tree-optimization/81396
15662 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
15663 (init_symbolic_number): Initialize it to 1.
15664 (perform_symbolic_merge): Add n_ops from both operands into the new
15666 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
15667 without base_addr as useless if they need more than one operation.
15668 (bswap_replace): Handle !bswap case for NULL base_addr.
15670 2017-07-17 Tom de Vries <tom@codesourcery.com>
15673 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
15676 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
15678 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
15679 conditional builtin define __FIX_LEON3FT_B2BST.
15681 2017-07-17 Daniel Cederman <cederman@gaisler.com>
15683 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
15684 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
15687 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
15689 PR rtl-optimization/81424
15690 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
15691 to remove potential trapping from operands if -fnon-call-exceptions.
15693 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15695 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
15696 profile_proability for scalling.
15697 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
15699 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15701 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
15703 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15705 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
15706 fixpoint arithmetics.
15708 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15710 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
15711 fixpoint arithmetics.
15713 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15715 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
15716 fixpoint arithmetics.
15718 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15720 * profile-count.h (profile_probability::from_reg_br_prob_note,
15721 profile_probability::to_reg_br_prob_note): New functions.
15722 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
15723 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
15724 * predict.c (probability_reliable_p): Update.
15725 (edge_probability_reliable_p): Update.
15726 (br_prob_note_reliable_p): Update.
15727 (invert_br_probabilities): Update.
15728 (add_reg_br_prob_note): New function.
15729 (combine_predictions_for_insn): Update.
15730 * asan.c (asan_clear_shadow): Update.
15731 * cfgbuild.c (compute_outgoing_frequencies): Update.
15732 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
15733 (update_br_prob_note): Update.
15734 (rtl_verify_edges): Update.
15735 (purge_dead_edges): Update.
15736 (fixup_reorder_chain): Update.
15737 * emit-rtl.c (try_split): Update.
15738 * ifcvt.c (cond_exec_process_insns): Update.
15739 (cond_exec_process_if_block): Update.
15740 (dead_or_predicable): Update.
15741 * internal-fn.c (expand_addsub_overflow): Update.
15742 (expand_neg_overflow): Update.
15743 (expand_mul_overflow): Update.
15744 * loop-doloop.c (doloop_modify): Update.
15745 * loop-unroll.c (compare_and_jump_seq): Update.
15746 * optabs.c (emit_cmp_and_jump_insn_1): Update.
15747 * predict.h: Update.
15748 * reorg.c (mostly_true_jump): Update.
15750 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
15751 * config/alpha/alpha.c (emit_unlikely_jump): Update.
15752 * config/arc/arc.c: (emit_unlikely_jump): Update.
15753 * config/arm/arm.c: (emit_unlikely_jump): Update.
15754 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
15755 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
15756 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
15757 (ix86_print_operand): Update.
15758 (ix86_split_fp_branch): Update.
15759 (predict_jump): Update.
15760 * config/ia64/ia64.c (ia64_print_operand): Update.
15761 * config/mmix/mmix.c (mmix_print_operand): Update.
15762 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
15763 (rs6000_expand_split_stack_prologue): Update.
15764 * config/rs6000/rs6000.c: Update.
15765 * config/s390/s390.c (s390_expand_vec_strlen): Update.
15766 (s390_expand_vec_movstr): Update.
15767 (s390_expand_cs_tdsi): Update.
15768 (s390_expand_split_stack_prologue): Update.
15769 * config/sh/sh.c (sh_print_operand): Update.
15770 (expand_cbranchsi4): Update.
15771 (expand_cbranchdi4): Update.
15772 * config/sparc/sparc.c (output_v9branch): Update.
15773 * config/spu/spu.c (get_branch_target): Update.
15774 (ea_load_store_inline): Update.
15775 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
15776 * config/tilepro/tilepro.c: Update.
15778 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
15780 * gimplify.c (mostly_copy_tree_r): Revert latest change.
15781 (gimplify_save_expr): Likewise.
15783 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15785 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
15787 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15789 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
15791 * timevar.def (TV_IPA_FNSUMMARY): Define.
15793 2017-07-16 Daniel Cederman <cederman@gaisler.com>
15795 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
15796 to back store errata sensitive sequence from being generated.
15797 (sqrtdf2_fix): Likewise.
15799 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15801 * tree-ssa-threadupdate.c (compute_path_counts,
15802 update_joiner_offpath_counts): Use profile_probability.
15804 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
15807 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15809 * config/arm/arm-c.c (arm_cpu_builtins): Define
15810 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
15812 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
15814 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15815 array entries to represent __ieee128 versions of the
15816 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
15817 scalar_extract_sig, and scalar_insert_exp built-in functions.
15818 (altivec_resolve_overloaded_builtin): Add special case handling
15819 for the __builtin_scalar_insert_exp function, as represented by
15820 the P9V_BUILTIN_VEC_VSIEDP constant.
15821 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
15822 exponent support for __ieee128 argument.
15823 (VSESQP): Add scalar extract signature support for __ieee128
15825 (VSTDCNQP): Add scalar test negative support for __ieee128
15827 (VSIEQP): Add scalar insert exponent support for __int128 argument
15828 with __ieee128 result.
15829 (VSIEQPF): Add scalar insert exponent support for __ieee128
15830 argument with __ieee128 result.
15831 (VSTDCQP): Add scalar test data class support for __ieee128
15833 (VSTDCNQP): Add overload support for scalar test negative with
15834 __ieee128 argument.
15835 (VSTDCQP): Add overload support for scalar test data class
15836 __ieee128 argument.
15837 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
15838 UNSPEC_VSX_SXSIGDP.
15839 (UNSPEC_VSX_SIEXPQP): New constant.
15840 (xsxexpqp): New insn for VSX scalar extract exponent quad
15842 (xsxsigqp): New insn for VSX scalar extract significand quad
15844 (xsiexpqpf): New insn for VSX scalar insert exponent quad
15845 precision with floating point argument.
15846 (xststdcqp): New expand for VSX scalar test data class quad
15848 (xststdcnegqp): New expand for VSX scalar test negative quad
15850 (xststdcqp): New insn to match expansions for VSX scalar test data
15851 class quad precision and VSX scalar test negative quad precision.
15852 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
15853 special case operand checking to enforce that second operand of
15854 VSX scalar test data class with quad precision argument is a 7-bit
15856 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
15857 prototypes and descriptions of __ieee128 versions of
15858 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
15859 scalar_test_data_class, and scalar_test_neg built-in functions.
15861 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15863 PR tree-optimization/81162
15864 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
15865 replace a negate with an add.
15867 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
15869 * doc/invoke.texi (arm/-mcpu): Document +crypto.
15871 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15873 * config/arm/arm-c.c (arm_cpu_builtins): Define
15874 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
15876 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15878 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
15879 (armv8-r): Set ARM Cortex-R52 as default CPU.
15880 * config/arm/arm-tables.opt: Regenerate.
15881 * config/arm/arm-tune.md: Regenerate.
15882 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
15884 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
15885 extension for -mcpu=cortex-r52.
15887 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15889 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
15890 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
15891 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
15892 (fp-armv8): Define it as FP_ARMv8 only.
15893 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
15894 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
15896 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
15897 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
15898 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
15899 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
15900 than TARGET_FPU_ARMV8.
15901 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
15902 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
15903 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
15905 * config/arm/neon.md (neon_vrint): Likewise.
15906 (neon_vcvt): Likewise.
15907 (neon_<fmaxmin_op><mode>): Likewise.
15908 (<fmaxmin><mode>3): Likewise.
15909 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
15910 * config/arm/predicates.md (arm_cond_move_operator): Check against
15911 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
15913 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
15915 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
15916 to top of function.
15918 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15920 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
15921 loop in comment with memset.
15923 2017-07-14 Martin Liska <mliska@suse.cz>
15925 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
15926 * dwarf2out.c (is_java): Remove the function.
15927 (output_pubname): Remove usage of the function.
15928 (lower_bound_default): Remove usage of DW_LANG_Java.
15929 (gen_compile_unit_die): Likewise.
15930 * gcc.c: Remove compiler defaults for .java and .zip files.
15931 * gimple-expr.c (remove_suffix): Change as there's no longer
15932 extension than 4-letter one.
15933 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
15934 (gimplify_save_expr): Likewise.
15935 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
15936 as it's possible even for other languages than Java.
15937 * langhooks.h (struct lang_hooks): Remove Java from a comment.
15938 * lto-opts.c (lto_write_options): Remove reference to Java.
15939 * opts.c (strip_off_ending): Update file extension handling.
15940 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
15941 * tree-eh.c (lower_resx): Likewise.
15942 * tree.c (free_lang_data_in_type): Remove dead code.
15943 (find_decls_types_r): Likewise.
15944 (build_common_builtin_nodes): Remove Java from a comment.
15945 (verify_type): Remove dead code.
15946 * varasm.c (assemble_external): Remove Java from a comment.
15948 2017-07-14 Martin Liska <mliska@suse.cz>
15950 * opts.c (finish_options): Add quotes.
15951 (common_handle_option): Likewise.
15953 2017-07-14 Martin Liska <mliska@suse.cz>
15955 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
15956 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
15957 Remove N_SO_PASCAL.
15958 * dwarf2out.c (lower_bound_default): Do not handle
15960 (gen_compile_unit_die): Likewise.
15961 * gcc.c: Remove default extension binding for GNU Pascal.
15962 * stmt.c: Remove Pascal language from a comment.
15963 * xcoffout.c: Likewise.
15965 2017-07-13 David Malcolm <dmalcolm@redhat.com>
15968 * diagnostic-show-locus.c (fixit_cmp): New function.
15969 (layout::layout): Sort m_fixit_hints.
15970 (column_range::column_range): Assert that the values are valid.
15971 (struct char_span): New struct.
15972 (correction::overwrite): New method.
15973 (struct source_line): New struct.
15974 (line_corrections::add_hint): Add assertions. Reimplement memcpy
15975 calls in terms of classes source_line and char_span, and
15976 correction::overwrite.
15977 (selftest::test_overlapped_fixit_printing_2): New function.
15978 (selftest::diagnostic_show_locus_c_tests): Call it.
15980 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
15982 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
15983 early if there is no lhs.
15985 2017-07-13 Martin Liska <mliska@suse.cz>
15987 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
15988 (gen_reference_type_die): Likewise.
15989 * stor-layout.c: Remove Pascal-related comment.
15991 2017-07-13 Martin Liska <mliska@suse.cz>
15993 * opts.c (finish_options): Add quotes to error messages.
15994 (parse_sanitizer_options): Likewise.
15996 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
15998 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
16000 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
16002 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
16004 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
16006 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
16008 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
16010 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
16013 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
16014 provides the hardware capability bits, define the macro
16015 __BUILTIN_CPU_SUPPORTS__.
16016 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
16017 if GLIBC does not provide the hardware capability bits. Add a
16018 gcc_unreachable call if the built-in cpu function is neither
16019 __builtin_cpu_is nor __builtin_cpu_supports.
16020 (rs6000_get_function_versions_dispatcher): Change the warning
16021 that an old GLIBC is used which does not export the capability
16022 bits to be an error.
16023 * doc/extend.texi (target_clones attribute): Document the
16024 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
16025 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
16026 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
16027 the macros defined by GCC if the newer GLIBC is available.
16029 2017-07-12 Jeff Law <law@redhat.com>
16031 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
16032 remaining includes slightly.
16033 * config/riscv/riscv-builtins.c: Include profile-count.h.
16035 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16038 * config/avr/avr.c (avr_set_current_function): In diagnostic
16039 messages: Quote keywords and (parts of) identifiers.
16040 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
16043 2017-07-12 Carl Love <cel@us.ibm.com>
16045 * config/rs6000/rs6000-c.c: Add support for built-in functions
16046 vector bool char vec_revb (vector bool char);
16047 vector bool short vec_revb (vector short char);
16048 vector bool int vec_revb (vector bool int);
16049 vector bool long long vec_revb (vector bool long long);
16050 * doc/extend.texi: Update the built-in documentation file for the
16051 new built-in functions.
16053 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16055 * config/s390/s390.md: Remove movcc splitter.
16057 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16059 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
16060 load/store on condition.
16062 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16065 * config/avr/avr.c (avr_encode_section_info)
16066 [progmem && !TREE_READONLY]: Error if progmem object needs
16069 2017-07-11 Michael Collison <michael.collison@arm.com>
16071 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
16074 2017-07-11 Carl Love <cel@us.ibm.com>
16076 * config/rs6000/rs6000-c.c: Add support for builtins
16077 vector unsigned int vec_parity_lsbb (vector signed int);
16078 vector unsigned int vec_parity_lsbb (vector unsigned int);
16079 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
16080 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
16081 vector unsigned long long vec_parity_lsbb (vector signed long long);
16082 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
16083 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
16084 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
16085 * doc/extend.texi: Update the built-in documentation file for the
16086 new built-in functions.
16088 2017-07-11 David Malcolm <dmalcolm@redhat.com>
16090 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
16091 (layout::m_primary_loc): New field.
16092 (layout::layout): Initialize new field. Move location filtering
16093 logic from here to...
16094 (layout::maybe_add_location_range): ...this new method. Add
16095 support for filtering to just the lines already specified by other
16097 (layout::will_show_line_p): New method.
16098 (gcc_rich_location::add_location_if_nearby): New method.
16099 (selftest::test_add_location_if_nearby): New test function.
16100 (selftest::diagnostic_show_locus_c_tests): Call it.
16101 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
16104 2017-07-11 Tom de Vries <tom@codesourcery.com>
16106 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
16107 (bb_first_real_insn): New function.
16108 (nvptx_single): Add extra initialization of broadcasted condition
16111 2017-07-11 Nathan Sidwell <nathan@acm.org>
16113 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
16115 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
16117 * doc/extend.texi (AVR Function Attributes): Remove weblink to
16118 Binutils doc as TEXI will mess them up.
16119 * doc/invoke.texi (AVR Options): Same here.
16121 2017-07-11 Daniel Cederman <cederman@gaisler.com>
16123 * config/sparc/sparc.opt (mfix-ut700): New option.
16124 (mfix-gr712rc): Likewise.
16125 (sparc_fix_b2bst): New variable.
16126 * doc/invoke.texi (SPARC options): Document them.
16127 (ARM options): Fix warnings.
16128 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
16129 instructions to prevent sequences that can trigger the store-store
16130 errata for certain LEON3FT processors.
16131 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
16132 (sparc_option_override): Set sparc_fix_b2bst appropriately.
16133 * config/sparc/sparc.md (fix_b2bst): New attribute.
16134 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
16136 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
16139 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
16141 (*rsqrtsf2_sse): Ditto.
16143 (div<mode>3): Macroize insn from divdf3 and divsf3
16144 using MODEF mode iterator.
16146 2017-07-10 Martin Sebor <msebor@redhat.com>
16148 PR tree-optimization/80397
16149 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
16150 instead of testing for equality to INTEGER_TYPE.
16152 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
16154 * config.gcc: Remove uclibc from arc target spec.
16156 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
16158 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
16160 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16163 * lto-wrapper.c (remove_option): New function.
16164 (merge_and_complain): Merge PIC/PIE options more realistically.
16166 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16168 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
16172 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
16173 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
16174 * config.in: Regenerate.
16175 * configure: Regenerate.
16176 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
16177 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
16178 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
16179 (TARGET_GASISR_PROLOGUES): ...target mask.
16180 * common/config/avr/avr-common.c
16181 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
16182 Set -mgas-isr-prologues.
16183 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
16184 INSERT_PASS_BEFORE for it.
16185 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
16186 * config/avr/avr.c (avr_option_override)
16187 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
16188 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
16189 (avr_attribute_table) <no_gccisr>: Add new function attribute.
16190 (avr_set_current_function) <is_no_gccisr>: Init machine field.
16191 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
16193 (make_avr_pass_pre_proep): New function.
16194 (emit_push_sfr) <treg>: Add argument to function and use it
16195 instead of TMP_REG.
16196 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
16197 and set machine->gasisr.yes.
16198 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
16199 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
16200 __gcc_isr.n_pushed to .L__stack_usage.
16201 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
16202 (avr_asm_final_postscan_insn): ...this new static function.
16203 * config/avr/avr.h (machine_function)
16204 <is_no_gccisr, use_L__stack_usage>: New fields.
16205 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
16206 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
16207 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
16208 (gasisr, *gasisr): New expander and insn.
16209 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
16210 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
16211 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
16213 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
16215 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
16218 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16220 Move jump-tables out of .text again.
16223 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
16224 (ASM_OUTPUT_ADDR_VEC): New function.
16225 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
16226 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
16227 INSN_ADDRESSes as asm comment.
16228 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
16229 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
16230 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
16231 * config/avr/avr.md (*tablejump): Adjust comment.
16232 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
16233 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
16235 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
16236 (avr_output_addr_vec): New proto.
16237 (avr_log_t) <insn_addresses>: New field.
16239 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
16242 * config/i386/i386.c (ix86_function_arg_advance): Set
16243 outgoing_args_on_stack to true if there are outgoing arguments
16245 (ix86_function_arg): Likewise.
16246 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
16247 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
16248 * config/i386/i386.h (machine_function): Add
16249 outgoing_args_on_stack.
16251 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
16253 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
16254 supporting pthreds.
16255 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
16257 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
16259 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
16260 (REAL_H): Remove $(MACHMODE_H).
16261 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
16263 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
16264 $(MACHMODE_H) and double-int.h.
16265 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
16267 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
16270 2017-07-07 Andrew Pinski <apinski@cavium.com>
16272 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
16273 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
16275 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16277 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
16278 Add warning if GCC was not configured to link against a GLIBC that
16279 exports the hardware capability bits.
16280 (make_resolver_func): Make resolver function private and not a
16281 COMDAT function. Create the name with clone_function_name instead
16282 of make_unique_name.
16285 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
16286 correct operand in doing the split.
16288 2017-07-07 Carl Love <cel@us.ibm.com>
16290 * config/rs6000/rs6000-c: Add support for built-in function
16291 vector unsigned short vec_pack_to_short_fp32 (vector float,
16293 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
16294 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
16295 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
16296 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
16297 (convert_4f32_8i16): Add define_expand.
16298 * doc/extend.texi: Update the built-in documentation file for the
16299 new built-in function.
16301 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16303 * config/sparc/m8.md: New file.
16304 * config/sparc/sparc.md: Include m8.md.
16306 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16308 * config/sparc/sparc.opt: New option -mvis4b.
16309 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
16310 (sparc_option_override): Handle VIS4B.
16311 (enum sparc_builtins): Define
16312 SPARC_BUILTIN_DICTUNPACK{8,16,32},
16313 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
16314 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
16315 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
16316 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
16317 (check_constant_argument): New function.
16318 (sparc_vis_init_builtins): Define builtins
16319 __builtin_vis_dictunpack{8,16,32},
16320 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
16321 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
16322 __builtin_vis_fpcmpde{8,16,32}shl and
16323 __builtin_vis_fpcmpur{8,16,32}shl.
16324 (sparc_expand_builtin): Check that the constant operands to
16325 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
16326 constant and in range.
16327 * config/sparc/sparc-c.c (sparc_target_macros): Handle
16329 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
16330 (SPARC_IMM5_P): Likewise.
16331 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
16332 (enabled): Handle vis4b.
16333 (UNSPEC_DICTUNPACK): New unspec.
16334 (UNSPEC_FPCMPSHL): Likewise.
16335 (UNSPEC_FPUCMPSHL): Likewise.
16336 (UNSPEC_FPCMPDESHL): Likewise.
16337 (UNSPEC_FPCMPURSHL): Likewise.
16338 (cpu_feature): New CPU feature `vis4b'.
16339 (dictunpack{8,16,32}): New insns.
16340 (FPCSMODE): New mode iterator.
16341 (fpcscond): New code iterator.
16342 (fpcsucond): Likewise.
16343 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
16344 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
16345 (fpcmpde{8,16,32}{si,di}shl): Likewise.
16346 (fpcmpur{8,16,32}{si,di}shl): Likewise.
16347 * config/sparc/constraints.md: Define constraints `q' for unsigned
16348 2-bit integer constants and `t' for unsigned 5-bit integer
16350 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
16352 (imm5_operand_dictunpack16): Likewise.
16353 (imm5_operand_dictunpack32): Likewise.
16354 (imm2_operand): Likewise.
16355 * doc/invoke.texi (SPARC Options): Document -mvis4b.
16356 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
16357 ditunpack* and fpcmp*shl builtins.
16359 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16361 * config.gcc: Handle m8 in --with-{cpu,tune} options.
16362 * config.in: Add HAVE_AS_SPARC6 define.
16363 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
16365 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
16367 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
16368 (CPP_CPU_SPEC): Handle m8.
16369 (ASM_CPU_SPEC): Likewise.
16370 * config/sparc/sparc-opts.h (enum processor_type): Add
16372 * config/sparc/sparc.c (m8_costs): New struct.
16373 (sparc_option_override): Handle TARGET_CPU_m8.
16374 (sparc32_initialize_trampoline): Likewise.
16375 (sparc64_initialize_trampoline): Likewise.
16376 (sparc_issue_rate): Likewise.
16377 (sparc_register_move_cost): Likewise.
16378 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
16379 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
16380 (ASM_CPU64_DEFAULT_SPEC): Likewise.
16381 (CPP_CPU_SPEC): Handle M8.
16382 (ASM_CPU_SPEC): Likewise.
16383 (AS_M8_FLAG): Define.
16384 * config/sparc/sparc.md: Add m8 to the cpu attribute.
16385 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
16386 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
16388 * configure: Regenerate.
16389 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
16392 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16394 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
16396 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
16397 ("*movdi_insn_sp32"): Do not set v3pipe.
16398 ("*movsi_insn"): Likewise.
16399 ("*movdi_insn_sp64"): Likewise.
16400 ("*movsf_insn"): Likewise.
16401 ("*movdf_insn_sp32"): Likewise.
16402 ("*movdf_insn_sp64"): Likewise.
16403 ("*zero_extendsidi2_insn_sp64"): Likewise.
16404 ("*sign_extendsidi2_insn"): Likewise.
16405 ("*mov<VM32:mode>_insn"): Likewise.
16406 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16407 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16408 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16409 ("<vlop:code><VL:mode>3"): Likewise.
16410 ("*not_<vlop:code><VL:mode>3"): Likewise.
16411 ("*nand<VL:mode>_vis"): Likewise.
16412 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
16413 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
16414 ("one_cmpl<VL:mode>2"): Likewise.
16415 ("faligndata<VM64:mode>_vis"): Likewise.
16416 ("alignaddrsi_vis"): Likewise.
16417 ("alignaddrdi_vis"): Likweise.
16418 ("alignaddrlsi_vis"): Likewise.
16419 ("alignaddrldi_vis"): Likewise.
16420 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16421 ("bmaskdi_vis"): Likewise.
16422 ("bmasksi_vis"): Likewise.
16423 ("bshuffle<VM64:mode>_vis"): Likewise.
16424 ("cmask8<P:mode>_vis"): Likewise.
16425 ("cmask16<P:mode>_vis"): Likewise.
16426 ("cmask32<P:mode>_vis"): Likewise.
16427 ("pdistn<P:mode>_vis"): Likewise.
16428 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16430 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16432 * config/sparc/sparc.md ("subtype"): New insn attribute.
16433 ("*wrgsr_sp64"): Set insn subtype.
16434 ("*rdgsr_sp64"): Likewise.
16435 ("alignaddrsi_vis"): Likewise.
16436 ("alignaddrdi_vis"): Likewise.
16437 ("alignaddrlsi_vis"): Likewise.
16438 ("alignaddrldi_vis"): Likewise.
16439 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16440 ("fexpand_vis"): Likewise.
16441 ("fpmerge_vis"): Likewise.
16442 ("faligndata<VM64:mode>_vis"): Likewise.
16443 ("bshuffle<VM64:mode>_vis"): Likewise.
16444 ("cmask8<P:mode>_vis"): Likewise.
16445 ("cmask16<P:mode>_vis"): Likewise.
16446 ("cmask32<P:mode>_vis"): Likewise.
16447 ("fchksm16_vis"): Likewise.
16448 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
16449 ("fmean16_vis"): Likewise.
16450 ("fp<plusminus_insn>64_vis"): Likewise.
16451 ("<plusminus_insn>v8qi3"): Likewise.
16452 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16453 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
16454 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
16455 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
16456 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
16457 ("*movqi_insn"): Likewise.
16458 ("*movhi_insn"): Likewise.
16459 ("*movsi_insn"): Likewise.
16460 ("movsi_pic_gotdata_op"): Likewise.
16461 ("*movdi_insn_sp32"): Likewise.
16462 ("*movdi_insn_sp64"): Likewise.
16463 ("movdi_pic_gotdata_op"): Likewise.
16464 ("*movsf_insn"): Likewise.
16465 ("*movdf_insn_sp32"): Likewise.
16466 ("*movdf_insn_sp64"): Likewise.
16467 ("*zero_extendhisi2_insn"): Likewise.
16468 ("*zero_extendqihi2_insn"): Likewise.
16469 ("*zero_extendqisi2_insn"): Likewise.
16470 ("*zero_extendqidi2_insn"): Likewise.
16471 ("*zero_extendhidi2_insn"): Likewise.
16472 ("*zero_extendsidi2_insn_sp64"): Likewise.
16473 ("ldfsr"): Likewise.
16474 ("prefetch_64"): Likewise.
16475 ("prefetch_32"): Likewise.
16476 ("tie_ld32"): Likewise.
16477 ("tie_ld64"): Likewise.
16478 ("*tldo_ldub_sp32"): Likewise.
16479 ("*tldo_ldub1_sp32"): Likewise.
16480 ("*tldo_ldub2_sp32"): Likewise.
16481 ("*tldo_ldub_sp64"): Likewise.
16482 ("*tldo_ldub1_sp64"): Likewise.
16483 ("*tldo_ldub2_sp64"): Likewise.
16484 ("*tldo_ldub3_sp64"): Likewise.
16485 ("*tldo_lduh_sp32"): Likewise.
16486 ("*tldo_lduh1_sp32"): Likewise.
16487 ("*tldo_lduh_sp64"): Likewise.
16488 ("*tldo_lduh1_sp64"): Likewise.
16489 ("*tldo_lduh2_sp64"): Likewise.
16490 ("*tldo_lduw_sp32"): Likewise.
16491 ("*tldo_lduw_sp64"): Likewise.
16492 ("*tldo_lduw1_sp64"): Likewise.
16493 ("*tldo_ldx_sp64"): Likewise.
16494 ("*mov<VM32:mode>_insn"): Likewise.
16495 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16496 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16498 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16500 * config/sparc/sparc.md ("type"): New insn type viscmp.
16501 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
16503 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
16504 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
16505 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16506 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
16508 ("n7_vis_logical_11cycle"): Likewise.
16509 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
16510 * config/sparc/niagara2.md ("niag3_vis": Likewise.
16511 * config/sparc/niagara.md ("niag_vis"): Likewise.
16512 * config/sparc/ultra3.md ("us3_fga"): Likewise.
16513 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
16515 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16517 * config/sparc/sparc.md: New instruction type `bmask'.
16518 (bmaskdi_vis): Use the `bmask' type.
16519 (bmasksi_vis): Likewise.
16520 * config/sparc/ultra3.md (us3_array): Likewise.
16521 * config/sparc/niagara7.md (n7_array): Likewise.
16522 * config/sparc/niagara4.md (n4_array): Likewise.
16523 * config/sparc/niagara2.md (niag2_vis): Likewise.
16524 (niag3_vis): Likewise.
16525 * config/sparc/niagara.md (niag_vis): Likewise.
16527 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16529 * ipa-comdats.c: Remove optimize check from gate.
16530 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
16531 for functions not optimized.
16532 (ipa_fn_summary_read): Skip optimize check.
16533 (ipa_fn_summary_write): Likewise.
16534 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
16536 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
16538 (can_inline_edge_p): Check flag_pcc_struct_return for match.
16539 (check_callers): Give up on caller which is not optimized.
16540 (inline_small_functions): Likewise.
16541 (ipa_inline): Do not give up when not optimizing.
16542 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
16543 away unoptimizes cdtors.
16544 (whole_program_function_and_variable_visibility): Do
16545 ipa_discover_readonly_nonaddressable_vars in LTO mode.
16546 * ipa.c (process_references): Do not check optimize.
16547 (symbol_table::remove_unreachable_nodes): Update optimize check.
16548 (set_writeonly_bit): Update optimize check.
16549 (pass_ipa_cdtor_merge::gate): Do not check optimize.
16550 (pass_ipa_single_use::gate): Remove.
16552 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16554 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
16555 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
16556 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
16557 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
16558 permute_load, permute_store, adjust_extract, adjust_splat,
16559 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
16560 replace_swap_with_copy, dump_swap_insn_table,
16561 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
16562 recombine_lvx_pattern, recombine_stvx_pattern,
16563 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
16564 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
16565 to file rs6000-p8swap.c.
16566 * config/rs6000/rs6000-p8swap.c: New file.
16567 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
16568 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
16569 and rs6000*-*-* targets.
16571 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16573 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
16575 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16577 * lto-wrapper.c (merge_and_complain): Do not merge
16578 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
16579 fsigned_zeros, ftrapping_math, fwrapv.
16580 (append_compiler_options): Do not track these options.
16581 (append_linker_options): Likewie
16583 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16585 * cgraphunit.c (cgraph_node::finalize_function): When
16586 !flag_toplevel_reorde set no_reorder flag.
16587 (varpool_node::finalize_decl): Likewise.
16588 (symbol_table::compile): Drop no toplevel reorder path.
16590 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16592 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
16593 edges; zero probability is not better than uninitialized.
16595 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
16597 * asan.h (asan_sanitize_allocas_p): Declare.
16598 * asan.c (asan_sanitize_allocas_p): New function.
16599 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
16600 (handle_builtin_alloca): Likewise.
16601 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
16602 if !asan_sanitize_allocas_p.
16603 * params.def (asan-instrument-allocas): Add new option.
16604 * params.h (ASAN_PROTECT_ALLOCAS): Define.
16605 * opts.c (common_handle_option): Disable allocas sanitization for
16608 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
16610 * asan.c: Include gimple-fold.h.
16611 (get_last_alloca_addr): New function.
16612 (handle_builtin_stackrestore): Likewise.
16613 (handle_builtin_alloca): Likewise.
16614 (asan_emit_allocas_unpoison): Likewise.
16615 (get_mem_refs_of_builtin_call): Add new parameter, remove const
16616 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
16617 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
16618 (instrument_builtin_call): Pass gimple iterator to
16619 get_mem_refs_of_builtin_call.
16620 (last_alloca_addr): New global.
16621 * asan.h (asan_emit_allocas_unpoison): Declare.
16622 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
16623 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
16624 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
16625 if function calls alloca.
16626 * gimple-fold.c (replace_call_with_value): Remove static keyword.
16627 * gimple-fold.h (replace_call_with_value): Declare.
16628 * internal-fn.c: Include asan.h.
16629 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
16630 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
16632 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16634 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
16635 (C_SELFTEST_FLAGS): New.
16636 (CPP_SELFTEST_FLAGS): New.
16637 (SELFTEST_DEPS): New, from deps of s-selftest.
16638 (C_SELFTEST_DEPS): New, from deps of s-selftest.
16639 (CPP_SELFTEST_DEPS): New.
16640 (selftest): Add dependency on s-selftest-c++.
16641 (s-selftest): Rename to...
16642 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
16643 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
16644 than SELFTEST_FLAGS.
16645 (selftest-gdb): Rename to...
16646 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
16648 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
16649 (selftest-valgrind): Rename to...
16650 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
16652 (selftest-valgrind): Reintroduce as an alias for
16653 selftest-c-valgrind.
16654 (s-selftest-c++): New.
16655 (selftest-c++-gdb): New.
16656 (selftest-c++-valgrind): New.
16658 2017-07-06 Olivier Hainque <hainque@adacore.com>
16660 * gcc.c (process_command): When deciding if undefined variables
16661 should be ignored when processing specs, accept "gcc -v" as well.
16663 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16665 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
16666 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
16668 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
16670 * config/arm/arm-cpus.in (armv8-r): Add new entry.
16671 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
16672 * config/arm/arm-tables.opt: Regenerate.
16673 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
16675 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
16677 2017-07-06 Carl Love <cel@us.ibm.com>
16679 * ChangeLog: Clean up from mid air collision
16681 2017-07-06 Carl Love <cel@us.ibm.com>
16683 * config/rs6000/rs6000-c.c: Add support for built-in functions
16684 vector signed int vec_subc (vector signed int, vector signed int);
16685 vector signed __int128 vec_subc (vector signed __int128,
16686 vector signed __int128);
16687 vector unsigned __int128 vec_subc (vector unsigned __int128,
16688 vector unsigned __int128);
16689 vector signed int vec_sube (vector signed int, vector signed int,
16690 vector signed int);
16691 vector unsigned int vec_sube (vector unsigned int,
16692 vector unsigned int,
16693 vector unsigned int);
16694 vector signed __int128 vec_sube (vector signed __int128,
16695 vector signed __int128,
16696 vector signed__int128);
16697 vector unsigned __int128 vec_sube (vector unsigned __int128,
16698 vector unsigned __int128,
16699 vector unsigned __int128);
16700 vector signed int vec_subec (vector signed int, vector signed int,
16701 vector signed int);
16702 vector unsigned int vec_subec (vector unsigned int,
16703 vector unsigned int,
16704 vector unsigned int);
16705 vector signed __int128 vec_subec (vector signed __int128,
16706 vector signed __int128,
16707 vector signed__int128);
16708 vector unsigned __int128 vec_subec (vector unsigned __int128,
16709 vector unsigned __int128,
16710 vector unsigned __int128);
16711 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
16712 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
16713 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
16714 BU_ALTIVEC_OVERLOAD_X definitions.
16715 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
16716 * doc/extend.texi: Update the built-in documentation file for the new
16717 built-in functions.
16719 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16722 * diagnostic-show-locus.c (layout::layout): Use start and finish
16723 spelling location for the start and finish of each range.
16724 * genmatch.c (linemap_client_expand_location_to_spelling_point):
16725 Add unused aspect param.
16726 * input.c (expand_location_1): Add "aspect" param, and use it
16727 to access the correct part of the location.
16728 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
16730 (expand_location_to_spelling_point): Likewise.
16731 (linemap_client_expand_location_to_spelling_point): Add "aspect"
16732 param, and pass it to expand_location_1.
16734 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
16736 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
16737 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
16738 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
16739 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
16740 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
16741 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
16742 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
16743 _mm_maskz_getmant_ss): New intrinsics.
16744 (__builtin_ia32_getexpss128_mask): Changed to ...
16745 __builtin_ia32_getexpss128_round ... this.
16746 (__builtin_ia32_getexpsd128_mask): Changed to ...
16747 __builtin_ia32_getexpsd128_round ... this.
16748 * config/i386/i386-builtin-types.def
16749 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
16750 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
16751 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
16752 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
16753 __builtin_ia32_getmantss_mask_round): New builtins.
16754 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
16755 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
16756 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
16757 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
16758 * config/i386/sse.md
16759 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
16760 avx512f_sgetexp<mode><mask_scalar_name>
16761 <round_saeonly_scalar_name> ... this.
16762 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
16763 %0, %1, %2<round_saeonly_op3>}): Changed to ...
16764 vgetexp<ssescalarmodesuffix>
16765 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16766 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
16767 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
16768 avx512f_vgetmant<mode><mask_scalar_name>
16769 <round_saeonly_scalar_name> ... this.
16770 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
16771 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
16772 vgetmant<ssescalarmodesuffix>
16773 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
16774 %0<mask_scalar_operand4>, %1, %2
16775 <round_saeonly_scalar_mask_op4>, %3} ... this.
16776 * config/i386/subst.md (mask_scalar_operand4,
16777 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
16778 round_saeonly_scalar_nimm_predicate): New subst attributes.
16780 2017-07-06 Julia Koval <julia.koval@intel.com>
16782 * config/i386/i386.c (ix86_erase_embedded_rounding):
16783 Remove code for old rounding pattern.
16785 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
16787 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
16789 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
16791 * doc/sourcebuild.texi (Test Directives, Variants of
16792 dg-require-support): Add documentation for dg-require-stack-check.
16794 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
16796 * config/i386/subst.md (mask_scalar, round_scalar,
16797 round_saeonly_scalar): New meta-templates.
16798 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
16799 round_scalar_mask_operand3, round_scalar_mask_op3,
16800 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
16801 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
16802 round_saeonly_scalar_constraint,
16803 round_saeonly_scalar_prefix): New subst attribute.
16804 * config/i386/sse.md
16805 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
16806 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
16807 <round_scalar_name> ... this.
16808 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
16809 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
16810 <round_scalar_name> ... this.
16811 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
16812 <sse>_vm<code><mode>3<mask_scalar_name>
16813 <round_saeonly_scalar_name> ... this.
16814 (v<plusminus_mnemonic><ssescalarmodesuffix>
16815 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
16816 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
16817 v<plusminus_mnemonic><ssescalarmodesuffix>
16818 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16819 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
16820 (v<multdiv_mnemonic><ssescalarmodesuffix>
16821 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
16822 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
16823 v<multdiv_mnemonic><ssescalarmodesuffix>
16824 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16825 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
16826 (v<maxmin_float><ssescalarmodesuffix>
16827 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
16828 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
16829 v<maxmin_float><ssescalarmodesuffix>
16830 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16831 %0<mask_scalar_operand3>, %1, %<iptr>2
16832 <round_saeonly_scalar_mask_op3>} ... this.
16834 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
16836 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
16837 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
16839 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
16840 Alan Hayward <alan.hayward@arm.com>
16841 David Sherwood <david.sherwood@arm.com>
16843 * combine.c (simplify_if_then_else): Remove "enum" before
16845 * compare-elim.c (can_eliminate_compare): Likewise.
16846 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
16848 (aarch64_lookup_simd_builtin_type): Likewise.
16849 (aarch64_simd_builtin_type): Likewise.
16850 (aarch64_init_simd_builtin_types): Likewise.
16851 (aarch64_simd_expand_args): Likewise.
16852 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
16854 (aarch64_reverse_mask): Likewise.
16855 (aarch64_simd_emit_reg_reg_move): Likewise.
16856 (aarch64_gen_adjusted_ldpstp): Likewise.
16857 (aarch64_ccmp_mode_to_code): Likewise.
16858 (aarch64_operands_ok_for_ldpstp): Likewise.
16859 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16860 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
16862 (aarch64_min_divisions_for_recip_mul): Likewise.
16863 (aarch64_reassociation_width): Likewise.
16864 (aarch64_get_condition_code_1): Likewise.
16865 (aarch64_simd_emit_reg_reg_move): Likewise.
16866 (aarch64_simd_attr_length_rglist): Likewise.
16867 (aarch64_reverse_mask): Likewise.
16868 (aarch64_operands_ok_for_ldpstp): Likewise.
16869 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16870 (aarch64_gen_adjusted_ldpstp): Likewise.
16871 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
16873 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
16874 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
16875 (arm_lookup_simd_builtin_type): Likewise.
16876 (arm_simd_builtin_type): Likewise.
16877 (arm_init_simd_builtin_types): Likewise.
16878 (arm_expand_builtin_args): Likewise.
16879 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
16880 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
16881 (ft32_setup_incoming_varargs): Likewise.
16882 (ft32_function_arg): Likewise.
16883 (ft32_function_arg_advance): Likewise.
16884 (ft32_pass_by_reference): Likewise.
16885 (ft32_arg_partial_bytes): Likewise.
16886 (ft32_valid_pointer_mode): Likewise.
16887 (ft32_addr_space_pointer_mode): Likewise.
16888 (ft32_addr_space_legitimate_address_p): Likewise.
16889 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
16891 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
16892 (ix86_emit_outlined_ms2sysv_restore): Likewise.
16893 (iamcu_alignment): Likewise.
16894 (canonicalize_vector_int_perm): Likewise.
16895 (ix86_noce_conversion_profitable_p): Likewise.
16896 (ix86_mpx_bound_mode): Likewise.
16897 (ix86_operands_ok_for_move_multiple): Likewise.
16898 * config/microblaze/microblaze-protos.h
16899 (microblaze_expand_conditional_branch_reg): Likewise.
16900 * config/microblaze/microblaze.c
16901 (microblaze_expand_conditional_branch_reg): Likewise.
16902 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
16904 (rs6000_reassociation_width): Likewise.
16905 (rs6000_invalid_binary_op): Likewise.
16906 (fusion_p9_p): Likewise.
16907 (emit_fusion_p9_load): Likewise.
16908 (emit_fusion_p9_store): Likewise.
16909 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
16911 (riscv_hard_regno_mode_ok_p): Likewise.
16912 (riscv_address_insns): Likewise.
16913 (riscv_split_symbol): Likewise.
16914 (riscv_legitimize_move): Likewise.
16915 (riscv_function_value): Likewise.
16916 (riscv_hard_regno_nregs): Likewise.
16917 (riscv_expand_builtin): Likewise.
16918 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
16919 (riscv_build_integer): Likewise.
16920 (riscv_split_integer): Likewise.
16921 (riscv_legitimate_constant_p): Likewise.
16922 (riscv_cannot_force_const_mem): Likewise.
16923 (riscv_regno_mode_ok_for_base_p): Likewise.
16924 (riscv_valid_base_register_p): Likewise.
16925 (riscv_valid_offset_p): Likewise.
16926 (riscv_valid_lo_sum_p): Likewise.
16927 (riscv_classify_address): Likewise.
16928 (riscv_legitimate_address_p): Likewise.
16929 (riscv_address_insns): Likewise.
16930 (riscv_load_store_insns): Likewise.
16931 (riscv_force_binary): Likewise.
16932 (riscv_split_symbol): Likewise.
16933 (riscv_force_address): Likewise.
16934 (riscv_legitimize_address): Likewise.
16935 (riscv_move_integer): Likewise.
16936 (riscv_legitimize_const_move): Likewise.
16937 (riscv_legitimize_move): Likewise.
16938 (riscv_address_cost): Likewise.
16939 (riscv_subword): Likewise.
16940 (riscv_output_move): Likewise.
16941 (riscv_canonicalize_int_order_test): Likewise.
16942 (riscv_emit_int_order_test): Likewise.
16943 (riscv_function_arg_boundary): Likewise.
16944 (riscv_pass_mode_in_fpr_p): Likewise.
16945 (riscv_pass_fpr_single): Likewise.
16946 (riscv_pass_fpr_pair): Likewise.
16947 (riscv_get_arg_info): Likewise.
16948 (riscv_function_arg): Likewise.
16949 (riscv_function_arg_advance): Likewise.
16950 (riscv_arg_partial_bytes): Likewise.
16951 (riscv_function_value): Likewise.
16952 (riscv_pass_by_reference): Likewise.
16953 (riscv_setup_incoming_varargs): Likewise.
16954 (riscv_print_operand): Likewise.
16955 (riscv_elf_select_rtx_section): Likewise.
16956 (riscv_save_restore_reg): Likewise.
16957 (riscv_for_each_saved_reg): Likewise.
16958 (riscv_register_move_cost): Likewise.
16959 (riscv_hard_regno_mode_ok_p): Likewise.
16960 (riscv_hard_regno_nregs): Likewise.
16961 (riscv_class_max_nregs): Likewise.
16962 (riscv_memory_move_cost): Likewise.
16963 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
16964 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
16965 (rl78_addr_space_address_mode): Likewise.
16966 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16968 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
16969 (rs6000_reassociation_width): Likewise.
16970 (rs6000_invalid_binary_op): Likewise.
16971 (fusion_p9_p): Likewise.
16972 (emit_fusion_p9_load): Likewise.
16973 (emit_fusion_p9_store): Likewise.
16974 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
16975 (ok_for_simple_move_operands): Likewise.
16976 (ok_for_simple_move_strict_operands): Likewise.
16977 (ok_for_simple_arith_logic_operands): Likewise.
16978 (visium_legitimize_reload_address): Likewise.
16979 (visium_select_cc_mode): Likewise.
16980 (output_cbranch): Likewise.
16981 (visium_split_double_move): Likewise.
16982 (visium_expand_copysign): Likewise.
16983 (visium_expand_int_cstore): Likewise.
16984 (visium_expand_fp_cstore): Likewise.
16985 * config/visium/visium.c (visium_pass_by_reference): Likewise.
16986 (visium_function_arg): Likewise.
16987 (visium_function_arg_advance): Likewise.
16988 (visium_libcall_value): Likewise.
16989 (visium_setup_incoming_varargs): Likewise.
16990 (visium_legitimate_constant_p): Likewise.
16991 (visium_legitimate_address_p): Likewise.
16992 (visium_legitimize_address): Likewise.
16993 (visium_secondary_reload): Likewise.
16994 (visium_register_move_cost): Likewise.
16995 (visium_memory_move_cost): Likewise.
16996 (prepare_move_operands): Likewise.
16997 (ok_for_simple_move_operands): Likewise.
16998 (ok_for_simple_move_strict_operands): Likewise.
16999 (ok_for_simple_arith_logic_operands): Likewise.
17000 (visium_function_value_1): Likewise.
17001 (rtx_ok_for_offset_p): Likewise.
17002 (visium_legitimize_reload_address): Likewise.
17003 (visium_split_double_move): Likewise.
17004 (visium_expand_copysign): Likewise.
17005 (visium_expand_int_cstore): Likewise.
17006 (visium_expand_fp_cstore): Likewise.
17007 (visium_split_cstore): Likewise.
17008 (visium_select_cc_mode): Likewise.
17009 (visium_split_cbranch): Likewise.
17010 (output_cbranch): Likewise.
17011 (visium_print_operand_address): Likewise.
17012 * expmed.c (flip_storage_order): Likewise.
17013 * expmed.h (emit_cstore): Likewise.
17014 (flip_storage_order): Likewise.
17015 * genrecog.c (validate_pattern): Likewise.
17016 * hsa-gen.c (gen_hsa_addr): Likewise.
17017 * internal-fn.c (expand_arith_overflow): Likewise.
17018 * ira-color.c (allocno_copy_cost_saving): Likewise.
17019 * lra-assigns.c (find_hard_regno_for_1): Likewise.
17020 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
17021 (process_invariant_for_inheritance): Likewise.
17022 * lra-eliminations.c (move_plus_up): Likewise.
17023 * omp-low.c (lower_oacc_reductions): Likewise.
17024 * simplify-rtx.c (simplify_subreg): Likewise.
17025 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
17026 (TARGET_CHKP_BOUND_MODE): Likewise..
17027 * targhooks.c (default_chkp_bound_mode): Likewise.
17028 (default_setup_incoming_vararg_bounds): Likewise.
17029 * targhooks.h (default_chkp_bound_mode): Likewise.
17030 (default_setup_incoming_vararg_bounds): Likewise.
17031 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
17032 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
17033 (have_whole_vector_shift): Likewise.
17034 * tree-vect-stmts.c (vectorizable_load): Likewise.
17035 * doc/tm.texi: Regenerate.
17037 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17039 Graceful degrade if Binutils PR21472 is not available.
17042 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
17043 .rodata in flash test fails.
17044 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
17045 * confgure: Regenerate.
17046 * config.in: Regenerate.
17047 * config/avr/avr.c (avr_asm_named_section)
17048 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
17049 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
17050 (avr_asm_init_sections): Same.
17052 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17054 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
17055 (fma<VH:mode>4_intrinsic): Likewise.
17056 (*fmsub<VCVTF:mode>4): Likewise.
17057 (*fmsub<VH:mode>4_intrinsic): Likewise.
17059 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17062 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
17063 Don't depend on "optimize > 0".
17064 (out_movhi_r_mr, out_movqi_mr_r): Same.
17065 (out_movhi_mr_r, out_movqi_r_mr): Same.
17066 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
17067 io_address_operand on "optimize > 0".
17069 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17071 * tree-loop-distribution.c: Add general explanantion on the pass.
17072 (generate_loops_for_partition): Mark distributed loop.
17073 (pg_add_dependence_edges): New parameter. Handle alias data
17074 dependence specially and record it in the parameter if asked.
17075 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
17076 (init_partition_graph_vertices, add_partition_graph_edge): New.
17077 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
17078 (free_partition_graph_vdata, build_partition_graph): New.
17079 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
17080 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
17081 (data_ref_segment_size, latch_dominated_by_data_ref): New.
17082 (compute_alias_check_pairs, version_loop_by_alias_check): New.
17083 (version_for_distribution_p, finalize_partitions): New.
17084 (distribute_loop): Handle alias data dependence specially. Factor
17085 out loop fusion code as functions and call these functions.
17087 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17089 * tree-loop-distribution.c (classify_partition): New parameter and
17090 better handle reduction statement.
17091 (rdg_build_partitions): Revise comment.
17092 (distribute_loop): Compute statements in all partitions and pass it
17093 to classify_partition.
17095 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17097 * tree-loop-distribution.c (enum partition_type): New.
17098 (struct partition): New field type.
17099 (partition_merge_into): Add parameter. Update partition type.
17100 (data_dep_in_cycle_p, update_type_for_merge): New functions.
17101 (build_rdg_partition_for_vertex): Compute partition type.
17102 (rdg_build_partitions): Dump partition type.
17103 (distribute_loop): Update calls to partition_merge_into.
17105 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17107 * tree-loop-distribution.c (struct ddr_hasher): New.
17108 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
17110 (classify_partition): Call get_data_dependence.
17111 (pg_add_dependence_edges): Ditto.
17112 (distribute_loop): Release data dependence hash table.
17114 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17116 * tree-loop-distribution.c (ref_base_address): Delete.
17117 (similar_memory_accesses): Rename ...
17118 (share_memory_accesses): ... to this. Check if partitions access
17119 the same memory reference.
17120 (distribute_loop): Call share_memory_accesses.
17122 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17124 * tree-loop-distribution.c (struct partition): New field recording
17125 its data reference.
17126 (partition_alloc, partition_free): Init and release data refs.
17127 (partition_merge_into): Merge data refs.
17128 (build_rdg_partition_for_vertex): Collect data refs for partition.
17129 (pg_add_dependence_edges): Change parameters from vector to bitmap.
17131 (distribute_loop): Remve data refs from vertice data of partition
17134 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17136 * tree-loop-distribution.c (params.h): Include header file.
17137 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
17138 (datarefs_vec): New global var.
17139 (create_rdg_vertices): Use datarefs_vec directly.
17140 (free_rdg): Don't free data references.
17141 (build_rdg): Update use. Don't free data references.
17142 (distribute_loop): Compute global variable for data references.
17143 Bail out if there are too many data references.
17145 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17147 * tree-loop-distribution.c (loop_nest): New global var.
17148 (build_rdg): Use loop directly, rather than loop nest.
17149 (pg_add_dependence_edges): Remove loop nest parameter. Use global
17151 (distribute_loop): Compute global variable loop nest. Update use.
17153 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17155 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
17156 (partition_merge_into): New parameter. Dump reason for fusion.
17157 (distribute_loop): Update use of partition_merge_into.
17159 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17161 * tree-loop-distribution.c (bb_top_order_index): New.
17162 (bb_top_order_index_size, bb_top_order_cmp): New.
17163 (stmts_from_loop): Use topological order.
17164 (pass_loop_distribution::execute): Compute and release topological
17165 order for basic blocks.
17167 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17169 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
17172 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17174 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
17175 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
17176 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
17177 * internal-fn.def (LOOP_DIST_ALIAS): New.
17178 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
17179 (fold_loop_internal_call): ... this.
17180 (vect_loop_dist_alias_call): New function.
17181 (set_uid_loop_bbs): Call fold_loop_internal_call.
17182 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
17185 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
17188 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
17189 Require dead FLAGS_REG at the beginning of a peephole.
17191 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
17194 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
17195 arguments in the call to __builtin_ia32_sbb_u32.
17196 (_subborrow_u64): Swap _X and _Y arguments in the call to
17197 __builtin_ia32_sbb_u64.
17199 2017-07-04 Jakub Jelinek <jakub@redhat.com>
17202 * tree-vrp.c (compare_assert_loc): Turn into a function template
17203 with stable template parameter. Only test if a->e is NULL,
17204 !a->e == !b->e has been verified already. Use e == NULL or
17205 e != NULL instead of e or ! e tests. If stable is true, don't use
17206 iterative_hash_expr, on the other side allow a or b or both NULL
17207 and sort the NULLs last.
17208 (process_assert_insertions): Sort using compare_assert_loc<false>
17209 instead of compare_assert_loc, later sort using
17210 compare_assert_loc<true> before calling process_assert_insertions_for
17211 in a loop. Use break instead of continue once seen NULL pointer.
17213 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
17215 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17216 Cortex-R7 and Cortex-R8 processors.
17218 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17220 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
17221 uninitialized while src is not.
17223 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
17225 * common/config/arm/arm-common.c: Adjust include path for
17227 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
17228 (arm-cpu.h): Create in build directory. Adjust dependency rules.
17229 (arm-cpu-data.h): Likewise.
17230 (arm-cpu-cdata.h): Likewise.
17231 * config/arm/arm-cpu.h: Delete.
17232 * config/arm/arm-cpu-cdata.h: Delete.
17233 * config/arm/arm-cpu-data.h: Delete.
17235 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
17237 * config/arm/arm-cpus.in (cortex-a55): New.
17238 (cortex-a75): Likewise.
17239 (cortex-a75.cortex-a55): Likewise.
17240 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
17242 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
17243 * config/arm/arm-cpu-cdata.h: Regenerate.
17244 * config/arm/arm-cpu-data.h: Regenerate.
17245 * config/arm/arm-cpu.h: Regenerate.
17246 * config/arm/arm-tables.opt: Regenerate.
17247 * config/arm/arm-tune.md: Regenerate.
17249 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17251 * haifa-sched.c (sched_create_recovery_edges): Update profile.
17253 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17255 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
17258 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
17260 PR tree-optimization/81292
17261 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
17262 full_string_p, also call adjust_related_strinfos if the adjustment
17263 is simple, otherwise invalidate related strinfos.
17265 2017-07-04 Martin Liska <mliska@suse.cz>
17268 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
17269 newly created variable as DECL_IGNORED_P.
17271 2017-07-04 Martin Liska <mliska@suse.cz>
17274 * ipa-inline.c (inline_small_functions):
17275 Use xstrdup_for_dump.
17277 2017-07-04 Tom de Vries <tom@codesourcery.com>
17279 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
17281 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
17284 * config/darwin.c (darwin_function_switched_text_sections):
17287 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
17289 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
17291 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17293 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
17295 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17297 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
17298 min_profitable_iters, and th as inclusive lower bounds.
17299 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
17300 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
17301 for min_profitable_iters and min_profitable_estimate.
17302 (vect_transform_loop): Treat th as an inclusive lower bound.
17303 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
17305 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
17308 * config/darwin.c (darwin_function_switched_text_sections):
17309 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
17310 in two pieces, and suppress the use of buf.
17312 2017-07-03 Nathan Sidwell <nathan@acm.org>
17314 * hash-table.h (hash_table_mod1): Fix indentation.
17316 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17318 PR middle-end/81290
17319 * predict.c (force_edge_cold): Be more careful about propagation
17321 * profile-count.h (profile_probability::guessed,
17322 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
17324 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
17326 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
17328 * doc/invoke.texi (rcpc architecture extension): Document it.
17330 2017-07-03 Richard Biener <rguenther@suse.de>
17332 PR tree-optimization/60510
17333 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
17334 the scalar reduction PHI and use it.
17335 (vectorizable_reduction): Properly guard the single_defuse_cycle
17336 path for non-SLP reduction chains where we cannot use it.
17337 Rework reduc_def/index and vector type deduction. Rework
17338 vector operand gathering during reduction op code-gen.
17339 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
17340 chains dissolve the chain and leave it to non-SLP reduction
17343 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17345 * tree-data-ref.h (dr_alignment): Declare.
17346 * tree-data-ref.c (dr_alignment): New function.
17347 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
17348 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
17350 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
17352 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17354 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
17355 and base_misalignment fields.
17356 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
17357 * tree-data-ref.c: Include builtins.h.
17358 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
17359 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
17360 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
17361 * tree-vect-data-refs.c: Include tree-cfg.h.
17362 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
17363 fields instead of calculating an alignment here.
17364 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
17365 innermost_loop_behavior fields.
17367 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17369 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
17371 (DR_STEP_ALIGNMENT): New macro.
17372 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
17373 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
17374 (create_data_ref): Print it.
17375 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
17376 to tell whether the step preserves vector (mis)alignment.
17377 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17378 Move the check for an integer step and generalise to all INTEGER_CST.
17379 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
17380 Print the outer step alignment.
17382 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17384 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
17385 with offset_alignment.
17386 (DR_ALIGNED_TO): Delete.
17387 (DR_OFFSET_ALIGNMENT): New macro.
17388 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
17389 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
17390 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
17391 (create_data_ref): Likewise.
17392 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17393 (vect_analyze_data_refs): Likewise.
17394 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
17395 creating dummy innermost behavior.
17397 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17399 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
17400 with a "innermost_loop_behavior *" and refeence tree.
17401 * tree-data-ref.c (dr_analyze_innermost): Likewise.
17402 (create_data_ref): Update call accordingly.
17403 * tree-predcom.c (find_looparound_phi): Likewise.
17405 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17407 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
17408 fields with dr_wrt_vec_loop.
17409 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
17410 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
17411 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
17412 (vect_dr_behavior): New function.
17413 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17414 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
17415 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
17416 track whether the step preserves the misalignment.
17417 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17418 Use vect_dr_behavior.
17419 (vect_setup_realignment): Update call accordingly.
17420 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
17421 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
17422 call to vect_create_addr_base_for_vector_ref.
17423 (vect_create_cond_for_align_checks): Likewise.
17424 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
17425 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
17426 (vect_recog_mask_conversion_pattern): Likewise.
17427 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
17428 (new_stmt_vec_info): Remove redundant zeroing.
17430 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17432 * common/config/arm/arm-common.c (arm_be8_option): New function.
17433 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
17434 (ISA_ARMv6): Add isa_bit_be8.
17435 * config/arm/arm.h (arm_be8_option): Add prototype.
17436 (BE8_SPEC_FUNCTION): New define.
17437 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
17438 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
17439 (mlittle-endian): Similarly.
17440 (mbe8, mbe32): New options.
17441 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
17442 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
17444 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17446 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
17448 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17450 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
17451 (cleanup_tree_cfg_bb): Use it.
17452 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
17454 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
17456 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17459 * loop-doloop.c (add_test): Update profile.
17461 2017-07-03 Martin Liska <mliska@suse.cz>
17464 * sanopt.c (rewrite_usage_of_param): New function.
17465 (sanitize_rewrite_addressable_params): Likewise.
17466 (pass_sanopt::execute): Call rewrite_usage_of_param.
17468 2017-07-03 Richard Biener <rguenther@suse.de>
17470 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
17471 back to using VIEW_CONVERT_EXPR.
17473 2017-07-03 Martin Liska <mliska@suse.cz>
17476 * doc/extend.texi: Document when a resolver function is
17477 generated for target_clones.
17479 2017-07-03 Martin Liska <mliska@suse.cz>
17481 * asan.c (asan_emit_stack_protection): Unpoison just red zones
17482 and shadow memory of auto variables which are subject of
17483 use-after-scope sanitization.
17484 (asan_expand_mark_ifn): Add do set only when is_poison.
17486 2016-07-03 Richard Biener <rguenther@suse.de>
17488 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
17490 (vect_force_simple_reduction): Record reduction def -> phi mapping.
17491 (vectorizable_reduction): Perform reduction PHI creation when
17492 visiting a reduction PHI and adjust and simplify code generation
17493 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
17494 (vect_transform_loop): Visit reduction PHIs.
17495 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
17496 defs into the SLP tree.
17497 (vect_build_slp_tree): Reduction defs terminate the recursion.
17498 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
17500 (vect_get_vec_defs_for_stmt_copy): Export.
17501 (vect_get_vec_defs): Likewise.
17502 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
17504 (vect_get_vec_defs_for_stmt_copy): Declare.
17505 (vect_get_vec_defs): Likewise.
17507 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17509 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
17510 parameter with a "loop" parameter and use it instead of the
17511 loop containing DR_STMT. Don't check simple_iv when doing
17512 BB analysis. Describe the two analysis modes in the comment.
17514 2017-07-03 Tom de Vries <tom@codesourcery.com>
17516 PR tree-optimization/69468
17517 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
17518 (find_same_succ_bb): Handle ignore_edge_flags.
17520 2017-07-03 Tom de Vries <tom@codesourcery.com>
17522 PR tree-optimization/81192
17523 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
17525 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
17527 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
17529 2017-07-03 Tom de Vries <tom@codesourcery.com>
17531 PR tree-optimization/81192
17532 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
17533 BB_SAME_SUCC (bb) == NULL.
17535 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17537 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
17540 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17542 * dumpfile.c: Include profile-count.h
17543 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
17545 (insert_cond_bb): Update profile.
17546 * tree-cfg.h (insert_cond_bb): Update prototype.
17547 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
17548 * tree-dump.c: Do not include tree-cfg.
17550 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17552 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
17554 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17556 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
17559 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17561 * tree-complex.c (expand_complex_div_wide): update profile.
17563 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17564 Alan Hayward <alan.hayward@arm.com>
17565 David Sherwood <david.sherwood@arm.com>
17567 * Makefile.in (MACHMODE_H): Remove insn-modes.h
17568 (CORETYPES_H): New define.
17569 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
17570 (insn-modes-inline.h, s-modes-inline-h): New rules.
17571 (generated_files): Add insn-modes-inline.h.
17572 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
17573 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
17574 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
17575 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
17576 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
17577 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
17578 (build/gencodes.o, build/genconditions.o): Likewise.
17579 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
17580 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
17581 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
17582 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
17583 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
17584 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
17585 * coretypes.h: Include everything up to real.h for generators.
17586 Include insn-modes.h first. Include wide-int-print.h after
17587 wide-int.h. Include insn-modes-inline.h and then machmode.h.
17588 * machmode.h: Don't include insn-modes.h here.
17589 * function-tests.c: Remove includes of signop.h, machmode.h,
17590 double-int.h and wide-int.h.
17592 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
17594 * optc-save-gen.awk: Likewise.
17595 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
17596 * godump.c: Remove include of wide-int-print.h.
17597 * pretty-print.h: Likewise.
17598 * wide-int-print.cc: Likewise.
17599 * wide-int.cc: Likewise.
17600 * hash-map-tests.c: Remove include of signop.h.
17601 * hash-set-tests.c: Likewise.
17602 * rtl-tests.c: Likewise.
17603 * mkconfig.sh: Remove include of machmode.h.
17604 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
17606 (emit_insn_modes_inline_h): ...this new function. Emit the code
17607 into an insn-modes-inline.h header file, adding appropriate
17608 include guards and end comments.
17609 (emit_insn_modes_c_header): Remove include of machmode.h.
17610 (emit_min_insn_modes_c_header): Include coretypes.h rather than
17612 (main): Handle -i flag and call emit_insn_modes_inline_h when
17615 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17617 * tree-ssa-strlen.c (strinfo): Rename the length field to
17618 nonzero_chars. Add a full_string_p field.
17619 (compare_nonzero_chars, zero_length_string_p): New functions.
17620 (get_addr_stridx): Add an offset_out parameter.
17621 Use compare_nonzero_chars.
17622 (get_stridx): Update accordingly. Use compare_nonzero_chars.
17623 (new_strinfo): Update after above changes to strinfo.
17624 (set_endptr_and_length): Set full_string_p.
17625 (get_string_length): Update after above changes to strinfo.
17626 (unshare_strinfo): Update call to new_strinfo.
17627 (maybe_invalidate): Likewise.
17628 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
17629 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
17630 as a uhwi instead of an shwi. Update after above changes to
17631 strinfo and new_strinfo.
17632 (zero_length_string): Assert that chainsi contains full strings.
17633 Use zero_length_string_p. Update call to new_strinfo.
17634 (adjust_related_strinfos): Update after above changes to strinfo.
17635 Copy full_string_p from origsi.
17636 (adjust_last_stmt): Use zero_length_string_p.
17637 (handle_builtin_strlen): Update after above changes to strinfo and
17638 new_strinfo. Install the lhs as the string length if the previous
17639 entry didn't describe a full string.
17640 (handle_builtin_strchr): Update after above changes to strinfo
17642 (handle_builtin_strcpy): Likewise.
17643 (handle_builtin_strcat): Likewise.
17644 (handle_builtin_malloc): Likewise.
17645 (handle_pointer_plus): Likewise.
17646 (handle_builtin_memcpy): Likewise. Track nonzero characters
17647 that aren't necessarily followed by a nul terminator.
17648 (handle_char_store): Likewise.
17650 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17652 PR tree-optimization/80769
17653 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
17654 for malloc and calloc. Document the new invariant that all related
17655 strinfos have delayed lengths or none do.
17656 (verify_related_strinfos): Move earlier in file.
17657 (set_endptr_and_length): New function, split out from...
17658 (get_string_length): ...here. Also set the lengths of related
17660 (zero_length_string): Assert that chainsi has known (rather than
17662 (adjust_related_strinfos): Likewise.
17664 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17666 PR tree-optimization/81136
17667 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
17668 assert that two references with the same misalignment have the same
17669 compile-time misalignment if those compile-time misalignments
17672 2017-07-01 Andi Kleen <ak@linux.intel.com>
17674 * print-tree.c (print_node): Print all attributes.
17676 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17678 * cfg.c (scale_bbs_frequencies): New function.
17679 * cfg.h (scale_bbs_frequencies): Declare it.
17680 * cfgloopanal.c (single_likely_exit): Cleanup.
17681 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
17683 (scale_loop_profile): Likewise.
17684 (loop_version): Likewise.
17685 (create_empty_loop_on_edge): Update.
17686 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
17687 scale_loop_frequencies, scale_loop_profile, loopify,
17688 loop_version): Update prototypes.
17689 * modulo-sched.c (sms_schedule): Update.
17690 * predict.c (unlikely_executed_edge_p): Also check probability.
17691 (probably_never_executed_edge_p): Fix typo.
17692 * tree-if-conv.c (version_loop_for_if_conversion): Update.
17693 * tree-parloops.c (gen_parallel_loop): Update.
17694 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
17695 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
17696 * tree-ssa-loop-split.c (split_loop): Update.
17697 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
17698 * tree-vect-loop-manip.c (vect_do_peeling): Update.
17699 (vect_loop_versioning): Update.
17700 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
17702 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17704 * trans-mem.c (split_bb_make_tm_edge): Update profile.
17706 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17708 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
17709 to keep profile consistent.
17711 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17713 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
17714 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
17715 * profile-count.h (max_safe_multiplier): Make unsigned.
17716 (profile_count::guessed_zero): New.
17718 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17720 * bb-reorder.c (fix_up_crossing_landing_pad,
17721 fix_crossing_conditional_branches): Use make_single_succ_edge
17722 to keep profile consistent.
17724 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17726 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
17729 2017-07-01 Jakub Jelinek <jakub@redhat.com>
17732 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
17733 the right scopes, make sure cond_jump isn't preserved between multiple
17734 iterations. Search for fallthru edge whenever there are 3+ edges and
17735 use find_fallthru_edge for it.
17737 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17739 Patch by Alexander Monakov <amonakov@ispras.ru>
17740 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
17741 probabilities consistently.
17743 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17745 * pa.c (pa_expand_compare_and_swap_loop): Update call of
17746 emit_cmp_and_jump_insns.
17748 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17751 * tree-inline.c (expand_call_inline): Combine profile statuses.
17753 2017-06-30 Andrew Pinski <apinski@cavium.com>
17755 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
17756 fold_stmt returned true.
17758 2017-06-30 Nathan Sidwell <nathan@acm.org>
17760 * ggc.h (empty_string): Delete.
17761 * cfgexpand.c (expand_asm_stmt): Use plain "".
17762 * optabs.c (expand_asm_memory_barrier): Likewise.
17763 * stringpool.c (empty_string): Delete.
17764 (digit_vector, digit_string): Delete.
17765 (ggc_alloc_string): Use plain "", don't optimize single digit
17766 strings. Use ggc_alloc_atomic.
17768 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
17770 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
17771 comparison set and one other set, use the cost of the non-comparison
17774 2017-06-30 Nathan Sidwell <nathan@acm.org>
17776 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
17779 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
17781 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
17782 loops. Remove now unneeded calls to gimple_switch_set_label() that
17783 just set removed labels to NULL_TREE.
17785 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
17787 * tree-ssanames.c (set_range_info_raw): Abstract from ...
17788 (set_range_info): ...here. Only call set_range_info_raw if domain
17790 (set_nonzero_bits): Call set_range_info_raw.
17791 * tree-ssanames.h (set_range_info_raw): New.
17793 2017-06-30 Jakub Jelinek <jakub@redhat.com>
17796 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
17797 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
17798 of nonimmediate_operand and <store_mask_constraint> instead of m
17799 for the input operand. For V8FI iterator, always split if input
17800 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
17801 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
17802 <store_mask_predicate> instead of register_operand and
17803 <store_mask_constraint> instead of v for the input operand. Make
17804 sure both operands aren't MEMs for if not <mask_applied>.
17806 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
17808 * lto-wrapper.c (copy_file) Close both file descriptors before
17811 2017-06-30 Martin Liska <mliska@suse.cz>
17814 * multiple_target.c (create_dispatcher_calls): Make ifunc
17815 also for function that don't have calls or are not referenced.
17817 2017-06-30 Richard Biener <rguenther@suse.de>
17819 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
17820 analyze the first scalar stmt. Move vector type computation
17821 for the BB case here from ...
17822 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
17823 live operation processing in the SLP case properly.
17825 2017-06-30 Richard Biener <rguenther@suse.de>
17827 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
17829 2017-06-30 Martin Liska <mliska@suse.cz>
17832 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
17833 before BUILT_IN_UNWIND_RESUME when ASAN is used.
17835 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
17837 * doc/invoke.texi (AArch64): Add missing options and remove redundant
17840 2017-06-30 Richard Biener <rguenther@suse.de>
17842 PR tree-optimization/81249
17843 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
17844 condition reduction result to original scalar type.
17846 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17848 * profile-count.h (enum profile_quality): Fix typos and whitespace
17851 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17853 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
17854 type for branch probabilities.
17856 2017-06-29 Julian Brown <julian@codesourcery.com>
17857 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17859 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
17860 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
17861 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
17862 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
17864 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17866 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
17867 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
17868 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
17869 CC usage from generic code to here.
17870 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
17871 CC usage into the target macros.
17873 2017-06-29 Maya Rashish <coypu@sdf.org>
17875 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
17878 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17880 * arm/arm-builtins.c: Include profile-count.h
17881 * except.c (sjlj_emit_function_enter): Use
17882 profile_probability::unlikely.
17884 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17886 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
17887 and tocrel_offset be pointer args rather than implicitly using
17889 (legitimate_constant_pool_address_p, rs6000_emit_move,
17890 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
17891 tocrel_offset and use in toc_relative_expr_p call.
17892 (print_operand, print_operand_address): Use static tocrel_base_oac
17893 and tocrel_offset_oac.
17894 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
17897 2017-06-29 Maya Rashish <coypu@sdf.org>
17899 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
17901 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
17903 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
17904 objects, take into account only the alignment of 'op0' and 'mode1' if
17907 2017-06-29 Steve Ellcey <sellcey@cavium.com>
17909 * ccmp.c (ccmp_tree_comparison_p): New function.
17910 (ccmp_candidate_p): Update to use above function.
17911 (get_compare_parts): New function.
17912 (expand_ccmp_next): Update to use new functions.
17913 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
17915 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
17916 take mode as argument.
17917 * ccmp.h (expand_ccmp_expr): Add mode as argument.
17918 * expr.c (expand_expr_real_1): Pass mode as argument.
17920 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
17922 * combine.c (combine_instructions): Print insns to dump_file, together
17925 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17927 * asan.c (asan_emit_stack_protection): Update.
17928 (create_cond_insert_point): Update.
17929 * auto-profile.c (afdo_propagate_circuit): Update.
17930 * basic-block.h (struct edge_def): Turn probability to
17931 profile_probability.
17932 (EDGE_FREQUENCY): Update.
17933 * bb-reorder.c (find_traces_1_round): Update.
17934 (better_edge_p): Update.
17935 (sanitize_hot_paths): Update.
17936 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
17937 (make_single_succ_edge): Update.
17938 (check_bb_profile): Update.
17939 (dump_edge_info): Update.
17940 (update_bb_profile_for_threading): Update.
17941 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
17942 probabilitycount to 0.
17943 * cfgbuild.c (compute_outgoing_frequencies): Update.
17944 * cfgcleanup.c (try_forward_edges): Update.
17945 (outgoing_edges_match): Update.
17946 (try_crossjump_to_edge): Update.
17947 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
17948 (expand_gimple_tailcall): Update.
17949 (construct_init_block): Use make_single_succ_edge.
17950 (construct_exit_block): Use make_single_succ_edge.
17951 * cfghooks.c (verify_flow_info): Update.
17952 (redirect_edge_succ_nodup): Update.
17953 (split_edge): Update.
17954 (account_profile_record): Update.
17955 * cfgloopanal.c (single_likely_exit): Update.
17956 * cfgloopmanip.c (scale_loop_profile): Update.
17957 (set_zero_probability): Remove.
17958 (duplicate_loop_to_header_edge): Update.
17959 * cfgloopmanip.h (loop_version): Update prototype.
17960 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
17961 (force_nonfallthru_and_redirect): Update.
17962 (update_br_prob_note): Update.
17963 (rtl_verify_edges): Update.
17964 (purge_dead_edges): Update.
17965 (rtl_lv_add_condition_to_bb): Update.
17966 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
17967 * cgraphunit.c (init_lowered_empty_function): Update.
17968 (cgraph_node::expand_thunk): Update.
17969 * cilk-common.c: Include profile-count.h
17970 * dojump.c (inv): Remove.
17971 (jumpifnot): Update.
17972 (jumpifnot_1): Update.
17973 (do_jump_1): Update.
17975 (do_jump_by_parts_greater_rtx): Update.
17976 (do_compare_rtx_and_jump): Update.
17977 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
17978 do_jump_1. do_compare_rtx_and_jump): Update prototype.
17979 * dwarf2cfi.c: Include profile-count.h
17980 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
17981 (sjlj_emit_dispatch_table): Likewise.
17982 * explow.c: Include profile-count.h
17983 * expmed.c (emit_store_flag_force): Update.
17984 (do_cmp_and_jump): Update.
17985 * expr.c (compare_by_pieces_d::generate): Update.
17986 (compare_by_pieces_d::finish_mode): Update.
17987 (emit_block_move_via_loop): Update.
17988 (store_expr_with_bounds): Update.
17989 (store_constructor): Update.
17990 (expand_expr_real_2): Update.
17991 (expand_expr_real_1): Update.
17992 * expr.h (try_casesi, try_tablejump): Update prototypes.
17993 * gimple-pretty-print.c (dump_probability): Update.
17994 (dump_profile): New.
17995 (dump_gimple_label): Update.
17996 (dump_gimple_bb_header): Update.
17997 * graph.c (draw_cfg_node_succ_edges): Update.
17998 * hsa-gen.c (convert_switch_statements): Update.
17999 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
18000 (find_if_case_1): Update.
18001 (find_if_case_2): Update.
18002 * internal-fn.c (expand_arith_overflow_result_store): Update.
18003 (expand_addsub_overflow): Update.
18004 (expand_neg_overflow): Update.
18005 (expand_mul_overflow): Update.
18006 (expand_vector_ubsan_overflow): Update.
18007 * ipa-cp.c (good_cloning_opportunity_p): Update.
18008 * ipa-split.c (split_function): Use make_single_succ_edge.
18009 * ipa-utils.c (ipa_merge_profiles): Update.
18010 * loop-doloop.c (add_test): Update.
18011 (doloop_modify): Update.
18012 * loop-unroll.c (compare_and_jump_seq): Update.
18013 (unroll_loop_runtime_iterations): Update.
18014 * lra-constraints.c (lra_inheritance): Update.
18015 * lto-streamer-in.c (input_cfg): Update.
18016 * lto-streamer-out.c (output_cfg): Update.
18017 * mcf.c (adjust_cfg_counts): Update.
18018 * modulo-sched.c (sms_schedule): Update.
18019 * omp-expand.c (expand_omp_for_init_counts): Update.
18020 (extract_omp_for_update_vars): Update.
18021 (expand_omp_ordered_sink): Update.
18022 (expand_omp_for_ordered_loops): Update.
18023 (expand_omp_for_generic): Update.
18024 (expand_omp_for_static_nochunk): Update.
18025 (expand_omp_for_static_chunk): Update.
18026 (expand_cilk_for): Update.
18027 (expand_omp_simd): Update.
18028 (expand_omp_taskloop_for_outer): Update.
18029 (expand_omp_taskloop_for_inner): Update.
18030 * omp-simd-clone.c (simd_clone_adjust): Update.
18031 * optabs.c (expand_doubleword_shift): Update.
18032 (expand_abs): Update.
18033 (emit_cmp_and_jump_insn_1): Update.
18034 (expand_compare_and_swap_loop): Update.
18035 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
18036 * predict.c (predictable_edge_p): Update.
18037 (edge_probability_reliable_p): Update.
18038 (set_even_probabilities): Update.
18039 (combine_predictions_for_insn): Update.
18040 (combine_predictions_for_bb): Update.
18041 (propagate_freq): Update.
18042 (estimate_bb_frequencies): Update.
18043 (force_edge_cold): Update.
18044 * profile-count.c (profile_count::dump): Add missing space into dump.
18045 (profile_count::debug): Add newline.
18046 (profile_count::differs_from_p): Explicitly convert to unsigned.
18047 (profile_count::stream_in): Update.
18048 (profile_probability::dump): New member function.
18049 (profile_probability::debug): New member function.
18050 (profile_probability::differs_from_p): New member function.
18051 (profile_probability::differs_lot_from_p): New member function.
18052 (profile_probability::stream_in): New member function.
18053 (profile_probability::stream_out): New member function.
18054 * profile-count.h (profile_count_quality): Rename to ...
18055 (profile_quality): ... this one.
18056 (profile_probability): New.
18057 (profile_count): Update.
18058 * profile.c (compute_branch_probabilities): Update.
18059 * recog.c (peep2_attempt): Update.
18060 * sched-ebb.c (schedule_ebbs): Update.
18061 * sched-rgn.c (find_single_block_region): Update.
18062 (compute_dom_prob_ps): Update.
18063 (schedule_region): Update.
18064 * sel-sched-ir.c (compute_succs_info): Update.
18065 * stmt.c (struct case_node): Update.
18066 (do_jump_if_equal): Update.
18067 (get_outgoing_edge_probs): Update.
18068 (conditional_probability): Update.
18069 (emit_case_dispatch_table): Update.
18070 (expand_case): Update.
18071 (expand_sjlj_dispatch_table): Update.
18072 (emit_case_nodes): Update.
18073 * targhooks.c: Update.
18074 * tracer.c (better_p): Update.
18075 (find_best_successor): Update.
18076 * trans-mem.c (expand_transaction): Update.
18077 * tree-call-cdce.c: Update.
18078 * tree-cfg.c (gimple_split_edge): Upate.
18079 (move_sese_region_to_fn): Upate.
18080 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
18081 * tree-eh.c (lower_resx): Upate.
18082 (cleanup_empty_eh_move_lp): Upate.
18083 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18084 * tree-inline.c (copy_edges_for_bb): Update.
18085 (copy_cfg_body): Update.
18086 * tree-parloops.c (gen_parallel_loop): Update.
18087 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
18088 (gimple_gen_time_profiler): Update.
18089 * tree-ssa-dce.c (remove_dead_stmt): Update.
18090 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
18091 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
18092 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
18093 (unloop_loops): Update.
18094 (try_peel_loop): Update.
18095 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18096 * tree-ssa-loop-split.c (connect_loops): Update.
18097 (split_loop): Update.
18098 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
18099 (hoist_guard): Update.
18100 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
18101 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
18102 (value_replacement): Update.
18103 * tree-ssa-reassoc.c (branch_fixup): Update.
18104 * tree-ssa-tail-merge.c (replace_block_by): Update.
18105 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
18106 (create_edge_and_update_destination_phis): Update.
18107 (compute_path_counts): Update.
18108 (recompute_probabilities): Update.
18109 (update_joiner_offpath_counts): Update.
18110 (freqs_to_counts_path): Update.
18111 (duplicate_thread_path): Update.
18112 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
18113 (struct switch_conv_info): Update.
18114 (gen_inbound_check): Update.
18115 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
18116 (vect_do_peeling): Update.
18117 (vect_loop_versioning): Update.
18118 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18119 (optimize_mask_stores): Update.
18120 * ubsan.c (ubsan_expand_null_ifn): Update.
18121 * value-prof.c (gimple_divmod_fixed_value): Update.
18122 (gimple_divmod_fixed_value_transform): Update.
18123 (gimple_mod_pow2): Update.
18124 (gimple_mod_pow2_value_transform): Update.
18125 (gimple_mod_subtract): Update.
18126 (gimple_mod_subtract_transform): Update.
18127 (gimple_ic): Update.
18128 (gimple_stringop_fixed_value): Update.
18129 (gimple_stringops_transform): Update.
18130 * value-prof.h: Update.
18132 2017-06-29 Carl Love <cel@us.ibm.com>
18134 * config/rs6000/rs6000-c.c: Add support for built-in functions
18135 vector signed int vec_signed (vector float);
18136 vector signed long long vec_signed (vector double);
18137 vector signed int vec_signed2 (vector double, vector double);
18138 vector signed int vec_signede (vector double);
18139 vector signed int vec_signedo (vector double);
18140 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
18141 instruction generator.
18142 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18143 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
18144 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
18146 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
18147 vunsignede_v2df): Add define_expands.
18148 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
18149 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
18150 VEC_UNSIGNEDO): Add definitions.
18151 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18152 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
18153 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
18154 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
18155 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
18156 * config/rs6000/altivec.h (vec_signed, vec_signed2,
18157 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
18158 vec_unsignede, vec_unsignedo): Add builtin defines.
18159 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
18161 * doc/extend.texi: Update the built-in documentation file for the
18162 new built-in functions.
18164 2017-06-29 Richard Biener <rguenther@suse.de>
18166 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
18167 reduction chains to LOOP_VINFO_REDUCTIONS.
18168 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
18169 SLP reductions after processing reduction chains.
18171 2017-06-29 Nathan Sidwell <nathan@acm.org>
18173 * builtins.c (fold_builtin_FUNCTION): Use
18174 lang_hooks.decl_printable_name.
18176 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
18178 PR middle-end/81194
18179 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
18180 with only one label.
18181 * stmt.c (expand_case): Assert NCASES is greater than one.
18183 2017-06-29 Richard Biener <rguenther@suse.de>
18185 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
18187 (group_case_labels): Likewise.
18188 (find_taken_edge): Push sanity checking on val to workers...
18189 (find_taken_edge_cond_expr): ... here
18190 (find_taken_edge_switch_expr): ... and here, handle cases
18191 with just a default label.
18192 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
18193 (group_case_labels): Likewise.
18194 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
18195 group_case_labels does anything cleanup the CFG again.
18197 2017-06-29 Bin Cheng <bin.cheng@arm.com>
18199 PR tree-optimization/81196
18200 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
18201 exit condition comparing two IVs.
18203 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
18205 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
18206 profile to the dummy entry at the end of the list of architectures.
18207 * config/arm/arm-cpu-cdata.h: Regenerated.
18209 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18210 Michael Collison <michael.collison@arm.com>
18213 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
18215 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
18216 (*aarch64_reg_<mode>3_minus_mask): New pattern.
18217 (*aarch64_<optab>_reg_di3_mask2): New pattern.
18218 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
18219 of shift when the shift amount is masked with constant equal to
18220 the size of the mode.
18221 * config/aarch64/predicates.md (subreg_lowpart_operator): New
18224 2017-06-29 Martin Liska <mliska@suse.cz>
18226 * config/i386/i386.opt: Change range from [1,5] to [0,5].
18228 2017-06-29 Yury Gribov <tetra2005@gmail.com>
18231 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
18234 (edge_growth_cache_entry::edge_growth_cache_entry): New
18236 (reset_edge_growth_cache): Update to use constructor.
18238 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18240 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
18241 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
18242 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
18244 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
18246 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
18247 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
18249 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
18251 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
18252 (*-linux-uclibc*): Add t-uclibc tmake_file.
18253 * config/t-musl: New.
18254 * config/t-uclibc: New.
18256 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
18258 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
18260 (gen_comm_data): Emit architectural setting of arch_prof.
18261 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
18263 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
18264 (armv8-m.base, armv8-m.main): Likewise.
18265 * arm-protos.h (arm_build_target): Add profile field.
18266 (arch_option): Likewise.
18267 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
18269 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
18270 arm_active_target.profile.
18272 2017-06-28 Richard Biener <rguenther@suse.de>
18274 PR middle-end/81227
18275 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
18276 TYPE_OVERFLOW_WRAPS.
18277 * match.pd (negate_expr_p): Likewise.
18278 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
18279 fold_build2, not fold_binary.
18281 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
18283 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
18284 Convert memory address to Pmode.
18285 (aarch64_print_operand): Assert MEM operands are always Pmode.
18287 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
18290 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
18291 Remove redundant if.
18292 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
18293 * config/arm/aarch-common-protos.h
18294 (aarch_forward_to_shift_is_not_shifted_re): Remove.
18295 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
18297 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
18300 * multiple_target.c (create_dispatcher_calls): Set the default
18301 clone to be static, not public.
18303 2017-06-28 Richard Biener <rguenther@suse.de>
18305 * tree-vect-loop.c (vectorizable_reduction): Move special
18306 cond reduction IV var creation ...
18307 (vect_create_epilog_for_reduction): ... here. Remove induction_index
18308 parameter. Use STMT_VINFO_VECTYPE.
18309 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
18312 2017-06-28 Martin Liska <mliska@suse.cz>
18315 * ipa-visibility.c (non_local_p): Handle visibility.
18317 2017-06-28 Martin Liska <mliska@suse.cz>
18320 * common.opt: Add IntegerRange to various options.
18321 * opt-functions.awk (integer_range_info): New function.
18322 * optc-gen.awk: Add integer_range_info to cl_options struct.
18323 * opts-common.c (decode_cmdline_option): Handle
18324 CL_ERR_INT_RANGE_ARG.
18325 (cmdline_handle_error): Likewise.
18326 * opts.c (print_filtered_help): Show valid interval in
18327 when --help is provided.
18328 * opts.h (struct cl_option): Add range_min and range_max fields.
18329 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
18331 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
18333 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
18334 (x * C EQ/NE y * C): New transformation.
18336 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
18338 * genmultilib (combination_space): Accept '+' in option names.
18340 2017-06-28 Martin Liska <mliska@suse.cz>
18343 * asan.c (instrument_derefs): Bail out inner references
18344 that are hard register variables.
18346 2017-06-28 Jakub Jelinek <jakub@redhat.com>
18349 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
18350 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
18352 2017-06-28 Richard Biener <rguenther@suse.de>
18354 * tree-vectorizer.h (vect_get_vec_defs): Remove.
18355 (vect_get_slp_defs): Adjust.
18356 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
18358 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
18360 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
18361 get_initial_defs_for_reduction instead of vect_get_vec_defs.
18362 (vectorizable_reduction): Adjust.
18363 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
18365 (vect_get_slp_defs): Likewise.
18366 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
18367 (vectorizable_bswap): Adjust.
18368 (vectorizable_call): Likewise.
18369 (vectorizable_conversion): Likewise.
18370 (vectorizable_assignment): Likewise.
18371 (vectorizable_shift): Likewise.
18372 (vectorizable_operation): Likewise.
18373 (vectorizable_store): Likewise.
18374 (vectorizable_condition): Likewise.
18375 (vectorizable_comparison): Likewise.
18377 2017-06-28 Michael Collison <michael.collison@arm.com>
18380 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
18382 (arm_gen_movmemqi): Removed unused variable 'i'.
18383 Convert 'for' loop into 'while' loop.
18384 (arm_expand_prologue): Remove last unnecessary set of insn.
18385 (thumb_pop): Remove unused variable 'pushed_words'.
18386 (thumb_exit): Remove last unnecessary set of regs_to_pop.
18388 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18390 * config/s390/predicates.md: Use s390_rel_address_ok_p.
18391 * config/s390/s390-protos.h: Add prototype of
18392 s390_rel_address_ok_p.
18393 * config/s390/s390.c (s390_got_symbol): New function.
18394 (s390_rel_address_ok_p): New function.
18395 (legitimize_pic_address): Use s390_rel_address_ok_p.
18396 (s390_load_got): Use s390_got_symbol.
18397 (s390_option_override): Issue error if
18398 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
18399 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
18401 * config/s390/s390.opt: New option mpic-data-is-text-relative.
18403 2017-06-27 Andrew Pinski <apinski@cavium.com>
18405 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
18406 (X * copysign (1.0, X)): New pattern.
18407 (X * copysign (1.0, -X)): New pattern.
18408 (copysign (-1.0, CST)): New pattern.
18410 2017-06-27 Joseph Myers <joseph@codesourcery.com>
18412 * genmultilib (combination_space): Remove variable.
18413 Validate reuse rules against regular expression for any sequence
18414 of multilib options in any order.
18416 2017-06-27 Michael Collison <michael.collison@arm.com>
18418 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
18419 call aarch64_split_simd_combine.
18420 * (aarch64_combine_internal<mode>): Delete pattern.
18421 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
18422 Allow register and subreg operands.
18424 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18426 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
18427 specific need, just fallback on defaults.
18428 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
18430 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18431 Olivier Hainque <hainque@adacore.com>
18433 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
18435 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
18436 targets. Pick a default if no particular attempt applied.
18437 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
18440 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18442 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
18443 (x86_64-wrs-vxworks7): Likewise.
18445 2017-06-27 Marek Polacek <polacek@redhat.com>
18448 * ubsan.c (instrument_null): Check get_base_address's result for null.
18450 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18452 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
18454 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18456 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
18457 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
18458 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
18459 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
18460 New function types.
18461 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
18462 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
18463 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
18464 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
18465 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
18466 BUILT_IN_FEUPDATEENV): New builtins.
18467 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
18468 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
18469 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
18470 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
18472 (builtin_structptr_types): Adjust size.
18473 * tree.c (builtin_structptr_types): Add four entries.
18475 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18476 Olivier Hainque <hainque@adacore.com>
18478 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
18479 (TLS_SYM): New local macro, forcing reference to __tls__ on
18480 link command lines for VxWorks 7 RTPs, triggering initialization
18482 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
18483 OS features TLS support, true for RTPs on VxWorks 7.
18484 * config/vxworks.c (vxworks_override_options): Setup emutls
18487 2017-06-27 Jakub Jelinek <jakub@redhat.com>
18489 * predict.c (test_prediction_value_range): Use -1U instead of -1
18490 to avoid narrowing conversion warning.
18491 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
18492 to avoid narrowing conversion warning.
18493 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
18495 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
18497 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18499 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
18500 64bit configurations.
18501 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
18502 (SIZE_TYPE): Likewise.
18503 * config/vxworks.c (vxworks_emutls_var_fields): Use
18504 long_unsigned_type_node instead of unsigned_type_node as the offset
18505 field type, which is "pointer" mode in emutls.c.
18507 2017-06-27 Jakub Jelinek <jakub@redhat.com>
18510 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
18512 PR middle-end/81207
18513 * gimple-fold.c (replace_call_with_call_and_fold): Handle
18514 gimple_vuse copying separately from gimple_vdef copying.
18516 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18518 * value-prof.c (free_hist): Remove call to memset and the enclosing if
18521 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
18522 Olivier Hainque <hainque@adacore.com>
18524 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
18525 for all vxworks7 targets.
18526 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
18527 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
18528 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
18529 variations for VX6/VX7 and 32/64bits later on in ...
18530 (VXWORKS_LIB_SPEC): Leverage new macros.
18531 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
18532 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
18534 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
18536 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
18537 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
18539 2017-06-26 Carl Love <cel@us.ibm.com>
18541 * config/rs6000/rs6000-c.c: Add support for built-in functions
18542 vector bool char vec_reve (vector bool char);
18543 vector signed char vec_reve (vector signed char);
18544 vector unsigned char vec_reve (vector unsigned char);
18545 vector bool int vec_reve (vector bool int);
18546 vector signed int vec_reve (vector signed int);
18547 vector unsigned int vec_reve (vector unsigned int);
18548 vector bool long long vec_reve (vector bool long long);
18549 vector signed long long vec_reve (vector signed long long);
18550 vector unsigned long long vec_reve (vector unsigned long long);
18551 vector bool short vec_reve (vector bool short);
18552 vector signed short vec_reve (vector signed short);
18553 vector double vec_reve (vector double);
18554 vector float vec_reve (vector float);
18555 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
18556 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
18557 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
18558 (altivec_vreve): New pattern.
18559 * config/rs6000/altivec.h (vec_reve): New define.
18560 * doc/extend.texi (vec_rev): Update the built-in documentation file
18561 for the new built-in functions.
18563 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18565 PR tree-optimization/71815
18566 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
18568 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
18570 (slsr_process_phi): Likewise.
18571 (replace_uncond_cands_and_profitable_phis): Don't replace a
18572 multiply candidate with a stride of 1 (copy or cast).
18573 (phi_incr_cost): Call uses_consumed_by_stmt rather than
18575 (lowest_cost_path): Likewise.
18576 (total_savings): Likewise.
18578 2017-06-26 Richard Biener <rguenther@suse.de>
18581 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
18582 Use def_builtin_pure for all gather builtins.
18584 2017-06-26 Richard Biener <rguenther@suse.de>
18586 PR tree-optimization/81203
18587 * tree-tailcall.c (find_tail_calls): Do not move stmts into
18588 non-dominating BBs.
18590 2017-06-26 Marek Polacek <polacek@redhat.com>
18593 * doc/invoke.texi: Document -Wmultistatement-macros.
18595 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
18597 * doc/sourcebuild.texi (ARM-specific attributes): Document new
18598 arm_neon_ok_no_float_abi effective target.
18600 2017-06-26 Richard Biener <rguenther@suse.de>
18602 PR tree-optimization/80928
18603 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
18604 (copy_bbs): Set BB_DUPLICATED flag early.
18605 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
18607 (execute_on_shrinking_pred): Likewise.
18608 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
18609 BB_DUPLICATED blocks.
18610 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
18611 iterate over all PHIs considering removal of *gsi.
18613 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
18615 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
18618 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18620 * config/rs6000/rs6000-string.c: (expand_block_clear,
18621 do_load_for_compare, select_block_compare_mode,
18622 compute_current_alignment, expand_block_compare,
18623 expand_strncmp_align_check, expand_strn_compare,
18624 expand_block_move, rs6000_output_load_multiple)
18625 Move functions related to string/block move/compare
18626 to a separate file.
18627 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
18628 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
18629 for this function which is now used in two files.
18630 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
18631 * config.gcc: Add rs6000-string.o to extra_objs for
18632 targets powerpc*-*-* and rs6000*-*-*.
18634 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
18637 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
18638 32-bit, since indexed is not valid for DImode.
18639 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
18640 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
18641 (define_peephole2 for Altivec d-form load): Add 32-bit support.
18642 (define_peephole2 for Altivec d-form store): Likewise.
18645 * multiple_target.c (create_dispatcher_calls): Only create the
18646 dispatcher call if the function is the default clone of a
18647 versioned function.
18649 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
18651 PR middle-end/80902
18652 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
18653 a call, force the call to not be a tail call.
18655 2017-06-23 Jeff Law <law@redhat.com>
18657 * doc/contrib.texi: Add entry for Steven Pemberton's work on
18660 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
18662 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
18663 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
18664 handling for early expansion of vector shifts (sl,sr,sra,rl).
18665 (builtin_function_type): Add vector shift right instructions
18666 to the unsigned argument list.
18668 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
18670 rtl-optimizatoin/79286
18671 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
18672 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
18673 trap. PIC register plus a const unspec without offset can never trap.
18675 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
18677 * tree.h (builtin_structptr_type): New type.
18678 (builtin_structptr_types): Declare new array.
18679 * tree.c (builtin_structptr_types): New array.
18680 (free_lang_data, build_common_tree_nodes): Use it.
18682 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
18685 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
18688 2017-06-22 Matt Turner <mattst88@gmail.com>
18690 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
18691 Lake models to skylake case. Assume skylake for unknown
18692 models with clflushopt.
18694 2017-06-22 Jeff Law <law@redhat.com>
18696 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
18697 frame sizes that do not satisfy aarch64_uimm12_shift.
18699 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
18701 * profile-count.h (apply_probability,
18702 apply_scale, probability_in): Fix checks for zero.
18704 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18706 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
18707 * doc/cppdiropts.texi (-I @var{dir}): Document it.
18709 2016-06-22 Richard Biener <rguenther@suse.de>
18711 * tree-vect-loop.c (vect_model_reduction_cost): Handle
18712 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
18713 REDUC_MAX_EXPR support.
18714 (vectorizable_reduction): Likewise.
18715 (vect_create_epilog_for_reduction): Likewise.
18717 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
18719 * match.pd (A / (1 << B) -> A >> B): New.
18720 * generic-match-head.c: Include optabs-tree.h.
18721 * gimple-match-head.c: Likewise.
18722 * optabs-tree.h (target_supports_op_p): New.
18723 * optabs-tree.c (target_supports_op_p): New.
18725 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18727 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
18728 $gcc_cv_ld --help output.
18729 (gcc_cv_ld_demangle): Likewise.
18730 (gcc_cv_ld_eh_frame_hdr): Likewise.
18731 (gcc_cv_ld_pie): Likewise.
18732 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
18733 (gcc_cv_ld_buildid): Likewise.
18734 (gcc_cv_ld_sysroot): Likewise.
18735 (ld_bndplt_support): Likewise.
18736 (ld_pushpopstate_support): Likewise.
18737 * configure: Regenerate.
18738 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
18740 2017-06-21 Jakub Jelinek <jakub@redhat.com>
18743 * config/i386/sse.md (round<mode>2): Renumber match_dup and
18744 operands indexes to avoid gap between operands and match_dups.
18746 2017-06-21 Andrew Pinski <apinski@cavium.com>
18748 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
18749 Increment Arith_shift and Arith_shift_reg by 1.
18750 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
18752 * config/aarch64/aarch64.c (thunderx_tunings): Enable
18753 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
18754 (aarch64_strip_extend): Add new argument and test for it.
18755 (aarch64_cheap_mult_shift_p): New function.
18756 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
18757 add a cost if it is true.
18758 Update calls to aarch64_strip_extend.
18759 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
18761 2017-06-21 Andrew Pinski <apinski@cavium.com>
18763 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
18765 (thunderxt88): Likewise.
18766 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
18767 (thunderx_prefetch_tune): New variable.
18768 (thunderx2t99_prefetch_tune): Update for the correct values.
18769 (thunderxt88_tunings): New variable.
18770 (thunderx_tunings): Use thunderx_prefetch_tune instead of
18771 generic_prefetch_tune.
18772 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
18774 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18776 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
18777 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
18778 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
18779 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
18780 (aarch64_atomic_cas<mode>, GPI): Likewise.
18782 2017-06-21 Martin Liska <mliska@suse.cz>
18784 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
18785 statements on cold and hot labels.
18786 * predict.c (tree_estimate_probability_bb): Remove the
18787 prediction from this place.
18789 2017-06-21 Martin Liska <mliska@suse.cz>
18791 PR tree-optimization/79489
18792 * gimplify.c (maybe_add_early_return_predict_stmt): New
18794 (gimplify_return_expr): Call the function.
18795 * predict.c (tree_estimate_probability_bb): Remove handling
18797 * predict.def: Update comment about early return predictor.
18798 * gimple-predict.h (is_gimple_predict): New function.
18799 * predict.def: Change default value of early return to 66.
18800 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
18802 * passes.def: Put pass_strip_predict_hints to the beginning of
18805 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
18807 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
18808 FUNCTION_DECL declarations.
18809 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
18811 (dwaf2out_decl): Likewise.
18812 * godump.c (go_early_global_decl): Skip call to the real debug hook
18813 for FUNCTION_DECL declarations.
18814 * passes.c (rest_of_decl_compilation): Skip call to the
18815 early_global_decl debug hook for FUNCTION_DECL declarations, unless
18816 -fdump-go-spec is passed.
18818 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
18820 * config/i386/i386.c (struct builtin_isa): New field pure_p.
18821 Reorder for compactness.
18822 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
18823 (def_builtin_pure, def_builtin_pure2): New functions.
18824 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
18826 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
18828 * match.pd (nop_convert): New predicate.
18829 ((A +- CST1) +- CST2): Allow some NOP conversions.
18831 2017-06-21 Jakub Jelinek <jakub@redhat.com>
18834 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
18835 with ctors/dtors if GOVD_SHARED is set.
18837 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
18839 * config/aarch64/aarch64.md (movti_aarch64):
18840 Emit mov rather than orr.
18841 (movtf_aarch64): Likewise.
18842 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
18843 Emit mov rather than orr.
18845 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
18847 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
18848 Swap alternatives, make integer dup more expensive.
18850 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
18852 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
18853 Return true for non-tls symbols.
18855 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
18857 * config/aarch64/aarch64-cores.def (cortex-a55): New.
18858 (cortex-a75): Likewise.
18859 (cortex-a75.cortex-a55): Likewise.
18860 * config/aarch64/aarch64-tune.md: Regenerate.
18861 * doc/invoke.texi (-mtune): Document new values for -mtune.
18863 2017-06-21 Tom de Vries <tom@codesourcery.com>
18865 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
18866 stack_size feature.
18867 (Effective-Target Keywords, Other attributes): Suggest using
18868 dg-add-options stack_size feature to get stack limit in stack_size
18869 effective target documentation.
18871 2017-06-21 Julian Brown <julian@codesourcery.com>
18872 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18874 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
18875 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
18876 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
18878 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
18879 attribute type list for neon_multiply.
18880 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
18881 attribute type list for neon_multiply.
18882 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
18883 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
18884 attribute type list for neon_multiply.
18885 * config/arm/types.md (crypto_pmull): Add.
18886 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
18887 attribute type list.
18889 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
18891 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
18894 2017-06-20 Jakub Jelinek <jakub@redhat.com>
18896 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
18897 to make sure not to dereference a NULL cost_classes_ptr pointer.
18899 2017-06-20 Carl Love <cel@us.ibm.com>
18901 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18902 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
18903 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
18904 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
18905 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
18906 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
18907 VMULOSW): New enum "unspec" values.
18908 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
18909 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
18910 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
18911 altivec_vmulosw): New patterns.
18912 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
18913 VMULOSW): Add definitions.
18915 2017-06-20 Julia Koval <julia.koval@intel.com>
18917 * config/i386/i386.c: Fix rounding expand for new pattern.
18918 * config/i386/subst.md: Fix pattern (parallel -> unspec).
18920 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
18922 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
18923 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
18925 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
18927 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
18930 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
18932 * config/aarch64/aarch64-cores.def: Rearrange to sort by
18933 architecture, then by implementer ID.
18934 * config/aarch64/aarch64-tune.md: Regenerate.
18936 2017-06-20 Richard Biener <rguenther@suse.de>
18938 PR middle-end/81097
18939 * fold-const.c (split_tree): Fold to type before negating.
18941 2017-06-20 David Malcolm <dmalcolm@redhat.com>
18943 * diagnostic-show-locus.c
18944 (selftest::test_fixit_deletion_affecting_newline): New function.
18945 (selftest::diagnostic_show_locus_c_tests): Call it.
18947 2017-06-20 Andreas Schwab <schwab@suse.de>
18950 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
18953 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18955 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
18956 __ARM_FEATURE_COPROC according to support.
18958 2017-06-20 Jakub Jelinek <jakub@redhat.com>
18960 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
18961 Rewritten to avoid overflow for > 32-bit pointers.
18964 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
18965 by removing enum keyword.
18966 (ubsan_type_descriptor): Likewise. Formatting fix.
18969 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
18970 splitter): Require TARGET_SSE2 in the condition.
18972 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
18975 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
18976 for doing vector set of SFmode on ISA 3.0.
18977 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
18978 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
18980 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
18981 SFmode value into a V4SF variable that was extracted from another
18982 V4SF variable without converting the element to double precision
18983 and back to single precision vector format.
18984 (vsx_insert_extract_v4sf_p9_2): Likewise.
18986 2017-06-19 Jakub Jelinek <jakub@redhat.com>
18988 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
18989 in UWHI to avoid undefined overflow.
18992 * ubsan.h (enum ubsan_encode_value_phase): New.
18993 (ubsan_encode_value): Change second argument to
18994 enum ubsan_encode_value_phase with default value of
18995 UBSAN_ENCODE_VALUE_GENERIC.
18996 * ubsan.c (ubsan_encode_value): Change second argument to
18997 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
18998 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
18999 create_tmp_var_raw instead of create_tmp_var and use a
19001 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
19002 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
19003 ubsan_encode_value callers.
19006 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
19007 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
19008 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
19010 2017-06-19 Richard Biener <rguenther@suse.de>
19012 PR middle-end/81118
19013 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
19014 estimates if we changed anything.
19016 2017-06-19 Richard Biener <rguenther@suse.de>
19018 PR tree-optimization/80887
19019 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
19020 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
19021 simplified lookups, then reset mprts_hook.
19022 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
19024 (try_to_simplify): Likewise.
19026 2017-06-19 Martin Liska <mliska@suse.cz>
19029 * gimplify.c (gimplify_switch_expr):
19030 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
19032 2017-06-19 Martin Liska <mliska@suse.cz>
19034 * doc/install.texi: Document that PGO runs in 4 stages.
19036 2017-06-19 Martin Liska <mliska@suse.cz>
19039 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
19040 to dispatcher function name.
19041 * multiple_target.c (replace_function_decl): New function.
19042 (create_dispatcher_calls): Redirect both edges and references.
19044 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
19046 * profile-count.c (profile_count::dump): Dump quality.
19047 (profile_count::differs_from_p): Update for unsigned val.
19048 * profile-count.h (profile_count_quality): New enum.
19049 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
19051 2017-06-19 Richard Biener <rguenther@suse.de>
19053 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
19054 struct function as arg.
19055 (estimate_numbers_of_iterations): Export overload with loop arg.
19056 (free_numbers_of_iterations_estimates_loop): Use an overload of
19057 free_numbers_of_iterations_estimates instead.
19058 * tree-cfg.c (remove_bb): Adjust.
19059 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
19060 * tree-parloops.c (gen_parallel_loop): Likewise.
19061 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
19063 (tree_unroll_loops_completely): Likewise.
19064 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
19065 Use an overload instead and export.
19066 (estimated_loop_iterations): Adjust.
19067 (max_loop_iterations): Likewise.
19068 (likely_max_loop_iterations): Likewise.
19069 (estimate_numbers_of_iterations): Take struct function as arg
19071 (loop_exits_before_overflow): Adjust.
19072 (free_numbers_of_iterations_estimates_loop): Use an overload.
19073 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
19074 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
19076 2017-06-19 Richard Biener <rguenther@suse.de>
19079 * ipa-prop.c (find_constructor_constant_at_offset): Handle
19080 RANGE_EXPR conservatively.
19082 2017-06-16 Carl Love <cel@us.ibm.com>
19084 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19085 definitions for vec_float, vec_float2, vec_floato,
19086 vec_floate built-ins.
19087 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
19088 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
19090 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
19091 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
19092 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
19093 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
19094 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
19095 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
19096 vec_floato): Add builtin defines.
19097 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
19098 Update the built-in documentation file for the new built-in
19101 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19103 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
19104 (mthumb): Mark as the negative of -marm.
19106 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19108 * doc/invoke.texi (ARM Options, -mcpu): Document supported
19110 (ARM Options, -mtune): Document that this accepts the same
19111 extension options as -mcpu.
19112 (ARM Options, -mfpu): Document addition of -mfpu=auto.
19114 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19116 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
19117 permitted extensions.
19119 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19121 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
19122 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
19123 (armv8-m.main): Add option +nodsp.
19124 * config/arm/arm-cpu-cdata.h: Regenerated.
19126 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19128 * config/arm/t-fuchsia: New file.
19129 * config.gcc (arm*-*-fuchsia*): Use it.
19131 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19133 * config/arm/t-symbian: Rewrite for new option infrastructure.
19135 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19137 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
19138 (MULTILIB_REQUIRED): Likewise.
19140 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19142 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
19143 (MULTILIB_RESUE): Likewise.
19144 (MULTILIB_MATCHES): Likewise.
19145 (MULTLIB_REQUIRED): Likewise.
19147 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19149 * config/arm/t-rtems: Rewrite for new option framework.
19151 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19153 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
19154 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
19155 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
19156 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
19157 * config/arm/t-multilib: ... here.
19158 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
19159 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
19160 armv7-a and armv8*-a when A-profile libraries have not been built.
19161 * config/arm/t-rmprofile: Rewrite.
19163 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19165 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
19166 with a backslash. Remove the backslash after substituting unescaped
19168 * doc/fragments.texi (MULTILIB_REUSE): Document it.
19170 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19172 * config.gcc: (arm*-*-*): When building a-profile libraries, force
19173 the driver to pass through the default setting of -mfloat-abi.
19174 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
19176 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
19177 (all_feat_combs): New rule.
19178 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
19180 * config/arm/t-aprofile: Rewrite.
19182 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19184 * config/arm/arm.h (FPUTYPE_AUTO): Define.
19185 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
19186 fpu is not specified by the user/command-line.
19187 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
19188 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
19189 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
19190 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
19191 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
19192 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
19194 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19196 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
19197 * config/arm/t-arm-elf: Rewritten.
19199 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19201 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
19202 have some floating-point instructions.
19203 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
19204 (TARGET_MAYBE_HARD_FLOAT): New macro.
19205 * config/arm/arm-builtins.c (arm_init_builtins): Use
19206 TARGET_MAYBE_HARD_FLOAT.
19207 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
19209 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19211 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
19212 (configargs.h): Include it.
19213 (arm_print_hint_for_fpu_option): New function.
19214 (arm_parse_fpu_option): New function.
19215 (candidate_extension): New class.
19216 (arm_canon_for_multilib): New function.
19217 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
19218 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
19219 (ARCH_CANONICAL_SPECS): New macro.
19220 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
19222 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19224 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
19225 are set after handling multilib fragments. Set target_cpu_default2
19228 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19230 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
19232 (arm*-*-*): Set target_cpu_default2 to a quoted string.
19233 * config/arm/parsecpu.awk (check_cpu): Validate any extension
19235 (check_arch): Likewise.
19236 * config/arm/arm.c (arm_configure_build_target): Handle
19237 TARGET_CPU_DEFAULT being a string constant. Scan any feature
19238 options in the default.
19240 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19242 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
19243 when an option is an alias of another.
19244 * config/arm/parsecpu.awk (optalias): New parser token.
19245 (gen_comm_data): Mark non-alias options as such. Emit entries
19246 for extension aliases.
19247 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
19248 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
19249 (armv6kz, armv6zk, armv6t2): Likewise.
19250 (armv7): Make vfpv3-d16 an alias.
19251 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
19253 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
19254 Sort in canonical order.
19255 (armv8-a): Sort in canonical order.
19256 (armv8.1-a, armv8.2-a): Likewise.
19257 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
19259 (cortex-a9): Sort in canonical order.
19260 * config/arm/arm.c (selftests.h): Include it.
19261 (arm_test_cpu_arch_data): New function.
19262 (arm_run_self_tests): New function.
19263 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
19264 (targetm): Move declaration to the end of the file.
19265 * arm-cpu-cdata.h: Regenerated.
19267 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19269 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
19270 call to target_mode_check describing the type of option passed.
19271 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
19272 (arm_target_thumb_only): Use arm_parse_arch_option_name or
19273 arm_parse_cpu_option_name to match parameters against list of
19275 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
19276 arm_arch_core_flags data structure.
19277 * config/arm/arm-cpu_cdata.h: Regenerated.
19279 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19281 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
19283 (arm_print_hint_for_cpu_option): Likewise.
19284 (arm_print_hint_for_arch_option): Likewise.
19285 (arm_parse_cpu_option_name): Likewise.
19286 (arm_parse_arch_option_name): Likewise.
19287 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
19288 of entries in the all_fpus list.
19289 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
19290 (arm_parse_cpu_option_name): Declare.
19291 (arm_parse_arch_option_name): Declare.
19292 (arm_parse_option_features): Declare.
19293 (arm_intialize_isa): Declare.
19294 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
19296 (gen_comm_data): ... here. Make definitions non-static.
19297 * config/arm/arm-cpu-data.h: Regenerated.
19298 * config/arm/arm-cpu-cdata.h: Regenerated.
19300 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19302 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
19303 (cpu_arch_extension): New structure.
19304 (cpu_arch_option, arch_option, cpu_option): New structures.
19305 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
19306 architecture types.
19307 (gen_data): Generate new format data tables.
19308 * config/arm/arm.c (cpu_tune): New structure.
19309 (cpu_option, processors): Delete.
19310 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
19311 (arm_print_hint_for_cpu_option): ... this and ...
19312 (arm_print_hint_for_arch_option): ... this.
19313 (arm_parse_arch_cpu_name): Delete. Replace with ...
19314 (arm_parse_cpu_option_name): ... this and ...
19315 (arm_parse_arch_option_name): ... this.
19316 (arm_unrecognized_feature): Change type of target parameter to
19318 (arm_parse_arch_cpu_features): Delete. Replace with ...
19319 (arm_parse_option_features): ... this.
19320 (arm_configure_build_target): Rework to use new configuration data
19322 (arm_print_tune_info): Rework for new configuration data tables.
19323 * config/arm/arm-cpu-data.h: Regenerated.
19324 * config/arm/arm-cpu.h: Regenerated.
19326 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19328 * Makefile.in (OBJS): Move sbitmap.o from here ...
19329 (OBJS-libcommon): ... to here.
19331 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19333 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
19334 (ISA_ALL_CRYPTO): New macro.
19335 (ISA_ALL_SIMD): New macro
19336 (ISA_ALL_FP): New macro.
19337 * config/arm/arm.c (fpu_bitlist): Update initializer.
19338 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
19340 (arm9e): Add fpu. Add option for nofp
19341 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
19342 (arm926ej-s, arm1026ej-s): Likewise.
19343 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
19344 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
19345 neon-fp16, neon-vfpv4, nofp and nosimd.
19346 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
19347 (cortex-a8): Add fpu. Add option for nofp.
19348 (cortex-a9): Add fpu. Add options for nosimd and nofp.
19349 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
19350 (cortex-r4f): Add fpu.
19351 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
19352 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
19354 (cortex-r8): Likewise.
19355 (cortex-m4): Add fpu. Add option for nofp.
19356 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
19357 (cortex-a17.cortex-a7): Likewise.
19358 (cortex-a32): Add fpu. Add options for crypto and nofp.
19359 (cortex-a35, cortex-a53): Likewise.
19360 (cortex-a57): Add fpu. Add option for crypto.
19361 (cortex-a72, cortex-a73): Likewise.
19362 (exynos-m1): Likewise.
19363 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
19364 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
19365 (cortex-m33): Add fpu. Add option for nofp.
19366 * config/arm/arm-cpu-cdata.h: Regenerated
19367 * config/arm/arm-cpu-data.h: Regenerated.
19369 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19371 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
19372 (armv5te, armv5tej): Likewise.
19373 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
19374 (armv7): Add options fp and vfpv3-d16.
19375 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
19376 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
19378 (armv7ve): Likewise.
19379 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
19380 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
19381 (armv8-a): Add nocrypto option.
19382 (armv8.1-a, armv8.2-a): Likewise.
19383 (armv8-m.main): add options fp, fp.dp and nofp.
19385 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19387 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
19389 (armv8-a+crc): Delete.
19390 (armv8.1-a): Add options simd, crypto and nofp.
19391 (armv8.2-a): Add options fp16, simd, crypto and nofp.
19392 (armv8.2-a+fp16): Delete.
19393 (armv8-m.main): Add option dsp.
19394 (armv8-m.main+dsp): Delete.
19395 (cortex-a8): Add fpu. Add nofp option.
19396 (cortex-a9): Add fpu. Add nofp and nosimd options.
19397 * config/arm/parsecpu.awk (gen_data): Generate option tables and
19398 link to main cpu and architecture data structures.
19399 (gen_comm_data): Only put isa attributes from the main architecture
19401 (option): New statement for architecture and CPU entries.
19402 * arm.c (struct cpu_option): New structure.
19403 (struct processors): Add entry for options.
19404 (arm_unrecognized_feature): New function.
19405 (arm_parse_arch_cpu_name): Ignore any characters after the first
19407 (arm_parse_arch_cpu_feature): New function.
19408 (arm_configure_build_target): Separate out any CPU and architecture
19409 features and parse separately. Don't error out if -mfpu=auto is
19410 used with only an architecture string.
19411 (arm_print_asm_arch_directives): New function.
19412 (arm_file_start): Call it.
19413 * config/arm/arm-cpu-cdata.h: Regenerated.
19414 * config/arm/arm-cpu-data.h: Likewise.
19415 * config/arm/arm-tables.opt: Likewise.
19417 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19419 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
19420 assembler when it is not -mfpu=auto.
19422 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19424 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
19425 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
19426 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
19427 (ASM_CPU_SPEC): Rewrite.
19428 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
19429 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
19430 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
19431 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
19432 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
19433 copied string is NUL-terminated. Also strip any characters prefixed
19435 (arm_rewrite_selected_arch): New function.
19436 (arm_rewrite_march): New function.
19438 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19440 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
19441 (x_arm_cpu_string, x_arm_tune_string): Likewise.
19442 (march, mcpu, mtune): Convert to string-based options.
19443 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
19444 (arm_parse_arch_cpu_name): New function.
19445 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
19446 identify selected architecture or CPU.
19447 (arm_option_save): New function.
19448 (TARGET_OPTION_SAVE): Redefine.
19449 (arm_option_restore): Restore string options.
19450 (arm_option_print): Print string options.
19452 2017-06-16 Martin Sebor <msebor@redhat.com>
19454 PR tree-optimization/80933
19455 PR tree-optimization/80934
19456 * builtins.c (fold_builtin_3): Do not handle bcmp here.
19457 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
19458 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
19459 (gimple_fold_builtin): Call them.
19461 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19463 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
19464 as unlikely; update profile.
19466 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19468 * predict.c (force_edge_cold): Handle declaring edges impossible
19471 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19473 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
19475 (try_unroll_loop_completely): Fix reporting.
19477 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19479 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
19481 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
19484 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
19485 if given a non-constant argument for an intrinsic which requires a
19488 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19490 * profile.c (compare_freqs): New function.
19491 (branch_prob): Sort edge list.
19492 (find_spanning_tree): Assume that the list is priority sorted.
19494 2017-06-16 Richard Biener <rguenther@suse.de>
19496 PR tree-optimization/81090
19497 * passes.def (pass_record_bounds): Remove.
19498 * tree-pass.h (make_pass_record_bounds): Likewise.
19499 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
19500 make_pass_record_bounds): Likewise.
19501 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
19502 not free niter estimates at the beginning but at the end.
19503 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
19505 2017-06-16 Richard Biener <rguenther@suse.de>
19507 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
19508 initializer to workaround ICE in host GCC 4.8.
19510 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19512 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
19514 (clone_inlined_nodes): Update.
19516 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19518 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
19519 prefetch settings, and enable prefetching by default at -O3.
19521 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19523 * config/aarch64/aarch64.c (aarch64_override_options_internal):
19524 Set flag_prefetch_loop_arrays according to tuning data.
19526 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19528 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
19529 New tune structure.
19530 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
19531 [Unrelated to main purpose of the patch] Place the pointer field last
19532 to enable type checking errors when tune structure are wrongly merged.
19533 * config/aarch64/aarch64.c (generic_prefetch_tune,)
19534 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
19535 (thunderx2t99_prefetch_tune): New tune constants.
19536 (tune_params *_tunings): Update all tunings (no functional change).
19537 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
19538 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
19539 from tunings structures.
19541 2017-06-16 Jakub Jelinek <jakub@redhat.com>
19544 * ubsan.c (instrument_null): Add T argument, use it instead
19545 of computing it based on IS_LHS.
19546 (instrument_object_size): Likewise.
19547 (pass_ubsan::execute): Adjust instrument_null and
19548 instrument_object_size callers to pass gimple_get_lhs or
19549 gimple_assign_rhs1 result to it. Use instrument_null instead of
19550 calling get_base_address and instrument_mem_ref. Handle
19551 aggregate call arguments for object-size sanitization.
19553 2017-06-16 Yury Gribov <tetra2005@gmail.com>
19555 PR tree-optimization/81089
19556 * tree-vrp.c (is_masked_range_test): Validate operands of
19559 2017-06-15 Martin Sebor <msebor@redhat.com>
19562 * dumpfile.c (dump_register): Avoid calling memset to initialize
19563 a class with a default ctor.
19564 * gcc.c (struct compiler): Remove const qualification.
19565 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
19566 * hash-table.h: Ditto.
19567 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
19569 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
19570 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
19572 * params.h (struct param_info): Make struct members non-const.
19573 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
19574 with default initialization.
19575 * vec.h (vec_copy_construct, vec_default_construct): New helper
19577 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
19578 with vec_copy_construct.
19579 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
19580 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
19581 * doc/invoke.texi (-Wclass-memaccess): Document.
19583 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19585 * emit-rtl.h (is_leaf): Update comment about local
19586 register allocator.
19588 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19591 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
19592 for a variable to have a section before checking if the section has a
19594 Set section to.persistent if persistent attribute is set.
19595 Warn if .persistent attribute is used on an automatic variable.
19597 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
19599 PR rtl-optimization/80474
19600 * reorg.c (update_block): Do not ignore instructions in a delay slot.
19602 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
19604 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
19607 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
19609 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
19610 (casesi): Emit bounds checking as RTL.
19611 (casesi_internal_mips16_<mode>): Remove bounds checking.
19613 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
19615 * config/xtensa/xtensa.c (xtensa_option_override): Append
19616 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
19617 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
19618 xtensa_doloop_hooks): Define unconditionally.
19619 (xtensa_reorg_loops): Only call reorg_loops in the presence of
19621 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
19622 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
19623 for it in xtensa_option_override.
19624 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
19625 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
19627 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
19629 * doc/cppopts.texi: Document '-' special value to -MF.
19631 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
19633 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
19634 (cortex_a53_fconst): Likewise.
19635 (cortex_a53_fpmul): Likewise.
19636 (cortex_a53_f_load_64): Likewise.
19637 (cortex_a53_f_load_many): Likewise.
19638 (cortex_a53_advsimd_alu): Likewise.
19639 (cortex_a53_advsimd_alu_q): Likewise.
19640 (cortex_a53_advsimd_mul): Likewise.
19641 (cortex_a53_advsimd_mul_q): Likewise.
19642 (fpmac bypass): Add new bypass for fpmac-fpmac case.
19643 Add missing fmul, r2f_cvt and fconst cases.
19645 2017-06-14 Richard Biener <rguenther@suse.de>
19647 PR middle-end/81088
19648 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
19650 (fold_binary_loc): When associating do not treat pre-existing
19651 TREE_OVERFLOW on literal constants as a reason to allow
19652 TREE_OVERFLOW on associated literal constants.
19654 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19656 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
19657 (MASK_FEATURES): New macro.
19658 * config/sparc/sparc.c (sparc_option_override): Remove the special
19659 handling of -mfpu and generalize it to all MASK_FEATURES switches.
19661 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19663 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
19664 a division of 0 if non-call exceptions are enabled.
19666 2017-06-14 Andrew Pinski <apinski@cavium.com>
19667 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19670 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
19671 Improve vector initialization code gen for only variable case.
19673 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19675 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
19677 2017-06-14 Richard Biener <rguenther@suse.de>
19679 PR tree-optimization/81083
19680 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
19683 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19685 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
19686 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
19687 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
19688 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
19689 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
19690 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
19692 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19694 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
19695 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
19697 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19699 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
19701 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19703 * config/rs6000/t-rtems: Don't handle SPE.
19705 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19707 * config/rs6000/t-linux: Don't handle SPE.
19709 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19711 * config/rs6000/eabispe.h: Delete file.
19713 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19715 * config/rs6000/t-spe: Delete file.
19717 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19719 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
19720 (rs6000_legitimate_offset_address_p): Return false for anything in
19721 V2SImode or V2SFmode.
19723 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19725 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
19726 except V2SF and V2SI. Rearrange the vector modes, and add comments.
19727 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
19729 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
19730 (rs6000_legitimate_offset_address_p): Ditto.
19731 (rs6000_emit_move): Ditto.
19732 (rs6000_init_builtins): Remove V4HI_type_node.
19734 2017-06-13 Martin Liska <mliska@suse.cz>
19737 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
19738 (gate_asan): Likewise.
19739 * asan.h (asan_no_sanitize_address_p): Remove the function.
19740 (sanitize_flags_p): New function.
19741 * builtins.def: Fix coding style.
19742 * common.opt: Use renamed enum value.
19743 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
19744 * doc/extend.texi: Document no_sanitize attribute.
19745 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
19746 to SANITIZE_UNDEFINED_NONDEFAULT.
19747 * gcc.c (sanitize_spec_function): Use the renamed enum value.
19748 * gimple-fold.c (optimize_atomic_compare_exchange_p):
19749 Use sanitize_flags_p.
19750 * gimplify.c (gimplify_function_tree): Likewise.
19751 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
19752 * opts.c (parse_no_sanitize_attribute): New function.
19753 (common_handle_option): Use renamed enum value.
19754 * opts.h (parse_no_sanitize_attribute): Declare.
19755 * tree.c (sanitize_flags_p): New function.
19756 * tree.h: Declared here.
19757 * tsan.c: Use sanitize_flags_p.
19758 * ubsan.c (ubsan_expand_null_ifn): Likewise.
19759 (instrument_mem_ref): Likewise.
19760 (instrument_bool_enum_load): Likewise.
19761 (do_ubsan_in_current_function): Remove the function.
19762 (pass_ubsan::execute): Use sanitize_flags_p.
19763 * ubsan.h: Remove do_ubsan_in_current_function
19764 * tree-cfg.c (print_no_sanitize_attr_value): New function.
19765 (dump_function_to_file): Use it here.
19767 2017-06-13 Martin Jambor <mjambor@suse.cz>
19769 PR tree-optimization/80803
19770 PR tree-optimization/81063
19771 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
19772 (propagate_subaccesses_across_link): Enqueue subtree whenever
19773 necessary instead of relying on the caller.
19775 2017-06-13 Martin Jambor <mjambor@suse.cz>
19777 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
19778 that have a first_link.
19779 (sort_and_splice_var_accesses): Do not check first_link before
19781 (subtree_mark_written_and_enqueue): Likewise.
19782 (propagate_all_subaccesses): Likewise and do not stop at first
19783 parent with a first_link.
19785 2017-06-13 Martin Jambor <mjambor@suse.cz>
19787 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
19790 2017-06-13 Yury Gribov <tetra2005@gmail.com>
19792 * match.pd: New pattern.
19794 2017-06-13 Yury Gribov <tetra2005@gmail.com>
19796 * tree-vrp.c (is_masked_range_test): New function.
19797 (register_edge_assert_for): Determine ranges for
19800 2017-06-13 Yury Gribov <tetra2005@gmail.com>
19802 PR tree-optimization/67328
19803 * fold-const.c (maskable_range_p): New function.
19804 (build_range_check): Generate bittests if possible.
19806 2017-06-13 Martin Liska <mliska@suse.cz>
19808 * gimple-pretty-print.c (dump_probability): Add new argument.
19809 (dump_edge_probability): Dump both probability and count.
19810 (dump_gimple_label): Likewise.
19811 (dump_gimple_bb_header): Likewise.
19813 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
19816 * config/avr/avr-devices.c: Fix indentation.
19817 * config/avr/gen-avr-mmcu-specs.c: Dito.
19819 2017-06-13 Richard Biener <rguenther@suse.de>
19821 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
19822 instead get vector type from stmt_info.
19823 (vectorizable_reduction): Adjust. Remove dead code.
19825 2017-06-13 Richard Biener <rguenther@suse.de>
19827 PR middle-end/81065
19828 * fold-const.c (extract_muldiv_1): Remove bogus distribution
19829 case of C * (x * C2 + C3).
19830 (fold_addr_of_array_ref_difference): Properly fold index difference.
19832 2017-06-12 David S. Miller <davem@davemloft.net>
19835 * config/sparc/sparc.md (return expander): Emit frame blockage if
19836 function uses alloca.
19838 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
19840 * combine.c (make_field_assignment): Check len rather than the mode
19841 precision when calling force_to_mode.
19843 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
19845 Support multilibs and devices that see flash in RAM address range.
19848 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
19849 (avr_mcu_t) <flash_pm_offset>: New field.
19850 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
19851 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
19852 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
19853 (AVR_TINY_PM_OFFSET): Remove macro.
19854 * config/avr/avr.opt (-mshort-calls): New option.
19855 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
19856 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
19857 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
19858 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
19859 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
19860 instead of avr_arch->have_jmp_call.
19861 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
19862 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
19863 avr_arch->flash_pm_offset to define.
19864 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
19865 new field flash_pm_offset. Add entry for avrxmega3.
19866 (avr_texinfo): Add entry for avrxmega3.
19867 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
19868 attiny212, attiny214,
19869 attiny412, attiny414, attiny416, attiny417,
19870 attiny814, attiny816, attiny817,
19871 attiny1614, attiny1616, attiny1617,
19872 attiny3214, attiny3216, attiny3217.
19873 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
19874 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
19875 (avr_print_operand_address) [AVR_TINY]: Same.
19876 (avr_asm_init_sections) <readonly_data_section>: Only patch
19877 callback if avr_arch->flash_pm_offset = 0.
19878 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
19879 for rodata if avr_arch->flash_pm_offset != 0.
19880 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
19881 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
19882 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
19883 (m_options): Append opt_rcall.
19884 (m_dirnames): Append dir_rcall.
19885 * config/avr/t-multilib: Regenerate.
19887 * configure.ac [target=avr]: Check whether avrxmega3 default
19888 linker description file works as needed.
19889 * configure: Regenerate.
19890 * doc/avr-mmcu.texi: Regenerate.
19891 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
19892 <__AVR_ARCH__>: Document avrxmega3 and 103.
19893 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
19894 <__AVR_SHORT_CALLS__>: Document it.
19895 <__AVR_PM_BASE_ADDRESS__>: Document it.
19896 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
19897 (AVR Variable Attributes) <progmem>: Document this is
19898 not needed for avrxmega3.
19899 (AVR Named Address Spaces) <__flash>: Dito.
19901 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
19903 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
19905 2017-06-12 Doug Rupp <rupp@adacore.com>
19907 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
19908 Append vxworks-stdint.h to the tm_file list.
19909 * config/vxworks-stdint.h: New file.
19911 2017-06-12 Martin Liska <mliska@suse.cz>
19913 PR tree-optimization/81041
19914 * tree-profile.c (gimple_gen_ic_func_profiler):
19915 Create an extra BB in profile-generate
19916 (gimple_gen_time_profiler): Likewise.
19918 2017-06-12 Jakub Jelinek <jakub@redhat.com>
19920 PR tree-optimization/81003
19921 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
19922 (update_range_test): Use it instead of force_gimple_operand_gsi.
19924 2017-06-12 Richard Biener <rguenther@suse.de>
19926 PR tree-optimization/81053
19927 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
19928 with backedge value not defined in loop. Simplify def stmt
19931 2017-06-11 Tom de Vries <tom@codesourcery.com>
19934 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
19936 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
19937 nvptx_cannot_force_const_mem.
19939 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19941 * opts.c (finish_options): Move test for flag_split_stack after
19942 it has been initialized.
19944 2017-06-11 Jason Merrill <jason@redhat.com>
19946 * tree.h (id_equal): New.
19947 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
19948 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
19949 instead of strcmp of IDENTIFIER_POINTER.
19951 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19953 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
19954 (mark_all_inlined_calls_cdtor): Fix formating.
19955 (inline_transform): Rescale profile before inlining.
19957 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19959 * cgraph.h (cgraph_edge::clone): Update prototype.
19960 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
19961 (cgraph_node::create_clone): Update.
19962 (cgraph_node::create_version_clone): Update.
19963 * tree-inline.c (copy_bb): Update.
19964 (expand_call_inline): Update.
19966 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
19968 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
19969 factored out from ...
19970 (rs6000_emit_prologue): ... here.
19972 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
19974 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
19975 factored out from ...
19976 (rs6000_emit_prologue): ... here.
19978 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19980 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
19982 (handle_missing_profiles): Fix computation of tp_first_run.
19983 (counts_to_freqs): Do not touch freqs when count is 0.
19985 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19987 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
19990 2017-06-10 Tom de Vries <tom@codesourcery.com>
19992 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
19993 attributes): Document signal effective target.
19995 2017-06-10 Tom de Vries <tom@codesourcery.com>
19997 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
19998 Document effective target stack_size.
20000 2017-06-09 David Malcolm <dmalcolm@redhat.com>
20002 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
20003 to the edit_context if they can be auto-applied.
20005 2017-06-9 Ian Lance Taylor <iant@golang.org>
20007 * opts.c (finish_options): If -fsplit-stack, disable implicit
20008 -forder-blocks-and-partition.
20009 * doc/invoke.texi (Optimize Options): Document that when using
20010 -fsplit-stack -forder-blocks-and-partition is not implicitly
20013 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20015 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
20016 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
20017 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
20018 * builtins.def (abort, trap, unreachable): Declare cold.
20019 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
20020 * tree-core.h (ECF_COLD): New.
20021 * tree.c (set_call_expr_flags): Handle ECF_COLD.
20022 (build_common_builtin_nodes): Mark unreachable and abort as cold.
20024 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20026 * predict.c (unlikely_executed_stmt_p): Cleanup.
20028 2017-06-09 Richard Biener <rguenther@suse.de>
20030 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
20031 model if the ref is always written to.
20033 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20035 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
20037 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20039 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
20042 2017-06-09 Tom de Vries <tom@codesourcery.com>
20045 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
20046 "target cannot support label values" when encountering LABEL_REF.
20048 2017-06-09 Martin Liska <mliska@suse.cz>
20050 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
20051 (gimple_gen_ic_func_profiler): Emit direct comparison
20052 of __gcov_indirect_call_callee with NULL.
20053 (gimple_gen_time_profiler): Change probability from
20054 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
20056 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20058 * profile.c (edge_gcov_counts): Turn to pointer.
20059 (compute_branch_probabilities, compute_branch_probabilities): Update.
20060 (branch_prob): Do not clear edge_gcov_count.
20061 * profile.h (edge_gcov_counts): Turn to pointer.
20062 (edge_gcov_count): Update.
20064 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20066 * gimple.h (gimple_check_failed): Mark cold.
20068 2017-06-09 Richard Biener <rguenther@suse.de>
20070 PR tree-optimization/66623
20071 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
20072 refactor check_reduction into two parts, properly computing
20073 whether we have to check reduction validity for outer loop
20076 2017-06-09 Richard Biener <rguenther@suse.de>
20078 PR tree-optimization/79483
20079 * graphite-scop-detection.c (order): New global.
20080 (get_order): Compute bb to order mapping that satisfies code
20081 generation constraints.
20082 (cmp_pbbs): New helper.
20083 (build_scops): Start domwalk at entry block, sort generated
20086 2017-06-09 Richard Biener <rguenther@suse.de>
20088 PR middle-end/81007
20089 * ipa-polymorphic-call.c
20090 (ipa_polymorphic_call_context::restrict_to_inner_class):
20091 Skip FIELD_DECLs with error_mark_node type.
20092 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
20095 2017-06-09 Martin Liska <mliska@suse.cz>
20097 * predict.c (struct branch_predictor): New struct.
20098 (test_prediction_value_range): New test.
20099 (predict_c_tests): New function.
20100 * selftest-run-tests.c (selftest::run_tests): Run the function.
20101 * selftest.h: Declare new tests.
20103 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
20106 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
20107 gen_add3_insn did not fail.
20108 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
20109 r0, construct that number in a temporary reg and add that reg to r0.
20110 If asked to put the result in r0 as well, fail.
20112 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
20114 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
20115 for early expansion of vec_eqv.
20117 2017-06-08 Jakub Jelinek <jakub@redhat.com>
20119 PR middle-end/81005
20120 * ubsan.c (instrument_null): Avoid pointless code temporary.
20121 (pass_ubsan::execute): Instrument aggregate arguments of calls.
20123 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
20127 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
20130 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
20133 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20135 * predict.c (unlikely_executed_edge_p): Move ahead.
20136 (probably_never_executed_edge_p): Use it.
20138 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20140 PR middle-end/79988
20141 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
20142 gimple_call_builtin_p call.
20144 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20146 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
20147 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
20148 rtl_check_failed_type2, rtl_check_failed_code1,
20149 rtl_check_failed_code2, rtl_check_failed_code_mode,
20150 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
20151 rtvec_check_failed_bounds, rtl_check_failed_flag,
20152 _fatal_insn_not_found, _fatal_insn): Likewise.
20153 * tree.h (tree_contains_struct_check_failed,
20154 tree_check_failed, tree_not_check_failed,
20155 tree_class_check_failed, tree_range_check_failed,
20156 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
20157 tree_vec_elt_check_failed, phi_node_elt_check_failed,
20158 tree_operand_check_failed, omp_clause_check_failed,
20159 omp_clause_operand_check_failed, omp_clause_range_check_failed):
20162 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20164 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
20165 flag_branch_probabilities.
20166 * ipa-inline.c (edge_badness): Likewise.
20167 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
20168 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
20169 * predict.c (maybe_hot_frequency_p): Likewise.
20170 (probably_never_executed): Likewise.
20171 * sched-ebb.c (schedule_ebbs): Likewise.
20172 * sched-rgn.c (find_single_block_region): Likewise.
20173 * tracer.c (tail_duplicate): Likewise.
20175 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20177 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
20178 longer requires x_flag_profile_use.
20180 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20182 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
20183 instead of flag_reorder_blocks_and_partition.
20184 * dbxout.c (dbxout_function_end): Likewise.
20185 * dwarf2out.c (gen_subprogram_die): Likewise.
20186 * haifa-sched.c (sched_create_recovery_edges): Likewise.
20187 * hw-doloop.c (reorg_loops): Likewise.
20188 * varasm.c (assemble_start_function,
20189 assemble_end_function): Likewise.
20190 (decide_function_section): Do not check for
20191 flag_reorder_blocks_and_partition.
20193 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20195 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20197 (chkp_get_hard_register_fake_addr_expr): Ditto.
20198 (chkp_build_addr_expr): Add check for hard reg case.
20199 (chkp_parse_array_and_component_ref): Ditto.
20200 (chkp_find_bounds_1): Ditto.
20201 (chkp_process_stmt): Don't generate bounds store for
20204 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20206 * predict.c (maybe_hot_bb_p): Do not check profile status.
20207 (maybe_hot_edge_p): Likewise.
20208 (probably_never_executed): Check for zero counts even if profile
20210 (unlikely_executed_edge_p): New function.
20211 (unlikely_executed_stmt_p): New function.
20212 (unlikely_executed_bb_p): New function.
20213 (set_even_probabilities): Use unlikely predicates.
20214 (combine_predictions_for_bb): Likewise.
20215 (predict_paths_for_bb): Likewise.
20216 (predict_paths_leading_to_edge): Likewise.
20217 (determine_unlikely_bbs): New function.
20218 (estimate_bb_frequencies): Use it.
20219 (compute_function_frequency): Use zero counts even if profile is
20221 * profile-count.h: Fix typo.
20223 2017-08-08 Julia Koval <julia.koval@intel.com>
20225 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
20226 _mm512_mask_cvtsepi16_storeu_epi8,
20227 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
20228 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
20229 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
20230 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
20231 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
20232 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
20233 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
20234 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
20235 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
20236 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
20237 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
20238 __builtin_ia32_pmovuswb256mem_mask,
20239 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
20240 __builtin_ia32_pmovwb512mem_mask): New builtins.
20242 2017-08-08 Julia Koval <julia.koval@intel.com>
20244 PR target/73350,80862
20245 * config/i386/subst.md (round): Fix round pattern.
20246 * config/i386/i386.c (ix86_erase_embedded_rounding):
20247 Fix erasing rounding for the fixed pattern.
20249 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20251 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
20253 2017-06-08 Martin Liska <mliska@suse.cz>
20255 PR gcov-profile/80911
20256 * gcov.c (block_info::block_info): New constructor.
20258 2017-06-07 Carl Love <cel@us.ibm.com>
20260 * config/rs6000/rs6000-c: The return type of the following
20261 built-in functions was implemented as int not long long. Fix sign
20262 of return value for the unsigned version of vec_mulo and vec_mule.
20263 vector unsigned long long vec_bperm (vector unsigned long long,
20264 vector unsigned char)
20265 vector signed long long vec_mule (vector signed int,
20267 vector unsigned long long vec_mule (vector unsigned int,
20268 vector unsigned int)
20269 vector signed long long vec_mulo (vector signed int,
20271 vector unsigned long long vec_mulo (vector unsigned int,
20272 vector unsigned int)
20273 * doc/extend.texi: Fix the documentation for the built-in
20276 2017-06-07 Carl Love <cel@us.ibm.com>
20279 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
20282 2017-06-07 Carl Love <cel@us.ibm.com>
20284 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
20285 support, Generate doublehv for signed int/float for BE case only.
20287 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
20289 * doc/invoke.texi (mcx16): Rewrite.
20291 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20293 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
20294 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
20295 *mov<mode>_softfloat, and an anonymous splitter): Use
20296 nonimmediate_operand instead of rs6000_nonimmediate_operand.
20298 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20300 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
20302 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
20303 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
20304 (rs6000_debug_reg_global): Adjust.
20305 (rs6000_init_hard_regno_mode_ok): Adjust.
20306 (rs6000_dbx_register_number): Adjust.
20307 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
20308 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
20309 Remove SPE_ACC and SPEFSCR.
20310 (REG_ALLOC_ORDER): Ditto.
20311 (FRAME_POINTER_REGNUM): Change to 111.
20312 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
20313 (REG_CLASS_NAMES): Ditto.
20314 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
20315 (REGISTER_NAMES): Ditto.
20316 (ADDITIONAL_REG_NAMES): Ditto.
20317 (rs6000_reg_names): Ditto.
20318 * config/rs6000/rs6000.md: Renumber some register number
20321 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20323 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
20325 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
20326 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
20328 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
20329 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
20330 Delete the SPE high registers.
20331 (REG_ALLOC_ORDER): Ditto.
20332 (enum reg_class): Remove SPE_HIGH_REGS.
20333 (REG_CLASS_NAMES): Ditto.
20334 (REG_CLASS_CONTENTS): Delete the SPE high registers.
20335 (REGISTER_NAMES): Ditto.
20336 (rs6000_reg_names): Ditto.
20337 * doc/tm.texi.in: Remove SPE as example.
20338 * doc/tm.texi: Regenerate.
20340 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20342 * config/rs6000/8540.md (ppc8540_brinc): Delete.
20343 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
20344 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
20345 * config/rs6000/rs6000.md (type): Remove "brinc".
20347 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20349 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
20350 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
20351 * config/rs6000/linuxspe.h: Delete file.
20352 * config/rs6000/rs6000.md: Don't include spe.md.
20353 * config/rs6000/spe.h: Delete file.
20354 * config/rs6000/spe.md: Delete file.
20355 * config/rs6000/t-rs6000: Remove spe.md.
20357 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20359 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
20360 (reg_or_none500mem_operand): Delete.
20361 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
20362 instead of reg_or_none500mem_operand.
20364 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20366 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
20367 handling of SPE flags.
20368 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
20370 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20372 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
20374 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
20375 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
20376 paired_divv2sf3): Similar.
20377 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
20378 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
20379 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
20381 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
20382 Rename the paired_* instruction patterns.
20383 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
20385 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
20386 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
20387 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
20388 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
20389 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
20390 PAIRED_VECTOR_MODE.
20391 (struct machine_function): Delete field spe_insn_chain_scanned_p.
20392 (spe_func_has_64bit_regs_p): Delete.
20393 (spe_expand_predicate_builtin): Delete.
20394 (spe_expand_evsel_builtin): Delete.
20395 (TARGET_DWARF_REGISTER_SPAN): Do not define.
20396 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
20397 (invalid_e500_subreg): Delete.
20398 (rs6000_legitimize_address): Always force_reg op2 as well, for
20399 paired single memory accesses.
20400 (rs6000_member_type_forces_blk): Delete.
20401 (rs6000_spe_function_arg): Delete.
20402 (rs6000_expand_unop_builtin): Delete SPE handling.
20403 (rs6000_expand_binop_builtin): Ditto.
20404 (spe_expand_stv_builtin): Delete.
20405 (bdesc_2arg_spe): Delete.
20406 (spe_expand_builtin): Delete.
20407 (spe_expand_predicate_builtin): Delete.
20408 (spe_expand_evsel_builtin): Delete.
20409 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
20410 (spe_init_builtins): Delete.
20411 (spe_func_has_64bit_regs_p): Delete.
20412 (savres_routine_name): Delete "info" parameter. Adjust callers.
20413 (rs6000_emit_stack_reset): Ditto.
20414 (rs6000_dwarf_register_span): Delete.
20415 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
20416 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
20417 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
20419 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
20421 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
20422 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
20423 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
20424 mulv2sf3, divv2sf3): Delete expanders.
20426 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20428 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
20430 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20432 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
20433 * config/rs6000/rs6000.c: Ditto.
20435 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20437 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
20438 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
20439 comparison_operator.
20441 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20443 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
20444 * config/rs6000/rs6000.opt: Ditto.
20445 * config/rs6000/t-rtems: Ditto.
20447 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20449 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
20450 TARGET_E500_SINGLE by 0, simplify.
20451 * config/rs6000/rs6000.c: Ditto.
20452 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
20453 (spe_build_register_parallel): Delete.
20454 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
20455 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
20456 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
20457 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
20458 (E500_CONVERT): Delete.
20459 * config/rs6000/spe.md: Remove many patterns and all define_constants.
20461 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20463 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
20464 * config/rs6000/dfp.md: Ditto.
20465 (negdd2, *negdd2_fpr): Merge.
20466 (absdd2, *absdd2_fpr): Merge.
20467 (negtd2, *negtd2_fpr): Merge.
20468 (abstd2, *abstd2_fpr): Merge.
20469 * config/rs6000/e500.h: Delete file.
20470 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
20471 TARGET_FPRS by 1 and simplify.
20472 * config/rs6000/rs6000-c.c: Ditto.
20473 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
20474 TARGET_DF_SPE by 0.
20475 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
20477 * config/rs6000/rs6000.md: Ditto.
20478 (floatdidf2, *floatdidf2_fpr): Merge.
20479 (move_from_CR_gt_bit): Delete.
20480 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
20481 (E500_CR_IOR_COMPARE): Delete.
20482 (All patterns that require !TARGET_FPRS): Delete.
20483 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
20485 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20487 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
20489 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20491 * graphds.c (add_edge): Intitialize edge's attached data.
20492 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
20493 pointer parameter. Call pointed function on each edge during
20494 graph traversing. Skip traversing the edge when the function
20496 (graphds_dfs, graphds_scc): Ditto.
20497 (for_each_edge): New parameter. Pass the new parameter to callback
20499 * graphds.h (skip_edge_callback): New function pointer type.
20500 (graphds_dfs, graphds_scc): New function pointer parameter.
20501 (graphds_edge_callback, for_each_edge): New parameter.
20503 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20505 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
20506 out code checking if runtime alias check is possible to below ...
20507 Call the new function.
20508 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
20509 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
20511 2017-06-07 Marek Polacek <polacek@redhat.com>
20514 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
20515 TYPE_OVERFLOW_WRAPS check.
20517 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20519 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
20520 if versioning is required.
20521 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
20522 peeling with the check for versioning.
20524 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20526 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
20527 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
20528 Set true to new parameter if new ssa variable is defined.
20529 (vect_gen_vector_loop_niters): Refactor. Set range information
20530 for the new vector loop bound variable.
20531 (vect_do_peeling): Ditto.
20533 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20535 * tree-affine.c (ssa.h): Include header file.
20536 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
20537 has wrapping overflow behavior.
20539 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20541 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
20543 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20545 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
20546 (tree_to_aff_combination): ... here.
20548 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20550 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
20551 reg_pressure model function.
20552 (ivopts_global_cost_for_size): Delete.
20553 (determine_set_costs, iv_ca_recount_cost): Call new model function
20554 ivopts_estimate_reg_pressure.
20556 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20558 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
20559 expensive than udiv. Remove floating point cases from mod.
20561 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20563 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
20564 Increase idiv cost.
20566 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20568 * config/aarch64/aarch64.md
20569 (copysignsf3): Fix mask generation.
20571 2017-06-07 Jakub Jelinek <jakub@redhat.com>
20573 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
20575 (class dump_manager): Add register_dumps method.
20576 * dumpfile.c: Include langhooks.h.
20577 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
20578 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
20579 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
20580 (dump_manager::dump_register): Start with 512 entries instead of 32.
20581 (dump_manager::register_dumps): New method.
20582 * toplev.c (general_init): Instead of invoking register_dumps
20583 langhook, invoke register_dumps method on the dump manager.
20584 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
20587 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
20589 * doc/md.texi: Clarify the restrictions on a define_insn condition.
20590 Say that # requires an associated define_split to exist, and that
20591 the define_split must be suitable for use after register allocation.
20593 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20595 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
20596 (compute_outgoing_frequencies): Also initialize zero counts.
20597 (find_many_sub_basic_blocks): Do not produce uninitialized profile
20598 around loops; preserve more of profile when nothing changes.
20600 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
20602 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
20604 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
20605 * config/arm/arm-cpu-cdata.h: Regenerate.
20606 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
20607 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
20608 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
20609 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
20610 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
20612 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
20614 * config/arm/t-rmprofile: Likewise.
20615 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
20617 2017-06-06 David S. Miller <davem@davemloft.net>
20620 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
20621 blockage if function uses alloca.
20623 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20625 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
20626 New "uid" fields to hold pretty-print IDs of group and ref.
20627 Memory references are now identified as <group_id>:<ref_id>
20628 instead of using [random] addresses.
20629 (dump_mem_details): Simplify, no functional change.
20630 (dump_mem_ref): Simplify and make output more concise.
20631 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
20632 (find_or_create_group): Initialize group uid.
20633 (record_ref): Initialize ref uid. Improve debug output.
20634 (prune_group_by_reuse, should_issue_prefetch_p,)
20635 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
20636 (mark_nontemporal_store, determine_loop_nest_reuse):
20637 Improve debug output.
20639 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20641 * dbgcnt.def (prefetch): New debug counter.
20642 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
20643 (schedule_prefetches): Stop issueing prefetches if debug counter
20646 2017-06-06 Tom de Vries <tom@codesourcery.com>
20648 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
20649 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
20651 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20653 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
20654 Use aarch64_reg_or_zero predicate for operand 4.
20655 (aarch64_compare_and_swap<mode> define_insn_and_split):
20656 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
20657 (aarch64_store_exclusive<mode>): Likewise for operand 2.
20659 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
20661 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
20662 (arm_compute_save_core_reg_mask): This.
20663 (thumb1_compute_save_reg_mask): Rename into ...
20664 (thumb1_compute_save_core_reg_mask): This.
20665 (arm_compute_save_reg0_reg12_mask): Adapt comment.
20666 (arm_compute_frame_layout): Likewise.
20668 2017-06-06 Richard Biener <rguenther@suse.de>
20670 PR tree-optimization/80974
20671 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
20672 keep or clear leaders SSA info.
20674 2017-06-06 Tom de Vries <tom@codesourcery.com>
20676 * config/nvptx/nvptx.c (split_mode_p): New function.
20677 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
20679 2017-06-06 Tom de Vries <tom@codesourcery.com>
20681 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
20683 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20686 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
20689 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20691 * shrink-wrap.c (handle_simple_exit): Update profile.
20692 (try_shrink_wrapping): Upate profile.
20694 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20696 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
20697 (tree_guess_outgoing_edge_probabilities): New.
20698 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
20699 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
20701 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20703 * ipa-split.c (split_function): Initialize return bb profile.
20705 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20707 * profile.c (compute_branch_probabilities): Also initialize
20708 EXIT_BLOCK profile.
20710 2017-06-06 Richard Biener <rguenther@suse.de>
20712 PR tree-optimization/80928
20713 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
20714 (vect_analyze_loop_operations): Properly guard analysis for
20716 (vect_transform_loop): Likewise.
20717 (vect_analyze_loop_2): Also reset SLP type on PHIs.
20718 (vect_model_induction_cost): Do not cost for pure SLP.
20719 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
20720 of induction in inner loop vectorization.
20721 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
20722 (vect_get_and_check_slp_defs): Handle vect_induction_def.
20723 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
20725 (vect_analyze_slp_cost_1): Cost induction.
20726 (vect_detect_hybrid_slp_stmts): Handle PHIs.
20727 (vect_get_slp_vect_defs): Likewise.
20728 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
20729 (vect_transform_stmt): Handle SLP reductions.
20730 * tree-vectorizer.h (vectorizable_induction): Adjust.
20732 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
20734 * config/rs6000/rs6000.c (make_resolver_func): Update
20735 init_lowered_empty_function call.
20737 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
20739 * doc/invoke.texi: Document the -fprofile-abs-path option.
20740 * common.opt (fprofile-abs-path): New option.
20741 * gcov-io.h (gcov_write_filename): Declare.
20742 * gcov-io.c (gcov_write_filename): New function.
20743 * coverage.c (coverage_begin_function): Use gcov_write_filename.
20744 * profile.c (output_location): Likewise.
20746 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20748 * shring-wrap.c: Revert accidental commit.
20750 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
20752 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
20754 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20756 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
20758 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
20759 profile in callgraph edge.
20760 * profile-count.h (apply_probability): If THIS is 0, then result is 0
20761 (apply_scale): Likewise.
20762 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
20763 Also scale profile when inlining function with zero profile.
20764 (initialize_cfun): Update exit block profile even when it is zero.
20765 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
20766 when profile is read.
20768 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
20770 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
20771 (CLONE_*): New constants to define the processors we can generate
20772 code for with the target_clone attribute.
20773 (rs6000_clone_map): New array to identify which clone processors
20774 the current program is running on.
20775 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
20776 target_clone attribute.
20777 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
20778 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
20779 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
20780 (cpu_expand_builtin): Add support for target_clone attribute.
20781 (rs6000_valid_attribute_p): Allow "default" attribute.
20782 (get_decl_name): New debug function to simplify printing the
20783 current function name in debugging statements.
20784 (rs6000_clone_priority): New functions to support the target_clone
20785 attribute, and be able to generate code to switch between ISA 2.05
20786 through ISA 3.0 (power6 through power9).
20787 (rs6000_compare_version_priority): Likewise.
20788 (rs6000_get_function_versions_dispatcher): Likewise.
20789 (make_resolver_func): Likewise.
20790 (add_condition_to_bb): Likewise.
20791 (dispatch_function_versions): Likewise.
20792 (rs6000_generate_version_dispatcher_body): Likewise.
20793 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
20794 (fusion_gpr_load_p): Fix a spacing issue.
20795 * doc/extend.texi (Common Function Attributes): Document that the
20796 PowerPC supports the target_clone attribute.
20798 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
20800 * config/arm/arm.h: explain F symbol found in description of ARM
20801 register allocation in its legend.
20803 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20805 * config/mips/frame-header-opt.c: Include profile-count.h.
20806 * config/riscv/riscv.c: Include profile-count.h
20808 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20810 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
20812 (sm_set_flag_if_changed): Add bbs field.
20813 (execute_sm_if_changed_flag_set): Pass BBS.
20814 (execute_sm): Update.
20816 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20818 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
20821 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20823 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
20824 (peephole2): New peephole2 to emit the above.
20825 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
20827 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20829 * config/aarch64/aarch64.c (define_peephole2 above
20830 *sub_<shift>_<mode>): New peephole.
20832 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
20834 * config/i386/i386.c (make_resolver_func): Update.
20835 * Makefile.in: Add profile-count.h and profile-count.o
20836 * auto-profile.c (afdo_indirect_call): Update to new API.
20837 (afdo_set_bb_count): Update.
20838 (afdo_propagate_edge): Update.
20839 (afdo_propagate_circuit): Update.
20840 (afdo_calculate_branch_prob): Update.
20841 (afdo_annotate_cfg): Update.
20842 * basic-block.h: Include profile-count.h
20843 (struct edge_def): Turn count to profile_count.
20844 (struct basic_block_def): Likewie.
20845 (REG_BR_PROB_BASE): Move to profile-count.h
20846 (RDIV): Move to profile-count.h
20847 * bb-reorder.c (max_entry_count): Turn to profile_count.
20848 (find_traces): Update.
20849 (rotate_loop):Update.
20850 (connect_traces):Update.
20851 (sanitize_hot_paths):Update.
20852 * bt-load.c (migrate_btr_defs): Update.
20853 * cfg.c (RDIV): Remove.
20854 (init_flow): Use alloc_block.
20855 (alloc_block): Uninitialize count.
20856 (unchecked_make_edge): Uninitialize count.
20857 (check_bb_profile): Update.
20858 (dump_edge_info): Update.
20859 (dump_bb_info): Update.
20860 (update_bb_profile_for_threading): Update.
20861 (scale_bbs_frequencies_int): Update.
20862 (scale_bbs_frequencies_gcov_type): Update.
20863 (scale_bbs_frequencies_profile_count): New.
20864 * cfg.h (update_bb_profile_for_threading): Update.
20865 (scale_bbs_frequencies_profile_count): Declare.
20866 * cfgbuild.c (compute_outgoing_frequencies): Update.
20867 (find_many_sub_basic_blocks): Update.
20868 * cfgcleanup.c (try_forward_edges): Update.
20869 (try_crossjump_to_edge): Update.
20870 * cfgexpand.c (expand_gimple_tailcall): Update.
20871 (construct_exit_block): Update.
20872 * cfghooks.c (verify_flow_info): Update.
20873 (dump_bb_for_graph): Update.
20874 (split_edge): Update.
20875 (make_forwarder_block): Update.
20876 (duplicate_block): Update.
20877 (account_profile_record): Update.
20878 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
20879 (get_estimated_loop_iterations): Update.
20880 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
20881 (single_likely_exit): Update.
20882 * cfgloopmanip.c (scale_loop_profile): Update.
20884 (set_zero_probability): Update.
20885 (lv_adjust_loop_entry_edge): Update.
20886 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
20887 (purge_dead_edges): Update.
20888 (rtl_account_profile_record): Update.
20889 * cgraph.c (cgraph_node::create): Uninitialize count.
20890 (symbol_table::create_edge): Uninitialize count.
20891 (cgraph_update_edges_for_call_stmt_node): Update.
20892 (cgraph_edge::dump_edge_flags): Update.
20893 (cgraph_node::dump): Update.
20894 (cgraph_edge::maybe_hot_p): Update.
20895 * cgraph.h: Include profile-count.h
20896 (create_clone), create_edge, create_indirect_edge): Update.
20897 (cgraph_node): Turn count to profile_count.
20898 (cgraph_edge0: Likewise.
20899 (make_speculative, clone): Update.
20900 (create_edge): Update.
20901 (init_lowered_empty_function): Update.
20902 * cgraphclones.c (cgraph_edge::clone): Update.
20903 (duplicate_thunk_for_node): Update.
20904 (cgraph_node::create_clone): Update.
20905 * cgraphunit.c (cgraph_node::analyze): Update.
20906 (cgraph_node::expand_thunk): Update.
20907 * final.c (dump_basic_block_info): Update.
20908 * gimple-streamer-in.c (input_bb): Update.
20909 * gimple-streamer-out.c (output_bb): Update.
20910 * graphite.c (print_global_statistics): Update.
20911 (print_graphite_scop_statistics): Update.
20912 * hsa-brig.c: Include basic-block.h.
20913 * hsa-dump.c: Include basic-block.h.
20914 * hsa-gen.c (T sum_slice): Update.
20915 (convert_switch_statements):Update.
20916 * hsa-regalloc.c: Include basic-block.h.
20917 * ipa-chkp.c (chkp_produce_thunks): Update.
20918 * ipa-cp.c (struct caller_statistics): Update.
20919 (init_caller_stats): Update.
20920 (gather_caller_stats): Update.
20921 (ipcp_cloning_candidate_p): Update.
20922 (good_cloning_opportunity_p): Update.
20923 (get_info_about_necessary_edges): Update.
20924 (dump_profile_updates): Update.
20925 (update_profiling_info): Update.
20926 (update_specialized_profile): Update.
20927 (perhaps_add_new_callers): Update.
20928 (decide_about_value): Update.
20929 (ipa_cp_c_finalize): Update.
20930 * ipa-devirt.c (struct odr_type_warn_count): Update.
20931 (struct decl_warn_count): Update.
20932 (struct final_warning_record): Update.
20933 (possible_polymorphic_call_targets): Update.
20934 (ipa_devirt): Update.
20935 * ipa-fnsummary.c (redirect_to_unreachable): Update.
20936 * ipa-icf.c (sem_function::merge): Update.
20937 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
20938 * ipa-inline.c (compute_uninlined_call_time): Update.
20939 (compute_inlined_call_time): Update.
20940 (want_inline_small_function_p): Update.
20941 (want_inline_self_recursive_call_p): Update.
20942 (edge_badness): Update.
20943 (lookup_recursive_calls): Update.
20944 (recursive_inlining): Update.
20945 (inline_small_functions): Update.
20946 (dump_overall_stats): Update.
20947 (dump_inline_stats): Update.
20948 * ipa-profile.c (ipa_profile_generate_summary): Update.
20949 (ipa_propagate_frequency): Update.
20950 (ipa_profile): Update.
20951 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
20952 * ipa-utils.c (ipa_merge_profiles): Update.
20953 * loop-doloop.c (doloop_modify): Update.
20954 * loop-unroll.c (report_unroll): Update.
20955 (unroll_loop_runtime_iterations): Update.
20956 * lto-cgraph.c (lto_output_edge): Update.
20957 (lto_output_node): Update.
20958 (input_node): Update.
20959 (input_edge): Update.
20960 (merge_profile_summaries): Update.
20961 * lto-streamer-in.c (input_cfg): Update.
20962 * lto-streamer-out.c (output_cfg): Update.
20963 * mcf.c (create_fixup_graph): Update.
20964 (adjust_cfg_counts): Update.
20965 (sum_edge_counts): Update.
20966 * modulo-sched.c (sms_schedule): Update.
20967 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
20968 * predict.c (maybe_hot_count_p): Update.
20969 (probably_never_executed): Update.
20970 (dump_prediction): Update.
20971 (combine_predictions_for_bb): Update.
20972 (propagate_freq): Update.
20973 (handle_missing_profiles): Update.
20974 (counts_to_freqs): Update.
20975 (rebuild_frequencies): Update.
20976 (force_edge_cold): Update.
20977 * predict.h: Include profile-count.h
20978 (maybe_hot_count_p, counts_to_freqs): UPdate.
20979 * print-rtl-function.c: Do not include cfg.h
20980 * print-rtl.c: Include basic-block.h
20981 * profile-count.c: New file.
20982 * profile-count.h: New file.
20983 * profile.c (is_edge_inconsistent): Update.
20984 (correct_negative_edge_counts): Update.
20985 (is_inconsistent): Update.
20986 (set_bb_counts): Update.
20987 (read_profile_edge_counts): Update.
20988 (compute_frequency_overlap): Update.
20989 (compute_branch_probabilities): Update; Initialize and deinitialize
20991 (branch_prob): Update.
20992 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
20993 (edge_gcov_count): New.
20994 (bb_gcov_count): New.
20995 * shrink-wrap.c (try_shrink_wrapping): Update.
20996 * tracer.c (better_p): Update.
20997 * trans-mem.c (expand_transaction): Update.
20998 (ipa_tm_insert_irr_call): Update.
20999 (ipa_tm_insert_gettmclone_call): Update.
21000 * tree-call-cdce.c: Update.
21001 * tree-cfg.c (gimple_duplicate_sese_region): Update.
21002 (gimple_duplicate_sese_tail): Update.
21003 (gimple_account_profile_record): Update.
21004 (execute_fixup_cfg): Update.
21005 * tree-inline.c (copy_bb): Update.
21006 (copy_edges_for_bb): Update.
21007 (initialize_cfun): Update.
21008 (freqs_to_counts): Update.
21009 (copy_cfg_body): Update.
21010 (expand_call_inline): Update.
21011 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
21012 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
21013 (try_unroll_loop_completely): Update.
21014 (try_peel_loop): Update.
21015 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
21016 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
21017 * tree-ssa-loop-split.c (connect_loops): Update.
21018 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
21019 * tree-ssa-reassoc.c (branch_fixup): Update.
21020 * tree-ssa-tail-merge.c (replace_block_by): Update.
21021 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
21022 (compute_path_counts): Update.
21023 (update_profile): Update.
21024 (recompute_probabilities): Update.
21025 (update_joiner_offpath_counts): Update.
21026 (estimated_freqs_path): Update.
21027 (freqs_to_counts_path): Update.
21028 (clear_counts_path): Update.
21029 (ssa_fix_duplicate_block_edges): Update.
21030 (duplicate_thread_path): Update.
21031 * tree-switch-conversion.c (case_bit_test_cmp): Update.
21032 (struct switch_conv_info): Update.
21033 * tree-tailcall.c (decrease_profile): Update.
21034 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
21035 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
21036 * value-prof.c (check_counter): Update.
21037 (gimple_divmod_fixed_value): Update.
21038 (gimple_mod_pow2): Update.
21039 (gimple_mod_subtract): Update.
21040 (gimple_ic_transform): Update.
21041 (gimple_stringop_fixed_value): Update.
21042 * value-prof.h (gimple_ic): Update.
21044 2017-06-02 Carl Love <cel@us.ibm.com>
21046 * config/rs6000/rs6000-c: Add support for built-in functions
21047 vector double vec_doublee (vector signed int);
21048 vector double vec_doublee (vector unsigned int);
21049 vector double vec_doublee (vector float);
21050 vector double vec_doubleh (vector signed int);
21051 vector double vec_doubleh (vector unsigned int);
21052 vector double vec_doubleh (vector float);
21053 vector double vec_doublel (vector signed int);
21054 vector double vec_doublel (vector unsigned int);
21055 vector double vec_doublel (vector float);
21056 vector double vec_doubleo (vector signed int);
21057 vector double vec_doubleo (vector unsigned int);
21058 vector double vec_doubleo (vector float);.
21059 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
21060 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
21062 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
21063 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
21064 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
21066 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
21067 vec_doublel, vec_doubleh.
21068 * doc/extend.texi: Update the built-in documentation file for the
21069 new built-in functions.
21071 2017-06-02 David Malcolm <dmalcolm@redhat.com>
21074 * ipa-inline-analysis.c (free_growth_caches): Set
21075 edge_removal_hook_holder to NULL after removing it.
21077 2017-06-02 Sudakshina Das <sudi.das@arm.com>
21079 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
21080 comparision with zero.
21082 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21083 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21084 for early expansion of vec_min and vec_max builtins.
21085 (builtin_function_type): Add min/max unsigned variants to those
21086 identified as having unsigned arguments.
21088 2017-06-02 Olivier Hainque <hainque@adacore.com>
21090 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
21092 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21094 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
21095 Use VALL_F16 iterator rather than VALL.
21097 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21099 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
21100 Emit CBNZ inside loop when doing a strong exchange and comparing
21101 against zero. Generate the CC flags after the loop.
21103 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
21105 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
21106 (dl_section_ref): New.
21107 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
21108 On AIX, append an expression to subtract the size of the
21109 section length to dl_section_ref.
21111 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21113 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21114 for early expansion of vector absolute builtins.
21116 2017-06-02 Richard Biener <rguenther@suse.de>
21118 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
21119 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
21121 2017-06-02 Richard Biener <rguenther@suse.de>
21123 PR tree-optimization/80948
21124 * tree-tailcall.c (find_tail_calls): Track stmts to move in
21125 stmt order as well.
21127 2017-06-02 Richard Biener <rguenther@suse.de>
21129 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
21131 * tree-vect-stmts.c (process_use): Do not mark backedge defs
21132 for inductions as relevant.
21134 2017-06-02 Richard Biener <rguenther@suse.de>
21136 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
21137 (vectorizable_induction): ... this. Remove dead code.
21139 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
21141 * builtins. (expand_builtin_alloca): Remove second parameter and
21142 infer its value from the first parameter instead.
21143 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
21145 2017-06-02 Jakub Jelinek <jakub@redhat.com>
21147 PR rtl-optimization/80903
21148 * loop-doloop.c (add_test): Unshare sequence.
21150 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
21152 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
21154 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
21156 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
21158 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
21159 xlogue_layout::get_instance, logue_layout::xlogue_layout,
21160 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
21161 (xlogue_layout::get_stub_rtx): Make static.
21162 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
21163 (xlogue_layout::compute_stub_managed_regs): Rename to...
21164 (xlogue_layout::count_stub_managed_regs): ...this.
21165 (xlogue_layout::is_stub_managed_reg): New function.
21166 (xlogue_layout::m_stub_names): Rename to...
21167 (xlogue_layout::s_stub_names): ...this, make static.
21168 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
21169 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
21170 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
21171 xlogue_layout::s_stub_names): Instantiate statics.
21172 (stub_managed_regs): Remove.
21173 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
21174 (disable_call_ms2sysv_xlogues): Rename to...
21175 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
21176 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
21178 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
21179 change after reload_completed.
21180 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
21182 (ix86_expand_prologue): Likewise.
21183 (ix86_expand_epilogue): Likewise.
21184 (ix86_expand_split_stack_prologue): Likewise.
21185 (ix86_compute_frame_layout): Remove frame parameter ...
21186 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
21187 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
21189 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
21190 (ix86_frame): Move from here ...
21191 * config/i386/i386.h (ix86_frame): ... to here.
21192 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
21193 complete ix86_frame data structure instead. Remove some_ld_name.
21195 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
21197 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
21198 symbols that hold a DECL_VALUE_EXPR.
21200 2017-06-01 Martin Jambor <mjambor@suse.cz>
21202 PR tree-optimization/80898
21203 * tree-sra.c (process_subtree_disqualification): Removed.
21204 (disqualify_candidate): Do not acll
21205 process_subtree_disqualification.
21206 (subtree_mark_written_and_enqueue): New function.
21207 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
21208 RHS has been disqualified and re-queue LHS if necessary. Apart
21209 from that, ignore disqualified RHS.
21211 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21213 * config/s390/s390.c (s390_emit_epilogue): Disable early return
21214 address fetch for z10 or later.
21216 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21218 * config/arc/arc.md (tst_movb): Add guard when splitting.
21220 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21222 * config/arc/arc.c (arc_can_eliminate): Test against
21223 arc_frame_pointer_needed.
21225 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21227 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
21228 to prevent store reordering.
21229 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
21230 (type): Add block type.
21231 (stack_tie): Define special instruction to be used in
21234 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21236 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
21237 constraint. It is not valid for the pattern.
21238 (noncommutative_binary_comparison): Likewise.
21240 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21242 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
21245 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21247 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
21248 be used by the reg-alloc.
21250 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21252 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
21253 reg-alloc when having mul64 or mul32x16 instructions.
21254 (mulsidi3): Likewise.
21255 (umulsidi3): Likewise.
21256 (mulsi32x16): New pattern.
21257 (mulsi64): Likewise.
21258 (mulsidi64): Likewise.
21259 (umulsidi64): Likewise.
21260 (MUL32x16_REG): Define.
21261 (mul64_600): Use MUL32x16_REG.
21262 (mac64_600): Likewise.
21263 (umul64_600): Likewise.
21264 (umac64_600): Likewise.
21266 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21268 * config/arc/arc.md (mulsi3_700): Make it commutative.
21270 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
21272 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
21274 (*sign_extendsidi2_insn): Likewise for movstosw.
21276 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
21278 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
21279 the type of the input discriminant value. Convert the
21280 discriminant value of signedness vary.
21282 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
21284 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
21285 Add to -Wall section.
21287 2017-06-01 Richard Biener <rguenther@suse.de>
21289 PR middle-end/66313
21290 * fold-const.c (fold_plusminus_mult_expr): If the factored
21291 factor may be zero use a wrapping type for the inner operation.
21292 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
21293 and handle moved defs.
21294 (process_assignment): Properly guard the unary op case. Return a
21295 tri-state indicating that moving the stmt before the call may allow
21296 to continue. Pass through to_move.
21297 (find_tail_calls): Handle moving unrelated defs before
21300 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
21303 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
21304 splitter result in the canonical way.
21306 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
21308 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
21309 also for 32bit target. Update insn attributes.
21310 (zero-extendsidi2 splitter): Allow all registers for operand 1.
21312 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
21314 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
21315 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
21316 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
21317 (_mm_maskz_min_ss): New intrinsics.
21319 2017-05-31 Martin Liska <mliska@suse.cz>
21321 * tree-vect-loop.c (vect_create_epilog_for_reduction):
21322 Change comment style to one we normally use.
21323 (vectorizable_reduction): Likewise.
21324 (vectorizable_induction): Likewise.
21325 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
21326 (vectorizable_call): Likewise.
21327 (vectorizable_simd_clone_call): Likewise.
21328 (vectorizable_conversion): Likewise.
21329 (vectorizable_assignment): Likewise.
21330 (vectorizable_shift): Likewise.
21331 (vectorizable_operation): Likewise.
21332 (vectorizable_store): Likewise.
21333 (vectorizable_load): Likewise.
21334 * tree-vectorizer.h: Likewise.
21336 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
21338 * passes.c (emergency_dump_function): New.
21339 * tree-pass.h (emergency_dump_function): Declare.
21340 * plugin.c (plugins_internal_error_function): Remove.
21341 * plugin.h (plugins_internal_error_function): Remove declaration.
21342 * toplev.c (internal_error_function): New static function. Use it...
21343 (general_init): ...here.
21345 2017-05-31 Graham Markall <graham.markall@embecosm.com>
21347 * config/arc/arc.c (arc_print_operand): Handle constant operands.
21348 (arc_rtx_costs): Add costs for new patterns.
21349 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
21350 * config/arc/predicates.md: Add _1_2_3_operand predicate.
21352 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21354 * tree-ssa-strlen.c (get_next_strinfo): New function.
21355 (get_stridx_plus_constant): Use it.
21356 (zero_length_string): Likewise.
21357 (adjust_related_strinfos): Likewise.
21358 (adjust_last_stmt): Likewise.
21360 2017-05-31 Richard Biener <rguenther@suse.de>
21363 * config/i386/i386.c (ix86_expand_builtin): Remove assert
21364 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
21366 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21368 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
21369 loop_vinfo argument and use of dependence distance vectors.
21370 Check instead whether the two references differ only in their
21371 initial value and assume that they have the same alignment if the
21372 difference is a multiple of the vector alignment.
21373 (vect_analyze_data_refs_alignment): Update call accordingly.
21375 2017-05-31 Martin Liska <mliska@suse.cz>
21378 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
21380 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21382 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
21383 (create_intersect_range_checks): Move from ...
21384 * tree-data-ref.c (create_intersect_range_checks_index)
21385 (create_intersect_range_checks): ... to here.
21386 (create_runtime_alias_checks): New function factored from ...
21387 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
21388 here. Call above function.
21389 * tree-data-ref.h (create_runtime_alias_checks): New function.
21391 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21393 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
21394 segment length for dr_b and compute it in wide_int.
21396 2017-05-31 Richard Biener <rguenther@suse.de>
21398 PR tree-optimization/80906
21399 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
21400 and pass through iv_map.
21401 (copy_bb_and_scalar_dependences): Adjust.
21402 (translate_pending_phi_nodes): Likewise.
21403 (copy_loop_close_phi_args): Handle code-generating IVs instead
21406 2017-05-30 David Malcolm <dmalcolm@redhat.com>
21408 * diagnostic-color.c (color_dict): Add "type-diff".
21409 (parse_gcc_colors): Update comment.
21410 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
21411 -fdiagnostics-show-template-tree and -fno-elide-type.
21412 (GCC_COLORS): Add type-diff to example.
21414 (-fdiagnostics-show-template-tree): New.
21415 (-fno-elide-type): New.
21416 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
21417 the pp_format_decoder callback. Call any m_format_postprocessor's
21419 (pretty_printer::pretty_printer): Initialize
21420 m_format_postprocessor.
21421 (pretty_printer::~pretty_printer): Delete any
21422 m_format_postprocessor.
21423 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
21424 (class format_postprocessor): New class.
21425 (struct pretty_printer::format_decoder): Document the new parameters.
21426 (struct pretty_printer::m_format_postprocessor): New field.
21427 * tree-diagnostic.c (default_tree_printer): Update for new
21428 bool and const char ** params.
21429 * tree-diagnostic.h (default_tree_printer): Likewise.
21431 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
21433 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
21434 (lwa_operand): Delete rs6000_gen_cell_microcode test.
21435 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
21436 rs6000_gen_cell_microcode code.
21437 (rs6000_final_prescan_insn): Delete.
21438 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
21439 "warn-cell-microcode" entries.
21440 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
21441 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
21442 throughout. Change cc_reg_not_micro_cr0_operand to
21443 cc_reg_not_cr0_operand throughout.
21444 (*extendhi<mode>2_noload): Delete.
21445 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
21446 (mwarn-cell-microcode): Delete.
21447 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
21448 -mgen-cell-microcode and -mwarn-cell-microcode.
21450 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
21453 * config/i386/constraints.md (Yd): New constraint.
21455 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
21456 and (?Yd, r) alternatives. Update insn attributes.
21457 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
21458 and (?*Yd, r) alternatives. Update insn attributes.
21459 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
21461 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
21463 * gimplify.c (gimplify_modify_expr): Don't create a
21464 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
21467 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
21469 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
21471 2017-05-30 Richard Biener <rguenther@suse.de>
21473 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
21474 and reduc_def fields.
21475 (STMT_VINFO_REDUC_TYPE): New define.
21476 (STMT_VINFO_REDUC_DEF): Likewise.
21477 (vect_force_simple_reduction): Adjust prototype.
21478 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
21479 (vect_is_simple_reduction): Remove check_reduction argument.
21480 (vect_force_simple_reduction): Adjust and set
21481 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21482 (vectorizable_reduction): Do not re-do reduction analysis
21483 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21484 * tree-parloops.c (gather_scalar_reductions): Adjust.
21486 2017-05-30 Richard Biener <rguenther@suse.de>
21488 PR middle-end/80901
21489 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
21492 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
21494 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
21495 Introduce unknown_misalignment parameter and remove vf.
21496 (vect_peeling_hash_get_lowest_cost):
21497 Pass unknown_misalignment parameter.
21498 (vect_enhance_data_refs_alignment):
21499 Fix unsupportable data ref treatment.
21501 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21503 * tree-vect-data-refs.c (vect_get_data_access_cost):
21504 Workaround for SLP handling.
21505 (vect_enhance_data_refs_alignment):
21506 Compute costs for doing no peeling at all, compare to the best
21507 peeling costs so far and avoid peeling if cheaper.
21509 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21511 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
21512 Return peeling info and set costs to zero for unlimited cost
21514 (vect_enhance_data_refs_alignment): Also inspect all datarefs
21515 with unknown misalignment. Compute and costs for unknown
21516 misalignment, compare them to the costs for known misalignment
21517 and choose the cheapest for peeling.
21519 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21521 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
21522 (vect_get_peeling_costs_all_drs): Create function.
21523 (vect_peeling_hash_get_lowest_cost):
21524 Use vect_get_peeling_costs_all_drs.
21525 (vect_peeling_supportable): Create function.
21526 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
21528 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21530 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
21531 DR_HAS_NEGATIVE_STEP.
21532 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
21533 (vect_enhance_data_refs_alignment): Use.
21534 (vect_duplicate_ssa_name_ptr_info): Use.
21535 * tree-vectorizer.h (dr_misalignment): Use.
21536 (known_alignment_for_access_p): Use.
21538 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21541 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
21543 (has_section_name): New function.
21545 2017-05-30 Martin Liska <mliska@suse.cz>
21548 * auto-profile.c (get_function_decl_from_block): Fix
21551 2017-05-30 Richard Biener <rguenther@suse.de>
21553 PR middle-end/80876
21554 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
21556 2017-05-30 Martin Liska <mliska@suse.cz>
21558 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
21559 * dumpfile.h (struct dump_file_info): Remove ctors.
21561 2017-05-30 Martin Liska <mliska@suse.cz>
21563 * predict.def: Fix GNU coding style.
21565 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
21567 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
21568 Mark 'to' argument with ATTRIBUTE_UNUSED.
21570 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
21572 * config/xtensa/xtensa.c (xtensa_emit_call): Use
21573 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
21574 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
21577 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
21579 * doc/install.texi (Options specification): Restore entry of
21580 --enable-sjlj-exceptions.
21582 2017-05-27 Michael Eager <eager@eagercon.com>
21585 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
21587 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
21589 * config/microblaze/microblaze.h
21590 (FIXED_REGISTERS): Update in macro.
21591 (CALL_USED_REGISTERS): Update in macro.
21593 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
21595 * doc/install.texi: Add links to macOS binary distributions.
21597 2017-05-27 Jakub Jelinek <jakub@redhat.com>
21601 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21603 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
21605 2017-05-26 Martin Liska <mliska@suse.cz>
21607 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
21609 2017-05-26 Martin Liska <mliska@suse.cz>
21611 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
21612 always leading ';; '.
21613 (dump_bb_info): Likewise.
21614 (brief_dump_cfg): Likewise.
21615 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
21616 * dumpfile.c: Remove usage of TDF_VERBOSE.
21617 * dumpfile.h (enum dump_kind): Likewise.
21618 (dump_gimple_bb_header): Do not use TDF_COMMENT.
21619 * print-tree.c (debug_verbose): Remove.
21620 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
21621 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
21622 * tree-diagnostic.c (default_tree_printer): Replace
21623 TDF_DIAGNOSTIC with TDF_SLIM.
21625 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21627 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
21628 in parameter loop, rather than loop_vinfo.
21629 (create_intersect_range_checks): Ditto.
21630 (vect_create_cond_for_alias_checks): Update call to above functions.
21632 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21634 PR tree-optimization/80815
21635 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
21636 for merging runtime alias checks. Handle negative DR_STEPs.
21638 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21640 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
21642 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
21643 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
21644 out code pruning runtime alias checks.
21645 * tree-data-ref.c (prune_runtime_alias_test_list): New function
21646 factored out from above.
21647 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
21649 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
21651 (prune_runtime_alias_test_list): New decalaration.
21653 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21655 * tree-vect-data-refs.c (compare_tree): Rename and move ...
21656 * tree-data-ref.c (data_ref_compare_tree): ... to here.
21657 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
21658 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
21659 (operator==, comp_dr_with_seg_len_pair): Ditto.
21660 (vect_prune_runtime_alias_test_list): Ditto.
21662 2017-05-26 Martin Liska <mliska@suse.cz>
21665 * params.def: Bound partial-inlining-entry-probability param.
21667 2017-05-26 Marek Polacek <polacek@redhat.com>
21670 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
21673 2017-05-26 Richard Biener <rguenther@suse.de>
21675 PR tree-optimization/80842
21676 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
21679 2017-05-26 Richard Biener <rguenther@suse.de>
21681 PR tree-optimization/80844
21682 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
21684 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
21686 * doc/md.texi (Machine Constraints): Update x86 family
21687 machine constraints section to match 'config/i386/constraints.md'.
21689 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
21691 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
21693 2017-05-25 Nathan Sidwell <nathan@acm.org>
21695 * doc/invoke.texi (--enable-languages): Update documentation.
21697 2017-05-25 Martin Liska <mliska@suse.cz>
21699 * dumpfile.c: Add TDF_FOLDING.
21700 * dumpfile.h (enum dump_kind): Likewise.
21701 * genmatch.c (dt_simplify::gen_1): Use it.
21703 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21705 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
21707 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21709 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
21710 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
21712 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21714 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
21715 * match.pd (X == C): Rewrite it here.
21716 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
21717 with_certain_nonzero_bits2): New predicates.
21718 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
21720 2017-05-24 Nathan Sidwell <nathan@acm.org>
21722 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
21725 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
21728 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
21730 * config/powerpcspe: New port. Files are copied from the rs6000
21731 port, with "rs6000" in filenames replaced by "powerpcspe".
21733 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
21735 PR rtl-optimization/80754
21736 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
21738 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
21740 * config/sparc/sparc.md (length): Return the correct value for -mflat
21741 sibcalls to match output_sibcall.
21743 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
21747 * config/rs6000/rs6000.c (struct machine_function): Add new field
21749 (rs6000_get_separate_components): Init that field, use it.
21750 (rs6000_components_for_bb): Use the field.
21752 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
21754 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
21756 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
21758 PR middle-end/80823
21759 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
21761 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21764 * config/s390/s390.c (s390_check_qrst_address): Check incoming
21765 address against address_operand predicate.
21766 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
21768 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
21770 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
21771 parameters passed indirectly.
21773 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
21775 * config/i386/i386.md (*movdi_internal): Remove SSE4
21776 alternative 18 (?r, *v). Update insn attributes.
21777 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
21778 Update insn attributes.
21779 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
21780 Update insn attributes.
21781 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
21782 alternative 1 (r, v). Remove isa attribute.
21783 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
21784 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
21785 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
21787 2017-05-23 Tom de Vries <tom@codesourcery.com>
21789 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
21792 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21794 * cgraphunit.c (symbol_table::process_new_functions): Update.
21795 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
21796 (inline_generate_summary): Rename to ...
21797 (ipa_fn_summary_generate): ... this one.
21798 (inline_read_summary): Rename to ...
21799 (ipa_fn_summary_read): ... this one.
21800 (inline_write_summary): Rename to ...
21801 (ipa_fn_summary_write): ... this one.
21802 (inline_free_summary): Rename to ...
21803 (ipa_free_fn_summary): ... this one.
21804 (pass_data_local_fn_summary, pass_local_fn_summary,
21805 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
21806 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
21807 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
21808 make_pass_ipa_fn_summary): New.
21809 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
21810 inline_write_summary, inline_free_summary): Remove.
21811 (ipa_free_fn_summary) : New.
21812 * ipa-inline.c (ipa_inline): Update.
21813 (pass_ipa_inline): Do not generate summaries.
21814 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
21816 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
21817 and add pass_ipa_fn_summary.
21818 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
21820 (make_pass_inline_parameters): Remove.
21822 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
21824 * omp-low.c (struct omp_context): Remove "default_kind" member.
21827 * omp-offload.c (execute_oacc_device_lower): Remove the
21828 parallelism dimensions function attributes for unparallelized
21829 OpenACC kernels constructs.
21831 2017-05-23 Martin Liska <mliska@suse.cz>
21833 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
21835 (cgraph_edge::make_speculative): Likewise.
21836 (cgraph_edge::resolve_speculation): Likewise.
21837 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
21838 (cgraph_node::dump): Likewise.
21839 * cgraph.h: Likewise.
21840 * cgraphunit.c (analyze_functions): Likewise.
21841 (symbol_table::compile): Likewise.
21842 * ipa-cp.c (print_all_lattices): Likewise.
21843 (determine_versionability): Likewise.
21844 (initialize_node_lattices): Likewise.
21845 (ipcp_verify_propagated_values): Likewise.
21846 (estimate_local_effects): Likewise.
21847 (update_profiling_info): Likewise.
21848 (create_specialized_node): Likewise.
21849 (perhaps_add_new_callers): Likewise.
21850 (decide_about_value): Likewise.
21851 (decide_whether_version_node): Likewise.
21852 (identify_dead_nodes): Likewise.
21853 (ipcp_store_bits_results): Likewise.
21854 * ipa-devirt.c (dump_targets): Likewise.
21855 (ipa_devirt): Likewise.
21856 * ipa-icf.c (sem_item::dump): Likewise.
21857 (sem_function::equals): Likewise.
21858 (sem_variable::equals): Likewise.
21859 (sem_item_optimizer::read_section): Likewise.
21860 (sem_item_optimizer::execute): Likewise.
21861 (congruence_class::dump): Likewise.
21862 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
21863 (dump_inline_summary): Likewise.
21864 (estimate_node_size_and_time): Likewise.
21865 (inline_analyze_function): Likewise.
21866 * ipa-inline-transform.c (inline_call): Likewise.
21867 * ipa-inline.c (report_inline_failed_reason): Likewise.
21868 (want_early_inline_function_p): Likewise.
21869 (edge_badness): Likewise.
21870 (update_edge_key): Likewise.
21871 (inline_small_functions): Likewise.
21872 * ipa-profile.c (ipa_profile): Likewise.
21873 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
21874 (ipa_make_edge_direct_to_target): Likewise.
21875 (remove_described_reference): Likewise.
21876 (ipa_impossible_devirt_target): Likewise.
21877 (propagate_controlled_uses): Likewise.
21878 (ipa_print_node_params): Likewise.
21879 (ipcp_transform_function): Likewise.
21880 * ipa-pure-const.c (pure_const_read_summary): Likewise.
21881 (propagate_pure_const): Likewise.
21882 * ipa-reference.c (generate_summary): Likewise.
21883 (read_write_all_from_decl): Likewise.
21884 (propagate): Likewise.
21885 (ipa_reference_read_optimization_summary): Likewise.
21886 * ipa-utils.c (ipa_merge_profiles): Likewise.
21887 * ipa.c (walk_polymorphic_call_targets): Likewise.
21888 (symbol_table::remove_unreachable_nodes): Likewise.
21889 (ipa_single_use): Likewise.
21890 * passes.c (execute_todo): Likewise.
21891 * predict.c (drop_profile): Likewise.
21892 * symtab.c (symtab_node::get_dump_name): New function.
21893 (symtab_node::dump_name): Likewise.
21894 (symtab_node::dump_asm_name): Likewise.
21895 (symtab_node::dump_references): Likewise.
21896 (symtab_node::dump_referring): Likewise.
21897 (symtab_node::dump_base): Likewise.
21898 (symtab_node::debug_symtab): Likewise.
21899 * tree-sra.c (convert_callers_for_node): Likewise.
21900 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
21901 * value-prof.c (init_node_map): Likewise.
21903 2017-05-23 Martin Liska <mliska@suse.cz>
21905 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
21906 and symtab_node::debug_symtab to symbol_table::debug.
21907 * cgraphunit.c (analyze_functions): Use the renamed function.
21908 (symbol_table::compile): Likewise.
21909 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
21910 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
21911 * passes.c (execute_todo): Likewise.
21912 * symtab.c (symbol_table::dump): New function.
21913 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
21915 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21917 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
21918 that nonconst implies exec.
21920 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21922 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
21923 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
21924 (inline_edge_summary_vec): Turn into ...
21925 (ipa_call_summaries): ... this one.
21926 (redirect_to_unreachable, edge_set_predicate,
21927 evaluate_properties_for_edge, inline_summary_alloc,
21928 reset_ipa_call_summary, reset_inline_summary,
21929 inline_summary_t::duplicate): Update.
21930 (inline_edge_duplication_hook): Turn to ...
21931 (ipa_call_summary_t::duplicate): ... this one.
21932 (inline_edge_removal_hook): Turn to ...
21933 (ipa_call_summary_t::remove): ... this one.
21934 (dump_inline_edge_summary): Turn to ...
21935 (dump_ipa_call_summary): ... this one.
21936 (estimate_function_body_sizes): Update.
21937 (inline_update_callee_summaries): Update.
21938 (remap_edge_change_prob): Update.
21939 (remap_edge_summaries): Update.
21940 (inline_merge_summary): Update.
21941 (do_estimate_edge_time): Update.
21942 (inline_generate_summary): Update.
21943 (inline_read_section): Update.
21944 (inline_read_summary): Update.
21945 (inline_free_summary): Update.
21946 * ipa-inline.c (can_inline_edge_p): Update.
21947 (compute_inlined_call_time): Update.
21948 (want_inline_small_function_p): Update.
21949 (edge_badness): Update.
21950 (early_inliner): Update.
21951 * ipa-inline.h (inline_edge_summary): Turn to ...
21952 (ipa_call_summary): ... this one.
21953 (ipa_call_summary_t): New class.
21954 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
21955 (ipa_call_summaries): New.
21956 (inline_edge_summary): Remove.
21957 (estimate_edge_growth): Update.
21958 * ipa-profile.c (ipa_propagate_frequency_1): Update.
21959 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
21960 * ipa-split.c (execute_split_functions): Update.
21961 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
21963 2017-05-23 Tom de Vries <tom@codesourcery.com>
21965 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
21966 attributes): Document rdrand effective target.
21968 2017-05-23 Tom de Vries <tom@codesourcery.com>
21970 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
21971 attributes): Sort alphabetically.
21973 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
21975 * config/avr/genmultilib.awk: Use gsub instead of gensub.
21977 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
21980 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
21981 V2DF/V2DI splat into two separate patterns, one that handles
21982 registers, and the other that only handles memory. Drop support
21983 for splatting from a GPR on ISA 2.07 and then splitting the
21984 splat into direct move and splat.
21985 (vsx_splat_<mode>_reg): Likewise.
21986 (vsx_splat_<mode>_mem): Likewise.
21988 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
21990 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
21992 2017-05-22 Jakub Jelinek <jakub@redhat.com>
21994 PR middle-end/80809
21995 * omp-low.c (finish_taskreg_remap): New function.
21996 (finish_taskreg_scan): If unit size of ctx->record_type
21997 is non-constant, unshare the size expression and replace
21998 decls in it with possible outer var refs.
22000 PR middle-end/80809
22001 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
22002 GOVD_SHARED rather than GOVD_PRIVATE with it.
22003 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
22004 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
22006 PR middle-end/80853
22007 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
22008 as last argument to build_outer_var_ref for pointer bases of array
22009 section reductions.
22011 2017-05-19 Martin Sebor <msebor@redhat.com>
22013 * print-tree.c (print_node): Print DECL_READ_P flag.
22015 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22017 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
22018 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
22019 * cgraph.c: Likewise.
22020 * cgraphunit.c: Likewise.
22021 * gengtype.c: Likewise.
22022 * ipa-cp.c: Likewise.
22023 * ipa-devirt.c: Likewise.
22024 * ipa-icf.c: Likewise.
22025 * ipa-predicate.c: Likewise.
22026 * ipa-profile.c: Likewise.
22027 * ipa-prop.c: Likewise.
22028 * ipa-split.c: Likewise.
22030 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
22031 edge_predicate_pool, dump_inline_hints,
22032 inline_summary::account_size_time, redirect_to_unreachable,
22033 edge_set_predicate, set_hint_predicate,
22034 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
22035 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
22036 inline_summary_t::remove, remap_hint_predicate_after_duplication,
22037 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
22038 ipa_call_summary_t::remove, initialize_growth_caches,
22039 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
22040 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
22041 mark_modified, unmodified_parm_1, unmodified_parm,
22042 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
22043 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
22044 compute_bb_predicates, will_be_nonconstant_expr_predicate,
22045 will_be_nonconstant_predicate, record_modified_bb_info,
22046 get_minimal_bb, record_modified, param_change_prob,
22047 phi_result_unknown_predicate, predicate_for_phi_result,
22048 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
22049 estimate_function_body_sizes, compute_inline_parameters,
22050 compute_inline_parameters_for_curren, pass_data_inline_parameters,
22051 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
22052 inline_update_callee_summaries, remap_edge_change_prob,
22053 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
22054 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
22055 inline_analyze_function, inline_summary_t::insert,
22056 inline_generate_summary, read_ipa_call_summary, inline_read_section,
22057 inline_read_summary, write_ipa_call_summary, inline_write_summary,
22058 inline_free_summary): Move to ipa-fnsummary.h
22059 (predicate_t): Remove.
22060 * ipa-fnsummary.c: New file.
22061 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
22062 (enum inline_hints_vals, inline_hints, agg_position_info,
22063 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
22064 inline_summaries, ipa_call_summary, ipa_call_summary_t,
22065 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
22066 dump_inline_summary, dump_inline_hints, inline_generate_summary,
22067 inline_read_summary, inline_write_summary, inline_free_summary,
22068 inline_analyze_function, initialize_inline_failed,
22069 inline_merge_summary, inline_update_overall_summary,
22070 compute_inline_parameters): Move to ipa-fnsummary.h
22071 * ipa-fnsummary.h: New file.
22072 * ipa-inline-transform.h: Include ipa-inline.h.
22073 * ipa-inline.c: LIkewise.
22075 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22077 * ipa-inline.c (edge_badness): Use inlined_time instead of
22078 inline_summaries->get.
22080 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22082 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
22084 2017-05-22 Nathan Sidwell <nathan@acm.org>
22086 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
22087 (fdump-lang): Document 'raw' option.
22088 * dumpfile.h (TDI_tu): Delete.
22089 * dumpfile.c (dump_files): Remove translation-unit.
22090 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
22092 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
22094 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
22095 command option from $(AWK) call.
22096 * config/avr/genmultilib.awk: Simplify and rewrite so that it
22097 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
22098 [FORMAT]: Remove handling of variable.
22099 * config/avr/t-multilib: Regenerate.
22101 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22103 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
22105 (dump_inline_summary): Do not print self_time.
22106 (estimate_function_body_sizes): Do not set self_time.
22107 (compute_inline_parameters): Likewise.
22108 (inline_read_section, inline_write_summary): Do not stream self_time.
22109 * ipa-inline.h (inline_summary): Drop self_time.
22111 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22113 * ipa-inline-analysis.c (account_size_time): Rename to ...
22114 (inline_summary::account_size_time): ... this one.
22115 (reset_ipa_call_summary): Turn to ...
22116 (ipa_call_summary::reset): ... this one.
22117 (reset_inline_summary): Turn to ...
22118 (inline_summary::reset): ... this one.
22119 (inline_summary_t::remove): Update.
22120 (inline_summary_t::duplicate): Update.
22121 (ipa_call_summary_t::remove): Update.
22122 (dump_inline_summary): Update.
22123 (estimate_function_body_sizes): Update.
22124 (compute_inline_parameters): Update.
22125 (estimate_node_size_and_time): Update.
22126 (inline_merge_summary): Update.
22127 (inline_update_overall_summary): Update.
22128 (inline_read_section): Update.
22129 (inline_write_summary): Update.
22130 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
22131 add account_size_time and reset member functions.
22132 (ipa_call_summary): Add reset function.
22133 * ipa-predicate.h (predicate::operator &): Constify.
22135 2017-05-22 Richard Biener <rguenther@suse.de>
22137 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
22139 2017-05-19 Jason Merrill <jason@redhat.com>
22141 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
22143 2017-05-19 Marek Polacek <polacek@redhat.com>
22146 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
22147 TYPE_OVERFLOW_WRAPS checks.
22149 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
22151 * tree-core.h (enum omp_clause_default_kind): Add
22152 "OMP_CLAUSE_DEFAULT_PRESENT".
22153 * tree-pretty-print.c (dump_omp_clause): Handle it.
22154 * gimplify.c (enum gimplify_omp_var_data): Add
22155 "GOVD_MAP_FORCE_PRESENT".
22156 (gimplify_adjust_omp_clauses_1): Map it to
22157 "GOMP_MAP_FORCE_PRESENT".
22158 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
22160 * gimplify.c (oacc_default_clause): Clarify.
22162 2017-05-19 Nathan Sidwell <nathan@acm.org>
22164 LANG_HOOK_REGISTER_DUMPS
22165 * toplev.c (general_init): Call register dump lang hook.
22166 * doc/invoke.texi: Document -fdump-lang option family.
22167 * dumpfile.c (dump_files): Remove class dump here.
22168 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
22169 * dumpfile.h (tree_dump_index): Remove TDI_class.
22170 * langhooks-def.h (lhd_register_dumps): Declare.
22171 (LANG_HOOKS_REGISTER_DUMPS): Define.
22172 (LANG_HOOKS_INITIALIZER): Add it.
22173 * langhooks.c (lhd_register_dumps): Define.
22174 * langhooks.h (struct lang_hooks): Add register_dumps.
22176 2017-05-19 Nathan Sidwell <nathan@acm.org>
22178 * context.h (context::set_passes): New.
22179 * context.c (context::context): Do not create pass manager.
22180 * toplev.c (general_init): Create pass manager here.
22182 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
22184 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
22185 use this splitter if two add or or instructions would also work for
22186 the constant we want to generate.
22188 2017-05-19 Richard Biener <rguenther@suse.de>
22191 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
22192 predicate evaluation.
22194 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
22196 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
22198 * ipa-inline.c (want_inline_small_function_p): Do not cast to
22201 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
22203 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22204 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22205 (inline_edge_summary_vec): Turn into ...
22206 (ipa_call_summaries): ... this one.
22207 (redirect_to_unreachable, edge_set_predicate,
22208 evaluate_properties_for_edge, inline_summary_alloc,
22209 reset_ipa_call_summary, reset_inline_summary,
22210 inline_summary_t::duplicate): Update.
22211 (inline_edge_duplication_hook): Turn to ...
22212 (ipa_call_summary_t::duplicate): ... this one.
22213 (inline_edge_removal_hook): Turn to ...
22214 (ipa_call_summary_t::remove): ... this one.
22215 (dump_inline_edge_summary): Turn to ...
22216 (dump_ipa_call_summary): ... this one.
22217 (estimate_function_body_sizes): Update.
22218 (inline_update_callee_summaries): Update.
22219 (remap_edge_change_prob): Update.
22220 (remap_edge_summaries): Update.
22221 (inline_merge_summary): Update.
22222 (do_estimate_edge_time): Update.
22223 (inline_generate_summary): Update.
22224 (inline_read_section): Update.
22225 (inline_read_summary): Update.
22226 (inline_free_summary): Update.
22227 * ipa-inline.c (can_inline_edge_p): Update.
22228 (compute_inlined_call_time): Update.
22229 (want_inline_small_function_p): Update.
22230 (edge_badness): Update.
22231 (early_inliner): Update.
22232 * ipa-inline.h (inline_edge_summary): Turn to ...
22233 (ipa_call_summary): ... this one.
22234 (ipa_call_summary_t): New class.
22235 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22236 (ipa_call_summaries): New.
22237 (inline_edge_summary): Remove.
22238 (estimate_edge_growth): Update.
22239 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22240 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22241 * ipa-split.c (execute_split_functions): Update.
22242 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22244 2017-05-19 Richard Biener <rguenther@suse.de>
22246 PR middle-end/80764
22247 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
22249 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
22251 * config/rs6000/rs6000.c (struct machine_function): Add field
22252 fpr_is_wrapped_separately.
22253 (rs6000_get_separate_components): Use 64 components. Handle the
22254 new FPR components.
22255 (rs6000_components_for_bb): Handle the FPR components.
22256 (rs6000_emit_prologue_components): Handle the FPR components.
22257 (rs6000_emit_epilogue_components): Handle the FPR components.
22258 (rs6000_set_handled_components): Handle the FPR components.
22259 (rs6000_emit_prologue): Don't output prologue code for those FPRs
22260 that are already separately shrink-wrapped.
22261 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
22262 that are already separately shrink-wrapped.
22264 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
22267 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
22270 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
22271 (define_peephole2 for Altivec d-form load): Add peepholes to catch
22272 cases where the register allocator uses a move and an offsettable
22273 memory operation to/from a FPR register on ISA 2.06/2.07.
22274 (define_peephole2 for Altivec d-form store): Likewise.
22276 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
22279 * config/i386/mmx.md (*mov<mode>_internal): Enable
22280 alternatives 11, 12, 13 and 14 also for 32bit targets.
22281 Remove alternatives 15, 16, 17 and 18.
22282 * config/i386/sse.md (vec_concatv2di): Change
22283 alternative (!x, *y) to (x, ?!*Yn).
22285 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
22287 * dumpfile.h (enum dump_kind): Remove stray comma.
22289 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
22291 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
22292 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
22293 predicate::num_conditions
22294 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
22295 (CHANGED): turn into predicate::changed.
22296 (agg_position_info): Move to ipa-predicate.h
22297 (add_condition, predicate::add_clause, predicate::operator &=,
22298 predicate::or_with, predicate::evaluate, predicate::probability,
22299 dump_condition, dump_clause, predicate::dump,
22300 predicate::remap_after_duplication, predicate::remap_after_inlining,
22301 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
22302 (evaluate_conditions_for_known_args): Update.
22303 (set_cond_stmt_execution_predicate): Update.
22304 * ipa-inline.h: Include ipa-predicate.h
22305 (condition, inline_param_summary, conditions, agg_position_info,
22306 predicate): Move to ipa-predicate.h
22307 * ipa-predicate.c: New file.
22308 * ipa-predicate.h: New file.
22310 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
22312 * final.c (leaf_function_p): Check we are not in a sequence.
22314 2017-05-18 Martin Liska <mliska@suse.cz>
22316 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
22317 * dumpfile.c (dump_register): Use new enum dump_kind.
22318 (get_dump_file_name): Likewise.
22319 (dump_enable_all): Likewise.
22320 (dump_switch_p_1): Likewise.
22321 (enable_rtl_dump_file): Remove usage of TDF_RTL.
22322 * dumpfile.h (enum dump_kind): New enum type.
22323 (struct dump_file_info): Create constructor and
22324 format fields and comments.
22325 * passes.c (pass_manager::register_one_dump_file):
22327 * statistics.c (statistics_early_init): Likewise.
22328 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
22329 TDF_TREE with TDF_SLIM.
22330 (gather_memory_references_ref): Likewise.
22332 2017-05-18 Martin Liska <mliska@suse.cz>
22334 * vec.h (struct vnull): Use it.
22336 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
22338 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
22339 (true_predicate, false_predicate, true_predicate_p,
22340 false_predicate_p): Remove.
22341 (single_cond_predicate, not_inlined_predicate): Turn to member function
22343 (add_condition): Update.
22344 (add_clause): Turn to...
22345 (predicate::add_clause): ... this one; update; allow passing NULL
22347 (and_predicates): Turn to ...
22348 (predicate::operator &=): ... this one.
22349 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
22350 (or_predicates): Turn to ...
22351 (predicate::or_with): ... this one.
22352 (evaluate_predicate): Turn to ...
22353 (predicate::evaluate): ... this one.
22354 (predicate_probability): Turn to ...
22355 (predicate::probability): ... this one.
22356 (dump_condition): Update.
22357 (dump_predicate): Turn to ...
22358 (predicate::dump): ... this one.
22359 (account_size_time): Update.
22360 (edge_set_predicate): Update.
22361 (set_hint_predicate): UPdate.
22362 (evaluate_conditions_for_known_args): Update.
22363 (evaluate_properties_for_edge): Update.
22364 (remap_predicate_after_duplication): Turn to...
22365 (predicate::remap_after_duplication): ... this one.
22366 (remap_hint_predicate_after_duplication): Update.
22367 (inline_summary_t::duplicate): UPdate.
22368 (dump_inline_edge_summary): Update.
22369 (dump_inline_summary): Update.
22370 (set_cond_stmt_execution_predicate): Update.
22371 (set_switch_stmt_execution_predicate): Update.
22372 (compute_bb_predicates): Update.
22373 (will_be_nonconstant_expr_predicate): Update.
22374 (will_be_nonconstant_predicate): Update.
22375 (phi_result_unknown_predicate): Update.
22376 (predicate_for_phi_result): Update.
22377 (array_index_predicate): Update.
22378 (estimate_function_body_sizes): Update.
22379 (estimate_node_size_and_time): Update.
22380 (estimate_ipcp_clone_size_and_time): Update.
22381 (remap_predicate): Rename to ...
22382 (predicate::remap_after_inlining): ... this one.
22383 (remap_hint_predicate): Update.
22384 (inline_merge_summary): Update.
22385 (inline_update_overall_summary): Update.
22386 (estimate_size_after_inlining): Update.
22387 (read_predicate): Rename to ...
22388 (predicate::stream_in): ... this one.
22389 (read_inline_edge_summary): Update.
22390 (write_predicate): Rename to ...
22391 (predicate::stream_out): ... this one.
22392 (write_inline_edge_summary): Update.
22393 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
22394 (clause_t): Turn to uint32_t
22395 (predicate): Turn to class; implement constructor and operators
22397 (size_time_entry): Update.
22398 (inline_summary): Update.
22399 (inline_edge_summary): Update.
22401 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
22403 * fold-const.c (fold_binary_loc): Move transformation...
22404 * match.pd (C - X CMP X): ... here.
22406 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
22408 * config/sparc/sparc.c (sparc_option_override): Set function
22409 alignment for -mcpu=niagara7 to 64 to match the I$ line.
22410 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
22412 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
22414 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
22416 2017-05-18 Marek Polacek <polacek@redhat.com>
22419 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
22420 (pass_ubsan::execute): Call gimple_assign_single_p instead of
22421 gimple_assign_load_p.
22423 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
22425 PR middle-end/80692
22426 * real.c (do_compare): Give decimal_do_compare preference over
22427 comparing just the signs.
22429 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
22431 * doc/md.texi (Canonicalization of Instructions): Describe the
22432 canonical form of instructions that inherently set a condition
22435 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
22437 PR middle-end/80775
22438 * tree-cfg.c: Move deletion of unreachable case statements to after
22439 the merging of consecutive case labels.
22441 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
22443 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
22444 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
22445 restoring of callee-saved registers.
22447 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
22449 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
22450 * config/visium/visium.c (single_set_and_flags): Likewise.
22451 * config/visium/visium.md (Substitutions): Likewise.
22453 2017-05-17 Martin Liska <mliska@suse.cz>
22455 * cfg.c: Introduce dump_flags_t type and
22456 use it instead of int type.
22458 * cfghooks.c: Likewise.
22459 * cfghooks.h (struct cfg_hooks): Likewise.
22460 * cfgrtl.c: Likewise.
22461 * cfgrtl.h: Likewise.
22462 * cgraph.c (cgraph_node::get_body): Likewise.
22463 * coretypes.h: Likewise.
22464 * domwalk.c: Likewise.
22465 * domwalk.h: Likewise.
22466 * dumpfile.c (struct dump_option_value_info): Likewise.
22467 (dump_enable_all): Likewise.
22468 (dump_switch_p_1): Likewise.
22469 (opt_info_switch_p): Likewise.
22470 * dumpfile.h (enum tree_dump_index): Likewise.
22471 (struct dump_file_info): Likewise.
22472 * genemit.c: Likewise.
22473 * generic-match-head.c: Likewise.
22474 * gengtype.c (open_base_files): Likewise.
22475 * gimple-pretty-print.c: Likewise.
22476 * gimple-pretty-print.h: Likewise.
22477 * graph.c (print_graph_cfg): Likewise.
22478 * graphite-scop-detection.c (dot_all_sese): Likewise.
22479 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
22480 * loop-unroll.c (report_unroll): Likewise.
22481 * passes.c (pass_manager::register_one_dump_file): Likewise.
22482 * print-tree.c: Likewise.
22483 * statistics.c: Likewise.
22484 * tree-cfg.c: Likewise.
22485 * tree-cfg.h: Likewise.
22486 * tree-dfa.c: Likewise.
22487 * tree-dfa.h: Likewise.
22488 * tree-dump.c (dump_function): Likewise.
22489 * tree-dump.h (struct dump_info): Likewise.
22490 * tree-pretty-print.c: Likewise.
22491 * tree-pretty-print.h: Likewise.
22492 * tree-ssa-live.c: Likewise.
22493 * tree-ssa-live.h: Likewise.
22494 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
22495 * tree-vect-loop.c: Likewise.
22496 * tree-vect-slp.c: Likewise.
22498 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
22499 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22501 PR tree-optimization/80457
22502 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
22503 of all arguments to a statement as scalar_to_vec operations.
22504 (vectorizable_call): Adjust call to vect_model_simple_cost for
22506 (vectorizable_conversion): Likewise.
22507 (vectorizable_assignment): Likewise.
22508 (vectorizable_shift): Likewise.
22509 (vectorizable_operation): Likewise.
22510 (vectorizable_comparison): Likewise.
22511 (vect_is_simple_cond): Record the def types for operands.
22512 (vectorizable_condition): Likewise, call vect_model_simple_cost.
22513 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
22514 for statement argument count.
22516 2017-05-16 Carl Love <cel@us.ibm.com>
22518 * config/rs6000/rs6000-c: Add support for built-in functions
22519 vector unsigned long long vec_bperm (vector unsigned long long,
22520 vector unsigned char)
22521 vector signed long long vec_mule (vector signed int,
22523 vector unsigned long long vec_mule (vector unsigned int,
22524 vector unsigned int)
22525 vector signed long long vec_mulo (vector signed int,
22527 vector unsigned long long vec_mulo (vector unsigned int,
22528 vector unsigned int)
22529 vector signed char vec_sldw (vector signed char,
22530 vector signed char,
22532 vector unsigned char vec_sldw (vector unsigned char,
22533 vector unsigned char,
22535 vector signed short vec_sldw (vector signed short,
22536 vector signed short,
22538 vector unsigned short vec_sldw (vector unsigned short,
22539 vector unsigned short,
22541 vector signed int vec_sldw (vector signed int,
22544 vector unsigned int vec_sldw (vector unsigned int,
22545 vector unsigned int,
22547 vector signed long long vec_sldw (vector signed long long,
22548 vector signed long long,
22550 vector unsigned long long vec_sldw (vector unsigned long long,
22551 vector unsigned long long,
22553 * config/rs6000/rs6000-c: Add support for built-in functions
22554 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
22555 * config/rs6000/altivec.h: Add defintion for vec_sldw.
22556 * doc/extend.texi: Update the built-in documentation for the
22557 new built-in functions.
22559 2017-05-16 Marek Polacek <polacek@redhat.com>
22563 * tree.c (save_expr): Don't fold the expression.
22565 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
22567 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
22568 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
22569 and (?*y,m). Update insn attributes.
22571 2017-05-16 Martin Liska <mliska@suse.cz>
22573 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
22574 flags argument of print_gimple_stmt, print_gimple_expr,
22575 print_generic_stmt and print_generic_expr.
22576 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
22577 * coretypes.h: Likewise.
22578 * except.c (dump_eh_tree): Likewise.
22579 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
22580 * gimple-pretty-print.h: Likewise.
22581 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
22582 (backprop::push_to_worklist): Likewise.
22583 (backprop::pop_from_worklist): Likewise.
22584 (backprop::process_use): Likewise.
22585 (backprop::intersect_uses): Likewise.
22586 (note_replacement): Likewise.
22587 * gimple-ssa-store-merging.c
22588 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
22589 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
22590 (pass_store_merging::execute): Likewise.
22591 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
22592 (ssa_base_cand_dump_callback): Likewise.
22593 (dump_incr_vec): Likewise.
22594 (replace_refs): Likewise.
22595 (replace_mult_candidate): Likewise.
22596 (create_add_on_incoming_edge): Likewise.
22597 (create_phi_basis): Likewise.
22598 (insert_initializers): Likewise.
22599 (all_phi_incrs_profitable): Likewise.
22600 (introduce_cast_before_cand): Likewise.
22601 (replace_one_candidate): Likewise.
22602 * gimplify.c (gimplify_expr): Likewise.
22603 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
22604 (set_rename): Likewise.
22605 (rename_uses): Likewise.
22606 (copy_loop_phi_nodes): Likewise.
22607 (add_close_phis_to_merge_points): Likewise.
22608 (copy_loop_close_phi_args): Likewise.
22609 (copy_cond_phi_args): Likewise.
22610 (graphite_copy_stmts_from_block): Likewise.
22611 (translate_pending_phi_nodes): Likewise.
22612 * graphite-poly.c (print_pdr): Likewise.
22613 (dump_gbb_cases): Likewise.
22614 (dump_gbb_conditions): Likewise.
22615 (print_scop_params): Likewise.
22616 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
22617 (build_cross_bb_scalars_use): Likewise.
22618 (gather_bbs::before_dom_children): Likewise.
22619 * hsa-dump.c (dump_hsa_immed): Likewise.
22620 * ipa-cp.c (print_ipcp_constant_value): Likewise.
22621 (get_replacement_map): Likewise.
22622 * ipa-inline-analysis.c (dump_condition): Likewise.
22623 (estimate_function_body_sizes): Likewise.
22624 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
22625 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
22626 * ipa-prop.c (ipa_dump_param): Likewise.
22627 (ipa_print_node_jump_functions_for_edge): Likewise.
22628 (ipa_modify_call_arguments): Likewise.
22629 (ipa_modify_expr): Likewise.
22630 (ipa_dump_param_adjustments): Likewise.
22631 (ipa_dump_agg_replacement_values): Likewise.
22632 (ipcp_modif_dom_walker::before_dom_children): Likewise.
22633 * ipa-pure-const.c (check_stmt): Likewise.
22634 (pass_nothrow::execute): Likewise.
22635 * ipa-split.c (execute_split_functions): Likewise.
22636 * omp-offload.c (dump_oacc_loop_part): Likewise.
22637 (dump_oacc_loop): Likewise.
22638 * trans-mem.c (tm_log_emit): Likewise.
22639 (tm_memopt_accumulate_memops): Likewise.
22640 (dump_tm_memopt_set): Likewise.
22641 (dump_tm_memopt_transform): Likewise.
22642 * tree-cfg.c (gimple_verify_flow_info): Likewise.
22643 (print_loop): Likewise.
22644 * tree-chkp-opt.c (chkp_print_addr): Likewise.
22645 (chkp_gather_checks_info): Likewise.
22646 (chkp_get_check_result): Likewise.
22647 (chkp_remove_check_if_pass): Likewise.
22648 (chkp_use_outer_bounds_if_possible): Likewise.
22649 (chkp_reduce_bounds_lifetime): Likewise.
22650 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
22651 (chkp_mark_completed_bounds): Likewise.
22652 (chkp_register_incomplete_bounds): Likewise.
22653 (chkp_mark_invalid_bounds): Likewise.
22654 (chkp_maybe_copy_and_register_bounds): Likewise.
22655 (chkp_build_returned_bound): Likewise.
22656 (chkp_get_bound_for_parm): Likewise.
22657 (chkp_build_bndldx): Likewise.
22658 (chkp_get_bounds_by_definition): Likewise.
22659 (chkp_generate_extern_var_bounds): Likewise.
22660 (chkp_get_bounds_for_decl_addr): Likewise.
22661 * tree-chrec.c (chrec_apply): Likewise.
22662 * tree-data-ref.c (dump_data_reference): Likewise.
22663 (dump_subscript): Likewise.
22664 (dump_data_dependence_relation): Likewise.
22665 (analyze_overlapping_iterations): Likewise.
22666 * tree-inline.c (expand_call_inline): Likewise.
22667 (tree_function_versioning): Likewise.
22668 * tree-into-ssa.c (dump_defs_stack): Likewise.
22669 (dump_currdefs): Likewise.
22670 (dump_names_replaced_by): Likewise.
22671 (dump_update_ssa): Likewise.
22672 (update_ssa): Likewise.
22673 * tree-object-size.c (pass_object_sizes::execute): Likewise.
22674 * tree-parloops.c (build_new_reduction): Likewise.
22675 (try_create_reduction_list): Likewise.
22676 (ref_conflicts_with_region): Likewise.
22677 (oacc_entry_exit_ok_1): Likewise.
22678 (oacc_entry_exit_single_gang): Likewise.
22679 * tree-pretty-print.h: Likewise.
22680 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
22681 (get_scalar_evolution): Likewise.
22682 (add_to_evolution): Likewise.
22683 (get_loop_exit_condition): Likewise.
22684 (analyze_evolution_in_loop): Likewise.
22685 (analyze_initial_condition): Likewise.
22686 (analyze_scalar_evolution): Likewise.
22687 (instantiate_scev): Likewise.
22688 (number_of_latch_executions): Likewise.
22689 (gather_chrec_stats): Likewise.
22690 (final_value_replacement_loop): Likewise.
22691 (scev_const_prop): Likewise.
22692 * tree-sra.c (dump_access): Likewise.
22693 (disqualify_candidate): Likewise.
22694 (create_access): Likewise.
22695 (reject): Likewise.
22696 (maybe_add_sra_candidate): Likewise.
22697 (create_access_replacement): Likewise.
22698 (analyze_access_subtree): Likewise.
22699 (analyze_all_variable_accesses): Likewise.
22700 (sra_modify_assign): Likewise.
22701 (initialize_constant_pool_replacements): Likewise.
22702 (find_param_candidates): Likewise.
22703 (decide_one_param_reduction): Likewise.
22704 (replace_removed_params_ssa_names): Likewise.
22705 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
22706 * tree-ssa-copy.c (dump_copy_of): Likewise.
22707 (copy_prop_visit_cond_stmt): Likewise.
22708 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
22709 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
22710 (record_equivalences_from_stmt): Likewise.
22711 * tree-ssa-dse.c (compute_trims): Likewise.
22712 (delete_dead_call): Likewise.
22713 (delete_dead_assignment): Likewise.
22714 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
22715 (forward_propagate_into_cond): Likewise.
22716 (pass_forwprop::execute): Likewise.
22717 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
22718 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
22720 (move_computations_worker): Likewise.
22721 (execute_sm): Likewise.
22722 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
22723 (remove_exits_and_undefined_stmts): Likewise.
22724 (remove_redundant_iv_tests): Likewise.
22725 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
22726 (adjust_iv_update_pos): Likewise.
22727 * tree-ssa-math-opts.c (bswap_replace): Likewise.
22728 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
22729 (value_replacement): Likewise.
22730 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
22731 * tree-ssa-pre.c (print_pre_expr): Likewise.
22732 (get_representative_for): Likewise.
22733 (create_expression_by_pieces): Likewise.
22734 (insert_into_preds_of_block): Likewise.
22735 (eliminate_insert): Likewise.
22736 (eliminate_dom_walker::before_dom_children): Likewise.
22737 (eliminate): Likewise.
22738 (remove_dead_inserted_code): Likewise.
22739 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
22740 * tree-ssa-reassoc.c (get_rank): Likewise.
22741 (eliminate_duplicate_pair): Likewise.
22742 (eliminate_plus_minus_pair): Likewise.
22743 (eliminate_not_pairs): Likewise.
22744 (undistribute_ops_list): Likewise.
22745 (eliminate_redundant_comparison): Likewise.
22746 (update_range_test): Likewise.
22747 (optimize_range_tests_var_bound): Likewise.
22748 (optimize_vec_cond_expr): Likewise.
22749 (rewrite_expr_tree): Likewise.
22750 (rewrite_expr_tree_parallel): Likewise.
22751 (linearize_expr): Likewise.
22752 (break_up_subtract): Likewise.
22753 (linearize_expr_tree): Likewise.
22754 (attempt_builtin_powi): Likewise.
22755 (attempt_builtin_copysign): Likewise.
22756 (transform_stmt_to_copy): Likewise.
22757 (transform_stmt_to_multiply): Likewise.
22758 (dump_ops_vector): Likewise.
22759 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
22760 (print_scc): Likewise.
22761 (set_ssa_val_to): Likewise.
22762 (visit_reference_op_store): Likewise.
22763 (visit_use): Likewise.
22764 (sccvn_dom_walker::before_dom_children): Likewise.
22765 (run_scc_vn): Likewise.
22766 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
22768 (expr_hash_elt::print): Likewise.
22769 (const_and_copies::pop_to_marker): Likewise.
22770 (const_and_copies::record_const_or_copy_raw): Likewise.
22771 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
22772 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
22773 (dump_predicates): Likewise.
22774 (find_uninit_use): Likewise.
22775 (warn_uninitialized_phi): Likewise.
22776 (pass_late_warn_uninitialized::execute): Likewise.
22777 * tree-ssa.c (verify_vssa): Likewise.
22778 (verify_ssa): Likewise.
22779 (maybe_optimize_var): Likewise.
22780 * tree-vrp.c (dump_value_range): Likewise.
22781 (dump_all_value_ranges): Likewise.
22782 (dump_asserts_for): Likewise.
22783 (register_edge_assert_for_2): Likewise.
22784 (vrp_visit_cond_stmt): Likewise.
22785 (vrp_visit_switch_stmt): Likewise.
22786 (vrp_visit_stmt): Likewise.
22787 (vrp_visit_phi_node): Likewise.
22788 (simplify_cond_using_ranges_1): Likewise.
22789 (fold_predicate_in): Likewise.
22790 (evrp_dom_walker::before_dom_children): Likewise.
22791 (evrp_dom_walker::push_value_range): Likewise.
22792 (evrp_dom_walker::pop_value_range): Likewise.
22793 (execute_early_vrp): Likewise.
22795 2017-05-16 Richard Biener <rguenther@suse.de>
22797 * dwarf2out.c (loc_list_from_tree_1): Do not create
22798 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
22800 2017-05-16 Richard Biener <rguenther@suse.de>
22802 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
22804 (note_variable_value_in_expr): If we resolved the decl ref
22805 do not push to the stack.
22807 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
22809 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
22810 operations in fast-math mode.
22811 (vaddq_f16): Likewise.
22812 (vmul_f16): Likewise.
22813 (vmulq_f16): Likewise.
22814 (vsub_f16): Likewise.
22815 (vsubq_f16): Likewise.
22816 * config/arm/neon.md (add<mode>3): New.
22818 (fma:<VH:mode>3): New. Also remove outdated comment.
22821 2017-05-16 Martin Liska <mliska@suse.cz>
22825 * ipa-devirt.c (warn_types_mismatch): Fix typo.
22826 (odr_types_equivalent_p): Likewise.
22828 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
22830 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
22832 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
22835 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
22836 non-interunit SSE move alternatives with '?'.
22837 (zero-extendsidi peephole2): New peephole to skip intermediate
22838 general register in SSE zero-extend sequence.
22840 2017-05-15 Jeff Law <law@redhat.com>
22842 * reorg.c (relax_delay_slots): Create a new variable to hold
22843 the temporary target rather than clobbering TARGET_LABEL.
22845 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
22846 missing argument to extract_bit_field call.
22847 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
22849 2017-05-15 Martin Liska <mliska@suse.cz>
22852 * gcc.c (process_command): Do not allow empty argument of -o option.
22854 2017-05-15 Renlin Li <renlin.li@arm.com>
22856 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
22857 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
22858 * config/aarch64/constraints.md (Usf): Add long call check.
22859 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
22860 (call_value): Likewise.
22861 (sibcall): Likewise.
22862 (sibcall_value): Likewise.
22864 (call_value_insn): New.
22865 (sibcall_insn): Update rtx pattern.
22866 (sibcall_value_insn): Likewise.
22867 (call_internal): Remove.
22868 (call_value_internal): Likewise.
22869 (sibcall_internal): Likewise.
22870 (sibcall_value_internal): Likewise.
22871 (call_reg): Likewise.
22872 (call_symbol): Likewise.
22873 (call_value_reg): Likewise.
22874 (call_value_symbol): Likewise.
22876 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
22879 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
22881 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
22883 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
22884 compatible with CCGOCmode and with CCZmode.
22886 2017-05-14 Martin Sebor <msebor@redhat.com>
22888 PR middle-end/77671
22889 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
22890 (gimple_fold_builtin_snprintf): Same.
22891 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
22892 (gimple_fold_builtin_snprintf): Same.
22893 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
22894 of character types.
22895 (is_call_safe): New function.
22896 (try_substitute_return_value): Call it.
22897 (try_simplify_call): New function.
22898 (pass_sprintf_length::handle_gimple_call): Call it.
22900 2017-05-14 Martin Sebor <msebor@redhat.com>
22902 PR middle-end/80669
22903 * builtins.c (expand_builtin_stpncpy): Simplify.
22905 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
22907 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
22908 * config/i386/i386.h
22909 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
22910 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
22911 (struct machine_function): Add new members call_ms2sysv,
22912 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
22913 (struct machine_frame_state): New fields sp_realigned and
22914 sp_realigned_offset.
22915 * config/i386/i386.c
22916 (enum xlogue_stub): New enum.
22917 (enum xlogue_stub_sets): New enum.
22918 (class xlogue_layout): New class.
22919 (struct ix86_frame): New fields stack_realign_allocate_offset,
22920 stack_realign_offset and outlined_save_offset. Modify comments to
22921 detail stack layout when using out-of-line stubs.
22922 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
22923 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
22924 -mcall-ms2sysv-xlogues.
22925 (stub_managed_regs): New static variable.
22926 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
22927 registers managed by out-of-line stub.
22928 (disable_call_ms2sysv_xlogues): New function.
22929 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
22930 m->call_ms2sysv when appropriate and compute frame layout for
22932 (sp_valid_at, fp_valid_at): New inline functions.
22933 (choose_basereg): New function.
22934 (choose_baseaddr): Add align parameter, use choose_basereg and modify
22936 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
22937 Use align parameter of choose_baseaddr to generated aligned SSE movs
22939 (pro_epilogue_adjust_stack): Modify to track
22940 machine_frame_state::sp_realigned.
22941 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
22942 (ix86_nsaved_sseregs): Likewise.
22943 (ix86_emit_save_regs): Likewise.
22944 (ix86_emit_save_regs_using_mov): Likewise.
22945 (ix86_emit_save_sse_regs_using_mov): Likewise.
22946 (get_scratch_register_on_entry): Likewise.
22947 (gen_frame_set): New function.
22948 (gen_frame_load): Likewise.
22949 (gen_frame_store): Likewise.
22950 (emit_outlined_ms2sysv_save): Likewise.
22951 (emit_outlined_ms2sysv_restore): Likewise.
22952 (ix86_expand_prologue): Modify stack re-alignment code and call
22953 emit_outlined_ms2sysv_save when appropriate.
22954 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
22955 parameter rtx_insn *insn, which allows the function to be used to only
22956 generate the notes.
22957 (ix86_expand_epilogue): Modify validity checks of frame and stack
22958 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
22959 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
22960 * config/i386/predicates.md
22961 (save_multiple): New predicate.
22962 (restore_multiple): Likewise.
22963 * config/i386/sse.md
22964 (save_multiple<mode>): New pattern.
22965 (save_multiple_realign<mode>): Likewise.
22966 (restore_multiple<mode>): Likewise.
22967 (restore_multiple_and_return<mode>): Likewise.
22968 (restore_multiple_leave_return<mode>): Likewise.
22969 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
22971 2017-05-14 Julia Koval <julia.koval@intel.com>
22973 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
22974 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
22975 (__builtin_ia32_xsetbv): New builtins.
22976 * config/i386/i386.c (ix86_expand_special_args_builtin):
22978 (ix86_expand_builtin): Special expand for new intrinsics.
22979 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
22980 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
22981 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
22983 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22985 * cfganal.c (inverted_post_order_compute): Change argument type
22987 * cfganal.h (inverted_post_order_compute): Adjust prototype.
22988 * df-core.c (rest_of_handle_df_initialize): Adjust.
22989 (rest_of_handle_df_finish): Likewise.
22990 (df_analyze_1): Likewise.
22991 (df_analyze): Likewise.
22992 (loop_inverted_post_order_compute): Change argument to be a vec *.
22993 (df_analyze_loop): Adjust.
22994 (df_get_n_blocks): Likewise.
22995 (df_get_postorder): Likewise.
22996 * df.h (struct df_d): Change field to be a vec.
22997 * lcm.c (compute_laterin): Adjust.
22998 (compute_available): Likewise.
22999 * lra-lives.c (lra_create_live_ranges_1): Likewise.
23000 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
23001 * tree-ssa-pre.c (compute_antic): Likewise.
23003 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23005 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
23006 (depth_first_search::depth_first_search): Change structure init
23007 function to this constructor.
23008 (depth_first_search::add_bb): Rename function to this member.
23009 (depth_first_search::execute): Likewise.
23010 (flow_dfs_compute_reverse_finish): Adjust.
23012 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23014 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
23015 (longest_simple_path): Likewise.
23016 * shrink-wrap.c (spread_components): Likewise.
23017 (disqualify_problematic_components): Likewise.
23018 (emit_common_heads_for_components): Likewise.
23019 (emit_common_tails_for_components): Likewise.
23020 (insert_prologue_epilogue_for_components): Likewise.
23022 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23024 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
23027 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23029 * df-core.c (df_set_blocks): Start using auto_bitmap.
23030 (df_compact_blocks): Likewise.
23031 * df-problems.c (df_rd_confluence_n): Likewise.
23032 * df-scan.c (df_insn_rescan_all): Likewise.
23033 (df_process_deferred_rescans): Likewise.
23034 (df_update_entry_block_defs): Likewise.
23035 (df_update_exit_block_uses): Likewise.
23036 (df_entry_block_bitmap_verify): Likewise.
23037 (df_exit_block_bitmap_verify): Likewise.
23038 (df_scan_verify): Likewise.
23039 * lra-constraints.c (lra_constraints): Likewise.
23040 (undo_optional_reloads): Likewise.
23041 (lra_undo_inheritance): Likewise.
23042 * lra-remat.c (calculate_gen_cands): Likewise.
23043 (do_remat): Likewise.
23044 * lra-spills.c (assign_spill_hard_regs): Likewise.
23045 (spill_pseudos): Likewise.
23046 * tree-ssa-pre.c (bitmap_set_and): Likewise.
23047 (bitmap_set_subtract_values): Likewise.
23049 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23051 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
23052 management with auto_bitmap.
23053 (fix_inter_tick): Likewise.
23054 (fix_recovery_deps): Likewise.
23055 * ira.c (add_store_equivs): Likewise.
23056 (find_moveable_pseudos): Likewise.
23057 (split_live_ranges_for_shrink_wrap): Likewise.
23058 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
23059 (rtx_reuse_manager::seen_def_p): Likewise.
23060 (rtx_reuse_manager::set_seen_def): Likewise.
23061 * print-rtl.h (class rtx_reuse_manager): Likewise.
23063 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23065 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
23067 (migrate_btr_def): Likewise.
23068 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
23069 * df-core.c (loop_post_order_compute): Likewise.
23070 (loop_inverted_post_order_compute): Likewise.
23071 * hsa-common.h: Likewise.
23072 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
23073 * init-regs.c (initialize_uninitialized_regs): Likewise.
23074 * ipa-inline.c (resolve_noninline_speculation): Likewise.
23075 (inline_small_functions): Likewise.
23076 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
23077 * ira.c (combine_and_move_insns): Likewise.
23078 (build_insn_chain): Likewise.
23079 * loop-invariant.c (find_invariants): Likewise.
23080 * lower-subreg.c (propagate_pseudo_copies): Likewise.
23081 * predict.c (tree_predict_by_opcode): Likewise.
23082 (predict_paths_leading_to): Likewise.
23083 (predict_paths_leading_to_edge): Likewise.
23084 (estimate_loops_at_level): Likewise.
23085 (estimate_loops): Likewise.
23086 * shrink-wrap.c (try_shrink_wrapping): Likewise.
23087 (spread_components): Likewise.
23088 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
23089 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
23090 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
23091 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
23092 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
23093 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
23094 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
23095 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
23096 (mark_threaded_blocks): Likewise.
23097 (thread_through_all_blocks): Likewise.
23098 * tree-ssa.c (verify_ssa): Likewise.
23099 (execute_update_addresses_taken): Likewise.
23100 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
23102 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23104 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
23106 (post_order_compute): Likewise.
23107 (inverted_post_order_compute): Likewise.
23108 (pre_and_rev_post_order_compute_fn): Likewise.
23110 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23112 * genrecog.c (int_set::int_set): Explicitly construct our
23113 auto_vec base class.
23114 * vec.h (auto_vec::auto_vec): New constructor.
23116 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23118 * bitmap.h (class auto_bitmap): New constructor taking
23119 bitmap_obstack * argument.
23121 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23123 * bitmap.h (class auto_bitmap): Change type of m_bits to
23124 bitmap_head, and adjust ctor / dtor and member operators.
23126 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
23128 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
23129 when returned register mode doesn't match original mode.
23131 2017-05-12 Jeff Law <law@redhat.com>
23132 Jakub Jelinek <jakub@redhat.com>
23134 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
23135 we look for cc setter after the compare-elim changes.
23136 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
23137 within the vector to match what compare-elim now expects.
23138 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
23139 (xorsi3_flags, one_cmplsi2_flags): Likewise.
23141 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
23142 after the compare-elim changes.
23143 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
23144 the vector to match what compare-elim now expects.
23145 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
23146 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
23147 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
23148 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
23149 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
23151 * config/visium/visium.c (single_set_and_flags): Fix where
23152 we look for cc setter after the compare-elim changes.
23153 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
23154 with the vector to match what compare-elim now expects.
23155 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
23156 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
23157 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
23158 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
23159 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
23160 (neg<mode>2_insn_set_overflow): Likewise.
23162 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
23164 PR middle-end/79794
23165 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
23166 maybe_expand_insn call, set ops[0].target. If still set after call,
23167 set alt_rtl. Add extra arg to recursive calls.
23168 (extract_bit_field): Add alt_rtl argument. Pass to
23170 * expmed.h (extract_bit_field): Fix prototype.
23171 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
23172 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
23173 to extract_bit_field_calls.
23174 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
23175 Pass alt_rtl to extract_bit_field calls.
23176 * calls.c (store_unaligned_arguments_into_psuedos)
23177 load_register_parameters): Pass extra NULL to extract_bit_field calls.
23178 * optabs.c (maybe_legitimize_operand): Clear op->target when call
23180 * optabs.h (struct expand_operand): Add target bitfield.
23182 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
23184 * compare-elim.c (try_eliminate_compare): Canonicalize
23185 operation with embedded compare to
23186 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
23187 (set (reg) (operation)].
23189 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
23191 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
23194 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
23195 cost of adding a carry flag for ADC instruction.
23196 [case MINUS]: Ignore the cost of subtracting a carry flag
23197 for SBB instruction.
23199 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
23201 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
23203 * config/rs6000/bmiintrin.h: New file.
23204 * config/rs6000/bmi2intrin.h: New file.
23205 * config/rs6000/x86intrin.h: New file.
23207 2017-05-12 Jeff Law <law@redhat.com>
23209 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
23212 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
23214 PR middle-end/80707
23215 * tree-cfg.c: Remove cfg edges of unreachable case statements.
23217 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23219 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
23220 early expansion of vector divide builtins.
23221 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
23222 builtins identified as having unsigned arguments.
23224 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23226 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
23227 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
23228 expansion of vector logical operations (and, andc, or, xor,
23231 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23233 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
23234 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
23236 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23238 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
23239 early GIMPLE expansion of vector multiplies.
23241 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23243 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
23244 TARGET_HAVE_MOVT conditional.
23245 (movt splitter): Likewise.
23247 2017-05-12 Richard Biener <rguenther@suse.de>
23249 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
23250 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
23251 Fold all stmts not inplace.
23253 2017-05-12 Richard Biener <rguenther@suse.de>
23255 PR tree-optimization/80713
23256 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
23257 inserted_exprs bit for not removed stmts.
23259 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
23261 PR middle-end/69921
23262 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
23263 parallelized" attribute for parallelized OpenACC kernels.
23264 * omp-offload.c (execute_oacc_device_lower): Use it.
23266 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
23267 Set "oacc kernels" attribute.
23268 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
23269 parameter. Adjust all users.
23270 (oacc_fn_attrib_kernels_p): Remove function.
23271 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
23272 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
23273 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
23274 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
23275 assert "oacc kernels" attribute is set.
23277 2017-05-11 Carl Love <cel@us.ibm.com>
23279 * config/rs6000/rs6000-c: Add support for built-in functions
23280 vector unsigned char vec_popcnt (vector signed char)
23281 vector unsigned char vec_popcnt (vector unsigned char)
23282 vector unsigned short vec_popcnt (vector signed short)
23283 vector unsigned short vec_popcnt (vector unsigned short)
23284 vector unsigned int vec_popcnt (vector signed int)
23285 vector unsigned int vec_popcnt (vector unsigned int)
23286 vector unsigned long long vec_popcnt (vector signed long long)
23287 vector unsigned long long vec_popcnt (vector unsigned long long)
23288 vector signed long long vec_slo (vector signed long long,
23289 vector signed char)
23290 vector signed long long vec_slo (vector signed long long,
23291 vector unsigned char)
23292 vector unsigned long long vec_slo (vector unsigned long long,
23293 vector signed char)
23294 vector unsigned long long vec_slo (vector unsigned long long,
23295 vector unsigned char)
23296 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
23297 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
23298 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
23299 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
23300 * doc/extend.texi: Update the built-in documentation file for the
23301 new built-in functions.
23303 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
23305 * attribs.h (sorted_attr_string): Move machine independent
23306 functions for target clone support from the i386 port to common
23307 code. Rename ix86_function_versions to common_function_versions.
23308 Rename make_name to make_unique_name.
23309 (common_function_versions): Likewise.
23310 (make_unique_name): Likewise.
23311 (make_dispatcher_decl): Likewise.
23312 (is_function_default_version): Likewise.
23313 * attribs.c (attr_strcmp): Likewise.
23314 (sorted_attr_string): Likewise.
23315 (common_function_versions): Likewise.
23316 (make_unique_name): Likewise.
23317 (make_dispatcher_decl): Likewise.
23318 (is_function_default_version): Likewise.
23319 * config/i386/i386.c (attr_strcmp): Likewise.
23320 (sorted_attr_string): Likewise.
23321 (ix86_function_versions): Likewise.
23322 (make_name): Likewise.
23323 (make_dispatcher_decl): Likewise.
23324 (is_function_default_version): Likewise.
23325 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
23327 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23330 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23331 Account for direct move costs for vec_construct of integer
23334 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
23337 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
23338 (UNSPEC_STX_ATOMIC): Ditto.
23339 (loaddi_via_sse): New insn.
23340 (storedi_via_sse): Ditto.
23341 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
23342 Update corresponding peephole2 patterns.
23343 (atomic_storedi_fpu): Ditto.
23345 2017-05-11 Julia Koval <julia.koval@intel.com>
23347 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
23348 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
23350 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
23351 (__builtin_ia32_rsqrt14ss_mask): New builtins.
23352 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
23354 2017-05-11 Nathan Sidwell <nathan@acm.org>
23356 * graphite-poly.c: Include dumpfile.h.
23358 * dumpfle.h (dump_function): Declare here ...
23359 * tree-dump.h (dump_function): ... not here.
23360 * dumpfile.c: #include tree-cfg.h.
23361 (dump_function): Move here from ...
23362 * tree-dump.c (dump_function): ... here.
23363 * gimplify.c: #include splay-tree.h, not tree-dump.h.
23364 * graphite-poly.c: Don't include tree-dump.h.
23365 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
23366 * print-tree.c: Likewise.
23367 * stor-layout.c: Likewise.
23368 * tree-nested.c: Likewise.
23370 * dumpfile.c (dump_start): Use TDF_FLAGS.
23371 (dump_enable_all): Fix TDF_KIND check thinko.
23373 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
23375 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23376 array entries to represent two legal parameterizations of the
23377 overloaded __builtin_cmpb function, as represented by the
23378 P6_OV_BUILTIN_CMPB constant.
23379 (altivec_resolve_overloaded_builtin): Add special case handling
23380 for the __builtin_cmpb function, as represented by the
23381 P6_OV_BUILTIN_CMPB constant.
23382 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
23383 (BU_P6_64BIT_2): New macro.
23384 (BU_P6_OVERLOAD_2): New macro
23385 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
23386 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
23387 (CMPB): Add overload support to represent both 32-bit and 64-bit
23388 compare-bytes function.
23389 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
23390 support for TARGET_CMPB.
23391 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
23392 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23393 documentation of the __builtin_cmpb overloaded built-in function.
23395 2017-05-11 Richard Biener <rguenther@suse.de>
23397 PR tree-optimization/80705
23398 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
23399 bases are not vectorizable.
23401 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23403 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
23404 when counting register pressure.
23406 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23408 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
23409 (iv_ca_more_deps): Renamed to ...
23410 (iv_ca_compare_deps): ... this.
23411 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
23413 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23415 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
23417 (determine_group_iv_costs): ... here.
23418 (find_inv_vars_cb): Record inv var if it's not recorded before.
23420 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23422 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
23423 (get_shiftadd_cost): Ditto.
23425 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23427 * tree-ssa-address.c: Include header file.
23428 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
23430 (add_to_parts): Refactor.
23431 (addr_to_parts): New parameter. Update use of move_hint_to_base.
23432 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
23435 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23437 PR tree-optimization/53090
23438 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
23440 (extract_cond_operands): Detect condition with IV on both sides
23441 and return COMP_IV_EXPR_2.
23442 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
23443 (rewrite_use_compare): Simplify by removing call to function
23444 extract_cond_operands.
23446 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23448 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
23449 (extract_cond_operands): Detect condition comparing against non-
23450 invariant bound and return appropriate enum value.
23451 (find_interesting_uses_cond): Update use of extract_cond_operands.
23452 Handle its return value accordingly.
23453 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
23455 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23457 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
23458 nonlinear iv_use computation in loop invariant sensitive way.
23460 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23462 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
23463 (find_iv_candidates): Call relate_compare_use_with_all_cands.
23465 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23467 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
23468 (dump_cand): Support iv_cand.inv_exprs.
23469 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
23471 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
23474 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23476 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
23478 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
23479 as local function. Include necessary header files.
23480 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
23482 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23484 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
23486 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23488 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
23489 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
23490 RSHIFT_EXPR and BIT_NOT_EXPR.
23492 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23494 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
23495 (adjust_setup_cost): New parameter supporting round up adjustment.
23496 (struct address_cost_data): Delete.
23497 (force_expr_to_var_cost): Don't bound cost with spill_cost.
23498 (split_address_cost, ptr_difference_cost): Delete.
23499 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
23500 (struct ainc_cost_data): New struct.
23501 (get_address_cost_ainc): New function.
23502 (get_address_cost, get_computation_cost): Reimplement.
23503 (determine_group_iv_cost_address): Record inv_expr for all uses of
23505 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
23506 (iv_ca_has_deps): Reimplemented to ...
23507 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
23509 (iv_ca_extend): Call iv_ca_more_deps.
23511 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23513 * tree-ssa-address.c (struct mem_address): Move to header file.
23514 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
23515 * tree-ssa-address.h (struct mem_address): Move from C file.
23516 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
23518 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23520 * tree-affine.h (aff_combination_type): New interface.
23521 (aff_combination_zero_p): Remove static.
23522 (aff_combination_const_p): New interface.
23523 (aff_combination_singleton_var_p): New interfaces.
23525 2017-05-11 Richard Biener <rguenther@suse.de>
23527 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
23528 Skip unreachable blocks and destinations.
23529 (eliminate): Move stmt removal and fixup ...
23530 (fini_eliminate): ... here. Skip inserted exprs.
23531 (pass_pre::execute): Move fini_pre after fini_eliminate.
23532 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
23533 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
23534 PRE to get rid of dead code that has invalid SSA form and
23535 split critical edges again.
23537 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23539 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
23541 2017-05-11 Richard Biener <rguenther@suse.de>
23543 * passes.c (execute_function_todo): Verify loops if they are
23544 said to be up-to-date.
23545 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
23546 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
23548 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
23551 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
23552 handle calling assemble_external ourself.
23555 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
23556 modes with zero size. Enhance comment.
23558 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23560 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
23561 built-ins for vec_xl and vec_xst with short and char pointer
23564 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
23566 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
23567 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
23568 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
23569 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
23570 (_mm_maskz_min_round_ss): New intrinsics.
23571 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
23572 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
23573 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
23574 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
23575 (__builtin_ia32_minss_mask_round): New builtins.
23576 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23577 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23578 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
23580 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
23581 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
23583 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
23586 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
23588 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
23589 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
23590 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
23591 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
23592 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
23593 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
23594 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
23595 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23596 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
23597 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
23598 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
23599 (__builtin_ia32_mulss_mask_round): New builtins.
23600 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23601 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23602 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
23604 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
23605 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
23607 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
23610 2017-05-10 Julia Koval <julia.koval@intel.com>
23612 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
23613 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
23614 (_mm256_setr_m128i): New intrinsics.
23616 2017-05-10 Julia Koval <julia.koval@intel.com>
23618 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
23619 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
23620 (_mm_maskz_rcp14_ss): New intrinsics.
23621 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
23622 (__builtin_ia32_rcp14ss_mask): New builtins.
23623 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
23625 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
23627 PR tree-optimization/51513
23628 * tree-cfg.c (gimple_seq_unreachable_p): New function.
23629 (assert_unreachable_fallthru_edge_p): Use it.
23630 (group_case_labels_stmt): Likewise.
23631 * tree-cfg.h: Prototype it.
23632 * stmt.c: Include cfghooks.h and tree-cfg.h.
23633 (emit_case_dispatch_table) <gap_label>: New local variable.
23634 Use it to fill dispatch table gaps.
23635 Test for default_label before updating probabilities.
23636 (expand_case) <default_label>: Remove unneeded initialization.
23637 Test for unreachable default case statement and remove its edge.
23638 Set default_label accordingly.
23639 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
23641 2017-05-10 Carl Love <cel@us.ibm.com>
23643 * config/rs6000/rs6000-c: Add support for built-in functions
23644 vector signed char vec_neg (vector signed char)
23645 vector signed short int vec_neg (vector short int)
23646 vector signed int vec_neg (vector signed int)
23647 vector signed long long vec_neg (vector signed long long)
23648 vector float vec_neg (vector float)
23649 vector double vec_neg (vector double)
23650 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
23652 * config/rs6000/altivec.h: Add define for vec_neg
23653 * doc/extend.texi: Update the built-in documentation for the
23654 new built-in functions.
23656 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23658 PR tree-optimization/77644
23659 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
23661 2017-05-10 Nathan Sidwell <nathan@acm.org>
23663 * dumpfile.h (TDI_lang_all): New.
23664 (TDF_KIND): New. Renumber others
23665 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
23667 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
23669 (get_dump_file_name): Adjust suffix generation.
23670 (dump_enable_all): Use TDF_KIND.
23671 * doc/invoke.texi (-fdump-lang-all): Document.
23673 * dumpfile.h: Tabify.
23675 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
23678 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
23679 Move member access before delete.
23681 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
23683 * tree-inline.c (expand_call_inline): Split block at stmt
23686 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
23689 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
23690 are now unused after splitting mov{sf,sd}_hardfloat.
23691 (f32_lr2): Likewise.
23692 (f32_lm): Likewise.
23693 (f32_lm2): Likewise.
23694 (f32_li): Likewise.
23695 (f32_li2): Likewise.
23696 (f32_lv): Likewise.
23697 (f32_sr): Likewise.
23698 (f32_sr2): Likewise.
23699 (f32_sm): Likewise.
23700 (f32_sm2): Likewise.
23701 (f32_si): Likewise.
23702 (f32_si2): Likewise.
23703 (f32_sv): Likewise.
23704 (f32_dm): Likewise.
23705 (f32_vsx): Likewise.
23706 (f32_av): Likewise.
23707 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
23708 For movsf, order stores so the VSX stores occur before the GPR
23709 store which encourages the register allocator to use a traditional
23710 FPR instead of a GPR. For movsd, order the stores so that the GPR
23711 store comes before the VSX stores to allow the power6 to work.
23712 This is due to the power6 not having a 32-bit integer store
23713 instruction from a FPR.
23714 (movsf_hardfloat): Likewise.
23715 (movsd_hardfloat): Likewise.
23717 2017-05-09 Martin Sebor <msebor@redhat.com>
23719 PR translation/80280
23720 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
23723 PR translation/80280
23724 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
23725 data member added in r247778.
23726 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
23728 2017-05-09 Nathan Sidwell <nathan@acm.org>
23730 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
23732 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
23735 2017-05-09 Marek Polacek <polacek@redhat.com>
23737 * doc/invoke.texi: Fix typo.
23739 2017-05-09 Richard Biener <rguenther@suse.de>
23741 * tree-vrp.c (vrp_val_is_max): Adjust comment.
23742 (vrp_val_is_min): Likewise.
23743 (set_value_range_to_value): Likewise.
23744 (set_value_range_to_nonnegative): Likewise.
23745 (gimple_assign_nonzero_p): Likewise.
23746 (gimple_stmt_nonzero_p): Likewise.
23747 (vrp_int_const_binop): Likewise. Remove unreachable case.
23748 (adjust_range_with_scev): Adjust comments.
23749 (compare_range_with_value): Likewise.
23750 (extract_range_from_phi_node): Likewise.
23751 (test_for_singularity): Likewise.
23753 2017-05-09 Richard Biener <rguenther@suse.de>
23755 * tree-vrp.c (get_single_symbol): Add assert that we don't
23756 get overflowed constants as invariant part.
23757 (compare_values_warnv): Add comment before the TREE_NO_WARNING
23758 checks. Use wi::cmp instead of recursing for integer constants.
23759 (compare_values): Just ignore whether we assumed undefined
23760 overflow instead of failing the compare.
23761 (extract_range_for_var_from_comparison_expr): Add comment before the
23762 TREE_NO_WARNING sets.
23763 (test_for_singularity): Likewise.
23764 (extract_range_from_comparison): Do not disable optimization
23765 when we assumed undefined overflow.
23766 (extract_range_basic): Remove init of unused var.
23768 2017-05-09 Richard Biener <rguenther@suse.de>
23770 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
23771 (extract_range_from_multiplicative_op_1): Adjust.
23772 (extract_range_from_binary_expr_1): Use int_const_binop.
23774 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
23777 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
23778 rs6000_store_data_bypass_p in seven define_bypass directives and
23779 in several comments.
23780 * config/rs6000/rs6000-protos.h: Add prototype for
23781 rs6000_store_data_bypass_p function.
23782 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
23783 function implements slightly different (rs6000-specific) semantics
23784 than store_data_bypass_p, returning false rather than aborting
23785 with assertion error when arguments do not satisfy the
23786 requirements of store data bypass.
23787 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
23788 rs6000_store_data_bypass_p.
23790 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
23792 * config/xtensa/xtensa-protos.h
23793 (xtensa_initial_elimination_offset): New declaration.
23794 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
23795 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
23796 macro definition, add case for FRAME_POINTER_REGNUM when
23797 FRAME_GROWS_DOWNWARD.
23798 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
23799 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
23800 xtensa_initial_elimination_offset.
23802 2017-05-08 Nathan Sidwell <nathan@acm.org>
23804 * doc/invoke.texi: Alphabetize -fdump options.
23806 2017-05-08 Martin Sebor <msebor@redhat.com>
23808 PR translation/80280
23809 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
23811 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
23813 * target.def (compute_frame_layout): New optional target hook.
23814 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
23815 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
23816 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
23818 * reload1.c (verify_initial_elim_offsets): Likewise.
23819 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
23820 (use_simple_return_p): Call arm_compute_frame_layout if needed.
23821 (arm_get_frame_offsets): Split up into this ...
23822 (arm_compute_frame_layout): ... and this function.
23824 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
23826 * config/aarch64/constraints.md (Usa): New constraint.
23827 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
23829 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
23831 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
23832 with_multilib_list after it has been checked.
23834 2017-05-08 Richard Biener <rguenther@suse.de>
23836 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
23837 (bitmap_set_subtract_values): Likewise.
23839 2017-05-08 Richard Biener <rguenther@suse.de>
23841 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
23842 (gimple_assign_nonzero): ... this and remove strict_overflow_p
23844 (gimple_stmt_nonzero_warnv_p): Rename to ...
23845 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
23847 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
23848 (extract_range_basic): Adjust, do not disable propagation on
23849 strict overflow sensitive simplification.
23850 (vrp_visit_cond_stmt): Likewise.
23852 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
23854 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
23855 body size unconditionally.
23857 2017-05-07 Jeff Law <law@redhat.com>
23860 2017-05-06 Jeff Law <law@redhat.com>
23861 PR tree-optimization/78496
23862 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
23865 PR tree-optimization/78496
23866 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
23867 (simplify_stmt_using_ranges): Call it.
23868 (vrp_dom_walker::before_dom_children): Extract equivalences
23869 from an ASSERT_EXPR with an equality comparison against a
23872 2017-05-06 Jeff Law <law@redhat.com>
23874 PR tree-optimization/78496
23875 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
23878 PR tree-optimization/78496
23879 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
23880 (simplify_stmt_using_ranges): Call it.
23881 (vrp_dom_walker::before_dom_children): Extract equivalences
23882 from an ASSERT_EXPR with an equality comparison against a
23885 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
23887 * lra-constraints.c (lra_copy_reg_equiv): New function.
23888 (split_reg): Use it to copy equivalence information from the
23889 original register to the spill register.
23891 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
23893 PR rtl-optimization/75964
23894 * simplify-rtx.c (simplify_const_relational_operation): Remove
23895 invalid handling of comparisons of integer ABS.
23897 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
23899 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
23900 initialize to zero.
23901 (init_regs): Remove declaration.
23902 (function_arg_advance_32): Initialize error_p as boolean variable.
23904 2017-05-05 Nathan Sidwell <nathan@acm.org>
23906 * store-motion.c (remove_reachable_equiv_notes): Reformat long
23907 lines. Use for (;;).
23909 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23911 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
23912 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
23913 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
23914 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
23915 VF=2 that require versioning.
23917 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23919 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
23922 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23924 * diagnostic.h (diagnostic_override_option_index): Convert from
23925 macro to inline function.
23927 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23929 * diagnostic.c (last_module_changed_p): New function.
23930 (set_last_module): New function.
23931 (diagnostic_report_current_module): Convert macro usage to
23932 the above functions.
23933 * diagnostic.h (diagnostic_context::last_module): Strengthen
23934 from const line_map * to const line_map_ordinary *.
23935 (diagnostic_last_module_changed): Delete macro.
23936 (diagnostic_set_last_module): Delete macro.
23938 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23940 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
23941 with diagnostic_report_diagnostic.
23942 (diagnostic_n_impl_richloc): Likewise.
23943 * diagnostic.h (report_diagnostic): Delete macro.
23944 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
23945 with diagnostic_report_diagnostic.
23946 * substring-locations.c (format_warning_va): Likewise.
23948 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23950 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
23951 save/restor of format_spec. Move option-printing code to...
23952 (print_option_information): ...this new function, and
23953 reimplement by simply printing to the pretty_printer,
23954 rather than appending to the format string.
23956 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23958 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
23959 handling logic into...
23960 (update_effective_level_from_pragmas): ...this new function.
23962 2017-05-04 Andrew Waterman <andrew@sifive.com>
23964 * config/riscv/riscv.opt (mstrict-align): New option.
23965 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
23966 (SLOW_UNALIGNED_ACCESS): Define.
23967 (riscv_slow_unaligned_access): Declare.
23968 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
23970 (riscv_slow_unaligned_access): New variable.
23971 (rocket_tune_info): Set slow_unaligned_access to true.
23972 (optimize_size_tune_info): Set slow_unaligned_access to false.
23973 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
23974 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
23975 (riscv_option_override): Set riscv_slow_unaligned_access.
23976 * doc/invoke.texi: Add -mstrict-align to RISC-V.
23978 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
23980 * config/riscv/riscv.md: Unify indentation.
23982 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
23987 * config/rs6000/rs6000.md (u code attribute): Add FIX and
23989 (extendsi<mode>2): Add support for doing sign extension via
23990 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
23991 don't have ISA 3.0 instructions.
23992 (extendsi<mode>2 splitter): Likewise.
23993 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
23994 generate the normal insns since SImode can now go in vector
23995 registers. Disallow the special UNSPECs needed for previous
23996 machines to hide SImode being used. Add new insns
23997 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
23998 (fix_trunc<mode>si2_stfiwx): Likewise.
23999 (fix_trunc<mode>si2_internal): Likewise.
24000 (fixuns_trunc<mode>si2): Likewise.
24001 (fixuns_trunc<mode>si2_stfiwx): Likewise.
24002 (fctiw<u>z_<mode>_smallint): Likewise.
24003 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
24004 of floating point to 32-bit integer from doing a direct move to
24005 the GPR registers to do a store.
24006 (fctiwz_<mode>): Break long line.
24008 2017-05-05 Bin Cheng <bin.cheng@arm.com>
24010 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
24011 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
24012 (addr_list, addr_offset_valid_p): New.
24013 (split_address_groups): Check offset validity with above function.
24014 (gt-tree-ssa-loop-ivopts.h): Include header file.
24016 2017-05-05 Nathan Sidwell <nathan@acm.org>
24018 * config.gcc (arm*-*-*): Add missing 'fi'.
24020 2017-05-05 Steve Ellcey <sellcey@cavium.com>
24022 * doc/invoke.texi (-fopt-info): Explicitly say order of options
24023 included in -fopt-info does not matter.
24024 * doc/optinfo.texi (-fopt-info): Fix description of default
24025 behavour. Explicitly say order of options included in -fopt-info
24028 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
24030 * config.gcc: Allow combinations of aprofile and rmprofile values for
24031 --with-multilib-list.
24032 * config/arm/t-multilib: New file.
24033 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
24034 variables. Remove setting of ISA and floating-point ABI in
24035 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
24036 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
24037 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
24038 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
24040 * config/arm/t-rmprofile: Likewise except for the matches changes.
24041 * doc/install.texi (--with-multilib-list): Document the combination of
24042 aprofile and rmprofile values and warn about pitfalls in doing that.
24044 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24046 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
24047 (movdi_aarch64): Likewise.
24049 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24051 PR tree-optimization/80632
24052 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
24054 (build_arrays): Initialize it for virtual phis.
24055 (fix_phi_nodes): Use it for virtual phis.
24057 PR tree-optimization/80558
24058 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
24059 [x, y] op z into [x op, y op z] for op & or | if conditions
24062 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
24063 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24066 * config/arm/arm.md (use_literal_pool): Remove.
24067 (64-bit immediate split): No longer takes cost into consideration
24068 if arm_disable_literal_pool is enabled.
24069 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
24070 used when arm_disable_literal_pool is enabled.
24071 (arm_max_const_double_inline_cost): Remove use of
24072 arm_disable_literal_pool.
24073 (push_minipool_fix): Add assert.
24074 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
24075 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
24076 (no_literal_pool_sf_immediate): New.
24078 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24080 PR tree-optimization/80613
24081 * tree-ssa-dce.c (propagate_necessity): Remove cases for
24082 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
24084 2017-05-05 Richard Biener <rguenther@suse.de>
24086 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
24088 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
24090 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
24091 of this flag from insn conditions due to removal from r247495.
24093 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24095 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
24097 (arm_early_store_addr_dep_ptr): Likewise.
24098 * config/arm/aarch-common-protos.h
24099 (arm_early_load_addr_dep_ptr): Add prototype.
24100 (arm_early_store_addr_dep_ptr): Likewise.
24101 * config/arm/cortex-a53.md: Add new bypasses.
24103 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24105 * tree.c (next_type_uid): Change type to unsigned.
24106 (type_hash_canon): Decrement back next_type_uid if
24107 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
24108 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
24111 2017-05-04 Martin Sebor <msebor@redhat.com>
24113 * builtins.c: Fix a trivial typo in a comment.
24115 PR middle-end/79234
24116 * builtins.c (check_sizes): Adjust to handle reading past the end.
24117 Avoid printing excessive upper bound of ranges. Use %E to print
24118 tree nodes instead of converting them to %wu.
24119 (expand_builtin_memchr): New function.
24120 (compute_dest_size): Rename...
24121 (compute_objsize): ...to this.
24122 (expand_builtin_memcpy): Adjust.
24123 (expand_builtin_mempcpy): Adjust.
24124 (expand_builtin_strcat): Adjust.
24125 (expand_builtin_strcpy): Adjust.
24126 (check_strncat_sizes): Adjust.
24127 (expand_builtin_strncat): Adjust.
24128 (expand_builtin_strncpy): Adjust and simplify.
24129 (expand_builtin_memset): Adjust.
24130 (expand_builtin_bzero): Adjust.
24131 (expand_builtin_memcmp): Adjust.
24132 (expand_builtin): Handle memcmp.
24133 (maybe_emit_chk_warning): Check strncat just once.
24135 2017-05-04 Martin Sebor <msebor@redhat.com>
24137 PR preprocessor/79214
24138 PR middle-end/79222
24139 PR middle-end/79223
24140 * builtins.c (check_sizes): Add inlining context and issue
24141 warnings even when -Wno-system-headers is set.
24142 (check_strncat_sizes): Same.
24143 (expand_builtin_strncat): Same.
24144 (expand_builtin_memmove): New function.
24145 (expand_builtin_stpncpy): Same.
24146 (expand_builtin): Handle memmove and stpncpy.
24148 2017-05-04 Bin Cheng <bin.cheng@arm.com>
24150 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
24151 which is not used any more.
24153 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24155 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
24157 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24159 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
24160 (cortexa53_tunings): Likewise.
24161 (cortexa57_tunings): Likewise.
24162 (cortexa72_tunings): Likewise.
24163 (cortexa73_tunings): Likewise.
24165 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24167 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
24168 Set loop alignment to 8.
24170 2017-05-04 Martin Sebor <msebor@redhat.com>
24172 PR translation/80280
24173 * builtins.c (expand_builtin_object_size): Add missing quoting to
24174 %D and like directives.
24175 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
24176 (hsa_type_for_tree_type): Same.
24177 (verify_function_arguments): Same.
24178 * symtab.c (symbol_table::change_decl_assembler_name): Same.
24179 * varasm.c (get_section): Same.
24182 2017-05-04 Martin Sebor <msebor@redhat.com>
24184 PR translation/80280
24185 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
24187 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24189 * config/aarch64/aarch64.c (generic_addrcost_table):
24190 Change HI/TI mode setting.
24192 2017-05-04 Martin Jambor <mjambor@suse.cz>
24194 PR tree-optimization/80622
24195 * tree-sra.c (comes_initialized_p): New function.
24196 (build_accesses_from_assign): Only set write lazily when
24197 comes_initialized_p is false.
24198 (analyze_access_subtree): Use comes_initialized_p.
24199 (propagate_subaccesses_across_link): Assert !comes_initialized_p
24200 instead of testing for PARM_DECL.
24202 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24204 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
24205 constraint on operand 0 to allow more general addressing modes.
24206 Adjust output template.
24207 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
24209 * config/aarch64/aarch64-protos.h
24210 (aarch64_address_valid_for_prefetch_p): Declare prototype.
24211 * config/aarch64/constraints.md (Dp): New address constraint.
24212 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
24215 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
24217 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
24218 update use of estimate_ipcp_clone_size_and_time.
24219 (estimate_local_effects): Update use of
24220 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
24221 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
24222 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
24223 Return nonspecialized time.
24225 2017-05-04 Richard Biener <rguenther@suse.de>
24227 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
24228 for the last VUSE which def dominates the PHI. Directly call
24230 (get_continuation_for_phi_1): Remove.
24232 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
24234 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
24235 to explain the use of truncating division. Cap the number of
24236 iterations to the maximum given by nb_iterations_upper_bound,
24239 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24241 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
24242 * configure: Regenerate.
24243 * config.in: Regenerate.
24244 * config/i386/driver-mingw32.c: new file.
24245 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
24246 * config.host: Link driver-mingw32.o on MinGW host.
24247 * doc/install.texi: Document new --enable-mingw-wildcard configure
24250 2017-05-04 Marek Polacek <polacek@redhat.com>
24252 PR tree-optimization/80612
24253 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
24255 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24256 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
24258 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
24259 (movt splitter): Likewise.
24260 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
24261 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
24262 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
24263 block for Thumb-1 with MOVT.
24264 (thumb2_legitimate_address_p): Move code block ...
24265 (can_avoid_literal_pool_for_label_p): ... into this new function.
24266 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
24268 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
24269 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
24270 "M-profile targets with the MOVT instruction".
24272 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24274 * config/arm/arm-builtins.c (arm_init_builtins): Rename
24275 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
24276 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
24278 2017-05-04 Martin Liska <mliska@suse.cz>
24280 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
24281 variable cond_code.
24283 2017-05-04 Richard Biener <rguenther@suse.de>
24285 * tree.c (array_at_struct_end_p): Handle arrays at struct
24286 end with flexarrays more conservatively. Refactor and treat
24287 arrays of arrays or aggregates more strict. Fix
24288 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
24289 * tree.c (array_at_struct_end_p): Adjust prototype.
24290 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
24291 * gimple-fold.c (get_range_strlen): Likewise.
24292 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
24294 2017-05-04 Richard Biener <rguenther@suse.de>
24296 PR tree-optimization/31130
24297 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
24299 (supports_overflow_infinity): Likewise.
24300 (is_negative_overflow_infinity): Likewise.
24301 (is_positive_overflow_infinity): Likewise.
24302 (is_overflow_infinity): Likewise.
24303 (stmt_overflow_infinity): Likewise.
24304 (overflow_infinity_range_p): Likewise.
24305 (usable_range_p): Remove as always returning true.
24306 (make_overflow_infinity): Remove.
24307 (negative_overflow_infinity): Likewise.
24308 (positive_overflow_infinity): Likewise.
24309 (avoid_overflow_infinity): Likewise.
24310 (set_value_range): Adjust accordingly.
24311 (set_value_range_to_nonnegative): Likewise, remove now unused
24312 overflow_infinity arg.
24313 (vrp_operand_equal_p): Adjust.
24314 (update_value_range): Likewise.
24315 (range_int_cst_singleton_p): Likewise.
24316 (operand_less_p): Likewise.
24317 (compare_values_warnv): Likewise.
24318 (extract_range_for_var_from_comparison_expr): Likewise.
24319 (vrp_int_const_binop): Likewise.
24320 (zero_nonzero_bits_from_vr): Likewise.
24321 (extract_range_from_multiplicative_op_1): Likewise.
24322 (extract_range_from_binary_expr_1): Likewise.
24323 (extract_range_from_unary_expr): Likewise.
24324 (extract_range_from_comparison): Likewise.
24325 (extract_range_basic): Likewise.
24326 (adjust_range_with_scev): Likewise.
24327 (compare_ranges): Likewise.
24328 (compare_range_with_value): Likewise.
24329 (dump_value_range): Likewise.
24330 (test_for_singularity): Likewise, remove strict_overflow_p parameter
24332 (simplify_cond_using_ranges): Adjust.
24334 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
24336 * brig-builtins.def: Added a builtin for class_f64.
24337 * builtin-types.def: Added a builtin type needed by class_f64.
24339 2017-05-03 Jason Merrill <jason@redhat.com>
24341 * timevar.def: Add TV_CONSTEXPR.
24343 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24345 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
24347 2017-05-03 Martin Jambor <mjambor@suse.cz>
24349 * ipa-prop.c (ipa_update_after_lto_read): Removed.
24350 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
24351 * ipa-cp.c (ipcp_propagate_stage): Do not call
24352 ipa_update_after_lto_read.
24353 * ipa-inline.c (ipa_inline): Likewise.
24355 2017-05-03 Martin Jambor <mjambor@suse.cz>
24357 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
24358 tag. Added a default constructor and a destructor.
24359 (ipa_edge_args_sum_t): New class;
24360 (ipa_edge_args_sum): Declare.
24361 (ipa_edge_args_vector): Remove declaration.
24362 (IPA_EDGE_REF): Use ipa_edge_args_sum.
24363 (ipa_free_edge_args_substructures): Remove declaration.
24364 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
24365 (ipa_edge_args_info_available_for_edge_p): Likewise.
24366 * ipa-prop.c (ipa_edge_args_vector): Removed.
24367 (edge_removal_hook_holder): Likewise.
24368 (edge_duplication_hook_holder): Likewise.
24369 (ipa_edge_args_sum): New variable.
24370 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
24371 ipa_edge_args_vector.
24372 (ipa_free_edge_args_substructures): Likewise.
24373 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
24374 ipa_edge_args_vector.
24375 (ipa_edge_removal_hook): Turned into method
24376 ipa_edge_args_sum_t::remove.
24377 (ipa_edge_duplication_hook): Turned into method
24378 ipa_edge_args_sum_t::duplicate.
24379 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
24380 registering edge hooks.
24381 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
24382 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
24383 ipa_edge_args_sum instead of ipa_edge_args_vector.
24384 * ipa-profile.c (ipa_profile): Likewise.
24386 2017-05-03 Martin Jambor <mjambor@suse.cz>
24388 * symbol-summary.h (function_summary): New method exists.
24389 (function_summary::symtab_removal): Deallocate through release.
24390 (call_summary): New class.
24391 (gt_ggc_mx): New overload.
24392 (gt_pch_nx): Likewise.
24393 (gt_pch_nx): Likewise.
24395 2017-05-03 Jeff Law <law@redhat.com>
24397 PR tree-optimization/78496
24398 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
24399 from simplify_cond_using_ranges. Split off code to walk
24400 backwards through casts into ...
24401 (simplify_cond_using_ranges_2): New function.
24402 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
24403 (execute_vrp): After identifying jump threads, call
24404 simplify_cond_using_ranges_2.
24406 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
24409 * ipa-inline.h (inline_summary): Add ctor.
24410 (create_ggc): Do not use ggc_cleared_alloc.
24412 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
24413 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24415 * gcc.c (handle_braces): Support escaping in switch matching
24417 * doc/invoke.texi (Spec Files): Document it.
24418 Remove superfluous @code markup in items.
24420 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24422 * diagnostic-show-locus.c (struct column_range): New struct.
24423 (get_affected_columns): New function.
24424 (get_printed_columns): New function.
24425 (struct correction): New struct.
24426 (correction::ensure_capacity): New function.
24427 (correction::ensure_terminated): New function.
24428 (struct line_corrections): New struct.
24429 (line_corrections::~line_corrections): New dtor.
24430 (line_corrections::add_hint): New function.
24431 (layout::print_trailing_fixits): Reimplement in terms of the new
24433 (selftest::test_overlapped_fixit_printing): New function.
24434 (selftest::diagnostic_show_locus_c_tests): Call it.
24436 2017-05-03 Nathan Sidwell <nathan@acm.org>
24438 Canonicalize canonical type hashing
24439 * tree.h (type_hash_canon_hash): Declare.
24440 * tree.c (type_hash_list, attribute_hash_list): Move into
24441 type_hash_canon_hash.
24442 (build_type_attribute_qual_variant): Break out hash code calc into
24443 type_hash_canon_hash.
24444 (type_hash_canon_hash): New. Generic type hash computation.
24445 (build_range_type_1, build_array_type_1, build_function_type,
24446 build_method_type_directly, build_offset_type, build_complex_type,
24447 make_vector_type): Call it.
24449 2017-05-03 Richard Biener <rguenther@suse.de>
24451 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
24452 When all DRs have unknown misaligned do not always peel
24453 when there is a store but apply the same costing model as if
24454 there were only loads.
24456 2017-05-03 Richard Biener <rguenther@suse.de>
24459 PR tree-optimization/80492
24460 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
24461 compare_base_decls returning dont-know properly.
24463 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
24465 * config/arm/iterators.md (CCSI): New mode iterator.
24466 (arch): New mode attribute.
24467 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
24468 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
24469 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
24470 code iterator for success result mode.
24471 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
24472 the corresponding new insn generators.
24474 2017-05-03 Bin Cheng <bin.cheng@arm.com>
24477 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24478 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24480 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
24482 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
24483 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
24484 (DDR_A): Wrap DDR argument in brackets.
24485 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
24486 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
24487 (DDR_REVERSED_P): Likewise.
24489 2017-05-03 Jakub Jelinek <jakub@redhat.com>
24491 PR tree-optimization/79472
24492 * tree-switch-conversion.c (struct switch_conv_info): Add
24493 contiguous_range and default_case_nonstandard fields.
24494 (collect_switch_conv_info): Compute contiguous_range and
24495 default_case_nonstandard fields, don't clear final_bb if
24496 contiguous_range and only the default case doesn't have the required
24498 (check_all_empty_except_final): Set default_case_nonstandard instead
24499 of failing if contiguous_range and the default case doesn't have empty
24501 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
24502 and only the default case doesn't have the required constants. Skip
24504 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
24505 if default_case_nonstandard.
24506 (build_constructors): Build constant 1 just once. Assert that default
24507 values aren't inserted in between cases if contiguous_range. Skip
24509 (build_arrays): Skip virtual phis.
24510 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
24511 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
24512 Handle virtual phis.
24513 (gen_inbound_check): Handle default_case_nonstandard case.
24514 (process_switch): Adjust check_final_bb caller. Call
24515 gather_default_values with the first non-default case instead of
24516 default case if default_case_nonstandard.
24518 2017-05-02 Nathan Sidwell <nathan@acm.org>
24520 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
24521 check. Fix formatting.
24523 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
24525 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
24526 errors when comparing specialized and unspecialized times.
24528 2017-05-02 David Malcolm <dmalcolm@redhat.com>
24530 * diagnostic-show-locus.c
24531 (layout::should_print_annotation_line_p): Make private.
24532 (layout::print_annotation_line): Make private.
24533 (layout::annotation_line_showed_range_p): Make private.
24534 (layout::show_ruler): Make private.
24535 (layout::print_source_line): Make private. Pass in line and
24536 line_width, rather than calling location_get_source_line. Drop
24538 (layout::print_leading_fixits): New method.
24539 (layout::print_any_fixits): Rename to...
24540 (layout::print_trailing_fixits): ...this, and make private.
24541 Don't print newline fixits.
24542 (diagnostic_show_locus): Move logic for printing one row into...
24543 (layout::print_line): ...this new function. Move the
24544 location_get_source_line call and error-handling from
24545 print_source_line to here. Call print_leading_fixits, and rename
24546 print_any_fixits to print_trailing_fixits.
24547 (selftest::test_fixit_insert_containing_newline): Update now that
24548 newlines are partially supported.
24549 (selftest::test_fixit_insert_containing_newline_2): New test.
24550 (selftest::test_fixit_replace_containing_newline): Update comments.
24551 (selftest::diagnostic_show_locus_c_tests): Call the new test.
24552 * edit-context.c (class added_line): New class.
24553 (class edited_line): Describe newline handling in comment.
24554 (edited_line::actually_edited_p): New method.
24555 (edited_line::print_content): Delete redundant decl.
24556 (edited_line::m_predecessors): New field.
24557 (edited_file::print_content): Call edited_line::print_content.
24558 (edited_file::print_diff): Update to support newlines.
24559 (edited_file::print_diff_hunk): Likewise.
24560 (edited_file::print_run_of_changed_lines): New function.
24561 (edited_file::print_diff_line): Convert to...
24562 (print_diff_line): ...this.
24563 (edited_file::get_effective_line_count): New function.
24564 (edited_line::edited_line): Initialize new field m_predecessors.
24565 (edited_line::~edited_line): Clean up m_predecessors.
24566 (edited_line::apply_fixit): Handle newlines.
24567 (edited_line::get_effective_line_count): New function.
24568 (edited_line::print_content): New function.
24569 (edited_line::print_diff_lines): New function.
24570 (selftest::test_applying_fixits_insert_containing_newline): New
24572 (selftest::test_applying_fixits_replace_containing_newline): New
24574 (selftest::insert_line): New function.
24575 (selftest::test_applying_fixits_multiple_lines): Add example of
24577 (selftest::edit_context_c_tests): Call the new tests.
24579 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24581 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
24582 parameter cand. Update dump information.
24583 (get_computation_cost): Update uses.
24585 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24587 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
24588 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
24589 (get_computation_at, rewrite_use_address): Update use of
24590 get_computation_aff.
24592 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24594 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
24595 (get_computation): Delete.
24596 (get_computation_cost): Implement like get_computation_cost_at.
24597 Use get_computation_at.
24598 (get_computation_cost_at): Delete.
24599 (rewrite_use_nonlinear_expr): Use get_computation_at.
24600 (rewrite_use_compare, remove_unused_ivs): Ditto.
24602 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24604 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
24606 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24608 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
24609 (ivopts_global_cost_for_size): Rename parameter and update uses.
24610 (iv_ca_recount_cost): Update uses.
24611 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
24612 candidates seperately in n_invs and n_cands.
24613 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
24615 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24617 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
24618 (find_inv_vars_cb): New.
24619 (find_depends): Renamed to ...
24620 (find_inv_vars): ... this.
24621 (add_candidate_1, force_var_cost): Call find_inv_vars.
24622 (split_address_cost, determine_group_iv_cost_cond): Ditto.
24624 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24626 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
24627 inv_vars. Add inv_exprs.
24628 (struct iv_cand): Rename depends_on to inv_vars.
24629 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
24630 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
24631 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
24632 (dump_cand): Dump inv_vars.
24633 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
24634 (record_invariant, find_depends, add_candidate_1): Ditto.
24635 (set_group_iv_cost, force_var_cost): Ditto.
24636 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
24637 (get_computation_cost_at, get_computation_cost): Ditto.
24638 (determine_group_iv_cost_generic): Ditto.
24639 (determine_group_iv_cost_address): Ditto.
24640 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
24641 (determine_group_iv_costs): Ditto.
24642 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
24643 (iv_ca_set_remove_invariants): Renamed to ...
24644 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
24645 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
24646 (iv_ca_set_add_invariants): Renamed to ...
24647 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
24648 (iv_ca_set_cp): Use iv_ca_set_add_invs.
24649 (iv_ca_has_deps): Support inv_vars and inv_exprs.
24650 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
24651 (create_new_ivs): Remove useless dump.
24653 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24655 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
24657 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
24658 (iv_ca_set_no_cp, create_new_iv): Ditto.
24660 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24662 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24664 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
24666 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
24667 function tree_check2.
24669 2017-05-02 Martin Liska <mliska@suse.cz>
24671 * doc/gcov.texi: Add missing preposition.
24672 * gcov.c (function_info::function_info): Properly fill up
24673 all member variables.
24675 2017-05-02 Tamar Christina <tamar.christina@arm.com>
24677 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
24679 2017-05-02 Tamar Christina <tamar.christina@arm.com>
24681 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
24683 2017-05-02 Martin Liska <mliska@suse.cz>
24686 * lto-streamer-in.c (lto_read_tree_1): Remove
24687 LTO_STREAMER_DEBUG.
24688 * lto-streamer.c (struct tree_hash_entry): Likewise.
24689 (struct tree_entry_hasher): Likewise.
24690 (tree_entry_hasher::hash): Likewise.
24691 (tree_entry_hasher::equal): Likewise.
24692 (lto_streamer_init): Likewise.
24693 (lto_orig_address_map): Likewise.
24694 (lto_orig_address_get): Likewise.
24695 (lto_orig_address_remove): Likewise.
24696 * lto-streamer.h: Likewise.
24697 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
24698 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
24700 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
24702 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
24703 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
24704 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
24705 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
24706 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
24707 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
24708 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
24709 (mm_maskz_sub_ss): New intrinsics.
24710 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24711 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
24712 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
24713 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
24714 (__builtin_ia32_subss_mask_round): New builtins.
24715 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24716 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24717 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
24719 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
24720 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
24722 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
24725 2017-05-02 Martin Jambor <mjambor@suse.cz>
24727 PR tree-optimization/78687
24728 * tree-sra.c (access): New field parent.
24729 (process_subtree_disqualification): New function.
24730 (disqualify_candidate): Call it.
24731 (build_accesses_from_assign): Reset write flag if creating an
24733 (build_access_subtree): Fill in parent field and also prpagate
24734 down grp_write flag.
24735 (create_artificial_child_access): New parameter set_grp_write, set
24736 grp_write to its value.
24737 (propagate_subaccesses_across_link): Also propagate grp_write flag
24739 (propagate_all_subaccesses): Push the closest parent back to work
24740 queue if add_access_to_work_queue returned true.
24742 2017-05-02 Richard Biener <rguenther@suse.de>
24744 * common.opt (fstrict-overflow): Alias negative to fwrapv.
24745 * doc/invoke.texi (fstrict-overflow): Remove all traces of
24746 -fstrict-overflow documentation.
24747 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
24748 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
24749 flag_strict_overflow.
24750 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
24751 * lto-opts.c (lto_write_options): Do not stream it.
24752 * lto-wrapper.c (merge_and_complain): Do not handle it.
24753 * opts.c (default_options_table): Do not set -fstrict-overflow.
24754 (finish_options): Likewise do not clear it when sanitizing.
24755 * simplify-rtx.c (simplify_const_relational_operation): Do not
24756 test flag_strict_overflow.
24758 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
24760 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
24761 using enabled attribute.
24762 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
24763 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
24764 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
24765 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
24766 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
24767 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
24768 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
24769 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
24770 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
24771 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
24773 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
24775 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
24777 2017-05-02 Richard Biener <rguenther@suse.de>
24779 PR tree-optimization/80591
24781 2017-04-10 Richard Biener <rguenther@suse.de>
24783 * tree-ssa-structalias.c (find_func_aliases): Properly handle
24786 2017-05-02 Richard Biener <rguenther@suse.de>
24788 PR tree-optimization/80549
24789 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
24790 (cleanup_tree_cfg_noloop): Create forwarders to known loop
24791 headers if they do not have a preheader.
24793 2017-05-02 Martin Liska <mliska@suse.cz>
24796 * common.opt: Fix typo.
24797 * doc/invoke.texi: Likewise.
24799 2017-05-01 Jan Beulich <jbeulich@suse.com>
24801 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
24802 swapping, add (x,x,m,x,n) alternative.
24804 2017-05-01 Nathan Sidwell <nathan@acm.org>
24806 * calls.c (combine_pending_stack_adjustment_and_call): Remove
24807 unnecessary unadjusted_alignment check.
24809 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
24812 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
24814 * gimplify.c (gimplify_cilk_detach): New function.
24815 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
24816 * tree-core.h: Document EXPR_CILK_SPAWN.
24817 * tree.h (EXPR_CILK_SPAWN): Define.
24819 2017-05-01 David Malcolm <dmalcolm@redhat.com>
24821 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
24822 to use new fixit_hint representation, using the "replace" logic.
24823 (get_line_span_for_fixit_hint): Likewise.
24824 (layout::print_any_fixits): Likewise.
24825 (selftest::test_one_liner_many_fixits): Rename to...
24826 (selftest::test_one_liner_many_fixits_1): ...this, and update
24827 comment and expected output to reflect that the multiple fix-it
24828 hints are now consolidated into one insertion.
24829 (selftest::test_one_liner_many_fixits_2): New test.
24830 (selftest::test_diagnostic_show_locus_one_liner): Update for
24832 (selftest::test_fixit_consolidation): Update for fix-it API
24834 * diagnostic.c (print_parseable_fixits): Likewise.
24835 * edit-context.c (edited_line::m_line_events): Convert from
24836 auto_vec <line_event *> to auto_vec <line_event>.
24837 (class line_event): Convert from abstract base class to a concrete
24838 class, taking over the role of replace_event.
24839 (class insert_event): Delete.
24840 (class replace_event): Rename to class line_event. Convert to
24842 (edit_context::add_fixits): Reimplement.
24843 (edit_context::apply_insert): Delete.
24844 (edit_context::apply_replace): Rename to...
24845 (edit_context::apply_fixit): ...this. Convert to half-open range.
24846 (edited_file::apply_insert): Delete.
24847 (edited_file::apply_replace): Rename to...
24848 (edited_file::apply_fixit): ...this.
24849 (edited_line::~edited_line): Drop deletion of events.
24850 (edited_line::apply_insert): Delete.
24851 (edited_line::apply_replace): Rename to...
24852 (edited_line::apply_fixit): ...this. Convert to half-open range.
24853 Update for change to type of m_line_events.
24854 * edit-context.h (edit_context::apply_insert): Delete.
24855 (edit_context::apply_replace): Rename to...
24856 (edit_context::apply_fixit): ...this.
24858 2017-05-01 Martin Sebor <msebor@redhat.com>
24860 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
24863 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
24866 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
24867 __get_cpuid_max returns 0.
24868 (__get_cpuid_count): Ditto.
24870 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
24872 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
24873 replacement expression is another instance of one of its arguments.
24875 2017-05-01 Jakub Jelinek <jakub@redhat.com>
24878 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
24879 check for stack push/pop autoinc.
24880 * config/i386/i386.c (ix86_agi_dependent): Return false
24881 if the only reason why modified_in_p returned true is that
24882 addr is SP based and set_insn is a push or pop.
24884 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
24886 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
24889 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
24892 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
24893 (account_size_time): Use two predicates - exec_pred and
24895 (evaluate_conditions_for_known_args): Compute both clause and
24897 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
24898 (inline_summary_t::duplicate): Update.
24899 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
24901 (compute_inline_parameters): Likewise.
24902 (estimate_edge_size_and_time): Update caluclation of time.
24903 (estimate_node_size_and_time): Compute both time and nonspecialized
24905 (estimate_ipcp_clone_size_and_time): Update.
24906 (inline_merge_summary): Update.
24907 (do_estimate_edge_time): Update.
24908 (do_estimate_edge_size): Update.
24909 (do_estimate_edge_hints): Update.
24910 (inline_read_section, inline_write_summary): Stream both new predicates.
24911 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
24913 (compute_inlined_call_time): Cleanup.
24914 (big_speedup_p): Update.
24915 (edge_badness): Update.
24916 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
24917 (size_time_entry): Replace predicate by exec_predicate and
24918 nonconst_predicate.
24919 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
24920 (estimate_edge_time): Return also nonspec_time.
24921 (reset_edge_growth_cache): Update.
24923 2017-04-29 Jakub Jelinek <jakub@redhat.com>
24925 PR rtl-optimization/80491
24926 * ifcvt.c (noce_process_if_block): When looking for x setter
24927 with missing else_bb, don't check only the insn right before
24928 cond_earliest, but look for the last insn that x is modified in
24929 within the same bb.
24931 PR rtl-optimization/80491
24932 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
24934 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
24936 PR tree-optimization/80487
24937 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
24939 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24941 PR tree-optimization/79697
24942 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
24943 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
24944 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
24946 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
24947 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
24949 2017-04-28 Martin Sebor <msebor@redhat.com>
24951 PR tree-optimization/80523
24952 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
24953 (init_target_to_host_charmap, target_to_host, target_strtol10): New
24955 (maybe_warn, format_directive, parse_directive): Use new functions.
24956 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
24958 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
24960 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
24962 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
24964 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
24965 target_header_dir): Set correctly.
24966 * configure: Regenerated.
24967 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
24968 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
24969 instead of SYSTEM_HEADER_DIR.
24971 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
24973 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
24974 (estimate_local_effects): Likewise.
24975 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
24976 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
24977 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
24978 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
24979 do_estimate_edge_time, estimate_edge_time): Likewise.
24980 * ipa-inline-analysis.c (estimate_node_size_and_time,
24981 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
24982 (estimate_time_after_inlining): Remove.
24984 2017-04-28 Martin Liska <mliska@suse.cz>
24986 * doc/gcov.texi: Enhance documentation of gcov.
24988 2017-04-28 Martin Liska <mliska@suse.cz>
24990 * doc/gcov.texi: Sort options in alphabetic order.
24991 * doc/gcov-dump.texi: Likewise.
24992 * doc/gcov-tool.texi: Likewise.
24993 * gcov.c (print_usage): Likewise.
24994 * gcov-dump.c (print_usage): Likewise.
24995 * gcov-tool.c (print_merge_usage_message): Likewise.
24996 (print_rewrite_usage_message): Likewise.
24997 (print_overlap_usage_message): Likewise.
24999 2017-04-28 Martin Liska <mliska@suse.cz>
25001 PR gcov-profile/53915
25002 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
25004 2017-04-28 Martin Liska <mliska@suse.cz>
25006 PR gcov-profile/79891
25007 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
25008 is marked by compiler as living on a line.
25009 (get_cycles_count): Remove usage of the union.
25010 (output_intermediate_file): Likewise.
25011 (find_source): Fix GNU coding style.
25012 (accumulate_line_counts): Remove old non-all block mode.
25013 (output_lines): Remove usage of the union.
25014 * profile.c (output_location): Include all BBs, even if
25015 belonging to a same line (and file) as a previous BB.
25017 2017-04-28 Martin Liska <mliska@suse.cz>
25019 * gcov.c (process_args): Handle new argument 'w'.
25020 (read_graph_file): Assign ID to BBs.
25021 (output_branch_count): Display BB # if verbose flag is set.
25022 (output_lines): Likewise for arcs.
25023 (print_usage): Add '--verbose' option help.
25024 * doc/gcov.texi: Document --verbose (-w) option.
25026 2017-04-28 Martin Liska <mliska@suse.cz>
25028 * gcov.c (struct block_location_info): New struct.
25029 (process_file): Fill up the new structure.
25030 (read_graph_file): Replace usage of encoding by the newly added
25032 (add_line_counts): Likewise.
25033 (accumulate_line_counts): Remove usage of the union.
25034 (function_info::function_info): New function.
25035 (function_info::~function_info): Likewise.
25036 (process_file): Call delete instead of release_function.
25037 (release_function): Release the function.
25038 (release_structures): Call delete instead of release_function.
25039 (solve_flow_graph): Replace usage of num_blocks.
25040 (find_exception_blocks): Likewise.
25041 (output_lines): Fix GNU coding style.
25043 2017-04-28 Martin Liska <mliska@suse.cz>
25046 * coverage.c (coverage_remove_note_file): New function.
25047 * coverage.h: Declare the function.
25048 * toplev.c (finalize): Clean if an error has been seen.
25050 2017-04-28 Martin Liska <mliska@suse.cz>
25052 PR gcov-profile/80031
25053 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
25054 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
25055 * gcov.c (read_graph_file): Read just number of blocks.
25056 * profile.c (branch_prob): Do not stream 0 flags per a basic
25059 2017-04-28 Martin Liska <mliska@suse.cz>
25061 * gcov-dump.c (tag_*): Add new argument to declarations.
25062 (dump_gcov_file): Likewise.
25063 (tag_blocks): Add and use new argument depth.
25064 (tag_arcs): Likewise.
25065 (tag_lines): Likewise.
25066 (tag_counters): Likewise.
25067 (tag_summary): Likewise.
25068 (dump_working_sets): Use depth to do a proper indentation.
25070 2017-04-28 Jakub Jelinek <jakub@redhat.com>
25073 * cgraph.h (symtab_node::debug_symtab): No longer inline.
25074 * symtab.c (symtab_node::debug_symtab): Move definition here.
25076 2017-04-28 Richard Biener <rguenther@suse.de>
25078 * lto-streamer.h (LTO_major_version): Bump to 7.
25080 2017-04-28 Richard Biener <rguenther@suse.de>
25082 * tree-vrp.c (assert_info): New struct.
25083 (add_assert_info): New helper.
25084 (register_edge_assert_for_2): Refactor to add asserts to a vector
25086 (register_edge_assert_for_1): Likewise.
25087 (register_edge_assert_for): Likewise.
25088 (finish_register_edge_assert_for): New helper actually registering
25089 asserts where live on edge.
25090 (find_conditional_asserts): Adjust.
25091 (find_switch_asserts): Likewise.
25092 (evrp_dom_walker::try_find_new_range): Generalize.
25093 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
25095 2017-04-27 Marek Polacek <polacek@redhat.com>
25098 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
25099 arg10 and arg11 to itype.
25101 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
25103 * doc/extend.texi (Object Size Checking): Improve grammar.
25105 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
25108 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
25109 that the logic for permitting reciprocal estimates matches that
25112 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25115 * tree.c (type_cache_hasher::equal): Only compare
25116 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
25117 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
25118 non-aggregate element types.
25119 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
25120 about the flag on ARRAY_TYPEs in the comment, formatting fix.
25122 2017-04-27 Richard Biener <rguenther@suse.de>
25124 PR middle-end/80533
25125 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
25126 stripping ARRAY_REFs from MEM_EXPR make sure we're not
25127 keeping a reference to a trailing array.
25129 2017-04-27 Richard Biener <rguenther@suse.de>
25131 PR middle-end/80539
25132 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
25133 being in loop-closed SSA form conservatively.
25134 (chrec_fold_multiply_poly_poly): Likewise.
25136 2017-04-27 Tamar Christina <tamar.christina@arm.com>
25138 PR middle-end/79665
25139 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
25140 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
25142 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25145 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
25146 (aarch64_function_arg_alignment): Return unsigned int again, but still
25147 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
25148 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
25149 Don't emit -Wpsabi note.
25150 (aarch64_function_arg_boundary): Likewise.
25151 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
25154 2017-04-26 Nathan Sidwell <nathan@acm.org>
25156 * tree.h (crc32_unsigned_n): Declare.
25157 (crc32_unsigned, crc32_unsigned): Make inline.
25158 * tree.c (crc32_unsigned_bits): Replace with ...
25159 (crc32_unsigned_n): ... this.
25160 (crc32_unsigned, crc32_byte): Remove.
25161 (crc32_string): Remove unnecessary braces.
25163 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25165 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
25166 * ipa-inline-analysis.c (MAX_TIME): Remove.
25167 (account_size_time): Use sreal for time.
25168 (dump_inline_summary): Update.
25169 (estimate_function_body_sizes): Update.
25170 (estimate_edge_size_and_time): Update.
25171 (estimate_calls_size_and_time): Update.
25172 (estimate_node_size_and_time): Update.
25173 (inline_merge_summary): Update.
25174 (inline_update_overall_summary): Update.
25175 (estimate_time_after_inlining): Update.
25176 (inline_read_section): Update.
25177 (inline_write_summary): Update.
25178 * ipa-inline.c (compute_uninlined_call_time): Update.
25179 (compute_inlined_call_time): Update.
25180 (recursive_inlining): Update.
25181 (inline_small_functions): Update.
25182 (dump_overall_stats): Update.
25183 * ipa-inline.h: Include sreal.h.
25184 (size_time_entry): Turn time to sreal.
25185 (inline_summary): Turn self_time nad time to sreal.
25187 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25189 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
25191 (sreal::stream_out, sreal::stream_in): New.
25192 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
25194 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25196 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
25199 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
25202 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
25204 (dimode_scalar_chain::compute_convert_gain): Ditto.
25205 (dimode_scalar_chain::make_vector_copies): Ditto.
25206 (dimode_scalar_chain::convert_reg): Ditto.
25207 (dimode_scalar_chain::convert_insn): Ditto.
25208 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
25209 (VI248_AVX512BW_1): New mode iterator.
25210 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
25211 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
25214 2017-04-25 Martin Sebor <msebor@redhat.com>
25216 PR tree-optimization/80497
25217 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
25218 constants are representable in HOST_WIDE_INT.
25219 (parse_directive): Ditto.
25221 2017-04-25 Martin Sebor <msebor@redhat.com>
25224 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
25225 (new_zero_array): Adjust signature.
25226 (dom_info::dom_init): Used unsigned rather that size_t.
25227 (dom_info::dom_info): Same.
25229 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25230 Jakub Jelinek <jakub@redhat.com>
25233 * config/arm/arm.c: Include gimple.h.
25234 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
25235 returns negative, increment ncrn only if it returned positive.
25236 (arm_needs_doubleword_align): Return int instead of bool,
25237 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
25238 members, but if there is any such non-FIELD_DECL
25239 > PARM_BOUNDARY aligned decl, return -1 instead of false.
25240 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
25241 returns negative, increment nregs only if it returned positive.
25242 (arm_setup_incoming_varargs): Likewise.
25243 (arm_function_arg_boundary): Emit -Wpsabi note if
25244 arm_needs_doubleword_align returns negative, return
25245 DOUBLEWORD_ALIGNMENT only if it returned positive.
25247 2017-04-25 Marek Polacek <polacek@redhat.com>
25250 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
25251 first argument to type.
25253 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
25256 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
25257 type checks to test for compatibility instead of equality.
25259 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25260 Jakub Jelinek <jakub@redhat.com>
25263 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
25265 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
25266 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
25267 the alignment computation, but return their maximum in warn_alignment.
25268 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
25269 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
25271 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
25272 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
25275 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25277 * config/arc/simdext.md (dmpyh): Fix typo.
25279 2017-04-25 Richard Biener <rguenther@suse.de>
25281 PR tree-optimization/80492
25282 * alias.c (compare_base_decls): Handle registers with asm
25283 specification conservatively.
25284 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
25285 compare_base_decls returning dont-know properly.
25287 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25289 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
25290 (legitimate_offset_address_p): New function.
25291 (arc_legitimate_address_p): Use above function.
25293 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25295 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
25297 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25299 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
25300 ACCH registers whenever they are available.
25302 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25304 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
25305 double regs fix when not used.
25307 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25309 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
25311 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
25312 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
25314 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25316 * config/arc/arc.c (arc_output_addsi): Check for h-register class
25317 when emitting short ADD instructions.
25319 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25321 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
25323 (cmpsi_cc_c_insn): Likewise.
25324 (cbranchsi4_scratch): Compute proper instruction length using
25325 compact_hreg_operand.
25326 * config/arc/predicates.md (compact_hreg_operand): New predicate.
25328 2017-04-25 Richard Biener <rguenther@suse.de>
25330 PR middle-end/80509
25331 * passes.c (pass_manager::pass_manager): Initialize
25332 m_name_to_pass_map.
25334 2017-04-25 Richard Biener <rguenther@suse.de>
25336 PR tree-optimization/79201
25337 * tree-ssa-sink.c (statement_sink_location): Handle calls.
25339 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25342 * config/s390/vector.md: Split MEM->GPR vector moves for
25343 non-s_operand addresses.
25345 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25348 * config/s390/predicates.md (reload_const_wide_int_operand): New
25350 * config/s390/s390.md ("movti"): Remove d/P alternative.
25351 ("movti_bigconst"): New pattern definition.
25353 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25356 * s390-protos.h (s390_expand_cs_hqi): Removed.
25357 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
25358 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
25359 modes as well as CCZ1mode and CCZmode.
25360 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
25361 signature of s390_emit_compare_and_swap.
25362 (s390_expand_cs_hqi): Likewise, make static.
25363 (s390_expand_cs_tdsi): Generate an explicit compare before trying
25364 compare-and-swap, in some cases.
25365 (s390_expand_cs): Wrapper function.
25366 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
25368 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
25369 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
25370 patterns for small and large integers. Forbid symref memory operands.
25371 Move expander to s390.c. Require cc register.
25372 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
25373 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
25374 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
25375 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
25376 symref memory operands. Remove CC mode and call s390_match_ccmode
25378 ("atomic_exchange<mode>"): Allow and implement all integer modes.
25380 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25382 * config/s390/s390.md (define_peephole2): New peephole to help
25383 combining the load-and-test pattern with volatile memory.
25385 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25387 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
25388 with CCZmode for TARGET_Z196.
25390 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25392 PR rtl-optimization/80501
25393 * combine.c (make_compound_operation_int): Set subreg_code to SET
25394 even for AND with mask of the sign bit of mode.
25396 PR rtl-optimization/80500
25397 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
25398 sum's initial value.
25400 2017-04-25 Julian Brown <julian@codesourcery.com>
25401 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25403 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
25405 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
25407 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
25409 2017-04-25 Julian Brown <julian@codesourcery.com>
25410 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25412 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
25413 (thunderx2t99_sha): New Reservation.
25415 2017-04-25 Julian Brown <julian@codesourcery.com>
25416 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25418 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
25419 type for 1-element load.
25421 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
25423 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
25425 2017-04-24 Martin Jambor <mjambor@suse.cz>
25427 PR tree-optimization/80293
25428 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
25429 char arrays not totally scalarizable if it is false.
25430 (analyze_all_variable_accesses): Pass correct value in the new
25431 parameter. Add a statistics counter.
25433 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
25435 PR middle-end/79931
25436 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
25438 2017-04-24 Richard Biener <rguenther@suse.de>
25440 PR tree-optimization/80494
25441 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
25442 out for complex types.
25444 2017-04-24 Richard Biener <rguenther@suse.de>
25446 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
25447 * tree-ssa-sccvn.c (print_scc): Print SCC size.
25448 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
25449 (DFS): Adjust and never fail.
25450 (sccvn_dom_walker::fail): Remove.
25451 (sccvn_dom_walker::before_dom_children): Adjust.
25452 (run_scc_vn): Likewise and never fail.
25453 * tree-ssa-pre.c (pass_pre::execute): Adjust.
25454 (pass_fre::execute): Likewise.
25456 2017-04-24 Richard Biener <rguenther@suse.de>
25458 PR tree-optimization/79725
25459 * tree-ssa-sink.c (statement_sink_location): Return whether
25460 failure reason was zero uses. Move that check later.
25461 (sink_code_in_bb): Deal with zero uses by removing the stmt
25464 2017-04-24 Richard Biener <rguenther@suse.de>
25467 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
25468 pointer-based references.
25470 2017-04-24 Richard Biener <rguenther@suse.de>
25473 * pass_manager.h (pass_manager::operator new): Remove.
25474 (pass_manager::operator delete): Likewise.
25475 * passes.c (pass_manager::operator new): Remove.
25476 (pass_manager::operator delete): Likewise.
25477 (pass_manager::pass_manager): Zero individual pass members.
25479 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
25482 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
25483 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
25484 Check "XEXP (src, 1)" operand here.
25485 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
25486 Check "XEXP (src, 1)" operand here.
25487 (dimode_scalar_chain::make_vector_copies): Detect count register
25488 of a shift instruction. Zero extend count register from QImode
25489 to DImode to satisfy vector shift pattern count operand predicate.
25490 Substitute vector shift count operand with a DImode copy.
25491 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
25494 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
25496 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
25497 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
25498 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
25499 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
25500 (UNSPEC_NOREX_MEM): Remove definition.
25502 2017-04-21 Richard Biener <rguenther@suse.de>
25504 PR tree-optimization/79547
25505 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
25506 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
25507 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
25508 without any constraints.
25510 2017-04-21 Richard Biener <rguenther@suse.de>
25512 PR tree-optimization/78847
25513 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
25515 2017-04-21 Richard Biener <rguenther@suse.de>
25517 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
25518 (build_distinct_type_copy): Likewise.
25519 (build_variant_type_copy): Likewise.
25520 * tree.c (build_qualified_type): Pass down mem-stat info.
25521 (build_distinct_type_copy): Likewise.
25522 (build_variant_type_copy): Likewise.
25524 2017-04-21 Richard Biener <rguenther@suse.de>
25526 PR tree-optimization/80237
25527 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
25529 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
25530 for a simplified result.
25532 2016-04-21 Richard Biener <rguenther@suse.de>
25534 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
25535 sth as strict as a simple_iv but a chrec without symbols and an
25536 operand defined in the loop we are peeling (and not some subloop).
25537 (propagate_constants_for_unrolling): Propagate all constants.
25539 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25542 * config/i386/i386.c (print_reg): Remove assert for disalowed
25543 regno values, call output_operand_lossage instead.
25545 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25548 * config/i386/constraints.md (Yc): New register constraint.
25549 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
25550 Use Yc constraint for alternative 2 of operand 0. Remove
25551 preferred_for_speed attribute.
25553 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
25555 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
25556 lastprivate clauses in SIMT case.
25558 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
25560 * doc/invoke.texi (-Wextra-semi): Document new warning option.
25562 2017-04-20 Richard Biener <rguenther@suse.de>
25564 PR tree-optimization/57796
25565 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
25566 as N scalar stores.
25567 (vect_model_load_cost): Cost gathers as N scalar loads.
25569 2017-04-20 Richard Biener <rguenther@suse.de>
25571 * ggc-page.c (ggc_allocated_p): Rename to ...
25572 (safe_lookup_page_table_entry): ... this and return the lookup
25574 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
25576 2017-04-20 Richard Biener <rguenther@suse.de>
25578 PR tree-optimization/80453
25579 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
25580 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
25581 from the conditions.
25582 (vn_phi_eq): Pass them down.
25583 (vn_phi_lookup): Record them.
25584 (vn_phi_insert): Likewise.
25586 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
25588 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
25589 uninitialized variable warning to avoid buffer overrun.
25591 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
25594 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
25595 is suppressed for '{ 0 }' in C.
25597 2017-04-20 Jakub Jelinek <jakub@redhat.com>
25599 * BASE-VER: Set to 8.0.0.
25601 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
25603 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
25604 priority .init_array and .fini_array section with SECTION_NOTYPE
25607 2017-04-20 Jakub Jelinek <jakub@redhat.com>
25609 PR middle-end/80423
25610 * tree.h (build_array_type): Add typeless_storage default argument.
25611 * tree.c (type_cache_hasher::equal): Also compare
25612 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
25613 (build_array_type): Add typeless_storage argument, set
25614 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
25616 (build_nonshared_array_type): Adjust build_array_type_1 caller.
25617 (build_array_type): Likewise. Add typeless_storage argument.
25619 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
25620 Jakub Jelinek <jakub@redhat.com>
25622 PR tree-optimization/80426
25623 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
25624 operation on symbolic operands, also compute the overflow for the
25625 invariant part when the operation degenerates into a negation.
25627 2017-04-19 Jakub Jelinek <jakub@redhat.com>
25630 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
25631 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
25634 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
25636 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
25639 * config/avr/avr.c (tree.h): Include it.
25640 (cgraph.h): Include it.
25641 (avr_encode_section_info): Don't warn for uninitialized progmem
25642 variable if it's just an alias.
25644 2017-04-19 Richard Biener <rguenther@suse.de>
25647 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
25648 when needed by AutoPGO.
25650 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
25653 * doc/lto.texi: Remove an extra 'that'.
25655 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
25657 PR rtl-optimization/80429
25658 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
25659 are only used in debug insns.
25661 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
25662 Vladimir Makarov <vmakarov@redhat.com>
25664 * config/sparc/predicates.md (input_operand): Add comment. Return
25665 true for any memory operand when LRA is in progress.
25666 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
25668 2017-04-18 Jeff Law <law@redhat.com>
25671 * mips.md ({return,simple_return}_internal): Do not overwrite
25674 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25676 PR tree-optimization/80443
25677 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
25678 instead of adding 1, subtract -1 and similarly instead of subtracting
25681 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
25683 PR rtl-optimization/80357
25684 * haifa-sched.c (tmp_bitmap): New variable.
25685 (model_recompute): Handle duplicate use records.
25686 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
25687 (free_global_sched_pressure_data): Free it.
25689 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
25692 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
25693 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25694 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25695 instead of SYSTEM_HEADER_DIR.
25697 2017-04-18 Jeff Law <law@redhat.com>
25699 PR middle-end/80422
25700 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
25701 predecessors after walking up the insn chain.
25703 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25706 * dwarf2out.c (modified_type_die): Try harder not to emit internal
25707 sizetype type into debug info.
25709 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
25712 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
25713 unneeded test for TARGET_UPPER_REGS_SF.
25714 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
25716 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25719 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
25720 instead of gsi_after_labels.
25722 2017-04-18 Jeff Law <law@redhat.com>
25724 * regcprop.c (maybe_mode_change): Avoid creating copies of the
25728 2017-04-13 Jeff Law <law@redhat.com>
25729 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
25730 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
25732 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
25735 * config/avr/avr.c (intl.h): Include it.
25736 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
25738 2017-04-18 Martin Liska <mliska@suse.cz>
25740 PR gcov-profile/78783
25741 * gcov-tool.c (gcov_output_files): Validate that destination
25742 file is either removed by the tool or by a user.
25744 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
25745 Guy Benyei <guybe@mellanox.com>
25747 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
25748 block, and do not negate it, the stored id is already negative.
25750 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
25752 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
25754 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
25757 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
25758 masks of options that should be turned off if the VSX vector
25759 options are turned off.
25760 (OTHER_P8_VECTOR_MASKS): Likewise.
25761 (OTHER_VSX_VECTOR_MASKS): Likewise.
25762 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
25763 rs6000_disable_incompatible_switches to validate no type switches
25765 (rs6000_incompatible_switch): New function to disallow turning on
25766 other vector options if -mno-vsx, -mno-power8-vector, or
25767 -mno-power9-vector are specified.
25769 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
25771 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
25773 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
25775 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
25776 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
25777 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
25778 (ARG_POINTER_CFA_OFFSET): Likewise.
25780 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
25782 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
25783 conditions to take advantage of various optimizations.
25785 2017-04-13 Jeff Law <law@redhat.com>
25787 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
25788 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
25789 (zero_extendsidi2_dext): Likewise.
25791 2017-04-13 Jakub Jelinek <jakub@redhat.com>
25794 * fold-const.c (fold_ternary_loc): Revert
25795 use op0 instead of fold_convert_loc (loc, type, arg0) part of
25798 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
25800 PR rtl-optimization/80343
25801 * lra-remat.c (update_scratch_ops): Assign original hard reg to
25802 new scratch pseudo.
25804 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
25807 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
25808 to ubsan_encode_value.
25810 2017-04-13 Jeff Law <law@redhat.com>
25812 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
25813 appearing in DEBUG_INSNs.
25815 2017-04-13 Martin Liska <mliska@suse.cz>
25817 PR gcov-profile/80413
25818 * gcov-io.c (gcov_write_string): Copy to buffer just when
25819 allocated size is greater than zero.
25821 2017-04-13 Jakub Jelinek <jakub@redhat.com>
25824 * dwarf2out.c (decls_for_scope): Ignore declarations of
25825 current_function_decl in BLOCK_NONLOCALIZED_VARS.
25827 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
25830 * ipa-visibility.c (non_local_p): Fix typos.
25831 (localize_node): When localizing symbol in same comdat group,
25832 dissolve the group only when we know external symbols are going
25834 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
25836 2017-04-12 Jakub Jelinek <jakub@redhat.com>
25838 PR tree-optimization/79390
25839 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
25840 order does not result in usable sequence, retry with reversed operand
25846 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
25847 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
25848 op0 instead of fold_convert_loc (loc, type, arg0).
25850 2017-04-12 Jeff Law <law@redhat.com>
25852 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
25853 has a delay slot in the generated code.
25855 * config/cris/cris.md (cris_preferred_reload_class): Return
25856 GENNONACR_REGS rather than GENERAL_REGS.
25858 2017-04-12 Jakub Jelinek <jakub@redhat.com>
25861 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
25862 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
25863 signedness of the result type.
25865 2017-04-12 Richard Biener <rguenther@suse.de>
25866 Jeff Law <law@redhat.com>
25868 PR tree-optimization/80359
25869 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
25870 trim stores to TARGET_MEM_REFs.
25872 2017-04-12 Richard Biener <rguenther@suse.de>
25874 PR tree-optimization/79390
25875 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
25876 threading case even more.
25878 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
25881 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
25882 for quad_address_p for TImode, instead of just not indexed_address.
25884 2017-04-12 Richard Biener <rguenther@suse.de>
25885 Bernd Edlinger <bernd.edlinger@hotmail.de>
25887 PR middle-end/79671
25888 * alias.c (component_uses_parent_alias_set_from): Handle
25889 TYPE_TYPELESS_STORAGE.
25890 (get_alias_set): Likewise.
25891 * tree-core.h (tree_type_common): Add typeless_storage flag.
25892 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
25893 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
25894 for types containing members with TYPE_TYPELESS_STORAGE.
25895 (place_field): Likewise.
25896 (layout_type): Likewise for ARRAY_TYPE.
25897 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
25898 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
25899 TYPE_TYPELESS_STORAGE.
25900 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
25902 2017-04-12 Jakub Jelinek <jakub@redhat.com>
25905 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
25906 first argument to type.
25908 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25912 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
25913 CONST0_RTX (mode) rather than const0_rtx where appropriate.
25914 (rs6000_expand_binop_builtin): Likewise.
25915 (rs6000_expand_ternop_builtin): Likewise; also add missing
25916 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
25917 vshasigma built-ins.
25918 * doc/extend.texi: Document that vec_xxpermdi's third argument
25919 must be a constant.
25921 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
25923 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
25924 Use shift_const cost parameter when calculating gain of STV shifts.
25926 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
25928 PR rtl-optimization/70478
25929 * lra-constraints.c (process_alt_operands): Check memory for
25930 disfavoring memory insn operand.
25932 2017-04-11 Jakub Jelinek <jakub@redhat.com>
25934 PR middle-end/80100
25935 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
25936 left shift in unsigned HOST_WIDE_INT type.
25938 PR rtl-optimization/80385
25939 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
25940 (not (neg X)) into (plus X -1) for complex or non-integral modes.
25943 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
25944 if they have any depend clauses.
25946 2017-04-11 Martin Liska <mliska@suse.cz>
25949 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
25950 * ipa-split.c (split_function): Create a local comdat symbol
25951 if caller is in a comdat group.
25953 2017-04-11 Martin Liska <mliska@suse.cz>
25956 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
25959 2017-04-11 Martin Sebor <msebor@redhat.com>
25961 PR middle-end/80364
25962 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
25963 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
25965 (directive::set_width, directive::set_precision, format_character):
25967 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
25970 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
25973 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
25974 conflict, set target->arch_name instead of target->cpu_name.
25976 2017-04-11 Richard Biener <rguenther@suse.de>
25978 PR tree-optimization/80374
25979 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
25980 build_zero_cst, remove fold_convertible_p check again.
25982 2017-04-11 Martin Liska <mliska@suse.cz>
25985 * ubsan.c (instrument_object_size): Do not instrument register
25988 2017-04-11 Jakub Jelinek <jakub@redhat.com>
25991 * config/i386/i386-builtin-types.def
25992 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
25993 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
25994 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
25995 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
25996 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
25997 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
25998 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
25999 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
26000 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
26001 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
26002 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
26003 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
26004 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
26005 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
26006 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
26007 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
26008 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
26009 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
26010 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
26011 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
26012 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
26013 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
26014 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
26015 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
26016 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
26017 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
26018 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
26019 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
26020 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
26021 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
26022 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
26023 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
26024 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
26025 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
26026 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
26027 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
26028 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
26029 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
26030 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
26031 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
26032 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
26033 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
26034 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
26035 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
26036 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
26037 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
26039 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
26040 flag to second_arg_count, handle 4 argument function type _COUNT
26041 aliases, handle second_arg_count on second argument rather than last.
26043 2017-04-10 Jeff Law <law@redhat.com>
26045 PR tree-optimization/80374
26046 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
26047 record anything if we can not convert integer_zero_node to the
26050 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
26053 * config/rs6000/rs6000.c (rs6000_option_override_internal):
26054 Enhance special handling given to the TARGET_P9_MINMAX option in
26055 relation to certain other options.
26057 2017-04-10 Bin Cheng <bin.cheng@arm.com>
26059 PR tree-optimization/80153
26060 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
26061 remove POINTER_PLUS_EXPR's base part directly, rather than through
26064 2017-04-10 Richard Biener <rguenther@suse.de>
26065 Bin Cheng <bin.cheng@arm.com>
26067 PR tree-optimization/80153
26068 * tree-affine.c (aff_combination_to_tree): Get base pointer from
26069 the first element of pointer type aff_tree. Build result expr in
26071 (add_elt_to_tree): Convert to type unconditionally. Remove other
26072 fold_convert calls.
26073 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
26074 (rewrite_use_nonlinear_expr): Check invariant using iv information.
26076 2017-04-10 Richard Biener <rguenther@suse.de>
26078 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26081 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
26083 PR rtl-optimization/70478
26084 * lra-constraints.c (curr_small_class_check): New.
26085 (update_and_check_small_class_inputs): New.
26086 (process_alt_operands): Update curr_small_class_check. Disfavor
26087 alternative insn memory operands. Check available regs for small
26090 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26093 * config/mips/mips.opt (-mvirt): Update description.
26094 * doc/invoke.texi (-mvirt): Likewise.
26096 2017-04-10 Richard Biener <rguenther@suse.de>
26098 PR middle-end/80362
26099 * fold-const.c (fold_binary_loc): Look at unstripped ops when
26100 looking for NEGATE_EXPR in -A / -B to A / B folding.
26102 2017-04-10 Martin Liska <mliska@suse.cz>
26104 PR gcov-profile/80224
26105 * gcov.c (print_usage): Fix usage string.
26106 (get_gcov_intermediate_filename): Remove.
26107 (output_gcov_file): Use both for normal and intermediate format.
26108 (generate_results): Do not initialize special file for
26109 intermediate format.
26111 2017-04-10 Richard Biener <rguenther@suse.de>
26113 PR tree-optimization/80304
26114 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
26117 2017-04-10 Nathan Sidwell <nathan@acm.org>
26120 * config/rs6000/rs6000.c (rs6000_vector_type): New.
26121 (rs6000_init_builtins): Use it.
26123 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26125 * config/arm/arm.md (<mrc>): Add mode to SET source.
26126 (<mrrc>): Likewise.
26128 2017-04-10 Richard Biener <rguenther@suse.de>
26130 PR middle-end/80344
26131 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
26133 2017-04-10 Jakub Jelinek <jakub@redhat.com>
26136 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
26137 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
26138 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
26139 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
26140 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
26141 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
26142 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
26143 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
26144 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
26145 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
26146 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
26147 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
26148 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
26149 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
26150 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
26151 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
26152 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
26153 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
26154 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
26155 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
26156 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
26157 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
26158 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
26160 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
26162 PR rtl-optimization/70478
26163 * lra-constraints.c: Reverse the last patch.
26165 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
26167 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
26168 Add comment for WCHAR_T.
26170 2017-04-08 Martin Liska <mliska@suse.cz>
26173 2017-04-07 Martin Liska <mliska@suse.cz>
26176 * ipa-split.c (split_function): Add function part to a same comdat
26179 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26182 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
26184 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
26186 * rs6000/rs6000.c (vec_load_pendulum): Rename...
26187 (vec_pairing): ...to this.
26188 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
26189 (rs6000_sched_init): Adjust for name change.
26190 (struct rs6000_sched_context): Likewise.
26191 (rs6000_init_sched_context): Likewise.
26192 (rs6000_set_sched_context): Likewise.
26194 2017-04-07 Jakub Jelinek <jakub@redhat.com>
26200 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
26202 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
26203 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
26204 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
26206 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
26208 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
26210 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
26212 PR rtl-optimization/70703
26213 * ira-color.c (update_conflict_hard_regno_costs): Use
26214 int64_t instead of HOST_WIDE_INT.
26216 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
26218 PR rtl-optimization/70478
26219 * lra-constraints.c (process_alt_operands): Disfavor alternative
26220 insn memory operands.
26222 2017-04-07 Jeff Law <law@redhat.com>
26224 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
26225 CALL and NOTE_INSN_CALL_ARG_LOCATION.
26227 2017-04-07 Martin Liska <mliska@suse.cz>
26230 * config/aarch64/aarch64.c (aarch64_process_target_attr):
26231 Show error message instead of an ICE.
26233 2017-04-07 Martin Liska <mliska@suse.cz>
26236 * ipa-split.c (split_function): Add function part to a same comdat
26239 2017-04-07 Richard Biener <rguenther@suse.de>
26241 PR middle-end/80341
26242 * tree.c (get_unwidened): Also handle ! for_type case for
26244 * convert.c (do_narrow): Split out from ...
26245 (convert_to_integer_1): ... here. Do not pass final truncation
26246 type to get_unwidened for TRUNC_DIV_EXPR.
26248 2017-04-07 Richard Biener <rguenther@suse.de>
26250 * tree-affine.c (wide_int_ext_for_comb): Take type rather
26252 (aff_combination_const): Adjust.
26253 (aff_combination_scale): Likewise.
26254 (aff_combination_add_elt): Likewise.
26255 (aff_combination_add_cst): Likewise.
26256 (aff_combination_convert): Likewise.
26257 (add_elt_to_tree): Likewise. Remove unused argument.
26258 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
26260 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
26262 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
26264 * config/arm/arm.c (arm_default_short_enums): Use
26265 ARM_DEFAULT_SHORT_ENUMS.
26266 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
26268 2017-04-06 Jakub Jelinek <jakub@redhat.com>
26271 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
26272 members with redundant out-of-class redeclaration.
26274 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
26277 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
26278 * config/i386/i386.md (*zero_extendsidi2):
26279 Add (?*x,*x) and (?*v,*v) alternatives.
26281 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
26284 * config/i386/i386.c (ix86_expand_builtin)
26285 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
26286 mode from insn data. Convert operands to insn operand mode.
26287 Copy operands that don't satisfy insn predicate to a register.
26289 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
26291 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
26294 2017-04-06 Richard Biener <rguenther@suse.de>
26296 PR tree-optimization/80334
26297 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
26298 preserve alignment of accesses.
26300 2017-04-06 Richard Biener <rguenther@suse.de>
26302 PR tree-optimization/80262
26303 * tree-sra.c (build_ref_for_offset): Preserve address-space
26305 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
26306 Drop useless address-space information on MEM_REF offsets.
26308 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
26310 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
26312 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
26314 PR rtl-optimization/70703
26315 * ira-color.c (update_conflict_hard_regno_costs): Use
26316 HOST_WIDE_INT instead of long.
26318 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
26321 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
26322 not defined for x86_64 target. Add -mmmx target option when __SSE2__
26324 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
26325 for x86_64 target. Handle -m3dnowa option.
26327 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
26329 PR rtl-optimization/70703
26330 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
26331 (update_conflict_hard_regno_costs): Use long instead of unsigned
26332 arithmetic for cost calculation.
26334 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26335 Bernd Edlinger <bernd.edlinger@hotmail.de>
26338 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
26341 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
26344 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
26345 ptr_mode with Pmode throughout.
26346 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
26347 into probe_stack_range and use DImode.
26349 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
26352 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
26353 call_eh_return is true.
26355 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26357 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
26358 Initialize last_match_fntype_index.
26360 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26363 * tree-nvr.c: Include internal-fn.h.
26364 (pass_return_slot::execute): Ignore internal calls without
26367 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26368 Richard Biener <rguenther@suse.de>
26371 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
26372 captures used multiple times, except for the last use.
26373 * generic-match-head.c: Include gimplify.h.
26375 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26377 PR tree-optimization/79390
26378 * target.h (struct noce_if_info): Declare.
26379 * targhooks.h (default_noce_conversion_profitable_p): Declare.
26380 * target.def (noce_conversion_profitable_p): New target hook.
26381 * ifcvt.h (struct noce_if_info): New type, moved from ...
26382 * ifcvt.c (struct noce_if_info): ... here.
26383 (noce_conversion_profitable_p): Renamed to ...
26384 (default_noce_conversion_profitable_p): ... this. No longer
26386 (noce_try_store_flag_constants, noce_try_addcc,
26387 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
26388 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
26389 instead of noce_conversion_profitable_p.
26390 * config/i386/i386.c: Include ifcvt.h.
26391 (ix86_option_override_internal): Don't override
26392 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
26393 (ix86_noce_conversion_profitable_p): New function.
26394 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
26395 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
26396 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
26397 * doc/tm.texi: Regenerated.
26399 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26401 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
26404 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26407 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
26408 instructions for small multiply cores.
26410 2017-04-04 Jeff Law <law@redhat.com>
26412 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
26414 (mips_expand_vec_perm_const): Initialize elements in orig_perm
26415 that are not set by the loop over the elements.
26417 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26420 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
26421 int mode, convert_modes it to mode as unsigned, otherwise use
26422 lowpart_subreg to mode rather than SImode.
26423 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
26424 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
26425 Use DImode instead of SImode for the shift count operand.
26426 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
26429 2017-04-04 Richard Biener <rguenther@suse.de>
26431 PR middle-end/80281
26432 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
26433 arithmetic done for the negate or the plus. Simplify.
26434 (A - (-B) -> A + B): Likewise.
26435 * fold-const.c (split_tree): Make sure to not negate pointers.
26437 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
26439 PR rtl-optimization/60818
26440 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
26441 a compare of comparisons with the thing compared if this results
26442 in a different machine mode.
26444 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
26446 * alias.c (base_alias_check): Fix typo in comment.
26447 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
26448 * cgraphunit.c (symbol_table::compile): Likewise.
26449 * collect2.c (maybe_run_lto_and_relink): Likewise.
26450 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
26451 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
26452 * config/avr/avr.c (avr_map_op_t): Likewise.
26453 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
26454 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
26455 * config/epiphany/epiphany.md (movcc): Likewise.
26456 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
26457 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
26459 * config/mips/mips.c (mips_save_restore_reg): Likewise.
26460 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
26461 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
26462 * config/sh/sh.c (sh_rtx_costs): Likewise.
26463 * fold-const.c (fold_truth_andor): Likewise.
26464 * genautomata.c (collapse_flag): Likewise.
26465 * gengtype.h (struct type::u::s): Likewise.
26466 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
26467 * input.c (FORMAT_AMOUNT): Likewise.
26468 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
26469 (known_aggs_to_agg_replacement_list): Likewise.
26470 * ipa-inline-analysis.c: Likewise.
26471 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
26472 * ipa-polymorphic-call.c
26473 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
26474 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
26475 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
26477 * modulo-sched.c (apply_reg_moves): Likewise.
26478 * omp-expand.c (build_omp_regions_1): Likewise.
26479 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
26480 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
26481 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
26482 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
26483 * value-prof.c: Likewise.
26484 * var-tracking.c (val_reset): Likewise.
26486 2017-04-03 Richard Biener <rguenther@suse.de>
26488 PR tree-optimization/80275
26489 * fold-const.c (split_address_to_core_and_offset): Handle
26492 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
26494 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
26495 descriptors is at least equal to that of functions.
26497 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
26499 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
26501 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
26504 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
26505 (mov<IMOD4:mode>): New expander.
26506 (*mov<IMOD4:mode>_internal): New insn and split pattern.
26508 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
26510 PR rtl-optimization/79405
26511 * fwprop.c (propagations_left): New variable.
26512 (forward_propagate_into): Decrement it.
26513 (fwprop_init): Initialize it.
26514 (fw_prop): If the variable has reached zero, stop propagating.
26515 (fwprop_addr): Ditto.
26517 2017-03-31 Jakub Jelinek <jakub@redhat.com>
26520 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
26521 a FUNCTION_DECL, pass it as decl instead of origin to
26524 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
26526 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
26529 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
26532 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
26533 TARGET_VSX_SMALL_INTEGER.
26535 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26537 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
26538 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
26540 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26542 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
26543 extraction from odd-numbered MSA register.
26545 2017-03-31 Jakub Jelinek <jakub@redhat.com>
26547 PR middle-end/80173
26548 * expmed.c (store_bit_field_1): Don't attempt to create
26549 a word subreg out of hard registers wider than word if they
26550 have HARD_REGNO_NREGS of 1 for their mode.
26552 PR middle-end/80163
26553 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
26554 conversions to integer types wider than word and pointer.
26557 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
26558 (rtx_equal_for_cselib_p): Pass 0 to it.
26559 * cselib.c (cselib_hasher::equal): Likewise.
26560 (rtx_equal_for_cselib_1): Add depth argument. If depth
26561 is 128, don't look up VALUE locs and punt. Increment
26562 depth in recursive calls when walking VALUE locs.
26564 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
26566 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
26567 (make_gcov_file_name): Use the canonical path name for generating
26569 (read_line): Fix handling of files with ascii null bytes.
26571 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
26573 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
26574 to initialise a vector register instead
26575 of using a const_int.
26577 2017-03-30 Jakub Jelinek <jakub@redhat.com>
26579 PR translation/80189
26580 * gimplify.c (omp_default_clause): Use %qs instead of %s in
26581 diagnostic messages.
26583 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
26586 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
26587 (dfp_diex_<mode>): Update mode of operand 1.
26588 * doc/extend.texi (dxex, dxexq): Document change to return type.
26589 (diex, diexq): Document change to argument type.
26591 2017-03-30 Martin Jambor <mjambor@suse.cz>
26594 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
26595 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
26596 it reflects the signature changes performed at the callee side.
26597 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
26598 to cgraph_build_function_type_skip_args.
26599 (build_function_decl_skip_args): Adjust call to the above function.
26601 2017-03-30 Jakub Jelinek <jakub@redhat.com>
26604 * config/i386/sse.md
26605 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
26606 register as dest whenever it is a MEM not rtx_equal_p to the
26607 corresponding dup operand, and when forcing into reg move the
26608 reg into the memory afterwards.
26609 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
26610 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
26611 for the force_reg mode.
26612 (avx512vl_vextractf128<mode>): Use register as dest either
26613 always when a MEM, or when it is a MEM not rtx_equal_p to the
26614 corresponding dup operand, or even not when it is a CONST_VECTOR
26615 depending on the mode and lo vs. hi.
26616 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
26618 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
26619 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
26620 Likewise. Require that operands[2] is even.
26621 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
26622 Remove extraneous parens. Require that operands[2] is a multiple
26624 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
26625 operands[0] is a MEM if <mask_applied>, the predicates/constraints
26626 disallow memory then.
26628 2017-03-30 Richard Biener <rguenther@suse.de>
26630 PR tree-optimization/77498
26631 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
26632 to non-constants over backedges.
26634 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
26636 PR rtl-optimization/80233
26637 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
26638 as last_combined_insn. Do not test for BARRIER_P separately.
26640 2017-03-29 Andreas Schwab <schwab@suse.de>
26643 * calls.c (prepare_call_address): Convert funexp to Pmode before
26644 copying to temp reg.
26646 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26648 PR tree-optimization/80158
26649 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
26650 Handle possible future case of more than one alternate
26652 (replace_rhs_if_not_dup): Likewise.
26653 (replace_one_candidate): Likewise.
26655 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
26657 PR rtl-optimization/80193
26658 * ira.c (ira): Do not check allocation for LRA.
26660 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
26662 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
26663 (nvptx_output_simt_exit): Declare.
26664 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
26665 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
26666 (init_softstack_frame): Move initialization of crtl->is_leaf to...
26667 (nvptx_declare_function_name): ...here. Emit declaration of local
26668 memory space buffer for omp_simt_enter insn.
26669 (nvptx_output_unisimt_switch): New.
26670 (nvptx_output_softstack_switch): New.
26671 (nvptx_output_simt_enter): New.
26672 (nvptx_output_simt_exit): New.
26673 * config/nvptx/nvptx.h (struct machine_function): New fields
26674 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
26675 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
26676 (UNSPECV_SIMT_EXIT): Ditto.
26677 (omp_simt_enter_insn): New insn.
26678 (omp_simt_enter): New expansion.
26679 (omp_simt_exit): New insn.
26680 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
26682 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
26683 (expand_GOMP_SIMT_ENTER_ALLOC): New.
26684 (expand_GOMP_SIMT_EXIT): New.
26685 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
26686 (GOMP_SIMT_ENTER_ALLOC): Ditto.
26687 (GOMP_SIMT_EXIT): Ditto.
26688 * target-insns.def (omp_simt_enter): New insn.
26689 (omp_simt_exit): Ditto.
26690 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
26692 (lower_rec_simd_input_clauses): Implement SIMT privatization.
26693 (lower_rec_input_clauses): Likewise.
26694 (lower_lastprivate_clauses): Handle SIMT privatization.
26696 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
26697 (ompdevlow_adjust_simt_enter): New.
26698 (find_simtpriv_var_op): New.
26699 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
26700 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
26702 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
26703 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
26704 (copy_decl_for_dup_finish): Ditto.
26706 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
26708 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
26711 * config/i386/i386.c (ix86_option_override_internal): Always
26712 allow -mpreferred-stack-boundary=3 for 64-bit targets.
26714 2017-03-28 Bin Cheng <bin.cheng@arm.com>
26716 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
26718 2017-03-28 Bin Cheng <bin.cheng@arm.com>
26720 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
26721 mark new edge's irreducible flag accordign to it.
26722 (vect_do_peeling): Check loop preheader edge's irreducible flag
26723 and pass it to function slpeel_add_loop_guard.
26725 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
26727 PR tree-optimization/80218
26728 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
26729 Update block frequencies and counts.
26731 2017-03-28 Richard Biener <rguenther@suse.de>
26733 PR tree-optimization/78644
26734 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
26735 of a simplification result we may not use it at all.
26737 2017-03-28 Richard Biener <rguenther@suse.de>
26740 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
26741 without arguments, generate default definition of a SSA name.
26743 2017-03-28 Richard Biener <rguenther@suse.de>
26745 PR middle-end/80222
26746 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
26747 TYPE_REF_CAN_ALIAS_ALL references.
26748 * fold-const.c (fold_indirect_ref_1): Likewise.
26750 2017-03-28 Martin Liska <mliska@suse.cz>
26753 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
26754 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
26756 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
26757 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
26759 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
26760 (EXTRA_SPECS): Define.
26761 (SUBTARGET_EXTRA_SPECS): Likewise.
26762 (SUBTARGET_CPP_SPEC): Likewise.
26763 * config/arc/elf.h (EXTRA_SPECS): Renamed to
26764 SUBTARGET_EXTRA_SPECS.
26765 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
26767 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
26769 * config/arc/simdext.md (vst64_insn): Update pattern.
26770 (vld32wh_insn): Likewise.
26771 (vld32wl_insn): Likewise.
26772 (vld64_insn): Likewise.
26773 (vld32_insn): Likewise.
26775 2017-03-28 Marek Polacek <polacek@redhat.com>
26778 * fold-const.c (fold_comparison): Use protected_set_expr_location
26779 instead of SET_EXPR_LOCATION.
26781 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
26783 * tree.c (add_expr): Avoid name lookup warning.
26785 2017-03-27 Jeff Law <law@redhat.com>
26787 PR tree-optimization/80216
26788 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
26789 function name. Limit recursion depth.
26790 (record_temporary_equivalences): Corresponding changes.
26792 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
26794 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
26797 2017-03-27 Jakub Jelinek <jakub@redhat.com>
26800 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
26802 * cfgcleanup.c (reg_note_cfa_p): New array.
26803 (insns_have_identical_cfa_notes): New function.
26804 (old_insns_match_p): Don't cross-jump in between /f
26805 and non-/f instructions. If both i1 and i2 are frame related,
26806 verify all CFA notes, their order and content.
26808 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
26811 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
26812 HImode and SImode with zero extend to DImode to one insn.
26813 (bswap<mode>2_extenddi): Likewise.
26814 (bswapsi2_extenddi): Likewise.
26815 (bswaphi2_extendsi): Likewise.
26816 (bswaphi2): Combine bswap HImode and SImode into one insn.
26817 Separate memory insns from swapping register.
26818 (bswapsi2): Likewise.
26819 (bswap<mode>2): Likewise.
26820 (bswaphi2_internal): Delete, no longer used.
26821 (bswapsi2_internal): Likewise.
26822 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
26823 store, and gpr<-gpr swap insns.
26824 (bswap<mode>2_store): Likewise.
26825 (bswaphi2_reg): Register only splitter, combine with the splitter.
26826 (bswaphi2 splitter): Likewise.
26827 (bswapsi2_reg): Likewise.
26828 (bswapsi2 splitter): Likewise.
26829 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
26830 the insns into load, store, and register/register insns.
26831 (bswapdi2_ldbrx): Likewise.
26832 (bswapdi2_load): Likewise.
26833 (bswapdi2_store): Likewise.
26834 (bswapdi2_reg): Likewise.
26836 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
26838 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
26839 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
26841 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
26844 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
26846 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
26847 special handling for target option conflicts between dform
26848 options (-mpower9-dform, -mpower9-dform-vector,
26849 -mpower9-dform-scalar) and -mno-direct-move.
26851 2017-03-27 Richard Biener <rguenther@suse.de>
26853 PR tree-optimization/80181
26854 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
26856 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
26858 * config/arc/predicates.md (move_double_src_operand): Replace the
26859 call to move_double_src_operand with a call to address_operand.
26861 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
26863 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
26864 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
26865 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
26867 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
26869 * config/arc/predicates.md (long_immediate_loadstore_operand):
26870 Consider scaled addresses cases.
26872 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
26874 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
26875 restored when in interrupt.
26876 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
26877 doesn't have delay slot.
26879 2017-03-27 Richard Biener <rguenther@suse.de>
26882 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
26883 inlined thunk clones.
26885 2017-03-27 Jakub Jelinek <jakub@redhat.com>
26888 * asan.c (instrument_derefs): Copy over last operand from
26889 original COMPONENT_REF to the new COMPONENT_REF with
26890 DECL_BIT_FIELD_REPRESENTATIVE.
26891 * ubsan.c (instrument_object_size): Likewise.
26893 2017-03-27 Richard Biener <rguenther@suse.de>
26895 PR tree-optimization/80170
26896 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
26897 sure DR/SCEV didnt fold in constants we do not see when looking
26898 at the reference base alignment.
26900 2017-03-27 Richard Biener <rguenther@suse.de>
26902 PR middle-end/80171
26903 * gimple-fold.c (fold_ctor_reference): Properly guard against
26904 NULL return value from canonicalize_constructor_val.
26906 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
26909 * config/i386/i386.c (ix86_expand_builtin)
26910 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
26911 flags reg setting and flags reg using instructions.
26912 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
26913 clobbering instructions to zero extend op2.
26915 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
26917 * doc/install.texi (Configuration) <--with-aix-soname>:
26918 Update link to AIX ld.
26920 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
26922 PR rtl-optimization/80160
26923 PR rtl-optimization/80159
26924 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
26925 reg_alternate_class into account.
26927 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
26930 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
26931 to consider in curr_insn_transform.
26933 2017-03-24 Jakub Jelinek <jakub@redhat.com>
26935 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
26936 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
26937 and emit_mode_inner.
26939 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26941 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
26942 argument to the overloaded builtin variants. Use the new flag to
26943 deprecate certain builtin variants.
26944 * config/s390/s390-builtin-types.def: Add new builtin types.
26945 * config/s390/s390-builtins.h: Support new flags field for
26946 overloaded builtins.
26947 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
26948 (s390_macro_to_expand): Enable vector float data type.
26949 (s390_cpu_cpp_builtins_internal): Indicate support of the new
26950 builtins by incrementing the __VEC__ version number.
26951 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
26953 (s390_resolve_overloaded_builtin): Emit error messages depending
26954 on the builtin flags.
26955 * config/s390/s390.c (s390_expand_builtin): Support additional
26956 flags argument. Change error message to match the messages
26957 emitted in s390-c.c.
26958 * config/s390/s390.md: New UNSPEC_* constants.
26959 (op_type): Add new instruction types.
26960 * config/s390/vecintrin.h: Add new builtins and test data class
26962 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
26963 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
26964 (VEC_INEXACT, VEC_NOINEXACT): New constants.
26965 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
26966 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
26967 ("vec_mergel<mode>"): V_HW -> VEC_HW.
26969 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
26970 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
26971 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
26972 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
26974 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
26975 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
26976 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
26977 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
26979 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
26980 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
26981 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
26982 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
26983 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
26984 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
26985 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
26987 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
26988 ("vec_scatter_element<V_HW_4:mode>_DI")
26989 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
26990 ("vec_fpint<mode>", "vflls")
26991 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
26992 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
26993 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
26994 ("*vec_cmphe<mode>_cc"): ... these.
26996 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
26997 mode constant instead of magic value.
26999 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27001 * config/s390/s390.c (s390_expand_vec_compare): Support other
27002 vector floating point modes than just V2DF.
27003 (s390_expand_vcond): Likewise.
27004 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
27005 (s390_cannot_change_mode_class): Prevent mode changes between TF
27006 and V1TF in vector registers.
27007 * config/s390/s390.md (DF, SF): New mode attributes.
27008 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
27009 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
27010 SFmode support for VRs.
27011 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
27013 (VFT, VF_HW): New mode iterators.
27014 (vw, sdx): New mode attributes.
27015 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
27016 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
27017 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
27018 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
27019 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
27020 also the new vector floating point modes. Renaming to ...
27022 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
27023 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
27024 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
27025 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
27026 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
27027 ("vec_unordered<mode>"): ... these.
27029 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
27030 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
27031 ("*vec_extendv2df"): New insn definitions.
27033 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27035 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
27036 ("mulditi3_2", "*muldi3_sign"): New patterns.
27037 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
27038 rename the pattern definition.
27040 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27042 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
27044 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
27046 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27048 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
27049 instruction if possible.
27050 * config/s390/vector.md (vec_halfnumelts): New mode
27052 ("*vec_vllezlf<mode>"): New pattern.
27054 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27056 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
27057 ("popcountv4si2", "popcountv2di2"): Rename to ...
27058 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
27059 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
27061 ("popcount<mode>2_vxe"): New pattern.
27063 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27065 * common/config/s390/s390-common.c (processor_flags_table): Add
27067 * config.gcc: Add arch12.
27068 * config/s390/driver-native.c (s390_host_detect_local_cpu):
27069 Default to arch12 for unknown CPU model numbers.
27070 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
27071 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
27072 PROCESSOR_max sanity check.
27073 * config/s390/s390-opts.h (enum processor_type): Add
27075 * config/s390/s390.c (processor_table): Add arch12.
27076 (s390_expand_builtin): Add check for B_VXE flag.
27077 (s390_issue_rate): Add PROCESSOR_ARCH12.
27078 (s390_get_sched_attrmask): Likewise.
27079 (s390_get_unit_mask): Likewise.
27080 (s390_sched_score): Enable z13 scheduling for arch12.
27081 (s390_sched_reorder): Likewise.
27082 (s390_sched_variable_issue): Likewise.
27083 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
27085 (s390_tune_attr): Use z13 scheduling also for arch12.
27086 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
27087 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
27088 (TARGET_VXE_P): New macros.
27089 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
27090 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
27091 * config/s390/s390.opt: Add arch12 as processor_type.
27093 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27095 * config/s390/s390.md
27096 ("fixuns_truncdddi2", "fixuns_trunctddi2")
27097 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
27098 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
27100 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
27101 Rename expanders to ...
27103 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
27104 ("fixuns_truncdddi2_emu"): ... these.
27106 ("fixuns_trunc<mode>si2_emu"): New expander.
27108 ("*fixuns_truncdfdi2_z13"): Rename to ...
27109 ("*fixuns_truncdfdi2_vx"): ... this.
27111 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27113 * config/s390/2964.md: Remove the single element vector compare
27114 instructions which are no longer used.
27115 * config/s390/s390.c (s390_select_ccmode): Remove handling of
27117 (s390_canonicalize_comparison): Remove handling of DFmode
27119 (s390_expand_vec_compare_scalar): Remove function.
27120 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
27121 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
27123 ("*cmp<mode>_ccs"): Add wfcdb instruction.
27125 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27127 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
27129 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
27130 will anyway by matched by mov<mode>_64dfp.
27132 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27134 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
27135 vlef/vstef. Add missing operand to vleif.
27137 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27139 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
27140 pair for all vector types with 64 bit elements.
27141 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
27142 * config/s390/vector.md (V_HW_64): ... here.
27143 (V_128_NOSINGLE): New mode iterator.
27144 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
27145 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
27146 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
27147 ("*vec_load_pairv2di"): Change to ...
27148 ("*vec_load_pair<mode>"): ... this one.
27150 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27152 * config/s390/constraints.md: Add comments.
27153 (jKK): Reject element sizes > 8 bytes.
27154 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
27156 * config/s390/s390.md: Add the s_operand checks formerly in
27157 s390_split_ok_p to various splitters where they are still
27159 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
27160 for 128 bit vectors. Plus two splitters.
27162 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27164 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
27167 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27170 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
27171 error if the boundary argument is not constant.
27173 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27175 PR rtl-optimization/80112
27176 * loop-doloop.c (doloop_condition_get): Don't check condition
27177 if cmp isn't SET with IF_THEN_ELSE src.
27179 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27181 PR tree-optimization/80158
27182 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
27183 replacing a candidate statement, also replace it for the
27184 candidate's alternate interpretation.
27185 (replace_rhs_if_not_dup): Likewise.
27186 (replace_one_candidate): Likewise.
27188 2017-03-24 Richard Biener <rguenther@suse.de>
27190 PR tree-optimization/80167
27191 * graphite-isl-ast-to-gimple.c
27192 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
27194 (translate_isl_ast_to_gimple::get_rename): Likewise.
27196 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
27198 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
27199 handling of certain combinations of target options, including the
27200 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
27201 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
27203 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27206 * config/arm/arm.md (*load_multiple): Add reload_completed to
27207 matching condition.
27209 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27210 Richard Biener <rguenth@suse.de>
27212 PR tree-optimization/79908
27213 PR tree-optimization/80136
27214 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27215 been cast away, gimplify_and_add suffices.
27217 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
27219 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
27221 2017-03-23 Richard Biener <rguenther@suse.de>
27223 PR tree-optimization/80032
27224 * gimplify.c (gimple_push_cleanup): Forced unconditional
27225 cleanups still have to go to the conditional_cleanups
27228 2017-03-22 Jakub Jelinek <jakub@redhat.com>
27230 PR tree-optimization/80072
27231 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
27233 (next_operand_entry_id): Change type to unsigned int.
27234 (sort_by_operand_rank): Make sure to return the right return value
27235 even if unsigned fields are bigger than INT_MAX.
27236 (struct oecount): Change cnt and id type to unsigned int.
27237 (oecount_hasher::equal): Formatting fix.
27238 (oecount_cmp): Make sure to return the right return value
27239 even if unsigned fields are bigger than INT_MAX.
27240 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
27243 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
27244 TREE_READONLY on result if writing it more than once.
27247 * doc/invoke.texi (-fsanitize=thread): Document that with
27248 -fnon-call-exceptions atomics are not able to throw
27252 * tsan.c: Include tree-eh.h.
27253 (instrument_builtin_call): Call maybe_clean_eh_stmt or
27254 maybe_clean_or_replace_eh_stmt where needed.
27255 (instrument_memory_accesses): Add cfg_changed argument.
27256 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
27257 if it returned true.
27258 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
27260 PR rtl-optimization/63191
27261 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
27262 wrapper function, moved the whole old content into ...
27263 (ix86_delegitimize_address_1): ... this. New inline function.
27264 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
27265 true as last argument instead of ix86_delegitimize_address.
27267 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
27269 * config/aarch64/aarch64.c (generic_branch_cost): Copy
27270 cortexa57_branch_cost.
27272 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
27274 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
27276 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27279 * doc/md.texi (Constraints): Document wA constraint.
27280 * config/rs6000/constraints.md (wA): New.
27281 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
27282 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
27283 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
27284 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
27286 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
27289 * gimplify.c (is_oacc_declared): New function.
27290 (oacc_default_clause): Use it to set default flags for acc declared
27291 variables inside parallel regions.
27292 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
27293 declared variables.
27294 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
27295 declare attribute to any decl as necessary.
27297 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
27300 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
27301 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
27302 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
27303 (arm_arch_lpae): This.
27304 * config/arm/arm.c (arm_arch7ve): Rename into ...
27305 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
27306 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
27309 2017-03-22 Martin Liska <mliska@suse.cz>
27312 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
27313 error message instead of an ICE.
27315 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27317 * doc/extend.texi (6.11 Additional Floating Types): Revise.
27319 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
27321 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
27323 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27326 2017-03-21 Martin Sebor <msebor@redhat.com>
27328 * doc/extend.texi: Use "cannot" instead of "can't."
27329 * doc/hostconfig.texi: Same.
27330 * doc/install.texi: Same.
27331 * doc/invoke.texi: Same.
27332 * doc/loop.texi: Same.
27333 * doc/md.texi: Same.
27334 * doc/objc.texi: Same.
27335 * doc/rtl.texi: Same.
27336 * doc/tm.texi: Same.
27337 * doc/tm.texi.in: Same.
27338 * doc/trouble.texi: Same.
27340 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
27343 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
27344 (collect_checksum_attributes): Set it.
27345 (die_checksum_ordered): Use it.
27347 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27349 PR tree-optimization/79908
27350 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
27351 change: For a VA_ARG whose LHS has been cast away, use
27352 force_gimple_operand to construct the side effects.
27354 2017-03-21 David Malcolm <dmalcolm@redhat.com>
27356 PR translation/80001
27357 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
27358 more amenable to translation.
27359 (oacc_loop_auto_partitions): Likewise.
27361 2017-03-21 Marek Polacek <polacek@redhat.com>
27362 Martin Sebor <msebor@redhat.com>
27364 PR tree-optimization/80109
27365 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
27366 on INTEGRAL_TYPE_P.
27368 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27369 Segher Boessenkool <segher@kernel.crashing.org>
27372 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
27373 check reg_used_between_p between insn and one of succ or succ2
27374 depending on if succ is artificial insn not inserted into insn
27377 2017-03-21 Martin Liska <mliska@suse.cz>
27379 PR gcov-profile/80081
27380 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
27381 * doc/gcc.texi: Include gcov-dump stuff.
27382 * doc/gcov-dump.texi: New file.
27384 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
27386 PR rtl-optimization/79150
27387 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
27388 conditional jump, if the jump is the last insn of the loop.
27390 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27391 Richard Biener <rguenth@suse.de>
27393 PR tree-optimization/79908
27394 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27395 been cast away, use force_gimple_operand to construct the side
27398 2017-03-21 Martin Liska <mliska@suse.cz>
27400 PR libfortran/79956
27401 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
27404 2017-03-21 Brad Spengler <spender@grsecurity.net>
27407 * plugin.c (htab_hash_plugin): New function.
27408 (add_new_plugin): Use it and adjust.
27409 (parse_plugin_arg_opt): Adjust.
27410 (init_one_plugin): Likewise.
27412 2017-03-21 Richard Biener <rguenther@suse.de>
27414 PR tree-optimization/80032
27415 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
27416 if set force the cleanup to happen unconditionally.
27417 (gimplify_target_expr): Push inserted clobbers with force_uncond
27418 to avoid them being removed by control-dependent DCE.
27420 2017-03-21 Richard Biener <rguenther@suse.de>
27422 PR tree-optimization/80122
27423 * tree-inline.c (copy_bb): Do not expans va-arg packs or
27424 va_arg_pack_len when the inlined call stmt requires pack
27426 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
27428 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27431 * tsan.c (instrument_builtin_call): If the memory model argument
27432 is not a constant, assume it is valid.
27435 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
27438 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27440 PR rtl-optimization/79910
27441 * combine.c (can_combine_p): Do not allow combining an I0 or I1
27442 if its dest is used by an insn before I2 (other than the combined
27443 insns themselves, which are properly handled already).
27445 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27448 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
27450 * combine.c (record_used_regs): New static function.
27451 (try_combine): Handle situations where there is an additional
27452 instruction between I2 and I3 which needs to have a LOG_LINK
27456 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27458 * combine.c (try_combine): Delete redundant i1 test. Call
27459 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27461 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27464 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
27465 alternatives 13/14.
27467 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27469 PR tree-optimization/80054
27470 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
27471 the optimization if a PHI or any of its arguments is not dominated
27472 by the candidate's basis. Use gphi* rather than gimple* as
27474 (replace_profitable_candidates): Clean up a gimple* variable that
27475 should be a gphi* variable.
27477 2017-03-20 Martin Sebor <msebor@redhat.com>
27480 * doc/extend.texi (attribute constructor): Document present limitation.
27482 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
27485 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
27486 __POWER9_VECTOR__ #ifdef control, change template definition to
27487 use Power9-specific built-in function.
27488 (vec_any_eq): Likewise.
27489 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
27490 to control outcomes from this test.
27491 (vector_ae_<mode>p): For VEC_F modes, likewise.
27493 2017-03-20 Ian Lance Taylor <iant@google.com>
27495 * config/i386/i386.c (ix86_function_regparm): Save an extra
27496 register for -fsplit-stack with DECL_STATIC_CHAIN.
27498 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
27501 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
27502 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
27504 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
27506 * config/riscv/riscv.c (riscv_print_operand): Use "fence
27508 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
27511 2017-03-20 Marek Polacek <polacek@redhat.com>
27514 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
27516 2017-03-20 Richard Biener <rguenther@suse.de>
27518 PR tree-optimization/80113
27519 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
27520 allocate extra SSA name for PHI def.
27521 (add_close_phis_to_outer_loops): Likewise.
27522 (add_close_phis_to_merge_points): Likewise.
27523 (copy_loop_close_phi_args): Likewise.
27524 (copy_cond_phi_nodes): Likewise.
27526 2017-03-20 Martin Liska <mliska@suse.cz>
27528 PR middle-end/79753
27529 * tree-chkp.c (chkp_build_returned_bound): Do not build
27530 returned bounds for a LHS that's not a BOUNDED_P type.
27532 2017-03-20 Martin Liska <mliska@suse.cz>
27536 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
27537 COMPLEX_CST and VECTOR_CST.
27539 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27542 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
27543 target operand. A new splitter adds the clobber statement in case
27544 the target operand is dead anyway.
27546 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
27548 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
27549 to age-old versions of binutils and glibc.
27551 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
27553 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
27555 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
27557 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
27559 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
27561 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
27562 requirement for binutils 2.13.
27564 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27566 * combine.c (try_combine): Delete redundant i1 test. Call
27567 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27569 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
27571 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
27572 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
27574 <riscv64-*-elf>: Re-arrange section
27575 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
27576 <riscv32-*-linux>: Likewise.
27577 <riscv64-*-elf>: Likewise
27578 <riscv64-*-linux>: Likewise.
27580 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
27583 * aarch64.opt(verbose-cost-dump): Fix typo.
27585 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
27588 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
27589 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
27591 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
27593 * reload.c (find_reloads): When reloading a nonoffsettable address,
27594 use RELOAD_OTHER for it and its address reloads.
27596 PR rtl-optimization/79910
27597 * combine.c (record_used_regs): New static function.
27598 (try_combine): Handle situations where there is an additional
27599 instruction between I2 and I3 which needs to have a LOG_LINK
27602 2017-03-17 Jeff Law <law@redhat.com>
27604 PR tree-optimization/71437
27605 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
27606 conditional in the hash table first.
27607 (vrp_dom_walker::before_dom_children): Extract condition from
27608 ASSERT_EXPR. Record condition, its inverion and any implied
27609 conditions as well.
27611 2017-03-17 Marek Polacek <polacek@redhat.com>
27612 Markus Trippelsdorf <markus@trippelsdorf.de>
27614 PR tree-optimization/80079
27615 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
27618 2017-03-17 Richard Biener <rguenther@suse.de>
27620 PR middle-end/80075
27621 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
27622 Properly verify the LHS before the RHS possibly claims to be
27624 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
27627 2017-03-17 Martin Jambor <mjambor@suse.cz>
27629 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
27630 (List of -O2 options): Likewise.
27631 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
27634 2017-03-17 Tom de Vries <tom@codesourcery.com>
27636 * gcov-dump.c (print_usage): Print bug_report_url.
27638 2017-03-17 Richard Biener <rguenther@suse.de>
27640 PR middle-end/80050
27641 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
27642 (parser::peek): Likewise.
27644 2017-03-17 Richard Biener <rguenther@suse.de>
27646 PR tree-optimization/80048
27647 * sese.c (free_sese_info): Properly release rename_map and
27648 copied_bb_map elements.
27650 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
27652 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
27653 Add linked-list forward and backlinks. Insert on
27654 construction, remove on destruction.
27655 (class pass_store_merging): Add m_stores_head field.
27656 (pass_store_merging::terminate_and_process_all_chains):
27657 Iterate over m_stores_head list.
27658 (pass_store_merging::terminate_all_aliasing_chains):
27660 (pass_store_merging::execute): Check for debug stmts first.
27661 Push new chains onto the m_stores_head stack.
27663 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
27666 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
27667 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
27668 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
27670 2017-03-16 Jeff Law <law@redhat.com>
27672 PR tree-optimization/71437
27673 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
27674 member function. Implementation moved into after_dom_children
27675 member function and into the threader's thread_outgoing_edges
27677 (dom_opt_dom_walker::after_dom_children): Simplify by moving
27678 some code into new thread_outgoing_edges.
27679 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
27680 definition. Simplify marker handling (do it here). Assume we always
27681 have the available expression and the const/copies tables.
27682 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
27684 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
27685 * tree-vrp.c (equiv_stack): No longer file scoped.
27686 (vrp_dom_walker): New class.
27687 (vrp_dom_walker::before_dom_children): New member function.
27688 (vrp_dom_walker::after_dom_children): Likewise.
27689 (identify_jump_threads): Setup domwalker. Use it rather than
27690 walking edges in a random order by hand. Simplify setup/finalization.
27691 (finalize_jump_threads): Remove.
27692 (vrp_finalize): Do not call identify_jump_threads here.
27693 (execute_vrp): Do it here instead and call thread_through_all_blocks
27696 PR tree-optimization/71437
27697 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
27699 (simplify_stmt_for_jump_threading): Add basic_block argument. All
27701 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
27702 (dom_opt_dom_walker::thread_across_edge): Remove
27703 handle_dominating_asserts argument. All callers changed.
27704 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
27705 changes. Remove calls to lhs_of_dominating_assert. Other
27706 uses of handle_dominating_asserts turn into unconditional code
27707 (simplify_control_stmt_condition_1): Likewise.
27708 (simplify_control_stmt_condition): Likewise.
27709 (thread_through_normal_block, thread_across_edge): Likewise.
27710 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
27711 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
27712 object if it is not an SSA_NAME.
27713 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
27714 before calling into the VRP specific simplifiers.
27715 (identify_jump_threads): Remove handle_dominating_asserts
27718 2017-03-16 Jakub Jelinek <jakub@redhat.com>
27721 * tree-diagnostic.c (default_tree_printer): No longer static.
27722 * tree-diagnostic.h (default_tree_printer): New prototype.
27724 2017-03-16 Tamar Christina <tamar.christina@arm.com>
27726 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
27727 Change ins into fmov.
27729 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27731 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
27732 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
27733 Use h_con constraint for operand 1.
27734 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
27735 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
27737 2017-03-15 Jeff Law <law@redhat.com>
27739 PR tree-optimization/71437
27740 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
27741 (record_temporary_equivalences): Use it.
27743 PR tree-optimization/71437
27744 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
27745 tree-ssa-scopedtables.
27746 (lookup_avail_expr, build_and_record_new_cond): Likewise.
27747 (record_conditions, record_cond, vuse_eq): Likewise.
27748 (record_edge_info): Adjust to API tweak of record_conditions.
27749 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
27750 (record_temporary_equivalences, optimize_stmt): Likewise.
27751 (eliminate_redundant_computations): Likewise.
27752 (record_equivalences_from_stmt): Likewise.
27753 * tree-ssa-scopedtables.c: Include options.h and params.h.
27754 (vuse_eq): New function, moved from tree-ssa-dom.c
27755 (build_and_record_new_cond): Likewise.
27756 (record_conditions): Likewise. Accept vector of conditions rather
27757 than edge_equivalence structure for first argument.
27758 for the first argument.
27759 (avail_exprs_stack::lookup_avail_expr): New member function, moved
27760 from tree-ssa-dom.c.
27761 (avail_exprs_stack::record_cond): Likewise.
27762 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
27763 from tree-ssa-dom.c.
27764 (avail_exprs_stack): Add new member functions lookup_avail_expr
27766 (record_conditions): Declare.
27768 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
27771 * lra-constraints.c (process_alt_operands): Increase reject for
27772 reloading an input/output operand.
27774 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
27777 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
27778 insns to convert from signed/unsigned char/short to IEEE 128-bit
27780 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
27782 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
27785 * config/i386/i386.c (ix86_vector_duplicate_value): Create
27786 subreg of inner mode for values already in registers.
27788 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
27790 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
27791 iteration reg is used after the loop.
27793 2017-03-14 Martin Sebor <msebor@redhat.com>
27795 PR tree-optimization/79800
27796 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
27797 precision in negative-positive range.
27798 (format_floating): Call non-const overload with adjusted precision.
27800 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
27803 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
27806 2017-03-14 Martin Sebor <msebor@redhat.com>
27808 PR middle-end/80020
27809 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
27810 * builtins.def (aligned_alloc): Use it.
27813 * Makefile.in (GTFILES): Add calls.c.
27814 * calls.c: Include "gt-calls.h".
27816 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
27818 PR rtl-optimization/79728
27819 * regs.h (struct target_regs): New field
27820 x_contains_allocatable_regs_of_mode.
27821 (contains_allocatable_regs_of_mode): New macro.
27822 * reginfo.c (init_reg_sets_1): Initialize it, and change
27823 contains_reg_of_mode so it includes global regs as well.
27824 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
27825 rather than contains_regs_of_mode.
27827 2017-03-14 Martin Liska <mliska@suse.cz>
27829 * doc/invoke.texi: Document options that can't be combined with
27830 -fcheck-pointer-bounds.
27832 2017-03-14 Martin Liska <mliska@suse.cz>
27834 PR middle-end/79831
27835 * doc/invoke.texi (-Wchkp): Document the option.
27837 2017-03-14 Martin Liska <mliska@suse.cz>
27839 * Makefile.in: Install gcov-dump.
27841 2017-03-14 Martin Liska <mliska@suse.cz>
27843 * multiple_target.c (expand_target_clones): Bail out for
27844 an invalid attribute.
27846 2017-03-14 Richard Biener <rguenther@suse.de>
27848 * alias.c (struct alias_set_entry): Pack properly.
27849 * cfgloop.h (struct loop): Likewise.
27850 * cse.c (struct set): Likewise.
27851 * ipa-utils.c (struct searchc_env): Likewise.
27852 * loop-invariant.c (struct invariant): Likewise.
27853 * lra-remat.c (struct cand): Likewise.
27854 * recog.c (struct change_t): Likewise.
27855 * rtl.h (struct address_info): Likewise.
27856 * symbol-summary.h (function_summary): Likewise.
27857 * tree-loop-distribution.c (struct partition): Likewise.
27858 * tree-object-size.c (struct object_size_info): Likewise.
27859 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
27860 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
27861 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
27862 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
27863 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
27864 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
27865 (struct _stmt_vec_info): Likewise.
27867 2017-03-14 Martin Liska <mliska@suse.cz>
27870 * multiple_target.c (create_dispatcher_calls): Check that
27871 a target can create a function dispatcher.
27873 2017-03-14 Martin Liska <mliska@suse.cz>
27876 * multiple_target.c (expand_target_clones): Drop local.local
27877 flag for default implementation.
27879 2017-03-14 Richard Biener <rguenther@suse.de>
27881 PR tree-optimization/80030
27882 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
27884 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
27886 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
27887 gcc_fallthrough() instead of __attribute__((fallthrough));
27889 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
27891 * doc/gcc.texi: Remove "up" link to (DIR).
27892 * doc/gccint.texi: Ditto.
27894 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
27896 * doc/install.texi (Specific) <avr>: Remove reference to
27899 2017-03-13 Jeff Law <law@redhat.com>
27901 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
27902 attribute rather than comments.
27904 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
27905 match_scratch operand is highest.
27907 2017-03-13 Martin Liska <mliska@suse.cz>
27909 PR middle-end/78339
27910 * ipa-pure-const.c (warn_function_noreturn): If the declarations
27911 is a CHKP clone, use original declaration.
27913 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27915 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
27916 (arc_conditional_register_usage): Use a different allocation order
27917 when optimizing for size.
27918 * common/config/arc/arc-common.c (arc_option_optimization_table):
27919 Section anchors default on when optimizing for size.
27921 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27923 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
27925 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27927 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
27928 * config/arc/arc.md (cpu_facility): Add cd variant.
27929 (*movqi_insn): Add code density variant.
27930 (*movhi_insn): Likewise.
27931 (*movqi_insn): Likewise.
27932 (*addsi3_mixed): Likewise.
27933 (subsi3_insn): Likewise.
27935 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27937 * config/arc/arc.md (movsi_cond_exec): Update constraint.
27939 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27941 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
27942 expressions with MINUS and UNARY ops.
27944 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27947 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
27949 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
27950 between vec_select and vector argument.
27951 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
27952 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
27953 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
27954 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
27955 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
27956 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
27958 2017-03-13 Richard Biener <rguenther@suse.de>
27961 * params.def (vect-max-peeling-for-alignment): Fix typo.
27963 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
27965 * doc/install.texi (Specific) <mips-*-*>: Remove description of
27966 issue that only occurred with binutils below 2.18.
27968 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
27970 * doc/install.texi (Specific) <cris-axis-elf>: No longer
27971 refer to binutils 2.11/2.12 minimum.
27973 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
27975 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
27976 ftp.kernel.org and simplify binutils requirement.
27978 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
27980 * doc/invoke.texi (Warning Options): Fix spelling of link-time
27982 (Optimize Options): Ditto. Also remove redundancy.
27984 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27986 PR translation/79848
27987 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
27989 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
27990 to G_ to avoid double translation.
27992 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27994 PR translation/79923
27995 * auto-profile.c (get_combined_location): Convert leading
27996 character of diagnostics to lower case and remove trailing period.
27997 (read_profile): Likewise for various diagnostics.
27998 * config/arm/arm.c (arm_option_override): Remove trailing period
27999 from various diagnostics.
28000 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
28001 (msp430_expand_delay_cycles): Likewise.
28003 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28006 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
28007 full command-line argument, rather than just "str".
28008 (aarch64_validate_march): Likewise.
28009 (aarch64_validate_mtune): Likewise.
28011 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
28013 PR rtl-optimization/78911
28014 * lra-assigns.c (must_not_spill_p): New function.
28015 (spill_for): Use it.
28017 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28019 PR tree-optimization/79981
28020 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
28021 ATOMIC_COMPARE_EXCHANGE ifn result.
28022 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
28023 IFN_ATOMIC_COMPARE_EXCHANGE.
28025 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28028 * opts.c (parse_sanitizer_options): Add missing question mark to
28029 "did you mean" message.
28031 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28033 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
28035 (VMULEUH_UNS): Likewise.
28036 (VMULOUB_UNS): Likewise.
28037 (VMULOUH_UNS): Likewise.
28038 * config/rs6000/rs6000.c (builtin_function_type): Remove
28039 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
28041 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28044 * read-rtl-function.c (function_reader::read_rtx_operand): Update
28045 x with result of extra_parsing_for_operand_code_0.
28046 (function_reader::extra_parsing_for_operand_code_0): Convert
28047 return type from void to rtx, returning x. When reading
28048 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
28049 larger size containing struct block_symbol.
28051 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
28053 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
28054 -mfloat128-hardware without -m64.
28056 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
28059 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
28060 entries to the case statement that marks unsigned arguments to
28061 overloaded functions.
28063 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
28065 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
28066 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
28068 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
28071 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
28072 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
28074 2017-03-10 Martin Liska <mliska@suse.cz>
28078 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
28079 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
28082 2017-03-10 Olivier Hainque <hainque@adacore.com>
28084 * tree-switch-conversion (array_value_type): Start by resetting
28085 candidate type to it's main variant.
28087 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28089 PR rtl-optimization/79909
28090 * combine.c (try_combine): Use simplify_replace_rtx on individual
28091 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
28092 of the whole CALL_INSN_FUNCTION_USAGE.
28094 PR tree-optimization/79972
28095 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
28096 get_range_info on SSA_NAMEs. Formatting fixes.
28098 2017-03-10 Richard Biener <rguenther@suse.de>
28099 Jakub Jelinek <jakub@redhat.com>
28101 PR tree-optimization/77975
28102 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
28103 edge to be constant.
28104 (get_val_for): For constant x return it. Formatting fix.
28105 (loop_niter_by_eval): Avoid pointless looping if the next iteration
28106 would use the same bases as the current one.
28108 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28110 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
28111 instead of vec_select for V1TImode.
28112 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
28114 (VSX_LE_128): Add V1TI to this mode iterator.
28115 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
28116 (*vsx_le_perm_store_<mode>): Likewise.
28117 (pre-reload splitter for VSX stores): Likewise.
28118 (post-reload splitter for VSX stores): Likewise.
28119 (*vsx_xxpermdi2_le_<mode>): Likewise.
28120 (*vsx_lxvd2x2_le_<mode>): Likewise.
28121 (*vsx_stxvd2x2_le_<mode>): Likewise.
28123 2017-03-09 Michael Eager <eager@eagercon.com>
28125 Correct failures with --enable-checking=yes,rtl.
28127 * config/microblaze/microblaze.c (microblaze_expand_shift):
28128 Replace GET_CODE test with CONST_INT_P and INTVAL test with
28129 test for const0_rtx.
28130 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
28131 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
28133 2017-03-09 Richard Biener <rguenther@suse.de>
28135 PR tree-optimization/79977
28136 * graphite-scop-detection.c (scop_detection::merge_sese):
28137 Handle the case of extra exits to blocks dominating the entry.
28139 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
28141 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28144 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
28146 PR rtl-optimization/79949
28147 * lra-constraints.c (process_alt_operands): Check memory when
28148 trying to predict a cycle. Print about the overall increase.
28150 2017-03-09 Richard Biener <rguenther@suse.de>
28152 PR middle-end/79971
28153 * gimple-expr.c (useless_type_conversion_p): Preserve
28154 TYPE_SATURATING for fixed-point types.
28156 2017-03-09 Richard Biener <rguenther@suse.de>
28159 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
28160 alignment of BLKmode params.
28162 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28165 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
28166 (VALL_NO_V2Q): Likewise.
28168 * config/aarch64/aarch64-simd.md
28169 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
28171 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
28172 VALL_NO_V2Q mode iterator.
28173 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
28175 2017-03-09 Martin Liska <mliska@suse.cz>
28177 PR tree-optimization/79631
28178 * tree-chkp-opt.c (chkp_is_constant_addr): Call
28179 tree_int_cst_sign_bit just for INTEGER constants.
28181 2017-03-09 Martin Liska <mliska@suse.cz>
28185 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
28188 2017-03-09 Marek Polacek <polacek@redhat.com>
28191 * tree.c (inchash::add_expr): Handle TREE_VEC.
28193 2017-03-09 Martin Liska <mliska@suse.cz>
28196 (chkp_narrow_size_and_offset): New function.
28197 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
28198 (void chkp_parse_bit_field_ref): New function.
28199 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
28200 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
28202 2017-03-09 Martin Liska <mliska@suse.cz>
28205 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
28206 (chkp_find_bounds_1): Remove gcc_unreachable.
28208 2017-03-09 Jakub Jelinek <jakub@redhat.com>
28211 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
28212 BUILT_IN_SYNC*, determine the access type from the size suffix and
28213 always build a MEM_REF with that type. Handle forgotten
28214 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
28217 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
28218 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
28219 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
28220 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
28221 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
28222 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
28223 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
28224 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
28225 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
28226 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
28227 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
28228 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
28229 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
28230 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
28231 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
28232 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
28233 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
28234 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
28235 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
28236 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
28237 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
28238 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
28239 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
28240 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
28241 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
28242 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
28243 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
28244 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
28245 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
28246 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
28247 definitions outside of __OPTIMIZE__ guarded section.
28250 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
28251 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
28252 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
28253 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
28256 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28258 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
28259 ("vfenez<mode>"): Add missing constraints.
28261 2017-03-08 Martin Sebor <msebor@redhat.com>
28264 * config/nds32/nds32.c (nds32_option_override):
28265 Fix misspelled diagnostic.
28267 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28270 * gimplify.c (gimplify_omp_for): Replace index var in outer
28271 taskloop statement with an artificial variable and add
28272 OMP_CLAUSE_PRIVATE clause for it.
28274 2017-03-08 Richard Biener <rguenther@suse.de>
28276 PR tree-optimization/79955
28277 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
28278 for accesses that are completely outside of the variable.
28280 2017-03-08 Andrew Haley <aph@redhat.com>
28282 PR tree-optimization/79943
28283 * tree-ssa-loop-split.c (compute_new_first_bound): When
28284 calculating the new upper bound, (END-BEG) should be added, not
28287 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28289 * config/avr/avr.md (setmemhi): Make sure match_dup
28290 operand number comes before match_scratch.
28292 2017-03-08 Richard Biener <rguenther@suse.de>
28294 PR tree-optimization/79920
28295 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
28296 with ncopies == 1 to ...
28297 (vect_transform_slp_perm_load): ... here. Properly compute
28298 all element loads by iterating VF times over the group. Do
28299 not handle ncopies (computed in a broken way) in
28300 vect_create_mask_and_perm.
28302 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28305 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
28306 is a uniform vector, use uniform_vector_p return value instead of
28307 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
28309 2017-03-07 Marek Polacek <polacek@redhat.com>
28311 PR middle-end/79809
28312 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
28313 (alloca_call_type): Likewise.
28315 2017-03-07 Martin Liska <mliska@suse.cz>
28317 * gcov.c (process_args): Put comment to correct location.
28319 2017-03-07 Martin Liska <mliska@suse.cz>
28321 PR middle-end/68270
28322 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
28323 Use array_at_struct_end_p instead of DECL_CHAIN (field).
28324 (chkp_narrow_bounds_for_field): Likewise.
28325 (chkp_parse_array_and_component_ref): Pass one more argument to
28328 2017-03-07 Richard Biener <rguenther@suse.de>
28330 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
28333 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
28335 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
28336 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
28338 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28341 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
28342 to end of description.
28343 (PARAM_MAX_STORES_TO_MERGE): Likewise.
28345 2017-03-07 Jakub Jelinek <jakub@redhat.com>
28347 PR rtl-optimization/79901
28348 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
28350 (*avx512f_<code><mode>3<mask_name>): ... this.
28351 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
28352 iterator instead of VI8_AVX2_AVX512BW.
28354 PR rtl-optimization/79901
28355 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
28356 min/max expander, expand it using expand_vec_cond_expr.
28359 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
28362 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28365 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
28366 to void * for PCH reasons.
28367 * dwarf2out.c (output_loc_operands, output_die): Cast
28368 v.val_vec.array to unsigned char *.
28370 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
28373 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
28376 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
28378 PR rtl-optimization/79571
28379 * lra-constraints.c (process_alt_operands): Calculate static
28380 reject and subtract it from overall when only addresses will be
28383 2017-03-06 Julia Koval <julia.koval@intel.com>
28386 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
28387 incoming stack boundary to 128 for 64-bit targets.
28389 2017-03-06 Richard Biener <rguenther@suse.de>
28391 PR tree-optimization/79894
28392 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
28393 to NULL after folding it.
28395 2017-03-06 Richard Biener <rguenther@suse.de>
28397 PR tree-optimization/79824
28398 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
28399 check disabling peeling for gaps.
28401 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
28403 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28404 attributes): Document gettimeofday.
28406 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28408 * config/s390/s390.c (s390_option_override_internal): Set
28409 PARAM_MIN_VECT_LOOP_BOUND
28411 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28413 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
28414 * config/s390/s390.md: Likewise.
28416 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28419 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
28420 (<avx2_avx512>_perm<mode>): Rename to ...
28421 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
28423 (<avx512>_perm<mode>_mask): Rename to ...
28424 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
28426 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
28427 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
28428 instead of VI8F_256_512.
28429 (avx512f_perm<mode>): New define_expand.
28430 (avx512f_perm<mode>_mask): Likewise.
28431 (avx512f_perm<mode>_1<mask_name>): New define_insn.
28432 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
28434 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28436 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
28437 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
28439 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
28441 2017-03-06 Martin Liska <mliska@suse.cz>
28444 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
28445 when having a SSA NAME w/o VAR_DECL assigned to it.
28447 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28449 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
28450 msa_dpsub_<su>_d): Fix MODE for vec_select.
28452 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28454 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
28456 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
28458 2017-03-06 Richard Biener <rguenther@suse.de>
28460 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
28461 * plugin.c (register_plugin_info): Likewise.
28462 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
28464 2017-03-05 Jakub Jelinek <jakub@redhat.com>
28466 * config/i386/sse.md (sse_storehps, sse_storelps,
28467 avx_<castmode><avxsizesuffix>_<castmode>,
28468 avx512f_<castmode><avxsizesuffix>_<castmode>,
28469 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
28470 in condition that at least one operand is not a MEM.
28472 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28474 PR middle-end/79805
28475 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
28476 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
28478 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
28479 gimple_call_nothrow_p flag based on whether original builtin can throw.
28480 If it can, emit following stmts on the fallthrough edge.
28481 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
28482 don't create new bb if inserting just debug stmts on the edge, try to
28483 insert them on the fallthru bb or just reset debug stmts.
28485 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
28488 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
28489 restore recog_data (including the operand rtxes inside it) around
28490 the call to get_insn_template.
28492 2017-03-03 Martin Sebor <msebor@redhat.com>
28494 PR tree-optimization/79699
28495 * context.c (context::~context): Free MPFR caches to avoid
28496 a memory leak on program exit.
28498 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28500 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
28501 Use wide_int::ulow () instead of .elt (0).
28503 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
28505 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
28506 (*pushxf): Limit oF constraint to 32bit targets and add oC
28507 constraint for 64bit targets.
28508 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
28509 (*pushdf): Change rmF constraint to rmC.
28511 2017-03-03 Martin Liska <mliska@suse.cz>
28513 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
28514 Remove unused variable.
28516 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28519 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
28520 is a memory operand, increase num_memory.
28521 (ix86_expand_args_builtin): Likewise.
28523 2017-03-03 Jan Hubicka <jh@suse.cz>
28526 * ipa-devirt.c (maybe_record_node): Properly handle
28527 __cxa_pure_virtual visibility.
28529 2017-03-03 Martin Liska <mliska@suse.cz>
28531 PR tree-optimization/79803
28532 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
28534 (pass_loop_prefetch::execute): Disabled optimization if an
28535 assumption about L1 cache size is not met.
28537 2017-03-03 Martin Liska <mliska@suse.cz>
28539 PR rtl-optimization/79574
28540 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
28541 (hash_scan_set): Likewise.
28542 (dump_hash_table): Likewise.
28543 (hoist_code): Likewise.
28545 2017-03-03 Richard Biener <rguenther@suse.de>
28547 * fixed-value.c (fixed_from_string): Restore use of elt (1)
28548 in place of uhigh ().
28549 (fixed_convert_from_real): Likewise.
28551 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
28554 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
28556 2017-03-03 Richard Biener <rguenther@suse.de>
28558 PR middle-end/79818
28559 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
28560 TYPE_OVERFLOW_UNDEFINED check.
28562 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28564 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
28566 (vector_ae_<mode>_p): Likewise.
28567 (vector_nez_<mode>_p): Likewise.
28568 (vector_ne_v2di_p): Likewise.
28569 (vector_ae_v2di_p): Likewise.
28570 (vector_ne_<mode>_p): Likewise.
28571 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
28573 (vsx_tsqrt<mode>2_fe): Likewise.
28575 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
28578 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
28580 2017-03-02 Jakub Jelinek <jakub@redhat.com>
28582 PR rtl-optimization/79780
28583 * cprop.c (one_cprop_pass): When second and further conditional trap
28584 in a single basic block is turned into an unconditional trap, turn it
28585 into a deleted note to avoid RTL verification failures.
28587 2017-03-02 Richard Biener <rguenther@suse.de>
28589 * fold-const.c (const_binop): Use ulow () instead of elt (0).
28591 2017-03-02 Richard Biener <rguenther@suse.de>
28593 PR tree-optimization/79345
28595 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
28596 param and abort the walk, returning -1 if it is hit.
28597 (walk_aliased_vdefs): Take a limit param and pass it on.
28598 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
28599 defaulting to 0 and return a signed int.
28600 * tree-ssa-uninit.c (struct check_defs_data): New struct.
28601 (check_defs): New helper.
28602 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
28603 about uninitialized memory.
28604 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
28605 bogus uninitialized warning.
28606 (fixed_convert_from_real): Likewise.
28608 2017-03-02 Bin Cheng <bin.cheng@arm.com>
28610 PR tree-optimization/66768
28611 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
28612 iv_use if base object can't be determined.
28614 2017-03-02 Jakub Jelinek <jakub@redhat.com>
28616 PR tree-optimization/79345
28617 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
28618 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
28619 (get_pattern_stats): Initialize it.
28620 * genemit.c (gen_expand): Verify match_scratch numbers come after
28621 match_operand/match_dup numbers.
28622 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
28623 match_scratch numbers.
28624 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
28626 * config/s390/s390.md (trunctdsd2): Likewise.
28628 2017-03-02 Richard Biener <rguenther@suse.de>
28630 * wide-int.h (wide_int_storage::operator=): Implement in terms
28633 2017-03-02 Richard Biener <rguenther@suse.de>
28635 PR tree-optimization/79777
28636 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
28637 the to insert expression to sth existing.
28639 2017-03-01 Martin Sebor <msebor@redhat.com>
28641 PR middle-end/79692
28642 * gimple-ssa-sprintf.c
28643 (directive::known_width_and_precision): New function.
28644 (format_integer): Use it.
28645 (get_mpfr_format_length): Consider the full range of precision
28646 when computing %g output with the # flag. Set the likely byte
28647 count to 3 rather than 1 when precision is indeterminate.
28648 (format_floating): Correct the lower bound of precision.
28650 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28652 * doc/invoke.texi: Document default code model for 64-bit Linux.
28654 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28657 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
28658 udiv rather than div since input pattern is unsigned.
28660 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
28662 * config/i386/i386.c (print_reg): Warn for values of
28663 unsupported size in integer register.
28665 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
28668 * config/rs6000/predicates.md (current_file_function_operand): Do
28669 not allow self calls to be local if the function is replaceable.
28671 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
28674 * config/rs6000/altivec.h (vec_ctz and others): Change the
28675 preprocessor macro that controls conditional compilation from
28676 _ARCH_PWR9 to __POWER9_VECTOR__.
28677 (vec_all_ne): Change parameterization of __altivec_scalar_pred
28678 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
28679 control (instead of _ARCH_PWR9 control) so that template
28680 definition uses power9-specific function.
28681 (vec_any_eq): Likewise.
28682 (vec_all_ne): Change macro definition to use a power9-specific
28683 expansion under #ifdef __POWER9_VECTOR__ control (instead of
28684 _ARCH_PWR9 control).
28685 (vec_any_eq) Likewise.
28686 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
28687 expansion for CMPNEF to remove support for xvcmpnesp instruction.
28688 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
28689 support for xvcmpnedp instruction.
28690 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
28691 macro expansion so that Power9 implementation of vec_all_ne does
28692 not use the AltiVec predicate framework.
28693 (VCMPNEH_P): Likewise.
28694 (VCMPNEW_P): Likewise.
28695 (VCMPNED_P): Likewise.
28696 (VCMPNEFP_P): Likewise.
28697 (VCMPNEDP_P): Likewise.
28698 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
28699 implementation of vec_any_eq to not use AltiVec predicate
28701 (VCMPAEH_P): Likewise.
28702 (VCMPAEW_P): Likewise.
28703 (VCMPAED_P): Likewise.
28704 (VCMPAEFP_P): Likewise.
28705 (VCMPAEDP_P): Likewise.
28706 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
28707 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
28708 not use the AltiVec predicate framework.
28709 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
28710 of vec_any_eq to not use AltiVec predicate framework.
28711 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
28712 support for predefined __POWER9_VECTOR__ macro to indicate that
28713 Power9 instruction selection is enabled.
28714 (altivec_overloaded_builtins): Remove extraneous
28715 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
28716 function argument types RS6000_BTI_bool_V16QI and
28717 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
28718 entry for overloaded function argument types RS6000_BTI_bool_V4SI
28719 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
28720 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
28721 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
28722 Power9 for implementations of vec_cmpne. Change the signature for
28723 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
28724 (representing vec_all_ne) to remove the previously described first
28725 argument of type RS6000_BTI_INTSI, as this was an artifact of
28726 reliance on the AltiVec predicate framework, which is no longer
28727 used in the implementation of these functions. Add
28728 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
28729 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
28730 since, unlike the AltiVec predicate framework implementation, we
28731 do not share function descriptors between vec_alle and vec_anyeq.
28732 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
28733 set of modes that receive special treatment even when
28734 TARGET_P9_VECTOR is true. The special treatment emits code that
28735 does not depend on Power9 instructions.
28736 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
28737 define_expand to not rely on AltiVec predicate framework.
28738 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
28740 (vector_ne_v2di_p): Change this define_expand to not rely on
28741 AltiVec predicate framework.
28742 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
28744 (vector_ne_<mode>_p): Change this define_expand to not rely on
28745 AltiVec predicate framework.
28746 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
28748 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
28749 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
28750 define_insn pattern.
28751 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
28752 define_insn pattern because the xvcmpne<VSs>. instruction is not
28754 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
28755 instruction is not supported.
28757 2017-03-01 Jakub Jelinek <jakub@redhat.com>
28759 * config/nvptx/nvptx.c: Include intl.h.
28761 2017-03-01 Martin Jambor <mjambor@suse.cz>
28764 * ipa-prop.h (ipa_bits): Removed field known.
28765 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
28766 to pointers. Adjusted their comments to warn about their sharing.
28767 (ipcp_transformation_summary): Change bits to a vector of pointers.
28768 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
28769 (ipa_get_ipa_bits_for_value): Declare.
28770 * tree-vrp.h (value_range): Mark as GTY((for_user)).
28771 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
28772 (ipa_bits_hash_table): Likewise.
28773 (ipa_vr_ggc_hash_traits): Likewise.
28774 (ipa_vr_hash_table): Likewise.
28775 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
28776 being pointers and vr_known being removed.
28777 (ipa_set_jf_unknown): Likewise.
28778 (ipa_get_ipa_bits_for_value): New function.
28779 (ipa_set_jfunc_bits): Likewise.
28780 (ipa_get_value_range): New overloaded functions.
28781 (ipa_set_jfunc_vr): Likewise.
28782 (ipa_compute_jump_functions_for_edge): Use the above functions to
28783 construct bits and vr parts of jump functions.
28784 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
28785 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
28787 (ipcp_grow_transformations_if_necessary): Also allocate
28788 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
28790 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
28791 them. Fix too long lines.
28792 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
28793 vr_known being removed.
28794 (ipa_read_jump_function): Use new setter functions to construct bits
28795 and vr parts of jump functions or set them to NULL.
28796 (write_ipcp_transformation_info): Adjust for bits being pointers.
28797 (read_ipcp_transformation_info): Likewise.
28798 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
28800 Include gt-ipa-prop.h.
28801 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
28803 (ipcp_store_bits_results): Likewise.
28804 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
28805 Do not write to existing jump functions but use a temporary instead.
28807 2017-03-01 Jakub Jelinek <jakub@redhat.com>
28810 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
28811 attempt to use its first operand as BIT_FIELD_REF base.
28813 2017-03-01 Richard Biener <rguenther@suse.de>
28815 PR middle-end/79721
28816 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
28817 interpolating formula in wrapping arithmetic.
28818 (chrec_apply): Convert chrec_evaluate return value to wanted type.
28820 2017-03-01 Jakub Jelinek <jakub@redhat.com>
28822 PR tree-optimization/79734
28823 * tree-vect-generic.c (expand_vector_condition): Optimize
28824 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
28825 Handle VEC_COND_EXPR where comparison has different inner width from
28826 type's inner width.
28828 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
28830 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
28831 markup, and similar issues. Remove @opindex entries for things
28832 that aren't options. Add missing -mmpy-option entries.
28834 2017-02-28 Jakub Jelinek <jakub@redhat.com>
28836 PR tree-optimization/79737
28837 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
28838 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
28839 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
28840 instead of byte_size. Formatting fix.
28841 (shift_bytes_in_array_right): Formatting fix.
28843 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
28846 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
28847 condition on optimize for the leaf function test.
28849 2017-02-28 Martin Liska <mliska@suse.cz>
28852 * read-rtl-function.c (function_reader::handle_unknown_directive):
28853 Bail out when one uses -flto.
28855 2017-02-28 Martin Liska <mliska@suse.cz>
28857 * common.opt: Replace space with tabular for options of <number>
28859 * config/i386/i386.opt: Show <number> value for
28860 -mlarge-data-threshold.
28861 * opts.c (print_filtered_help): Do not display number in hexadecimal
28864 2017-02-28 Martin Liska <mliska@suse.cz>
28866 * common.opt: Fix --help=option -Q for options which are of
28869 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
28871 * config/i386/i386.c (print_reg): Error out for values
28872 of 8-bit size in invalid integer register.
28874 2017-02-28 Martin Sebor <msebor@redhat.com>
28876 PR tree-optimization/79691
28877 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
28879 2017-02-28 Jakub Jelinek <jakub@redhat.com>
28882 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
28883 gcc_unreachable with output_operand_lossage.
28885 2017-02-28 Richard Biener <rguenther@suse.de>
28887 PR tree-optimization/79740
28888 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
28890 (visit_nary_op): Insert the nary into the hashtable if we
28891 pattern-matched sth.
28892 * tree-ssa-pre.c (eliminate_insert): Robustify.
28894 2017-02-28 Richard Biener <rguenther@suse.de>
28896 PR middle-end/79731
28897 * fold-const.c (decode_field_reference): Reject out-of-bound
28900 2017-02-28 Jakub Jelinek <jakub@redhat.com>
28902 * config/i386/i386.c: Include intl.h.
28903 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
28904 instead of just cond ? "..." : "...".
28905 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
28906 * coverage.c (read_counts_file): Likewise.
28907 * omp-offload.c: Include intl.h.
28908 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
28909 of just cond ? "..." : "...".
28910 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
28911 of just cond ? "..." : "...".
28913 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
28916 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
28918 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
28919 'tune for' CPU name.
28920 * config/arm/arm-cpu-data.h: Regenerated.
28922 2017-02-28 Richard Biener <rguenther@suse.de>
28924 PR tree-optimization/79732
28925 * tree-inline.c (expand_call_inline): Do not shadow var.
28927 2017-02-28 Richard Biener <rguenther@suse.de>
28929 PR tree-optimization/79723
28930 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
28931 address-space properly.
28933 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
28935 * doc/optinfo.texi (Optimization groups): Fix option used for
28937 * doc/invoke.texi (-fopt-info): Document "omp".
28938 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
28939 (OPTGROUP_ALL): Add OPTGROUP_OMP.
28940 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
28941 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
28942 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
28944 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
28946 * dumpfile.c (optgroup_options): Instead of "openmp", associate
28947 OPTGROUP_OMP with "omp".
28949 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
28952 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
28953 for arithmetic shift of unsigned V2DI.
28955 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
28957 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
28958 arc/linux.h headers.
28959 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
28960 (LINK_SPEC): Likewise.
28961 (ARC_TLS_EXTRA_START_SPEC): Likewise.
28962 (EXTRA_SPECS): Likewise.
28963 (STARTFILE_SPEC): Likewise.
28964 (ENDFILE_SPEC): Likewise.
28965 (LIB_SPEC): Likewise.
28966 (TARGET_SDATA_DEFAULT): Likewise.
28967 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
28968 (MULTILIB_DEFAULTS): Likewise.
28969 (DWARF2_UNWIND_INFO): Likewise.
28970 * config/arc/big.h: New file.
28971 * config/arc/elf.h: Likewise.
28972 * config/arc/linux.h: Likewise.
28973 * config/arc/t-uClibc: Remove.
28975 2017-02-27 Bin Cheng <bin.cheng@arm.com>
28977 PR tree-optimization/77536
28978 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
28979 (tree_transform_and_unroll_loop): Use above function to compute the
28980 estimated niter of unrolled loop and use it when scaling profile.
28981 Also use count info rather than frequency if it's non-zero.
28982 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
28983 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
28984 (vect_transform_loop): Call above function.
28986 2017-02-27 Richard Biener <rguenther@suse.de>
28988 PR tree-optimization/45397
28989 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
28990 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
28991 (visit_nary_op): Add pattern matching for CSEing sign-changed
28992 or truncated operations with wider ones.
28994 2017-02-27 Richard Biener <rguenther@suse.de>
28996 PR tree-optimization/79690
28997 * tree-vect-stmts.c (vectorizable_store): Use vector type
28998 built from the DR with address-space.
29000 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
29002 * doc/invoke.texi (Optimize Options): Refine the description
29003 of asan-use-after-return.
29005 2017-02-25 Alan Modra <amodra@gmail.com>
29007 PR rtl-optimization/79584
29008 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
29009 base, not ad->base_term, the reg within base. Remove assertion
29010 that ad->base == ad->base_term. Replace gen_int_mode using
29011 bogus mode with const0_rtx.
29013 2017-02-25 Jakub Jelinek <jakub@redhat.com>
29015 PR middle-end/79396
29016 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
29017 FMA_EXPR like tcc_binary or tcc_unary.
29019 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
29022 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
29024 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
29025 (output_loc_operands): Handle DW_OP_call_ref and
29026 DW_OP_GNU_variable_value.
29027 (struct variable_value_struct): New type.
29028 (struct variable_value_hasher): Likewise.
29029 (variable_value_hash): New variable.
29030 (string_types): Remove.
29031 (copy_loc_descr): New function.
29032 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
29033 (prepend_loc_descr_to_each): New function.
29034 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
29035 instead of add_loc_descr_to_each if the first argument is single
29036 location list and the second has multiple.
29037 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
29038 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
29039 when looking for variable value which doesn't have other location info.
29040 (loc_list_from_tree): Formatting fix.
29041 (gen_array_type_die): Simplify DW_AT_string_length handling.
29042 (adjust_string_types): Remove.
29043 (gen_subprogram_die): Don't call adjust_string_types nor test/set
29044 string_types. Call resolve_variable_values.
29045 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
29046 (resolve_addr_in_expr): Likewise. Add A argument.
29047 (copy_deref_exprloc): Remove deref argument. Adjust for the
29048 original expression being DW_OP_GNU_variable_value with optionally
29049 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
29050 optionally after it.
29051 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
29052 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
29053 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
29054 (variable_value_hasher::hash, variable_value_hasher::equal): New
29056 (resolve_variable_value_in_expr, resolve_variable_value,
29057 resolve_variable_values, note_variable_value_in_expr,
29058 note_variable_value): New functions.
29059 (dwarf2out_early_finish): Call note_variable_value on all toplevel
29062 2017-02-24 Jakub Jelinek <jakub@redhat.com>
29065 * opts.h (handle_generated_option): Add GENERATED_P argument.
29066 * opts-common.c (handle_option): Adjust function comment.
29067 (handle_generated_option): Add GENERATED_P argument, pass it to
29069 (control_warning_option): Pass false to handle_generated_option
29071 * opts.c (maybe_default_option): Pass true to handle_generated_option
29073 * optc-gen.awk: Likewise.
29075 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29077 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
29078 a REG, look at the REG it is a SUBREG of.
29079 (splitter for cmpeqsi_t): Ditto.
29081 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29083 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
29084 the special USEs with the pattern of the insn, not the insn itself.
29086 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
29089 * doc/invoke.texi: Document -mload-store-pairs.
29091 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29092 Sandra Loosemore <sandra@codesourcery.com>
29094 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
29095 argument isn't a CONST_INT.
29096 (nios2_alternate_compare_const): Assert op is a CONST_INT.
29097 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
29098 (nios2_validate_compare): Bypass alternate compare logic if *op2
29099 is not a CONST_INT.
29100 (ldstwm_operation_p): Return false if first_base is not a REG or
29101 if first_offset is not a CONST_INT.
29103 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29105 * config/cris/cris.md: Use correct operand in a define_peephole2.
29107 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29109 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
29111 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29113 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
29114 this_insn if it is an INSN or JUMP_INSN.
29115 (force_offsettable): Look at base, not at addr.
29116 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
29117 on things that aren't necessarily CONST_INTs.
29119 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
29121 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
29122 -mfpmath=sse is the default also for x86-32 targets with SSE2
29123 instruction set when @option{-ffast-math} is enabled
29125 2017-02-24 Jeff Law <law@redhat.com>
29127 PR rtl-optimizatoin/79286
29128 * ira.c (update_equiv_regs): Drop may_trap_p exception to
29131 2017-02-24 Richard Biener <rguenther@suse.de>
29133 PR tree-optimization/79389
29134 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
29137 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
29139 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
29140 function comment to reflect reality.
29141 (loop_exits_before_overflow): Fix typo in function description.
29143 2017-02-24 Richard Biener <rguenther@suse.de>
29145 PR tree-optimization/79389
29146 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
29147 properly that a threading opportunity exists. Detect conditional
29148 copy/constant propagation opportunities.
29150 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
29152 * config/visium/visium.md (type): Add trap.
29153 (b): New mode attribute.
29154 (*btst): Rename into...
29155 (*btst<mode>): ...this and adjust.
29156 (*cbranchsi4_btst_insn): Rename into...
29157 (*cbranch<mode>4_btst_insn): ...this and adjust.
29158 (trap): New define_insn.
29160 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29162 PR tree-optimization/79389
29163 * ifcvt.c (struct noce_if_info): Add rev_cond field.
29164 (noce_reversed_cond_code): New function.
29165 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
29166 reversed_comparison_code. Formatting fix.
29167 (noce_try_store_flag): Test rev_cond != NULL in addition to
29168 reversed_comparison_code.
29169 (noce_try_store_flag_constants): Likewise.
29170 (noce_try_store_flag_mask): Likewise.
29171 (noce_try_addcc): Use rev_cond if non-NULL instead of
29172 reversed_comparison_code.
29173 (noce_try_cmove_arith): Likewise. Formatting fixes.
29174 (noce_try_minmax, noce_try_abs): Clear rev_cond.
29175 (noce_find_if_block): Initialize rev_cond.
29176 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
29177 instead of false as last argument never attempt to reverse it
29180 2017-02-23 Bin Cheng <bin.cheng@arm.com>
29182 PR tree-optimization/79663
29183 * tree-predcom.c (combine_chains): Process refs in reverse order
29184 only for ZERO length chains, and add explaining comment.
29186 2017-02-23 Jeff Law <law@redhat.com>
29188 PR tree-optimization/79578
29189 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
29190 in call to operand_equal_p.
29192 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
29195 * config/i386/cpuid.h: Fix another undefined behavior.
29197 2017-02-23 Richard Biener <rguenther@suse.de>
29199 PR tree-optimization/79683
29200 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
29201 vector types for data-refs.
29203 2017-02-23 Martin Liska <mliska@suse.cz>
29205 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
29207 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29209 PR middle-end/79665
29210 * internal-fn.c (get_range_pos_neg): Moved to ...
29211 * tree.c (get_range_pos_neg): ... here. No longer static.
29212 * tree.h (get_range_pos_neg): New prototype.
29213 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
29214 are known to be in between 0 and signed maximum inclusive, try to
29215 expand both unsigned and signed divmod and use the cheaper one from
29218 2017-02-22 Jeff Law <law@redhat.com>
29220 PR tree-optimization/79578
29221 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
29222 to compare base operands.
29224 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
29227 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
29228 gpc_reg_operand instead of fpr_reg_operand.
29230 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
29232 * config/mips/mips.c (mips_return_in_memory): Force FP
29233 vector types to be returned in memory for o32 ABI.
29235 2017-02-22 Jakub Jelinek <jakub@redhat.com>
29237 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
29238 instead of DW_TAG_member for static data member declarations and don't
29239 set no_linkage_name for static inline data members.
29240 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
29243 2017-02-22 Martin Liska <mliska@suse.cz>
29245 * doc/invoke.texi: Replace inequality signs with square brackets
29248 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
29251 * lra-constraints.c (simplify_operand_subreg): Handle
29252 WORD_REGISTER_OPERATIONS targets.
29254 2017-02-22 Jakub Jelinek <jakub@redhat.com>
29257 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
29258 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
29259 elimination by swapping fld*.
29261 2017-02-22 Richard Biener <rguenther@suse.de>
29263 PR tree-optimization/79673
29264 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
29265 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
29266 irrelevant address-space qualifiers and avoiding a
29267 ADDR_SPACE_CONVERT_EXPR from fold_convert.
29269 2017-02-22 Richard Biener <rguenther@suse.de>
29271 PR tree-optimization/79666
29272 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
29273 to not symbolically negate if that may introduce undefined
29276 2017-02-22 Martin Liska <mliska@suse.cz>
29279 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
29280 * data-streamer-out.c (streamer_write_gcov_count_stream):
29282 * value-prof.c (stream_out_histogram_value): Make assert more
29283 precise based on type of counter.
29285 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
29288 * config/i386/i386.md (standard_x87sse_constant_load splitter):
29289 Use nonimmediate_operand instead of memory_operand for operand 1.
29290 (float-extend standard_x87sse_constant_load splitter): Ditto.
29292 2017-02-21 Jeff Law <law@redhat.com>
29294 PR tree-optimization/79621
29295 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
29296 blocks with edges to themselves.
29298 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29301 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
29302 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
29303 Use gimple_call_builtin_p.
29306 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
29307 on temporarily removed DEBUG_INSNs.
29309 PR tree-optimization/79649
29310 * tree-loop-distribution.c (classify_partition): Give up on
29311 non-generic address space loads/stores.
29313 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
29315 * doc/loop.texi (Loop manipulation): Remove nonexistent
29316 tree_ssa_loop_version from the documentation.
29317 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
29319 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29322 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
29323 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
29324 * config/rs6000/rs6000.c: Include except.h.
29325 (rs6000_expand_split_stack_prologue): Call
29326 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
29328 2017-02-21 Martin Jambor <mjambor@suse.cz>
29331 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
29332 have been analyzed.
29334 2017-02-21 Martin Jambor <mjambor@suse.cz>
29336 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
29337 for backward compatibility only.
29338 * doc/invoke.texi (Option Summary): Remove all references to
29339 -fipa-cp-alignment.
29341 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
29345 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29347 * lra-constraints.c (curr_insn_transform): Handle
29348 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29350 2017-02-21 Martin Liska <mliska@suse.cz>
29352 * config/i386/i386.opt: Replace -masm-dialect with -masm.
29354 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
29356 PR translation/79638
29357 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
29359 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
29362 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
29363 (arm_function_ok_for_sibcall): Return false for an indirect call by
29364 descriptor if all the argument registers are used.
29365 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
29366 alignment of the function.
29368 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29370 PR tree-optimization/61441
29371 * simplify-rtx.c (simplify_const_unary_operation): For
29372 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
29373 the sNaN unmodified.
29375 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
29377 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29378 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29379 instead of SYSTEM_HEADER_DIR.
29381 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
29382 Martin Liška <mliska@suse.cz>
29384 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
29385 Fix typos and grammar, use active voice, and clarify.
29387 2017-02-20 Marek Polacek <polacek@redhat.com>
29389 PR middle-end/79537
29390 * gimplify.c (gimplify_expr): Handle unused *&&L;.
29393 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
29395 2017-02-20 Jakub Jelinek <jakub@redhat.com>
29398 * config/i386/i386.c (ix86_expand_builtin): Handle
29399 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
29400 ix86_builtins_isa[fcode].isa as a requirement of those
29401 flags and any other flag in the bitmask.
29402 (ix86_init_mmx_sse_builtins): Use 0 instead of
29403 ~OPTION_MASK_ISA_64BIT as mask.
29404 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
29405 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
29406 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
29407 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
29409 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29412 * lra-constraints.c (split_reg): Check requested split mode
29413 is supported by the register.
29415 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29417 * lra-constraints.c (simplify_operand_subreg): Remove early
29420 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29423 * lra-constraints.c (curr_insn_transform): Tighten condition
29424 for converting SUBREG reloads from OP_OUT to OP_INOUT.
29426 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29429 * lra-constraints.c (curr_insn_transform): Handle
29430 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29432 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
29435 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
29437 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
29439 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
29442 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
29445 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29447 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
29450 2017-02-18 Jakub Jelinek <jakub@redhat.com>
29452 * final.c (last_columnnum, override_columnnum): New variables.
29453 (final_start_function): Set last_columnnum, pass it to begin_prologue
29454 hook and pass 0 to dwarf2out_begin_prologue.
29455 (final_scan_insn): Update override_columnnum. Pass last_columnnum
29456 to source_line debug hook.
29457 (notice_source_line): Compute last_columnnum and for debug_column_info
29458 return true on column changes.
29459 * debug.h (struct gcc_debug_hooks): Add column argument to
29460 source_line and begin_prologue hooks.
29461 (debug_nothing_int_charstar_int_bool): Remove prototype.
29462 (debug_nothing_int_int_charstar,
29463 debug_nothing_int_int_charstar_int_bool): New prototypes.
29464 (dwarf2out_begin_prologue): Add column argument.
29465 * debug.c (do_nothing_debug_hooks): Adjust source_line and
29466 begin_prologue hooks.
29467 (debug_nothing_int_charstar_int_bool): Remove.
29468 (debug_nothing_int_int_charstar,
29469 debug_nothing_int_int_charstar_int_bool): New functions.
29470 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
29471 through to dwarf2out_source_line.
29472 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
29473 (dwarf2out_source_line): Add column argument, emit it if requested.
29474 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
29476 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29477 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29478 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
29479 through to dwarf2out_begin_prologue.
29480 (vmsdbgout_source_line): Add column argument, pass it through to
29481 dwarf2out_source_line.
29482 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
29483 dbxout_source_line caller.
29484 (dbxout_source_line): Add column argument.
29486 * common.opt (gno-column-info, gcolumn-info): New options.
29487 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
29488 (check_die): Also test for multiple DW_AT_decl_column attributes.
29489 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
29490 DW_AT_decl_column if requested.
29491 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
29493 (gen_variable_die): Likewise.
29494 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
29495 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
29498 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
29499 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
29500 (ix86_handle_option): Handle OPT_m3dnowa.
29501 * doc/invoke.texi (-m3dnowa): Document.
29502 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
29503 -m3dnowa instead of -m3dnow -march=athlon.
29506 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
29507 instead of gcc_assert for K, r and R code checks. Formatting fixes.
29509 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29512 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
29513 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
29514 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
29515 generator for vsx_xxpermdi_<mode>_be.
29516 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
29517 force big-endian semantics.
29518 (vsx_xxpermdi_<mode>_be): New define_expand with same
29519 implementation as previous version of vsx_xxpermdi_<mode>.
29521 2017-02-17 Jakub Jelinek <jakub@redhat.com>
29523 PR tree-optimization/79327
29524 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
29525 variable, its initialization and use.
29527 2017-02-17 Julia Koval <julia.koval@intel.com>
29529 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
29530 (OPTION_MASK_ISA_PKU_UNSET): New.
29531 (ix86_handle_option): Handle -mrdpid.
29532 * config/i386/cpuid.h (bit_RDPID): New.
29533 * config/i386/driver-i386.c (host_detect_local_cpu):
29534 Detect RDPID feature.
29535 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
29536 * config/i386/i386-c.c (ix86_target_macros_internal):
29538 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
29539 (ix86_valid_target_attribute_inner_p): Add "rdpid".
29540 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
29541 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
29542 * config/i386/i386.md (define_insn "rdpid"): New.
29543 * config/i386/i386.opt Add -mrdpid.
29544 * config/i386/immintrin.h (_rdpid_u32): New.
29546 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
29548 PR rtl-optimization/79541
29549 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
29550 instead of transforming it into USE.
29552 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
29554 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
29555 If HONOR_SNANS (SFmode) force the input to a register.
29556 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
29557 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
29558 an frsp or similar insn.
29560 2017-02-17 Martin Liska <mliska@suse.cz>
29562 PR rtl-optimization/79577
29563 * params.def (selsched-max-sched-times): Increase minimum to 1.
29565 2017-02-17 Martin Liska <mliska@suse.cz>
29567 PR rtl-optimization/79574
29568 * gcse.c (want_to_gcse_p): Prevent integer overflow.
29570 2017-02-17 Martin Liska <mliska@suse.cz>
29572 PR tree-optimization/79529
29573 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
29574 ssa_defined_default_def_p to handle cases which are implicitly
29576 * tree-ssa.c (ssa_defined_default_def_p): New function.
29577 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
29578 which are implicitly defined.
29579 * tree-ssa.h (ssa_defined_default_def_p): Declare.
29581 2017-02-17 Richard Biener <rguenther@suse.de>
29583 PR middle-end/79576
29584 * params.def (max-ssa-name-query-depth): Limit to 10.
29586 2017-02-17 Richard Biener <rguenther@suse.de>
29588 PR tree-optimization/79552
29589 * tree-ssa-structalias.c (visit_loadstore): Properly verify
29592 2017-02-17 Richard Biener <rguenther@suse.de>
29595 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
29597 2017-02-17 Marek Polacek <polacek@redhat.com>
29599 PR middle-end/79536
29600 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
29601 (fold_negate_expr): New wrapper.
29603 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
29605 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
29606 Correct terminology and de-emphasize pre-standard behavior.
29608 2017-02-16 Alan Modra <amodra@gmail.com>
29610 PR rtl-optimization/79286
29611 * ira.c (def_dominates_uses): New function.
29612 (update_equiv_regs): Don't create an equivalence for insns that
29613 may trap where the register def does not dominate the use.
29615 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
29617 PR rtl-optimization/78127
29618 * lra.c (lra): Call lra_eliminate before finish the loop after
29621 2017-02-16 Richard Biener <rguenther@suse.de>
29623 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
29625 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
29626 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
29627 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
29628 (isl_val_int_from_wi): New function.
29629 (extract_affine_gmp): Rename to ...
29630 (extract_affine_wi): ... this, take a widest_int.
29631 (extract_affine_int): Just wrap extract_affine_wi.
29632 (add_param_constraints): Use isl_val_int_from_wi.
29633 (add_loop_constraints): Likewise, and extract_affine_wi.
29635 2017-02-15 Jeff Law <law@redhat.com>
29637 PR middle-end/79521
29638 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
29639 ira_init_register_move_cost_if_necessary.
29641 2017-02-15 Martin Sebor <msebor@redhat.com>
29643 PR middle-end/32003
29644 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
29645 removed in a prior commit.
29647 2017-02-15 Bin Cheng <bin.cheng@arm.com>
29649 PR tree-optimization/79347
29650 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
29651 counters during peeling.
29653 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
29655 * Makefile.in (site.exp): Remove "set ISLVER".
29657 2017-02-15 Jakub Jelinek <jakub@redhat.com>
29660 * real.c (real_from_integer): Call real_convert even for decimal.
29662 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
29665 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
29667 2017-02-14 Andrew Pinski <apinski@cavium.com>
29669 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
29670 cores and change the partno/implementer to be correct.
29671 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
29672 the 'B" as the implementer.
29673 * config/aarch64/aarch64-tune.md: Regenerate.
29675 2017-02-14 Carl Love <cel@us.ibm.com>
29677 * config/rs6000/rs6000.c: Add case statement entry to make the
29678 xvcvuxdsp built-in argument unsigned.
29679 * config/rs6000/vsx.md: Fix the source and return operand types so they
29680 match the instruction definitions from the ISA document. Fix typo
29681 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
29684 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
29687 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
29688 member early_clobber_alts.
29689 * lra-lives.c (reg_early_clobber_p): New.
29690 (process_bb_lives): Use it.
29691 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
29692 (debug_operand_data): Initialize early_clobber_alts.
29693 (setup_operand_alternative): Set up early_clobber_alts.
29694 (collect_non_operand_hard_regs): Ditto. Pass early clobber
29695 alternatives to new_insn_reg.
29696 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
29698 (lra_update_insn_regno_info): Pass the new arg.
29700 2017-02-14 Jakub Jelinek <jakub@redhat.com>
29702 PR middle-end/79505
29703 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
29704 (new_oacc_loop_raw): Don't clear already cleared fields.
29707 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
29708 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
29709 _mm512_prefetch_i64gather_ps): New inline functions and macros.
29711 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
29714 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
29716 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
29719 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
29720 the extra instruction to the right place to store 128-bit constant
29723 2017-02-14 Martin Sebor <msebor@redhat.com>
29725 PR middle-end/79448
29726 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
29727 warning for strings of unknown length.
29729 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
29731 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
29733 2017-02-14 Jeff Law <law@redhat.com>
29736 * ira-costs.c (scan_one_insn): Initialize register move costs
29737 for pseudos seen in USE/CLOBBER insns.
29739 PR tree-optimization/79095
29740 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
29741 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
29742 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
29743 if the operands are known to be not equal, then the resulting range
29745 (intersect_ranges): If the new range is ~[0,0] and the old range is
29746 wide, then prefer ~[0,0].
29747 * tree-vrp.c (overflow_comparison_p_1): New function.
29748 (overflow_comparison_p): New function.
29749 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
29750 if NAME is used in an overflow test.
29751 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
29752 overflow check that can be expressed as an equality test, then adjust
29753 ops to be that equality test.
29755 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29757 * config/s390/s390-builtin-types.def: Remove flags argument.
29758 * config/s390/s390.c (s390_init_builtins): Likewise.
29760 2017-02-14 Martin Liska <mliska@suse.cz>
29762 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
29763 vector. Fix trailing white spaces.
29765 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
29767 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
29770 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29772 PR rtl-optimization/68664
29773 * config/arm/arm.c (arm_sched_can_speculate_insn):
29774 New function. Declare prototype.
29775 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
29777 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29779 PR rtl-optimization/68664
29780 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
29782 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
29784 2017-02-14 Amit Pawar <amit.pawar@amd.com>
29786 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
29787 max skip bytes for function, loop and jump.
29789 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29791 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
29792 ABS_EXPR for gimple dump.
29794 2017-02-14 Jakub Jelinek <jakub@redhat.com>
29797 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
29799 PR tree-optimization/79408
29800 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
29801 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
29802 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
29803 also if rhs1 is INTEGER_CST.
29805 2017-02-14 Richard Biener <rguenther@suse.de>
29807 PR middle-end/79432
29808 * tree-into-ssa.c (insert_phi_nodes): When the function can
29809 have abnormal edges rewrite SSA names with broken use-def
29810 dominance out of SSA and register them for PHI insertion.
29812 2017-02-13 Martin Sebor <msebor@redhat.com>
29814 PR middle-end/79496
29815 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
29816 clearing info.nowrite flag when snprintf size argument is a range.
29818 2017-02-13 Jakub Jelinek <jakub@redhat.com>
29820 * cprop.c (cprop_jump): Add missing space in string literal.
29821 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
29822 (get_constraint_for_component_ref): Likewise.
29823 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
29824 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
29825 * lra-constraints.c (process_alt_operands): Likewise.
29826 * ipa-inline.c (inline_small_functions): Likewise.
29827 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
29828 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
29829 * trans-mem.c (diagnose_tm_1_op): Likewise.
29830 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
29831 (grid_parallel_clauses_gridifiable): Likewise.
29833 * config/nvptx/mkoffload.c (process): Add space in between
29836 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
29837 "MOD4_SSE_REGS" and "ALL_REGS".
29839 * spellcheck.c (test_data): Add , in between "foo" and "food".
29841 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29844 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
29845 boundary crossing check and subsequent code generation agree.
29847 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29849 * config/aarch64/aarch64.c (has_memory_op): Delete.
29850 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
29853 2017-02-13 Jakub Jelinek <jakub@redhat.com>
29855 PR rtl-optimization/79388
29856 PR rtl-optimization/79450
29857 * combine.c (distribute_notes): When removing TEM_INSN for which
29858 corresponding dest has last value recorded, invalidate that last
29861 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29863 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
29864 of explicit '@'. Add missing assembly comment marker on branch costs
29867 2017-02-13 Nathan Sidwell <nathan@acm.org>
29869 * gengtype-lex.l (<in_struct>): Add '/'.
29871 2017-02-13 Martin Liska <mliska@suse.cz>
29874 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
29876 2017-02-13 Richard Biener <rguenther@suse.de>
29878 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
29880 * configure: Re-generate.
29881 * config.in: Likewise.
29882 * graphite-dependences.c: Simplify as if
29883 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
29884 * graphite-isl-ast-to-gimple.c: Likewise.
29885 * graphite-optimize-isl.c: Likewise.
29886 * graphite-poly.c: Likewise.
29887 * graphite-sese-to-poly.c: Likewise.
29888 * graphite.h: Likewise.
29889 * toplev.c: Include isl/version.h and use isl_version () for
29890 printing the ISL version.
29891 * doc/install.texi: Update ISL requirement.
29893 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
29895 * doc/standards.texi (Standards): Update reference to
29898 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
29900 * doc/extend.texi (Named Address Spaces): sourceware.org now
29902 * doc/install.texi (Binaries): Ditto.
29905 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
29907 * doc/cpp.texi: Replace "stringify"/"stringification" with C
29908 standard terminology "stringize"/"stringizing" throughout.
29909 * doc/cppinternals.texi: Likewise.
29911 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
29913 * doc/extend.texi: Fix some spelling mistakes and typos.
29914 * doc/invoke.texi: Likewise.
29916 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
29919 * params.def (inline-min-speedup) Change from 10 to 8.
29921 2017-02-11 Jakub Jelinek <jakub@redhat.com>
29923 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
29926 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
29929 * ipa-inline-analysis.c (get_minimal_bb): New function.
29930 (record_modified): Use it.
29931 (remap_edge_change_prob): Handle also ancestor functions.
29933 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
29935 * doc/contrib.texi (Contributors): Remove broken link into
29936 the Mauve CVS repository.
29938 2017-02-11 Jakub Jelinek <jakub@redhat.com>
29940 PR middle-end/79454
29941 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
29942 result computation whenever lhs doesn't have vector mode, not
29943 just when it has BLKmode.
29945 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
29947 * doc/makefile.texi (profiledbootstrap): Refer to the
29948 installation instructions only in textual form.
29950 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29953 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
29955 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
29957 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
29958 (Specific): Update mingw-w64 reference.
29960 (Specific): Remove broken link to Renesas RX processor.
29962 2017-02-10 Richard Biener <rguenther@suse.de>
29964 * toplev.c (process_options): Do not mention obsolete graphite
29965 options when printing sorry message about missing graphite support.
29966 Mention -floop-nest-optimize.
29968 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
29970 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
29971 (vtst_p16): Likewise.
29972 (vtstq_p8): Likewise.
29973 (vtstq_p16): Likewise.
29975 (vtstq_p64): Likewise.
29976 * config/arm/arm_neon.h (vgetq_lane_p64): New.
29977 (vset_lane_p64): New.
29978 (vsetq_lane_p64): New.
29980 2017-02-10 Jakub Jelinek <jakub@redhat.com>
29982 PR tree-optimization/79411
29983 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
29984 stmt operands are SSA_NAMEs used in abnormal phis.
29985 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
29988 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
29991 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
29994 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
29996 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
29998 2017-02-09 Jakub Jelinek <jakub@redhat.com>
30000 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
30004 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
30005 not arbitrary TREE_CONSTANT.
30008 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
30009 "omp declare target link" attribute unless is_global_var.
30010 * omp-offload.c (find_link_var_op): Likewise.
30012 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
30013 Chung-Lin Tang <cltang@codesourcery.com>
30015 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
30017 (gimplify_adjust_omp_clauses): Don't delete TILE.
30018 (gimplify_omp_for): Deal with TILE.
30019 * internal-fn.c (expand_GOACC_TILE): New function.
30020 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
30022 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
30023 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
30025 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
30026 avoid DIV for outermost collapse var.
30027 (expand_oacc_for): Insert tile element loop as needed. Adjust.
30028 Remove out of date comments, fix whitespace.
30029 * omp-general.c (omp_extract_for_data): Deal with tiling.
30030 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
30031 adjust OLF_DIM_BASE value.
30032 (struct omp_for_data): Add tiling field.
30033 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
30034 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
30035 for auto loops. Remove default auto determining, moved to
30036 oacc_loop_fixed_partitions.
30037 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
30038 stmts, add e_mask field.
30039 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
30040 (oacc_thread_numbers): Use oacc_dim_call.
30041 (oacc_xform_tile): New.
30042 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
30043 (finish_oacc_loop): Adjust for ifns vector.
30044 (oacc_loop_discover_walk): Append loop abstraction sites to list,
30045 add case for GOACC_TILE fns.
30046 (oacc_loop_xform_loop): Delete.
30047 (oacc_loop_process): Iterate over call list directly, and add
30048 handling for GOACC_TILE fns.
30049 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
30051 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
30052 vector partitioning to outer loops. Assign 2 partitions to loops
30053 when available. Add TILE handling.
30054 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
30055 (execite_oacc_device_lower): Process GOACC_TILE fns,
30056 ignore unknown specs.
30057 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
30058 * tree.c (omp_clause_num_ops): Adjust TILE ops.
30059 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
30061 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
30063 * configure.ac (ACX_BUGURL): Update.
30064 * configure: Regenerate.
30066 2017-02-09 Richard Biener <rguenther@suse.de>
30068 PR tree-optimization/69823
30069 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
30070 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
30072 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30074 * config/arc/arc-c.def: Add __NPS400__ definition.
30075 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
30076 (TARGET_NPS400): Define.
30078 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30080 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
30082 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
30083 pointer, arch_info.
30084 (arc_cpu_types): Fill the arch_info field with a pointer into the
30085 arc_arch_types table.
30086 (arc_selected_cpu): Declare.
30087 * config/arc/arc.c (arc_selected_cpu): Make global.
30088 (arc_selected_arch): Delete.
30089 (arc_base_cpu): Delete.
30090 (arc_override_options): Remove references to deleted variables,
30091 update access to arch information.
30092 (ARC_OPT): Update access to arch information.
30093 (ARC_OPTX): Likewise.
30094 * config/arc/arc.h (arc_base_cpu): Remove declaration.
30095 (TARGET_ARC600): Update access to arch information.
30096 (TARGET_ARC601): Likewise.
30097 (TARGET_ARC700): Likewise.
30098 (TARGET_EM): Likewise.
30099 (TARGET_HS): Likewise.
30100 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
30103 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
30106 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
30107 condition/operands for integer GE/LE/GEU/LEU operations.
30109 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
30111 PR translation/79397
30112 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
30115 2017-02-08 Martin Jambor <mjambor@suse.cz>
30118 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
30119 whether allocation happened.
30120 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
30121 nothing was allocated.
30123 2017-02-08 Jakub Jelinek <jakub@redhat.com>
30125 PR tree-optimization/79408
30126 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
30127 constant, but SSA_NAME with a known integer range, use the minimum
30128 of that range instead of op1 to determine if modulo can be replaced
30129 with its first operand.
30131 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30133 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
30135 2017-02-08 Richard Biener <rguenther@suse.de>
30137 PR tree-optimization/71824
30138 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
30139 Check all loops contained in the merged region.
30141 2017-02-07 Andrew Pinski <apinski@cavium.com>
30143 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
30145 2017-02-07 Andrew Pinski <apinski@cavium.com>
30147 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
30148 (thunderxt88): Likewise.
30149 (thunderxt81): Disable LSE and change v8.1 to v8.
30150 (thunderxt83): Likewise.
30152 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30153 Richard Biener <rguenther@suse.de>
30155 PR middle-end/79399
30156 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
30157 type from int to size_t.
30158 * ira-costs.c (struct_costs_size): Change type from int to size_t.
30160 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30162 PR rtl-optimization/79386
30163 * cprop.c (bypass_conditional_jumps): Initialize
30164 bypass_last_basic_block already before splitting bbs after
30165 unconditional traps...
30166 (bypass_conditional_jumps): ... rather than here.
30169 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
30170 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
30171 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
30172 fix -masm=intel patterns.
30174 2017-02-07 Richard Biener <rguenther@suse.de>
30176 PR tree-optimization/79256
30177 PR middle-end/79278
30178 * builtins.c (get_object_alignment_2): Use min_align_of_type
30179 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
30180 and ADJUST_FIELD_ALIGN.
30182 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
30184 * doc/tm.texi: Regenerate.
30185 * stor-layout.c (layout_decl): Adjust.
30186 (update_alignment_for_field): Likewise.
30187 (place_field): Likewise.
30188 (min_align_of_type): Likewise.
30189 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
30190 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
30191 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
30192 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
30193 * config/frv/frv.c (frv_adjust_field_align): Likewise.
30194 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
30195 * config/i386/i386.c (x86_field_alignment): Likewise.
30196 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
30197 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
30198 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
30199 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
30200 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
30201 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
30205 2017-01-30 Richard Biener <rguenther@suse.de>
30207 PR tree-optimization/79256
30208 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
30209 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
30212 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
30214 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
30215 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
30216 builtins to SImode and emit a zero-extend, if necessary.
30218 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30220 * docs/invoke.texi (RISC-V Options): Alphabetize.
30222 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30224 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
30227 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30229 * config/riscv/riscv.c: New file.
30230 * common/config/riscv/riscv-common.c: Likewise.
30231 * config.gcc: Likewise.
30232 * config/riscv/constraints.md: Likewise.
30233 * config/riscv/elf.h: Likewise.
30234 * config/riscv/generic.md: Likewise.
30235 * config/riscv/linux.h: Likewise.
30236 * config/riscv/multilib-generator: Likewise.
30237 * config/riscv/peephole.md: Likewise.
30238 * config/riscv/pic.md: Likewise.
30239 * config/riscv/predicates.md: Likewise.
30240 * config/riscv/riscv-builtins.c: Likewise.
30241 * config/riscv/riscv-c.c: Likewise.
30242 * config/riscv/riscv-ftypes.def: Likewise.
30243 * config/riscv/riscv-modes.def: Likewise.
30244 * config/riscv/riscv-opts.h: Likewise.
30245 * config/riscv/riscv-protos.h: Likewise.
30246 * config/riscv/riscv.h: Likewise.
30247 * config/riscv/riscv.md: Likewise.
30248 * config/riscv/riscv.opt: Likewise.
30249 * config/riscv/sync.md: Likewise.
30250 * config/riscv/t-elf-multilib: Likewise.
30251 * config/riscv/t-linux: Likewise.
30252 * config/riscv/t-linux-multilib: Likewise.
30253 * config/riscv/t-riscv: Likewise.
30254 * configure.ac: Likewise.
30255 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
30256 Waterman as RISC-V maintainers.
30257 * doc/install.texi: Add RISC-V entries.
30258 * doc/invoke.texi: Add RISC-V options section.
30259 * doc/md.texi: Add RISC-V constraints section.
30260 * configure: Regenerated.
30262 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
30265 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
30266 false values to be constant vectors with all 0 or all 1 bits set.
30267 (vcondu<mode><mode>): Likewise.
30268 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
30270 (fpmask_comparison_operator): Update comment.
30271 (vecint_comparison_operator): New predicate.
30272 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
30273 vector conditionals when the true and false values are constant
30274 vectors with all 0 bits or all 1 bits set.
30276 2017-02-06 Martin Sebor <msebor@redhat.com>
30278 PR tree-optimization/79376
30279 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
30281 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
30283 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
30284 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
30285 to simplify split condition.
30287 2017-02-06 Jakub Jelinek <jakub@redhat.com>
30289 * omp-expand.c (oxpand_omp_atomic_fetch_op,
30290 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
30293 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
30295 PR rtl-optimization/68664
30296 * target.def (can_speculate_insn): New hook.
30297 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
30298 * doc/tm.texi: Regenerate.
30299 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
30300 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
30301 (rs6000_sched_can_speculate_insn): New function.
30303 2017-02-06 Jakub Jelinek <jakub@redhat.com>
30305 PR tree-optimization/79284
30306 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
30307 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
30308 vectorizable_mask_load_store, vectorizable_operation,
30309 vect_is_simple_cond, get_same_sized_vectype): Use it instead
30310 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
30311 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
30312 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
30313 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
30314 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
30315 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
30316 is_gimple_assign (stmt). Replace another such test with
30317 is_gimple_assign (stmt).
30319 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
30322 * config/avr/avr.c (rtl-iter.h): Include it.
30323 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
30324 (avr_legitimate_combined_insn): ...and implementation.
30326 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
30328 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
30329 * config/s390/s390.c (s390_const_operand_ok)
30330 (s390_canonicalize_comparison, s390_extract_part)
30331 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
30332 (s390_contiguous_bitmask_p, s390_rtx_costs)
30333 (legitimize_pic_address): Likewise.
30334 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
30335 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
30336 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
30337 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
30338 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
30340 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
30342 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
30343 REGNO($0) == REGNO($1).
30345 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30347 * config/s390/linux.h(SIZE_TYPE): Add comment.
30349 2017-02-06 Julian Brown <julian@codesourcery.com>
30350 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30351 Virendra Pathak <virendra.pathak@broadcom.com>
30353 * config/aarch64/aarch64-cores.def: Change the scheduler
30355 * config/aarch64/aarch64.md: Include thunderx2t99.md.
30356 * config/aarch64/thunderx2t99.md: New file.
30358 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30360 * doc/standards.texi (Go Language): Update link to language
30363 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30365 * tree-eh.c (lower_resx): Sanitize profile.
30366 (cleanup_empty_eh_move_lp): Likewise.
30368 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30371 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
30373 * cfgloopmanip.h (loop_version): Update prototype.
30374 * modulo-sched.c (sms_schedule): Update call of loop_version.
30375 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
30376 * tree-parloops.c (gen_parallel_loop): Likewise.
30377 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
30378 * tree-ssa-loop-split.c (split_loop): Likewise.
30379 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
30380 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
30382 2017-02-05 Martin Liska <mliska@suse.cz>
30385 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
30387 (print_operand_address): Initialize a struct to zero.
30389 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30391 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
30392 garbage collector only in textual form.
30394 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30396 * doc/extend.texi (x86 specific memory model extensions for
30397 transactional memory): Simplify a phrase.
30399 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
30402 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
30403 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
30404 (atomic_storedi_1): Likewise.
30406 2017-02-04 Jakub Jelinek <jakub@redhat.com>
30408 PR tree-optimization/79338
30409 * tree-parloops.c (gather_scalar_reductions): Don't call
30410 vect_analyze_loop_form for loop->inner before destroying loop's
30413 2017-02-03 Martin Sebor <msebor@redhat.com>
30415 PR tree-optimization/79327
30416 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
30417 when precision has resulted in leading zeros.
30418 (format_integer): Adjust the likely counter to assume an unknown
30419 argument that may be zero is non-zero.
30421 2017-02-03 Jason Merrill <jason@redhat.com>
30424 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
30425 avoid copying non-taken branch.
30427 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30429 PR tree-optimization/79340
30430 * tree-vect-loop.c (vectorizable_reduction): Release
30431 vec_defs elements after safe_splicing them into other vectors.
30434 PR tree-optimization/79327
30435 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
30436 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
30438 (format_integer): Use wide_int_to_tree instead of build_int_cst
30439 + to_?hwi. If argmin is NULL, just set argmin and argmax to
30440 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
30441 of shortest and longest sequence.
30443 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
30445 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
30446 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
30448 2017-02-03 Walter Lee <walt@tilera.com>
30451 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
30452 after initial stackframe link reg save.
30453 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
30455 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30458 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
30459 wu for stxssp alternative.
30461 2017-02-03 Martin Sebor <msebor@redhat.com>
30463 PR tree-optimization/79352
30464 * gimple-fold.c (get_range_strlen): Add argument.
30465 (get_range_strlen): Change return type to bool.
30466 (get_maxval_strlen): Pass in a dummy argument.
30467 * gimple-fold.h (get_range_strlen): Change return type to bool.
30468 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
30469 * tree.h (array_at_struct_end_p): Add argument.
30470 * tree.c (array_at_struct_end_p): Handle it.
30472 2017-02-03 Martin Liska <mliska@suse.cz>
30475 * multiple_target.c (create_dispatcher_calls): Redirect edge
30476 from a caller of a dispatcher.
30477 (expand_target_clones): Make the clones local.
30478 (ipa_target_clone): Do both target clones and resolvers.
30479 (ipa_dispatcher_calls): Remove the pass.
30480 (pass_dispatcher_calls::gate): Likewise.
30481 (make_pass_dispatcher_calls): Likewise.
30482 * passes.def (pass_target_clone): Put as very first IPA early
30485 2017-02-03 Martin Liska <mliska@suse.cz>
30487 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
30488 in case of a function with ifunc attribute.
30490 2017-02-03 Martin Liska <mliska@suse.cz>
30492 * cgraph.c (cgraph_node::dump): Dump function version info.
30493 * symtab.c (symtab_node::dump_base): Add missing new line.
30495 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
30497 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
30498 (ifcombine_ifandif): Use it.
30500 2017-02-03 Martin Liska <mliska@suse.cz>
30502 * doc/invoke.texi: Document default value for
30503 use-after-scope-direct-emission-threshold.
30505 2017-02-03 Martin Liska <mliska@suse.cz>
30507 PR tree-optimization/79339
30508 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
30509 (format_floating): Likewise.
30511 2017-02-03 Martin Liska <mliska@suse.cz>
30514 * ipa-prop.c (ipa_node_params_t::insert): Remove current
30516 (ipa_node_params_t::remove): Likewise.
30517 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
30518 initialization from removed ipa_node_params_t::insert.
30519 (ipa_node_params::~ipa_node_params): Move from removed
30520 ipa_node_params_t::release.
30521 * symbol-summary.h (symbol_summary::m_released): New member.
30522 Do not release a summary twice. Do not allow to call finalizer
30523 for types of a summary that live in GGC memory.
30525 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30527 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
30530 2017-02-02 Martin Sebor <msebor@redhat.com>
30532 PR middle-end/79275
30533 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
30534 (format_string): Tighten up the range of output for non-constant
30535 strings and correct the expected range for wide non-constant strings.
30537 2017-02-02 Martin Sebor <msebor@redhat.com>
30539 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
30541 PR middle-end/32003
30542 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
30544 (-fdump-tree-@var): Add to index and document how to come up
30545 with pass-specific option and dump file names.
30546 (-fdump-passes): Clarify where to look for output.
30548 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
30550 PR middle-end/77445
30551 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
30552 statistics of the analyzed path; allow threading for speed when
30553 any of BBs along the path are optimized for speed.
30555 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
30557 PR middle-end/78468
30558 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
30559 settings of the virtual registers.
30562 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
30564 * explow.c (get_dynamic_stack_size): Take known alignment of stack
30565 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
30568 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30570 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
30571 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
30573 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30575 * config/s390/s390.md: Add missing comments with the expanded
30577 * config/s390/vector.md: Likewise.
30578 * config/s390/vx-builtins.md: Likewise.
30580 2017-02-02 Jakub Jelinek <jakub@redhat.com>
30583 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
30584 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
30585 conditions on a single line.
30587 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30589 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
30590 __S390_VX__ to __VX__.
30592 2017-02-01 Andrew Pinski <apinski@cavium.com>
30594 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
30595 stmt_info to record_stmt_cost.
30596 (vect_get_known_peeling_cost): Pass stmt_info if known to
30598 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
30599 cpu_vector_cost field into
30600 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
30601 field into vec_int_stmt_cost and vec_fp_stmt_cost.
30602 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
30603 splitting of scalar_stmt_cost and vec_stmt_cost.
30604 (thunderx_vector_cost): Likewise.
30605 (cortexa57_vector_cost): LIkewise.
30606 (exynosm1_vector_cost): Likewise.
30607 (xgene1_vector_cost): Likewise.
30608 (thunderx2t99_vector_cost): Improve after the splitting of the two
30610 (aarch64_builtin_vectorization_cost): Update for the splitting of
30611 scalar_stmt_cost and vec_stmt_cost.
30613 2017-02-01 Torvald Riegel <triegel@redhat.com>
30614 Richard Henderson <rth@redhat.com>
30616 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
30617 conditional on existance of a fast atomic load.
30618 * optabs-query.c (can_atomic_load_p): New function.
30619 * optabs-query.h (can_atomic_load_p): Declare it.
30620 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
30621 no fast atomic load is available for the particular size of access.
30622 (expand_atomic_compare_and_swap): Likewise.
30623 (expand_atomic_load): Likewise.
30624 (expand_atomic_store): Likewise.
30625 (expand_atomic_fetch_op): Likewise.
30626 * testsuite/lib/target-supports.exp
30627 (check_effective_target_sync_int_128): Remove x86 because it provides
30628 no fast atomic load.
30629 (check_effective_target_sync_int_128_runtime): Likewise.
30631 2017-02-01 Richard Biener <rguenther@suse.de>
30633 * graphite.c: Include tree-vectorizer.h for find_loop_location.
30634 (graphite_transform_loops): Provide opt-info for optimized nests.
30635 * tree-parloop.c (parallelize_loops): Provide opt-info for
30636 parallelized loops.
30638 2017-02-01 Richard Biener <rguenther@suse.de>
30640 PR middle-end/79315
30641 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
30642 was not set before.
30644 2017-02-01 Richard Biener <rguenther@suse.de>
30646 PR tree-optimization/71824
30647 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
30648 Verify the loops are valid in the merged SESE region.
30649 (scop_detection::can_represent_loop_1): Check analyzing the
30650 evolution of the number of iterations in the region succeeds.
30652 2017-01-31 Ian Lance Taylor <iant@golang.org>
30654 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
30655 REG_ARGS_SIZE note to 32-bit push insns and call insn.
30657 2017-01-31 David Malcolm <dmalcolm@redhat.com>
30659 PR preprocessor/79210
30660 * input.c (get_substring_ranges_for_loc): Replace line_width
30661 assertion with error-handling.
30663 2017-01-31 Richard Biener <rguenther@suse.de>
30665 PR tree-optimization/77318
30666 * graphite-sese-to-poly.c (extract_affine): Fix assert.
30667 (create_pw_aff_from_tree): Take loop parameter.
30668 (add_condition_to_pbb): Pass loop of the condition to
30669 create_pw_aff_from_tree.
30671 2017-01-31 Jakub Jelinek <jakub@redhat.com>
30673 * config/s390/s390.c (s390_asan_shadow_offset): New function.
30674 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
30676 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
30680 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
30682 (convert_int_to_float128): Likewise.
30683 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
30684 (convert_int_to_float128): Likewise.
30685 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
30686 (UNSPEC_IEEE128_CONVERT): Likewise.
30687 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
30688 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
30689 Use local variables for IBM extended format.
30690 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
30691 (fix_trunc<mode>si2_fprs): Likewise.
30692 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
30693 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
30694 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
30695 to know that we can now have integers of all sizes in vector
30697 (fix<uns>_<mode>di2_hw): Likewise.
30698 (float<uns>_<mode>si2_hw): Likewise.
30699 (fix_<mode>si2_hw): Likewise.
30700 (fixuns_<mode>si2_hw): Likewise.
30701 (float<uns>_<mode>di2_hw): Likewise.
30702 (float_<mode>di2_hw): Likewise.
30703 (float_<mode>si2_hw): Likewise.
30704 (floatuns_<mode>di2_hw): Likewise.
30705 (floatuns_<mode>si2_hw): Likewise.
30706 (xscvqp<su>wz_<mode>): Delete, no longer used.
30707 (xscvqp<su>dz_<mode>): Likewise.
30708 (xscv<su>dqp_<mode>): Likewise.
30709 (ieee128_mfvsrd_64bit): Likewise.
30710 (ieee128_mfvsrd_32bit): Likewise.
30711 (ieee128_mfvsrwz): Likewise.
30712 (ieee128_mtvsrw): Likewise.
30713 (ieee128_mtvsrd_64bit): Likewise.
30714 (ieee128_mtvsrd_32bit): Likewise.
30716 2017-01-31 Martin Liska <mliska@suse.cz>
30719 * ipa-prop.c (ipa_free_all_node_params): Call release method
30720 instead of ~sumbol_summary to not to trigger double times
30723 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
30725 PR tree-optimization/71691
30726 * bitmap.h (class auto_bitmap): New.
30727 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
30728 is_maybe_undefined instead of ssa_undefined_value_p.
30730 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30732 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
30733 __S390_ARCH_LEVEL__ to __ARCH__.
30735 2017-01-31 Jakub Jelinek <jakub@redhat.com>
30737 PR tree-optimization/79267
30738 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
30739 if should_remove_lhs_p is true.
30741 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
30744 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
30745 (add_alignment_attribute): New.
30746 (base_type_die): Add alignment attribute.
30747 (subrange_type_die): Likewise.
30748 (modified_type_die): Likewise.
30749 (gen_array_type_die): Likewise.
30750 (gen_descr_array_type_die: Likewise.
30751 (gen_enumeration_type_die): Likewise.
30752 (gen_subprogram_die): Likewise.
30753 (gen_variable_die): Likewise.
30754 (gen_field_die): Likewise.
30755 (gen_ptr_to_mbr_type_die): Likewise.
30756 (gen_struct_or_union_type_die): Likewise.
30757 (gen_subroutine_type_die): Likewise.
30758 (gen_typedef_die): Likewise.
30759 (base_type_cmp): Compare alignment attribute.
30761 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30764 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
30765 (setb_unsigned) New pattern for setb with CCUNS.
30766 * config/rs6000/rs6000.c (expand_block_compare): Use a different
30767 subfc./subfe sequence to avoid overflow problems. Generate a
30768 shorter sequence with cmpld/setb for power9.
30769 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
30770 for generating subfc. instruction.
30771 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
30772 now uses this instruction.
30774 2017-01-30 Ian Lance Taylor <iant@google.com>
30777 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
30778 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
30780 2017-01-30 Martin Sebor <msebor@redhat.com>
30782 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
30783 Move constant to the right of a relational operator.
30784 (get_mpfr_format_length, format_character, format_string): Ditto.
30785 (should_warn_p, maybe_warn): Same.
30787 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
30789 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
30792 * asan.c (get_translation_unit_decl): Remove function.
30793 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
30795 2017-01-30 Martin Liska <mliska@suse.cz>
30797 PR gcov-profile/79259
30798 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
30799 -fprofile-generate.
30801 2017-01-30 Martin Liska <mliska@suse.cz>
30804 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
30805 Initialize variables with NULL value.
30807 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
30810 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
30812 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
30814 2017-01-30 Richard Biener <rguenther@suse.de>
30816 PR tree-optimization/79276
30817 * tree-vrp.c (process_assert_insertions): Properly adjust common
30818 when removing a duplicate.
30820 2017-01-30 Richard Biener <rguenther@suse.de>
30822 PR tree-optimization/79256
30823 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
30824 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
30826 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
30828 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
30831 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
30832 ("*r<noxa>sbg_<mode>_sll_bitmask")
30833 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
30834 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
30835 Use contiguous_bitmask_nowrap_operand.
30837 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30840 * config/rs6000/altivec.h (vec_xl): Revise #define.
30841 (vec_xst): Likewise.
30843 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
30845 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
30847 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
30849 PR rtl-optimization/79194
30850 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
30851 traps before call to bypass_conditional_jumps.
30853 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
30855 PR tree-optimization/71374
30856 * lra-constraints.c (check_conflict_input_operands): New.
30857 (match_reload): Use it.
30859 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
30862 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
30863 account to calculate conflict_set.
30865 2017-01-27 Bin Cheng <bin.cheng@arm.com>
30867 PR rtl-optimization/78559
30868 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
30869 other_insn in combine.
30871 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
30873 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
30874 uint16_type_node for BT_UINT16.
30876 2017-01-27 David Malcolm <dmalcolm@redhat.com>
30878 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
30879 "RTL Tests" to menu.
30880 (GIMPLE Tests): New node.
30881 (RTL Tests): New node.
30883 2017-01-27 Richard Biener <rguenther@suse.de>
30885 PR tree-optimization/79245
30886 * tree-loop-distribution.c (distribute_loop): Apply cost
30887 modeling also to detected patterns.
30889 2017-01-27 Richard Biener <rguenther@suse.de>
30891 PR tree-optimization/71433
30892 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
30893 (compare_assert_loc): New function.
30894 (process_assert_insertions): Sort and optimize assert locations
30895 to remove duplicates and push down identical assertions on
30896 edges to their destination block.
30898 2017-01-27 Richard Biener <rguenther@suse.de>
30900 PR tree-optimization/79244
30901 * tree-vrp.c (remove_range_assertions): Forcefully propagate
30902 out SSA names even if abnormal.
30904 2017-01-27 Jakub Jelinek <jakub@redhat.com>
30906 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
30907 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
30908 instead of MPFR_RNDN.
30910 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
30913 * arm.c (arm_option_override): Don't call build_target_option_node
30914 until after doing all option overrides.
30915 (arm_valid_target_attribute_tree): Likewise.
30917 2017-01-27 Martin Liska <mliska@suse.cz>
30919 * doc/invoke.texi (-fprofile-arcs): Document profiling support
30920 for {cd}tors and C++ {cd}tors.
30922 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
30924 * config/s390/s390.md ("*setmem_long_and")
30925 ("*setmem_long_and_31z"): Use zero_extend instead of and.
30927 2017-01-26 Martin Sebor <msebor@redhat.com>
30929 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
30932 2017-01-26 Martin Sebor <msebor@redhat.com>
30934 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
30935 HAVE_DFmode before using XFmode or DFmode.
30936 (parse_directive): Avoid using the z length modifier to avoid
30937 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
30939 PR middle-end/78703
30940 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
30941 to accept adjustment as an array.
30942 (get_int_range): New function.
30943 (struct directive): Make width and prec arrays.
30944 (directive::set_width, directive::set_precision): Call get_int_range.
30945 (format_integer, format_floating): Handle width and precision ranges.
30946 (format_string, parse_directive): Same.
30948 2017-01-26 Jakub Jelinek <jakub@redhat.com>
30951 * dwarf2out.c (generate_skeleton_bottom_up): For children with
30952 comdat_type_p set, just clone them, but keep the children in the
30956 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
30957 which have direct callers with -fvar-tracking-assignments enabled
30959 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
30960 inside of type units.
30962 2017-01-26 Martin Sebor <msebor@redhat.com>
30964 PR middle-end/78703
30965 * gimple-ssa-sprintf.c (struct result_range): Add likely and
30967 (struct format_result): Replace number_chars, number_chars_min,
30968 and number_chars_max with a single member of struct result_range.
30970 (format_result::operator+=): Adjust.
30971 (struct fmtresult): Remove bounded. Handle likely and unlikely
30973 (fmtresult::adjust_for_width_or_precision): New function.
30974 (fmtresult:type_max_digits): New function.
30975 (bytes_remaining): Handle likely and unlikely counters.
30976 (min_bytes_remaining): Remove.
30977 (format_percent): Simplify.
30978 (format_integer, format_floating): Set likely and unlikely counters.
30979 (get_string_length, format_character, format_string): Same.
30980 (format_plain, should_warn_p): New function.
30981 (maybe_warn): Call should_warn_p. Update diagnostic messages
30982 and handle those for all directives, including plain strings.
30983 (format_directive): Handle likely and unlikely counters.
30984 Remove unnecessary quoting from diagnostics. Add an informational
30986 (add_bytes): Remove.
30987 (pass_sprintf_length::compute_format_length): Simplify.
30988 (try_substitute_return_value): Handle likely and unlikely counters.
30990 2017-01-26 Carl Love <cel@us.ibm.com>
30992 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
30993 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
30995 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
30998 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
30999 endianess for subregs into account.
31000 * lra-constraints.c (lra_constraints): Do risky transformations
31001 always on the first iteration.
31002 * lra-lives.c (check_pseudos_live_through_calls): Add arg
31003 last_call_used_reg_set.
31004 (process_bb_lives): Define and use last_call_used_reg_set.
31005 * lra.c (lra): Always continue after lra_constraints on the first
31008 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
31010 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
31012 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
31014 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31016 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
31017 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
31018 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
31019 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
31020 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
31021 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
31022 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
31023 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
31024 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
31026 2017-01-26 Marek Polacek <polacek@redhat.com>
31029 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
31030 for the third operand.
31032 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31034 PR middle-end/79236
31035 * omp-low.c (struct omp_context): Add simt_stmt field.
31036 (scan_omp_for): Return omp_context *.
31037 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
31038 context to the _simt_ SIMD stmt.
31039 (lower_omp_for): For combined SIMD with sibling _simt_
31040 SIMD, make sure to use the same decls in _looptemp_
31041 clauses as in the sibling.
31043 2017-01-26 David Sherwood <david.sherwood@arm.com>
31045 PR middle-end/79212
31046 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
31049 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31052 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
31053 emit fld b; fld a; if possible.
31055 * brig-builtins.def: Update copyright years.
31056 * config/arm/arm_acle_builtins.def: Update copyright years.
31058 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
31061 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
31062 constraint instead of o for the stxsd instruction.
31064 2017-01-25 Carl Love <cel@us.ibm.com>
31066 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
31067 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
31069 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
31071 * doc/invoke.texi (C++ Dialect Options): Fix typo.
31073 2017-01-25 Richard Biener <rguenther@suse.de>
31075 PR tree-optimization/69264
31076 * target.def (vector_alignment_reachable): Improve documentation.
31077 * doc/tm.texi: Regenerate.
31078 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
31080 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
31081 earlier changes with respect to TYPE_USER_ALIGN.
31082 (vector_alignment_reachable_p): Likewise. Improve dumping.
31084 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31087 * config/arm/arm.md (xordi3): Force constant operand into a register
31090 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31092 * doc/invoke.texi (-fstore-merging): Correct default optimization
31093 levels at which it is enabled.
31094 (-O): Move -fstore-merging from list to...
31097 2017-01-25 Richard Biener <rguenther@suse.de>
31100 * omp-expand.c: Include debug.h.
31101 (expand_omp_taskreg): Make sure to generate early debug before
31102 outlining anything from a function.
31103 (expand_omp_target): Likewise.
31104 (grid_expand_target_grid_body): Likewise.
31106 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
31109 * asan.c (get_translation_unit_decl): New function.
31110 (asan_add_global): Extract modules file name from globals
31111 TRANSLATION_UNIT_DECL name.
31113 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
31116 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
31117 for long calls with APCS frame and VFP.
31119 2017-01-24 David Malcolm <dmalcolm@redhat.com>
31121 * cfg.c (original_copy_tables_initialized_p): New function.
31122 * cfg.h (original_copy_tables_initialized_p): New decl.
31123 * cfgrtl.c (relink_block_chain): Guard the call to
31124 free_original_copy_tables with a call to
31125 original_copy_tables_initialized_p.
31126 * cgraph.h (symtab_node::native_rtl_p): New decl.
31127 * cgraphunit.c (symtab_node::native_rtl_p): New function.
31128 (symtab_node::needed_p): Don't assert for early assembly output
31129 for __RTL functions.
31130 (cgraph_node::finalize_function): Set "force_output" for __RTL
31132 (cgraph_node::analyze): Bail out early for __RTL functions.
31133 (analyze_functions): Update assertion to support __RTL functions.
31134 (cgraph_node::expand): Bail out early for __RTL functions.
31135 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
31137 * function.h (struct function): Update comment for field
31139 * gimple-expr.c: Include "tree-pass.h".
31140 (gimple_has_body_p): Return false for __RTL functions.
31141 * Makefile.in (OBJS): Add run-rtl-passes.o.
31142 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
31144 (gcc::pass_manager::get_clean_slate): New accessor.
31145 * passes.c: Include "insn-addr.h".
31146 (should_skip_pass_p): Add logging. Update logic for running
31147 "expand" to be compatible with both __GIMPLE and __RTL. Guard
31148 property-provider override so it is only done for gimple passes.
31150 (skip_pass): New function.
31151 (execute_one_pass): Call skip_pass when skipping passes.
31152 * read-md.c (md_reader::read_char): Support filtering
31153 the input to a subset of line numbers.
31154 (md_reader::md_reader): Initialize fields
31155 m_first_line and m_last_line.
31156 (md_reader::read_file_fragment): New function.
31157 * read-md.h (md_reader::read_file_fragment): New decl.
31158 (md_reader::m_first_line): New field.
31159 (md_reader::m_last_line): New field.
31160 * read-rtl-function.c (function_reader::create_function): Only
31161 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
31162 curr_properties. Set DECL_INITIAL to a dummy block.
31163 (read_rtl_function_body_from_file_range): New function.
31164 * read-rtl-function.h (read_rtl_function_body_from_file_range):
31166 * run-rtl-passes.c: New file.
31167 * run-rtl-passes.h: New file.
31169 2017-01-24 Jeff Law <law@redhat.com>
31171 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
31174 2017-01-24 Bin Cheng <bin.cheng@arm.com>
31176 PR tree-optimization/79159
31177 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
31178 (record_nonwrapping_iv): Improve boundary using above function if no
31179 value range information.
31181 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
31182 Martin Jambor <mjambor@suse.cz>
31184 * brig-builtins.def: New file.
31185 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
31186 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
31187 (DEF_HSAIL_SAT_BUILTIN): Likewise.
31188 (DEF_HSAIL_INTR_BUILTIN): Likewise.
31189 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
31190 * builtin-types.def (BT_INT8): New.
31191 (BT_INT16): Likewise.
31192 (BT_UINT8): Likewise.
31193 (BT_UINT16): Likewise.
31194 (BT_FN_ULONG): Likewise.
31195 (BT_FN_UINT_INT): Likewise.
31196 (BT_FN_UINT_ULONG): Likewise.
31197 (BT_FN_UINT_LONG): Likewise.
31198 (BT_FN_UINT_PTR): Likewise.
31199 (BT_FN_ULONG_PTR): Likewise.
31200 (BT_FN_INT8_FLOAT): Likewise.
31201 (BT_FN_INT16_FLOAT): Likewise.
31202 (BT_FN_UINT32_FLOAT): Likewise.
31203 (BT_FN_UINT16_FLOAT): Likewise.
31204 (BT_FN_UINT8_FLOAT): Likewise.
31205 (BT_FN_UINT64_FLOAT): Likewise.
31206 (BT_FN_UINT16_UINT32): Likewise.
31207 (BT_FN_UINT32_UINT16): Likewise.
31208 (BT_FN_UINT16_UINT16_UINT16): Likewise.
31209 (BT_FN_INT_PTR_INT): Likewise.
31210 (BT_FN_UINT_PTR_UINT): Likewise.
31211 (BT_FN_LONG_PTR_LONG): Likewise.
31212 (BT_FN_ULONG_PTR_ULONG): Likewise.
31213 (BT_FN_VOID_UINT64_UINT64): Likewise.
31214 (BT_FN_UINT8_UINT8_UINT8): Likewise.
31215 (BT_FN_INT8_INT8_INT8): Likewise.
31216 (BT_FN_INT16_INT16_INT16): Likewise.
31217 (BT_FN_INT_INT_INT): Likewise.
31218 (BT_FN_UINT_FLOAT_UINT): Likewise.
31219 (BT_FN_FLOAT_UINT_UINT): Likewise.
31220 (BT_FN_ULONG_UINT_UINT): Likewise.
31221 (BT_FN_ULONG_UINT_PTR): Likewise.
31222 (BT_FN_ULONG_ULONG_ULONG): Likewise.
31223 (BT_FN_UINT_UINT_UINT): Likewise.
31224 (BT_FN_VOID_UINT_PTR): Likewise.
31225 (BT_FN_UINT_UINT_PTR: Likewise.
31226 (BT_FN_UINT32_UINT64_PTR): Likewise.
31227 (BT_FN_INT_INT_UINT_UINT): Likewise.
31228 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
31229 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
31230 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
31231 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
31232 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
31233 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
31234 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
31235 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
31236 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
31237 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
31238 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
31239 * doc/frontends.texi: List BRIG FE.
31240 * doc/install.texi (Testing): Add BRIG tesring requirements.
31241 * doc/invoke.texi (Overall Options): Mention BRIG.
31242 * doc/standards.texi (Standards): Doucment BRIG HSA version.
31244 2017-01-24 Richard Biener <rguenther@suse.de>
31246 PR translation/79208
31247 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
31249 2017-01-24 Martin Jambor <mjambor@suse.cz>
31252 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
31253 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
31254 and known_contexts.
31256 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
31258 PR middle-end/79123
31259 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
31260 casts from signed to unsigned really don't have a range.
31262 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
31264 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
31265 GMP_RNDx for compatiblity.
31267 2017-01-24 Martin Liska <mliska@suse.cz>
31270 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
31271 that would prevent us to call alloca with -1 as argument.
31273 2017-01-24 Jakub Jelinek <jakub@redhat.com>
31275 * dwarf2out.c (output_compilation_unit_header, output_file_names):
31276 Avoid -Wformat-security warning.
31278 2017-01-23 Andrew Pinski <apinski@cavium.com>
31280 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
31283 2017-01-23 Martin Sebor <msebor@redhat.com>
31285 PR middle-end/78703
31286 * gimple-ssa-sprintf.c (warn_level): New global.
31287 (format_integer): Use it here and throughout the rest of the file.
31288 Use the same switch to compute sign as base.
31289 (maybe_warn): New function.
31290 (format_directive): Factor out warnings into maybe_warn.
31291 Add debugging output. Use warn_level.
31292 (add_bytes): Use warn_level.
31293 (pass_sprintf_length::compute_format_length): Add debugging output.
31294 (try_substitute_return_value): Same.
31295 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
31297 PR middle-end/78703
31298 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
31299 (struct fmtresult, format_integer, format_floating): Adjust.
31300 (fmtresult::fmtresult): Set max correctly in two argument ctor.
31301 (get_string_length, format_string,format_directive): Same.
31302 (pass_sprintf_length::compute_format_length): Same.
31303 (try_substitute_return_value): Simplify slightly.
31305 PR middle-end/78703
31306 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
31307 (fmtresult::operator+=): Outlined.
31308 (struct fmtresult): Add ctors.
31309 (struct conversion_spec): Rename...
31310 (struct directive): ...to this. Add and remove data members.
31311 (directive::set_width, directive::set_precision): New functions.
31312 (format_percent): Use fmtresult ctor.
31313 (get_width_and_precision): Remove.
31314 (format_integer): Make naming changes. Avoid computing width and
31316 (format_floating): Same. Adjust indentation.
31317 (format_character, format_none): New functions.
31318 (format_string): Moved character handling to format_character.
31319 (format_directive): Remove arguments, change return type.
31320 (parse_directive): New function.
31321 (pass_sprintf_length::compute_format_length): Move directive
31322 parsing to parse_directive.
31324 2017-01-23 Jakub Jelinek <jakub@redhat.com>
31326 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
31327 (assign_assembler_name_if_needed): ... this.
31328 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
31329 (assign_assembler_name_if_needed): ... this.
31330 (free_lang_data_in_cgraph): Adjust callers.
31331 * cgraphunit.c (cgraph_node::analyze): Likewise.
31332 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
31335 2017-01-23 Richard Biener <rguenther@suse.de>
31337 PR tree-optimization/79088
31338 PR tree-optimization/79188
31339 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
31340 resetting loop bounds after last path deletion. Reset loop
31341 bounds of the target loop, make code match the comments.
31342 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
31343 Make sure loops need no fixups.
31345 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
31347 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
31348 exponent support with double type for first argument.
31349 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
31350 type returned by __builtin_vec_extract_sig,
31351 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
31352 functions from "vector int" to "vector unsigned int" or from
31353 "vector long long int" to "vector unsigned long long int".
31354 Changed type returned by __builtin_vec_extract_exp,
31355 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
31356 functions from "vector int" to "vector unsigned int" or from
31357 "vector long long int" to "vector unsigned long long int".
31358 Changed return type of __builtin_vec_test_data_class,
31359 __builtin_vec_test_data_class_sp, and
31360 __builtin_vec_test_data_class_dp from "vector int" to
31361 "vector bool int" or from "vector long long int" to "vector bool
31362 long long int" and changed second argument type from "unsigned
31363 int" to "int". Added new overloaded function forms "vector float
31364 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
31365 "vector float __builtin_vec_insert_exp_sp (vector float, vector
31366 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
31367 double, vector unsigned long long int)" and "vector double
31368 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
31369 long int)". Changed return type of
31370 __builtin_scalar_test_data_class and
31371 __builtin_scalar_test_data_class_sp and
31372 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
31373 int" and changed second argument from "unsigned int" to "int".
31374 Changed type returned by __builtin_scalar_test_neg,
31375 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
31376 from "int" to "bool int". Added new overloaded function form
31377 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
31378 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
31379 exponent double-precision with floating point first argument.
31380 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
31381 documentation of scalar_test_data_class, scalar_test_neg,
31382 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
31383 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
31384 vec_test_data_class built-in functions to reflect refinements in
31385 their type signatures.
31387 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
31389 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
31391 (aarch64_elf_asm_destructor): Likewise.
31393 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
31395 PR rtl-optimization/78634
31396 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
31397 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
31398 * ifcvt.c (noce_try_cmove): Add missing cost check.
31400 PR rtl-optimization/71724
31401 * combine.c (if_then_else_cond): Look for situations where it is
31402 beneficial to undo the work of one of the recursive calls.
31404 2017-01-23 Bin Cheng <bin.cheng@arm.com>
31406 PR tree-optimization/70754
31407 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
31408 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
31409 combined stmt before it if not NULL.
31410 (combine_chains): Process refs reversely and compute dominance point
31413 2017-01-23 Martin Liska <mliska@suse.cz>
31415 PR tree-optimization/79196
31416 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
31417 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
31419 (strlen_optimize_stmt): Call the renamed function.
31421 2017-01-23 Michael Matz <matz@suse.de>
31423 PR tree-optimization/78384
31424 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
31426 2017-01-23 Richard Biener <rguenther@suse.de>
31428 PR tree-optimization/79186
31429 * tree-vrp.c (register_new_assert_for): Make sure we've seen
31430 both incoming edges before moving an assert.
31432 2017-01-23 Martin Jambor <mjambor@suse.cz>
31434 * ipa-prop.c (load_from_param_1): Removed.
31435 (load_from_unmodified_param): Bits from load_from_param_1 put back
31437 (load_from_param): Removed.
31438 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
31439 with stmt. Reverted back to use of load_from_unmodified_param.
31441 2017-01-23 Martin Jambor <mjambor@suse.cz>
31444 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
31445 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
31446 field a pointer to garbage collected vector, mark lattices and
31447 ipcp_orig_node with GTY((skip)).
31448 (ipa_get_param_count): Adjust to descriptors being a pointer.
31449 (ipa_get_param): Likewise.
31450 (ipa_get_type): Likewise.
31451 (ipa_get_param_move_cost): Likewise.
31452 (ipa_set_param_used): Likewise.
31453 (ipa_get_controlled_uses): Likewise.
31454 (ipa_set_controlled_uses): Likewise.
31455 (ipa_is_param_used): Likewise.
31456 (ipa_node_params_t): Move into garbage collector. New methods insert
31458 (ipa_node_params_sum): Annotate wth GTY(()).
31459 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
31461 (ipa_load_from_parm_agg): Adjust declaration.
31462 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
31463 * ipa-profile.c (ipa_profile): Likewise.
31464 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
31465 (ipa_populate_param_decls): Make descriptors parameter garbage
31467 (ipa_dump_param): Adjust to descriptors being a pointer.
31468 (ipa_alloc_node_params): Likewise.
31469 (ipa_initialize_node_params): Likewise.
31470 (load_from_param_1): Make descriptors parameter garbage collected.
31471 (load_from_unmodified_param): Likewise.
31472 (load_from_param): Likewise.
31473 (ipa_load_from_parm_agg): Likewise.
31474 (ipa_node_params::~ipa_node_params): Removed.
31475 (ipa_free_all_node_params): Remove call to delete operator.
31476 (ipa_node_params_t::insert): New.
31477 (ipa_node_params_t::remove): Likewise.
31478 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
31479 copy known_csts and known_contexts vectors.
31480 (ipa_read_node_info): Adjust to descriptors being a pointer.
31481 (ipcp_modif_dom_walker): Make m_descriptors field garbage
31483 (ipcp_transform_function): Make descriptors variable garbage
31486 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
31488 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
31489 * config/i386/avx512dqintrin.h: Ditto.
31490 * config/i386/avx512fintrin.h: Ditto.
31491 * config/i386/i386.c: Handle new builtins.
31492 * config/i386/i386-builtin.def: Add new builtins.
31493 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
31494 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
31496 2017-01-23 Jakub Jelinek <jakub@redhat.com>
31497 Martin Liska <mliska@suse.cz>
31499 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
31500 * asan.c (asan_expand_poison_ifn): Support stores and use
31501 appropriate ASAN report function.
31502 * internal-fn.c (expand_ASAN_POISON_USE): New function.
31503 * internal-fn.def (ASAN_POISON_USE): Declare.
31504 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
31505 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
31506 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
31507 ASAN_POISON calls w/o LHS.
31508 * tree-ssa.c (execute_update_addresses_taken): Create clobber
31509 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
31510 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
31511 * gimplify.c (asan_poison_variables): Add attribute
31512 use_after_scope_memory to variables that really needs to live
31514 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
31515 having the attribute.
31517 2017-01-23 Martin Liska <mliska@suse.cz>
31519 * asan.c (create_asan_shadow_var): New function.
31520 (asan_expand_poison_ifn): Likewise.
31521 * asan.h (asan_expand_poison_ifn): New declaration.
31522 * internal-fn.c (expand_ASAN_POISON): Likewise.
31523 * internal-fn.def (ASAN_POISON): New builtin.
31524 * sanopt.c (pass_sanopt::execute): Expand
31525 asan_expand_poison_ifn.
31526 * tree-inline.c (copy_decl_for_dup_finish): Make function
31528 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
31529 * tree-ssa.c (is_asan_mark_p): New function.
31530 (execute_update_addresses_taken): Rewrite local variables
31531 (identified just by use-after-scope as addressable) into SSA.
31533 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
31535 * doc/install.texi (Specific): opensource.apple.com uses https
31536 now. Remove trailing slash.
31538 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
31540 * README.Portability: Remove note on an Irix compatibility issue.
31542 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
31544 * gcov.c (INCLUDE_ALGORITHM): Define.
31545 (INCLUDE_VECTOR): Define.
31546 No longer include <vector> and <algorithm> directly.
31548 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
31550 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
31552 * doc/invoke.texi (Code Gen Options): Ditto.
31554 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
31557 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
31559 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
31561 rtl-optimization/79125
31562 * cprop.c (local_cprop_pass): Handle cases where we make an
31563 unconditional trap.
31565 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
31569 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
31570 read from, for big endian.
31572 2017-01-20 Jiong Wang <jiong.wang@arm.com>
31574 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
31575 register pauth builtins for LP64 only.
31577 2017-01-20 Marek Polacek <polacek@redhat.com>
31580 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
31583 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
31585 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
31587 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
31588 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
31589 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
31591 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31594 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
31595 in big-endian mode when they are not a single duplicated value.
31597 2017-01-20 Richard Biener <rguenther@suse.de>
31599 * BASE-VER: Bump to 7.0.1.
31601 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
31603 * omp-low.c (omplow_simd_context): New struct. Use it...
31604 (lower_rec_simd_input_clauses): ...here and...
31605 (lower_rec_input_clauses): ...here to hold common data. Adjust all
31606 references to idx, lane, max_vf, is_simt.
31608 2017-01-20 Graham Markall <graham.markall@embecosm.com>
31610 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
31613 2017-01-20 Martin Jambor <mjambor@suse.cz>
31615 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
31616 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
31618 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
31619 (GTFILES): Rename hsa.c to hsa-common.c.
31620 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
31621 * hsa-dump.c: Likewise.
31622 * hsa-gen.c: Likewise.
31623 * hsa-regalloc.c: Likewise.
31624 * ipa-hsa.c: Likewise.
31625 * omp-expand.c: Likewise.
31626 * omp-low.c: Likewise.
31627 * toplev.c: Likewise.
31629 2017-01-20 Marek Polacek <polacek@redhat.com>
31632 * doc/invoke.texi: Document -Wduplicated-branches.
31633 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
31634 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
31635 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
31636 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
31637 return 0 only when not OEP_LEXICOGRAPHIC.
31638 (fold_build_cleanup_point_expr): Use the expression
31639 location when building CLEANUP_POINT_EXPR.
31640 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
31641 * tree.c (add_expr): Handle error_mark_node.
31643 2017-01-20 Martin Liska <mliska@suse.cz>
31646 * tree-profile.c (init_ic_make_global_vars): Do not call
31648 (gimple_init_gcov_profiler): Likewise.
31650 2017-01-20 Martin Liska <mliska@suse.cz>
31653 * cgraph.h (maybe_create_reference): Remove argument and
31655 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
31657 * ipa-cp.c (create_specialized_node): Likewise.
31658 * symtab.c (symtab_node::maybe_create_reference): Handle
31659 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
31661 2017-01-20 Martin Liska <mliska@suse.cz>
31663 * read-rtl-function.c (function_reader::create_function): Use
31664 build_decl instread of build_decl_stat.
31666 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
31668 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
31669 * config/i386/avx512dqintrin.h: Ditto.
31670 * config/i386/avx512fintrin.h: Ditto.
31671 * config/i386/i386-builtin-types.def: Add new types.
31672 * config/i386/i386.c: Handle new types.
31673 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
31674 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
31675 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
31676 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
31677 (__builtin_ia32_kshiftridi): New.
31678 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
31680 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
31684 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
31685 define to rs6000_init_stack_protect_guard.
31686 (rs6000_init_stack_protect_guard): New function.
31688 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
31689 Yunqiang Su <yunqiang.su@imgtec.com>
31691 * config.gcc (supported_defaults): Add madd4.
31692 (with_madd4): Add validation.
31693 (all_defaults): Add madd4.
31694 * config/mips/mips.opt (mmadd4): New option.
31695 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
31697 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
31699 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
31700 (ISA_HAS_FUSED_MADD4): Likewise.
31701 * doc/invoke.texi (-mmadd4): Document the new option.
31702 * doc/install.texi (--with-madd4): Document the new option.
31704 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31706 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
31707 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
31708 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
31709 (aarch64_init_pauth_hint_builtins): New.
31710 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
31711 (aarch64_expand_builtin): Expand new builtins.
31713 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31715 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
31716 * combine-stack-adj.c (no_unhandled_cfa): Handle
31717 REG_CFA_TOGGLE_RA_MANGLE.
31718 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
31719 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
31720 info for return address signing.
31721 (aarch64_expand_epilogue): Likewise.
31723 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31725 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
31726 * config/aarch64/aarch64-protos.h
31727 (aarch64_return_address_signing_enabled): New declaration.
31728 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
31730 (aarch64_expand_prologue): Sign return address before it's pushed onto
31732 (aarch64_expand_epilogue): Authenticate return address fetched from
31734 (aarch64_override_options): Sanity check for ILP32 and ISA level.
31735 (aarch64_attributes): New function attributes for "sign-return-address".
31736 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
31737 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
31738 ("*do_return"): Generate combined instructions according to key index.
31739 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
31740 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
31742 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
31743 * config/aarch64/aarch64.opt (msign-return-address=): New.
31744 * doc/extend.texi (AArch64 Function Attributes): Documents
31745 "sign-return-address=".
31746 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
31748 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
31750 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
31751 overall option summary.
31753 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31755 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
31756 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
31757 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
31758 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
31760 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
31762 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
31763 -mpower9-minmax by default for -mcpu=power9.
31764 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
31765 128-bit floating point.
31767 2017-01-20 Alan Modra <amodra@gmail.com>
31769 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
31770 optimizing for size.
31772 2017-01-20 Alan Modra <amodra@gmail.com>
31775 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
31776 for strcmp and strncmp from corresponding builtin decl.
31778 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
31780 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
31781 instead of i386/rtems-64.h.
31782 * config/i386/rtems-64.h: Remove.
31784 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
31788 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
31790 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
31792 2017-01-19 Tamar Christina <tamar.christina@arm.com>
31794 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
31795 Change int to HOST_WIDE_INT.
31796 * config/aarch64/aarch64-protos.h
31797 (aarch64_simd_gen_const_vector_dup): Likewise.
31798 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
31800 2017-01-19 David Malcolm <dmalcolm@redhat.com>
31802 * langhooks-def.h (lhd_type_for_size): New decl.
31803 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
31804 * langhooks.c (lhd_type_for_size): New function, taken from
31807 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
31809 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
31810 define_bypass for CR latency.
31811 (power9-cracked-alu): Update bypass latency and remove power9-branch.
31812 (power9-alu2): Add define_bypass for CR latency.
31814 (power9-mul): Update insn latency.
31815 (power9-mul-compare): Update insn latency, bypass latency and remove
31818 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31820 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
31822 * config/aarch64/aarch64.md
31823 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
31824 aarch64_nopcrelative_literal_loads.
31825 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
31827 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
31829 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
31830 TARGET_LOONGSON_3A.
31831 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
31833 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
31836 * config.gcc (supported_defaults): Add lxc1-sxc1.
31837 (with_lxc1_sxc1): Add validation.
31838 (all_defaults): Add lxc1-sxc1.
31839 * config/mips/mips.opt (mlxc1-sxc1): New option.
31840 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
31842 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
31843 __mips_no_lxc1_sxc1.
31844 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
31845 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
31846 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
31848 2017-01-19 Richard Biener <rguenther@suse.de>
31850 PR tree-optimization/72488
31851 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
31852 sure to restore SSA info.
31853 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
31855 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
31857 PR rtl-optimization/79121
31858 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
31859 of the inner type when shifting an extended value.
31861 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
31864 * symtab.c (symtab_node::equal_address_to): Fix comparing of
31865 interposable aliases.
31867 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
31870 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
31871 Use the evmergelohi instruction.
31872 (mov_si<mode>_e500_subreg4_2_le): Likewise.
31873 (mov_sitf_e500_subreg8_2_be): Likewise.
31874 (mov_sitf_e500_subreg12_2_le): Likewise.
31875 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
31876 (mov_si<mode>_e500_subreg4_2_be): Likewise.
31877 (mov_sitf_e500_subreg8_2_le): Likewise.
31878 (mov_sitf_e500_subreg12_2_be): Likewise.
31880 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31882 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
31883 attribute from vecsimple to vecperm.
31884 (altivec_vbpermq2): Likewise.
31886 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31889 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
31891 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31892 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
31893 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
31894 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
31895 case where N arg is SIZE_MAX.
31896 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
31897 (cmpstrsi): Add pattern.
31899 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
31901 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
31902 __builtin_vec_revb builtins.
31903 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
31904 built-in functions to support generation of the ISA 3.0 XXBR<x>
31905 vector byte reverse instructions.
31906 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
31907 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
31908 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
31909 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
31910 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
31911 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
31912 (P9V_BUILTIN_VEC_REVB): Likewise.
31913 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
31914 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
31915 (p9_xxbrq_v16qi): Likewise.
31916 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
31917 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
31918 (p9_xxbrh_v8hi): Likewise.
31919 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
31920 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
31921 vec_revb built-in functions.
31923 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
31925 PR rtl-optimization/78952
31926 * config/i386/i386.md (any_extract): New code iterator.
31927 (*insvqi_2): Use any_extract for source operand.
31928 (*insvqi_3): Use any_shiftrt for source operand.
31930 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
31932 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
31934 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
31936 2017-01-18 Matthias Klose <doko@ubuntu.com>
31938 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
31940 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31942 * config/rs6000/altivec.h (vec_bperm): Change #define.
31943 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
31944 (altivec_vbpermq2): New define_insn.
31945 (altivec_vbpermd): Likewise.
31946 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
31947 function interface.
31948 (VBPERMD): Likewise.
31949 (VBPERM): New polymorphic function interface.
31950 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
31951 Add entries for P9V_BUILTIN_VEC_VBPERM.
31952 * doc/extend.texi: Add interfaces for vec_bperm.
31954 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31956 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
31957 first letter of error messages.
31958 (s390_resolve_overloaded_builtin): Likewise.
31959 * config/s390/s390.c (s390_expand_builtin): Likewise.
31960 (s390_invalid_arg_for_unprototyped_fn): Likewise.
31961 (s390_valid_target_attribute_inner_p): Likewise.
31962 * config/s390/s390.md ("tabort"): Likewise.
31964 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
31966 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
31967 (ISA_AVOID_DIV_HILO): New macro.
31968 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
31969 (ISA_HAS_DDIV): Likewise.
31971 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
31973 * doc/invoke.texi (fabi-version): Correct number of occurrences.
31975 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
31977 * doc/invoke.texi (fabi-version): Spelling fix.
31979 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
31982 * doc/invoke.texi (fabi-version): Mention mangling fix for
31985 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
31988 * doc/invoke.texi (fabi-version): Document discriminator mangling.
31990 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
31993 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
31994 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
31996 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
31998 (stack_protect_test): Ditto.
31999 * config/rs6000/rs6000.opt (mstack-protector-guard=,
32000 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
32002 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
32003 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
32004 -mstack-protector-guard-offset=.
32005 (RS/6000 and PowerPC Options): Ditto.
32007 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32009 * config/i386/i386.h (MASK_CLASS_P): New define.
32010 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
32011 there are no registers from different register sets also when
32012 mask registers are used. Update function comment.
32013 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
32014 to (*k/*r) and (*k/*km) alternatives.
32016 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
32018 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
32019 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
32020 (EH_RETURN_HANDLER_RTX): New define.
32021 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
32022 Force frame pointer in EH return functions.
32023 (aarch64_expand_epilogue): Add barrier for eh_return.
32024 (aarch64_final_eh_return_addr): Remove.
32025 (aarch64_eh_return_handler_rtx): New function.
32026 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
32028 (aarch64_eh_return_handler_rtx): New prototype.
32030 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32032 * config/rs6000/altivec.h (vec_rlmi): New #define.
32033 (vec_vrlnm): Likewise.
32034 (vec_rlnm): Likewise.
32035 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
32036 (UNSPEC_VRLNM): Likewise.
32037 (VIlong): New mode iterator.
32038 (altivec_vrl<VI_char>mi): New define_insn.
32039 (altivec_vrl<VI_char>nm): Likewise.
32040 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
32042 (VRLDNM): Likewise.
32043 (RLNM): New polymorphic function entry.
32044 (VRLWMI): New monomorphic function entry.
32045 (VRLDMI): Likewise.
32046 (RLMI): New polymorphic function entry.
32047 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
32048 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
32049 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
32052 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32055 * dwarf2out.c (field_byte_offset): Restore the
32056 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
32057 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
32058 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
32061 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32064 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
32066 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32069 * dwarf2out.c (add_data_member_location_attribute): For constant
32070 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
32071 instead of DW_AT_data_member_location, DW_AT_bit_offset and
32072 DW_AT_byte_size attributes.
32074 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32076 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
32077 after forcing to constant memory when the code model is medium.
32079 2017-01-17 Julia Koval <julia.koval@intel.com>
32082 * config/i386/avx512fintrin.h
32083 (_mm512_i32gather_ps): Change __addr type to void const*.
32084 (_mm512_mask_i32gather_ps): Ditto.
32085 (_mm512_i32gather_pd): Ditto.
32086 (_mm512_mask_i32gather_pd): Ditto.
32087 (_mm512_i64gather_ps): Ditto.
32088 (_mm512_mask_i64gather_ps): Ditto.
32089 (_mm512_i64gather_pd): Ditto.
32090 (_mm512_mask_i64gather_pd): Ditto.
32091 (_mm512_i32gather_epi32): Ditto.
32092 (_mm512_mask_i32gather_epi32): Ditto.
32093 (_mm512_i32gather_epi64): Ditto.
32094 (_mm512_mask_i32gather_epi64): Ditto.
32095 (_mm512_i64gather_epi32): Ditto.
32096 (_mm512_mask_i64gather_epi32): Ditto.
32097 (_mm512_i64gather_epi64): Ditto.
32098 (_mm512_mask_i64gather_epi64): Ditto.
32099 (_mm512_i32scatter_ps): Change __addr type to void*.
32100 (_mm512_mask_i32scatter_ps): Ditto.
32101 (_mm512_i32scatter_pd): Ditto.
32102 (_mm512_mask_i32scatter_pd): Ditto.
32103 (_mm512_i64scatter_ps): Ditto.
32104 (_mm512_mask_i64scatter_ps): Ditto.
32105 (_mm512_i64scatter_pd): Ditto.
32106 (_mm512_mask_i64scatter_pd): Ditto.
32107 (_mm512_i32scatter_epi32): Ditto.
32108 (_mm512_mask_i32scatter_epi32): Ditto.
32109 (_mm512_i32scatter_epi64): Ditto.
32110 (_mm512_mask_i32scatter_epi64): Ditto.
32111 (_mm512_i64scatter_epi32): Ditto.
32112 (_mm512_mask_i64scatter_epi32): Ditto.
32113 (_mm512_i64scatter_epi64): Ditto.
32114 (_mm512_mask_i64scatter_epi64): Ditto.
32115 * config/i386/avx512pfintrin.h
32116 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
32117 (_mm512_mask_prefetch_i32gather_ps): Ditto.
32118 (_mm512_mask_prefetch_i64gather_pd): Ditto.
32119 (_mm512_mask_prefetch_i64gather_ps): Ditto.
32120 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
32121 (_mm512_prefetch_i32scatter_ps): Ditto.
32122 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
32123 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
32124 (_mm512_prefetch_i64scatter_pd): Ditto.
32125 (_mm512_prefetch_i64scatter_ps): Ditto.
32126 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
32127 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
32128 * config/i386/avx512vlintrin.h
32129 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
32130 (_mm_mmask_i32gather_ps): Ditto.
32131 (_mm256_mmask_i32gather_pd): Ditto.
32132 (_mm_mmask_i32gather_pd): Ditto.
32133 (_mm256_mmask_i64gather_ps): Ditto.
32134 (_mm_mmask_i64gather_ps): Ditto.
32135 (_mm256_mmask_i64gather_pd): Ditto.
32136 (_mm_mmask_i64gather_pd): Ditto.
32137 (_mm256_mmask_i32gather_epi32): Ditto.
32138 (_mm_mmask_i32gather_epi32): Ditto.
32139 (_mm256_mmask_i32gather_epi64): Ditto.
32140 (_mm_mmask_i32gather_epi64): Ditto.
32141 (_mm256_mmask_i64gather_epi32): Ditto.
32142 (_mm_mmask_i64gather_epi32): Ditto.
32143 (_mm256_mmask_i64gather_epi64): Ditto.
32144 (_mm_mmask_i64gather_epi64): Ditto.
32145 (_mm256_i32scatter_ps): Change __addr type to void*.
32146 (_mm256_mask_i32scatter_ps): Ditto.
32147 (_mm_i32scatter_ps): Ditto.
32148 (_mm_mask_i32scatter_ps): Ditto.
32149 (_mm256_i32scatter_pd): Ditto.
32150 (_mm256_mask_i32scatter_pd): Ditto.
32151 (_mm_i32scatter_pd): Ditto.
32152 (_mm_mask_i32scatter_pd): Ditto.
32153 (_mm256_i64scatter_ps): Ditto.
32154 (_mm256_mask_i64scatter_ps): Ditto.
32155 (_mm_i64scatter_ps): Ditto.
32156 (_mm_mask_i64scatter_ps): Ditto.
32157 (_mm256_i64scatter_pd): Ditto.
32158 (_mm256_mask_i64scatter_pd): Ditto.
32159 (_mm_i64scatter_pd): Ditto.
32160 (_mm_mask_i64scatter_pd): Ditto.
32161 (_mm256_i32scatter_epi32): Ditto.
32162 (_mm256_mask_i32scatter_epi32): Ditto.
32163 (_mm_i32scatter_epi32): Ditto.
32164 (_mm_mask_i32scatter_epi32): Ditto.
32165 (_mm256_i32scatter_epi64): Ditto.
32166 (_mm256_mask_i32scatter_epi64): Ditto.
32167 (_mm_i32scatter_epi64): Ditto.
32168 (_mm_mask_i32scatter_epi64): Ditto.
32169 (_mm256_i64scatter_epi32): Ditto.
32170 (_mm256_mask_i64scatter_epi32): Ditto.
32171 (_mm_i64scatter_epi32): Ditto.
32172 (_mm_mask_i64scatter_epi32): Ditto.
32173 (_mm256_i64scatter_epi64): Ditto.
32174 (_mm256_mask_i64scatter_epi64): Ditto.
32175 (_mm_i64scatter_epi64): Ditto.
32176 (_mm_mask_i64scatter_epi64): Ditto.
32177 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
32178 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
32179 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
32180 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
32181 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
32182 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
32183 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
32184 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
32185 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
32186 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
32187 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
32188 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
32189 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
32190 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
32191 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
32192 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
32193 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
32194 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
32195 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
32196 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
32197 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
32198 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
32199 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
32200 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
32201 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
32202 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
32203 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
32204 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
32205 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
32206 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
32207 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
32208 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
32209 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
32210 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
32211 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
32212 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
32213 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
32214 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
32215 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
32216 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
32217 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
32218 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
32219 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
32220 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
32221 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
32222 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
32223 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
32224 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
32225 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
32226 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
32227 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
32228 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
32229 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
32230 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
32231 definitions accordingly.
32233 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
32234 Kuan-Lin Chen <kuanlinchentw@gmail.com>
32237 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
32240 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
32243 * ira-conflicts.c (ira_build_conflicts): Update total conflict
32244 hard regs for inner regno.
32246 2017-01-17 Martin Liska <mliska@suse.cz>
32249 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
32250 assumption and add comment.
32252 2017-01-17 Nathan Sidwell <nathan@acm.org>
32254 * ipa-visibility.c (localize_node): New function, broken out of ...
32255 (function_and_variable_visibility): ... here. Call it.
32257 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
32259 PR middle-end/77445
32260 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
32261 correctly set frequency of oudgoing edge.
32262 (duplicate_thread_path): Fix profile updating.
32264 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32267 * configure.ac: Add GCC_BASE_VER.
32268 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
32269 version from BASE-VER file.
32270 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
32271 (gcc.o): Depend on $(BASEVER).
32272 * common.opt (dumpfullversion): New option.
32273 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
32274 * doc/invoke.texi: Document -dumpfullversion.
32275 * doc/install.texi: Document --with-gcc-major-version-only.
32276 * configure: Regenerated.
32278 2017-01-17 Richard Biener <rguenther@suse.de>
32280 PR tree-optimization/71433
32281 * tree-vrp.c (register_new_assert_for): Merge same asserts
32282 on all incoming edges.
32283 (process_assert_insertions_for): Handle insertions at the
32286 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
32288 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
32289 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
32291 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
32294 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
32297 2017-01-17 Alan Modra <amodra@gmail.com>
32300 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
32301 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
32302 symbolic stack limit when pic.
32304 2017-01-16 Martin Sebor <msebor@redhat.com>
32306 PR tree-optimization/78608
32307 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
32309 2017-01-16 Jeff Law <law@redhat.com>
32312 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
32313 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
32314 for several include directories that may be relative to sysroot.
32315 * config/i386/x-mingw32 (gplus_includedir): Define.
32316 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
32317 (native_system_includedir): Likewise.
32318 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
32319 override if TARGET_SYSTEM_ROOT is defined.
32320 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
32322 PR tree-optimization/79090
32323 PR tree-optimization/33562
32324 PR tree-optimization/61912
32325 PR tree-optimization/77485
32326 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
32327 and computed trims into the dump file.
32329 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32331 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
32333 2017-01-16 Jakub Jelinek <jakub@redhat.com>
32336 * gimplify.c (gimplify_init_constructor): If want_value and
32337 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
32341 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
32342 sequence. Formatting fixes.
32343 (doloop_optimize): Formatting fixes.
32346 * gcc.c (debug_level_greater_than_spec_func): New function.
32347 (static_spec_functions): Add debug-level-gt spec function.
32348 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
32350 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32351 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32352 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
32353 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
32354 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
32355 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
32357 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
32359 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
32360 QImode fixups to general and mask registers only.
32362 2017-01-16 Carl Love <cel@us.ibm.com>
32364 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
32365 for built-in functions
32366 vector signed char vec_nabs (vector signed char)
32367 vector signed short vec_nabs (vector signed short)
32368 vector signed int vec_nabs (vector signed int)
32369 vector signed long long vec_nabs (vector signed long long)
32370 vector float vec_nabs (vector float)
32371 vector double vec_nabs (vector double)
32372 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
32374 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
32375 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
32376 * doc/extend.texi: Update the documentation file for the new built-in
32379 2017-01-16 Martin Sebor <msebor@redhat.com>
32381 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
32384 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32386 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
32387 UNSPEC_VSX__XXSPLTD to require special splat handling.
32389 2017-01-16 David Malcolm <dmalcolm@redhat.com>
32392 * system.h: Poison strndup.
32394 2017-01-16 Alan Modra <amodra@gmail.com>
32397 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
32400 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
32402 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
32404 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
32406 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
32407 call recog here. Assert that INSN_CODE (insn) is non-negative.
32409 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32412 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
32414 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
32415 in the currently scheduled RTL fragment.
32417 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32419 PR rtl-optimization/78751
32420 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
32423 2017-01-14 Jeff Law <law@redhat.com>
32425 PR tree-optimization/79090
32426 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
32427 variable length stores.
32428 (compute_trims): Delete dead assignment to *trim_tail.
32429 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
32432 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
32434 PR rtl-optimization/78626
32435 PR rtl-optimization/78727
32436 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
32437 of a block, and split such blocks after everything else is finished.
32439 2017-01-14 Alan Modra <amodra@gmail.com>
32442 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
32443 target legitimate_combined_insn.
32444 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
32445 (rs6000_legitimate_combined_insn): New function.
32446 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
32448 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
32449 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
32450 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
32452 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
32454 * doc/frontends.texi (G++ and GCC): Remove references to Java.
32456 2017-01-13 Jeff Law <law@redhat.com>
32458 PR tree-optimization/33562
32459 PR tree-optimization/61912
32460 PR tree-optimization/77485
32461 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
32463 (delete_dead_assignment): Likewise.
32464 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
32465 statement to delete_dead_call and delete_dead_assignment.
32467 2017-01-13 David Malcolm <dmalcolm@redhat.com>
32470 * substring-locations.c (format_warning_va): Strengthen case 1 so
32471 that both endpoints of the substring must be within the format
32472 range for just the substring to be printed.
32474 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
32476 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
32477 * config/i386/i386.c (ix86_target_string): Add missing options
32478 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
32479 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
32480 flags_other and ix86_target_other to flags2_other. Display unknown
32482 (ix86_valid_target_attribute_inner_p): Add missing options and
32483 reorder options by implied ISAs, as in ix86_target_string.
32485 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
32487 * hash-table.h (hash_table::too_empty_p): New function.
32488 (hash_table::expand): Use it.
32489 (hash_table::traverse): Likewise.
32490 (hash_table::empty_slot): Use sizeof (value_type) instead of
32491 sizeof (PTR) to convert bytes to elements. Shrink the table
32492 if the current size is excessive for the current number of
32495 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
32497 * ira-costs.c (record_reg_classes): Break from the inner loop
32498 early once alt_fail is known to be true. Update outer loop
32499 handling accordingly.
32501 2017-01-13 Jeff Law <law@redhat.com>
32503 * tree-ssa-dse.c (decrement_count): New function.
32504 (increment_start_addr, maybe_trim_memstar_call): Likewise.
32505 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
32506 when we know the partially dead statement is a mem* function.
32508 PR tree-optimization/61912
32509 PR tree-optimization/77485
32510 * tree-ssa-dse.c: Include expr.h.
32511 (maybe_trim_constructor_store): New function.
32512 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
32514 PR tree-optimization/33562
32515 PR tree-optimization/61912
32516 PR tree-optimization/77485
32517 * doc/invoke.texi: Document new dse-max-object-size param.
32518 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
32519 * tree-ssa-dse.c: Include params.h.
32520 (dse_store_status): New enum.
32521 (initialize_ao_ref_for_dse): New, partially extracted from
32523 (valid_ao_ref_for_dse, normalize_ref): New.
32524 (setup_live_bytes_from_ref, compute_trims): Likewise.
32525 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
32526 (maybe_trim_partially_dead_store): Likewise.
32527 (maybe_trim_complex_store): Likewise.
32528 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
32529 Track what bytes live from the original store. Return tri-state
32530 for dead, partially dead or live.
32531 (dse_dom_walker): Add constructor, destructor and new private members.
32532 (delete_dead_call, delete_dead_assignment): New extracted from
32534 (dse_optimize_stmt): Make a member of dse_dom_walker.
32535 Use initialize_ao_ref_for_dse.
32537 PR tree-optimization/33562
32538 PR tree-optimization/61912
32539 PR tree-optimization/77485
32540 * sbitmap.h (bitmap_count_bits): Prototype.
32541 (bitmap_clear_range, bitmap_set_range): Likewise.
32542 * sbitmap.c (bitmap_clear_range): New function.
32543 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
32545 2017-01-13 Martin Liska <mliska@suse.cz>
32548 * function.c (set_cfun): Add new argument force.
32549 * function.h (set_cfun): Likewise.
32550 * ipa-inline-transform.c (inline_call): Use the function when
32551 strict alising from is dropped for function we inline to.
32553 2017-01-13 Richard Biener <rguenther@suse.de>
32555 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
32556 for dumping GIMPLE INTEGER_CSTs.
32558 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
32560 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
32561 to 201112L since C++17.
32563 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
32566 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
32567 if -fsanitize=kernel-address is present.
32569 2017-01-13 Richard Biener <rguenther@suse.de>
32571 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
32572 as _Literal ( type ) number in case usual suffixes do not
32573 preserve all information.
32575 2017-01-13 Richard Biener <rguenther@suse.de>
32577 PR tree-optimization/77283
32578 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
32579 and ssa-iterators.h.
32580 (is_feasible_trace): Implement a cost model based on joiner
32583 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
32586 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
32587 char or short to __float128/_Float128 directly.
32589 2017-01-12 Martin Sebor <msebor@redhat.com>
32591 to -Wformat-overflow.
32592 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
32593 (min_bytes_remaining): Same.
32594 (get_string_length): Same.
32595 (format_string): Same.
32596 (format_directive): Same.
32598 (pass_sprintf_length::handle_gimple_call): Same.
32600 2017-01-12 Jakub Jelinek <jakub@redhat.com>
32602 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
32603 info.nowrite calls with no lhs that can't throw. Return bool
32604 whether gsi_remove has been called or not.
32605 (pass_sprintf_length::handle_gimple_call): Return bool whether
32606 try_substitute_return_value called gsi_remove. Formatting fix.
32607 (pass_sprintf_length::execute): Don't use gsi_remove if
32608 handle_gimple_call returned true.
32611 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
32612 be removed due to side-effects, don't remove following barrier nor
32613 turn the successor edge into fallthru edge.
32615 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32618 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
32619 element-reversing loads and stores as not swappable.
32621 2017-01-12 Nathan Sidwell <nathan@acm.org>
32622 Nicolai Stange <nicstange@gmail.com>
32624 * combine.c (try_combine): Don't ignore result of overlap checking
32625 loop. Combine overlap & asm check into single loop.
32627 2017-01-12 Richard Biener <rguenther@suse.de>
32629 * tree-pretty-print.c (dump_generic_node): Provide -gimple
32630 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
32632 2017-01-12 Richard Biener <rguenther@suse.de>
32634 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
32635 and TS_TARGET_OPTION directly derive from TS_BASE.
32636 * tree-core.h (tree_optimization_option): Derive from tree_base.
32637 (tree_target_option): Likewise.
32639 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
32641 * config/i386/i386.c (memory_address_length): Increase len
32642 only when rip_relative_addr_p returns false.
32644 2017-01-11 Julia Koval <julia.koval@intel.com>
32646 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
32647 (OPTION_MASK_ISA_SGX_SET): New.
32648 (ix86_handle_option): Handle OPT_msgx.
32649 * config.gcc: Added sgxintrin.h.
32650 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
32651 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
32652 * config/i386/i386.c (ix86_target_string): Add -msgx.
32654 (ix86_option_override_internal): Handle new options.
32655 (ix86_valid_target_attribute_inner_p): Add sgx.
32656 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
32657 * config/i386/i386.opt: Add msgx.
32658 * config/i386/sgxintrin.h: New file.
32659 * config/i386/x86intrin.h: Add sgxintrin.h.
32661 2017-01-11 Jakub Jelinek <jakub@redhat.com>
32664 * fold-const.c (maybe_nonzero_address): Return 1 for function
32666 (tree_single_nonzero_warnv_p): Don't handle function local objects
32670 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
32673 2017-01-11 David Malcolm <dmalcolm@redhat.com>
32676 * opts.c: Include "spellcheck.h"
32677 (struct string_fragment): New struct.
32678 (struct edit_distance_traits<const string_fragment &>): New
32680 (get_closest_sanitizer_option): New function.
32681 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
32683 2017-01-11 Jakub Jelinek <jakub@redhat.com>
32685 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
32687 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
32688 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
32689 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
32690 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
32691 for initial die_offset if dwarf_split_debug_info.
32692 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
32693 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
32695 (output_skeleton_debug_sections): Formatting fix. Use
32696 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
32697 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
32699 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
32701 * config/arm/cortex-a53.md: Add bypasses for
32702 cortex_a53_r2f_cvt.
32703 (cortex_a53_r2f): Only use for transfers.
32704 (cortex_a53_f2r): Likewise.
32705 (cortex_a53_r2f_cvt): Add reservation for conversions.
32706 (cortex_a53_f2r_cvt): Likewise.
32708 2017-01-11 Tamar Christina <tamar.christina@arm.com>
32710 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
32711 to all inlined functions, change static to extern.
32713 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
32716 * config/arm/arm.c (legitimize_pic_address): Handle reference to
32718 (arm_assemble_integer): Likewise.
32720 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
32722 * config.gcc: Use new awk script to check CPU, FPU and architecture
32723 parameters for --with-... options.
32724 * config/arm/parsecpu.awk: New file
32725 * config/arm/arm-cpus.in: New file.
32726 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
32728 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
32730 * config/arm/t-arm: Update dependency rules.
32731 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
32732 of processing .def files.
32733 * config/arm/genopt.sh: Deleted.
32734 * config/arm/gentune.sh: Deleted.
32735 * config/arm/arm-cores.def: Deleted.
32736 * config/arm/arm-arches.def: Deleted.
32737 * config/arm/arm-fpus.def: Deleted.
32738 * config/arm/arm-tune.md: Regenerated.
32739 * config/arm/arm-tables.opt: Regenerated.
32740 * config/arm/arm-cpu.h: New generated file.
32741 * config/arm/arm-cpu-data.h: New generated file.
32742 * config/arm/arm-cpu-cdata.h: New generated file.
32744 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
32747 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
32749 (input_varpool_node): Unpack dynamically_initialized bit.
32751 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
32753 PR rtl-optimization/79032
32754 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
32755 the alignment of the adjusted memory reference against that of MODE,
32756 instead of the alignment of the original memory reference.
32758 2017-01-11 Martin Jambor <mjambor@suse.cz>
32760 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
32762 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
32763 decorated functions.
32765 2017-01-11 Richard Biener <rguenther@suse.de>
32767 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
32768 set range/nonnull info for PHI results. Do not set it on
32769 stmts marked for removal.
32771 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
32773 * expr.c (store_field): In the bitfield case, fetch the return value
32774 from the registers before applying a single big-endian adjustment.
32775 Always do a final load for a BLKmode value not larger than a word.
32777 2017-01-10 David Malcolm <dmalcolm@redhat.com>
32780 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
32781 that we correctly handle column numbers greater than
32782 LINE_MAP_MAX_COLUMN_NUMBER.
32784 2017-01-10 Martin Sebor <msebor@redhat.com>
32786 PR middle-end/78245
32787 * gimple-ssa-sprintf.c (get_destination_size): Call
32788 {init,fini}object_sizes.
32789 * tree-object-size.c (addr_object_size): Adjust.
32790 (pass_through_call): Adjust.
32791 (pass_object_sizes::execute): Adjust.
32792 * tree-object-size.h (fini_object_sizes): Declare.
32794 2017-01-10 Martin Sebor <msebor@redhat.com>
32796 PR tree-optimization/78775
32797 * builtins.c (get_size_range): Move...
32798 * calls.c: ...to here.
32799 (alloc_max_size): Accept zero argument.
32800 (operand_signed_p): Remove.
32801 (maybe_warn_alloc_args_overflow): Call get_size_range.
32802 * calls.h (get_size_range): Declare.
32804 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
32806 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
32807 from TI's devices.csv file as of September 2016.
32808 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
32810 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
32812 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
32813 * doc/invoke.texi: Likewise.
32814 * doc/md.texi: Likewise.
32815 * doc/objc.texi: Likewise.
32817 2017-01-10 Joshua Conner <joshconner@google.com>
32819 * config/arm/fuchsia-elf.h: New file.
32820 * config/fuchsia.h: New file.
32821 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
32822 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
32824 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
32826 2016-01-10 Richard Biener <rguenther@suse.de>
32828 PR tree-optimization/79034
32829 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
32830 Propagate out degenerate PHIs in the joiner.
32832 2017-01-10 Martin Liska <mliska@suse.cz>
32834 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
32835 (sort_congruence_classes_by_decl_uid): Likewise.
32836 (sort_congruence_class_groups_by_decl_uid): Likewise.
32837 (sem_item_optimizer::merge_classes): Sort class, groups in these
32838 classes and members in the groups by DECL_UID of declarations.
32839 This would make merge operations stable.
32841 2017-01-10 Martin Liska <mliska@suse.cz>
32843 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
32844 usage of m_classes_vec.
32845 (sem_item_optimizer::~sem_item_optimizer): Likewise.
32846 (sem_item_optimizer::get_group_by_hash): Likewise.
32847 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
32848 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
32849 (sem_item_optimizer::verify_classes): Likewise.
32850 (sem_item_optimizer::process_cong_reduction): Likewise.
32851 (sem_item_optimizer::dump_cong_classes): Likewise.
32852 (sem_item_optimizer::merge_classes): Likewise.
32853 * ipa-icf.h (congruence_class_hash): Rename from
32854 congruence_class_group_hash. Remove declaration of m_classes_vec.
32856 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
32858 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
32859 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
32860 * config.gcc: Add avx512vpopcntdqintrin.h.
32861 * config/i386/avx512vpopcntdqintrin.h: New.
32862 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
32863 * config/i386/i386-builtin-types.def: Add new types.
32864 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
32865 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
32866 __builtin_ia32_vpopcountq_v8di_mask): New.
32867 * config/i386/i386-c.c (ix86_target_macros_internal): Define
32868 __AVX512VPOPCNTDQ__.
32869 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
32870 (PTA_AVX512VPOPCNTDQ): Define.
32871 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
32872 TARGET_AVX512VPOPCNTDQ_P): Define.
32873 * config/i386/i386.opt: Add mavx512vpopcntdq.
32874 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
32875 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
32877 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32879 PR middle-end/77484
32880 * predict.def (PRED_CALL): Set to 67.
32882 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
32884 * expr.c (store_field): In the bitfield case, if the value comes from
32885 a function call and is of an aggregate type returned in registers, do
32886 not modify the field mode; extract the value in all cases if the mode
32887 is BLKmode and the size is not larger than a word.
32889 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
32892 * config/i386/cpuid.h: Fix undefined behavior.
32894 2017-01-04 Jeff Law <law@redhat.com>
32896 PR tree-optimization/79007
32897 PR tree-optimization/67955
32898 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
32899 conservative for pt.null when flag_non_call_exceptions is on.
32901 2017-01-09 Jakub Jelinek <jakub@redhat.com>
32903 PR translation/79019
32904 PR translation/79020
32905 * params.def (PARAM_INLINE_MIN_SPEEDUP,
32906 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
32907 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
32909 * config/avr/avr.opt (maccumulate-args): Likewise.
32910 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
32911 * common.opt (freport-bug): Likewise.
32912 * cif-code.def (CIF_FINAL_ERROR): Likewise.
32913 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
32914 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
32915 translatable string.
32916 * config/i386/i386.c (function_value_32): Likewise.
32917 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
32918 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
32920 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
32921 * common/config/msp430/msp430-common.c (msp430_handle_option):
32923 * symtab.c (symtab_node::verify_base): Likewise.
32924 * opts.c (set_debug_level): Likewise.
32925 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
32926 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
32927 missing whitespace to translatable strings.
32928 * config/avr/avr.md (bswapsi2): Fix typo in comment.
32929 * config/sh/superh.h: Likewise.
32930 * config/i386/xopintrin.h: Likewise.
32931 * config/i386/znver1.md: Likewise.
32932 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
32933 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
32934 * double-int.h (struct double_int): Likewise.
32935 * double-int.c (div_and_round_double): Likewise.
32936 * wide-int.cc: Likewise.
32937 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
32938 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
32939 * cfgcleanup.c (crossjumps_occured): Renamed to ...
32940 (crossjumps_occurred): ... this.
32941 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
32944 PR tree-optimization/78899
32945 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
32946 returning bool return struct loop *, NULL for failure and the new
32948 (versionable_outer_loop_p): Don't version outer loop if it has
32949 dont_vectorized bit set.
32950 (tree_if_conversion): When versioning outer loop, ensure
32951 tree_if_conversion is performed also on the inner loop of the
32952 non-vectorizable outer loop copy.
32953 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
32954 LOOP_VECTORIZED in inner loop of the scalar outer loop and
32955 prevent vectorization of it.
32956 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
32957 the outer loop vectorization of the non-scalar version is attempted
32958 before vectorization of the inner loop in scalar version. If
32959 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
32960 vectorization of its inner loop.
32961 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
32962 has 2 inner loops, rename also on edges from bb whose single pred
32963 is outer_loop->header. Fix typo in function comment.
32965 2017-01-09 Martin Sebor <msebor@redhat.com>
32968 * asan.c (asan_emit_stack_protection): Increase local buffer size
32969 to avoid snprintf truncation warning.
32971 2017-01-09 Andrew Pinski <apinski@cavium.com>
32973 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
32974 to reference thunderx2t99 for the tuning structure
32975 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
32977 (thunderx2t99_extra_costs): This.
32978 * config/aarch64/aarch64-tune.md: Regenerate.
32979 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
32980 (vulcan_addrcost_table): This.
32981 (vulcan_regmove_cost): Rename to ...
32982 (thunderx2t99_regmove_cost): This.
32983 (vulcan_vector_cost): Rename to ...
32984 (thunderx2t99_vector_cost): this.
32985 (vulcan_branch_cost): Rename to ...
32986 (thunderx2t99_branch_cost): This.
32987 (vulcan_tunings): Rename to ...
32988 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
32989 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
32991 2017-01-09 Martin Jambor <mjambor@suse.cz>
32995 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
32996 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
32997 (propagate_vr_accross_jump_function): Use the above function for all
32998 value range computations for pass-through jump functions and type
32999 converasion from explicit value range values.
33000 (ipcp_propagate_stage): Do not attempt to deduce types of formal
33001 parameters from TYPE_ARG_TYPES.
33002 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
33003 (ipa_write_node_info): Stream type of the actual argument.
33004 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
33006 2017-01-09 Martin Liska <mliska@suse.cz>
33009 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
33010 (lookup_compiler): Do not show error message with have_E.
33012 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33014 PR tree-optimization/78938
33015 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
33016 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
33017 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
33018 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
33021 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33023 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
33026 2017-01-09 Richard Biener <rguenther@suse.de>
33028 PR tree-optimization/78997
33029 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
33030 name condition properly.
33032 2017-01-09 Richard Biener <rguenther@suse.de>
33035 * dwarf2out.c (is_cxx): New overload with context.
33036 (is_naming_typedef_decl): Use it.
33038 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33040 * invoke.texi (Option Summary): Correct spacing in option lists
33041 and add line breaks to fix over-long lines.
33043 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33045 PR middle-end/17660
33047 * extend.texi (Common Variable Attributes): Add xref to GCC
33048 Internals manual to explain mode attribute keywords.
33050 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33053 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
33054 and Preprocessor Options.
33055 (Options for Linking): Document -pthread here....
33056 (RS/6000 and PowerPC Options): ...not here.
33057 (Solaris 2 Options): ...or here.
33058 * doc/cppopts.texi: Document -pthread.
33060 2017-01-08 Martin Sebor <msebor@redhat.com>
33062 PR middle-end/77708
33063 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
33064 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
33065 New member functions.
33066 (format_directive): Used them.
33068 (pass_sprintf_length::handle_gimple_call): Same.
33069 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
33070 to avoid truncation for any argument.
33071 (extract_affine_mul): Same.
33072 * tree.c (get_file_function_name): Same.
33074 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33076 PR middle-end/77484
33077 * predict.def (PRED_INDIR_CALL): Set to 86.
33079 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33081 PR preprocessor/54124
33082 * doc/cppopts.texi: Reformat -d subtable to list the full name
33083 of the options. Add cross-reference to the docs for the general
33084 compiler -d options.
33085 * doc/invoke.texi (Developer Options): Add cross-reference to the
33086 preprocessor-specific -d option documentation.
33088 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33090 PR preprocessor/13498
33091 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
33092 redudant material, and reflect new command-line options.
33093 (System Headers): Likewise.
33095 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33097 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
33098 -isystem, and -idirafter. Copy-edit.
33099 * doc/cppopts.texi: Copy-edit. Remove contradiction about
33100 default for -ftrack-macro-expansion. Delete obsolete and
33101 badly-formatted implementation details about -fdebug-cpp output.
33102 * doc/cppwarnopts.texi: Copy-edit.
33104 2017-01-07 David Malcolm <dmalcolm@redhat.com>
33107 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
33108 that the transition from a max line width >= 1<<10 to narrower
33109 lines works correctly.
33111 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
33113 * doc/options.texi (PerFunction): New.
33114 * opt-functions.awk (switch_flags): Map both Optimization and
33115 PerFunction to CL_OPTIMIZATION.
33116 * opth-gen.awk: Test for PerFunction flag along with
33118 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
33119 it only when the latter is present. Skip those that don't in
33120 the hash function generator.
33121 * common.opt (fvar-tracking): Mark as PerFunction instead of
33123 (fvar-tracking-assignments): Likewise.
33124 (fvar-tracking-assignments-toggle): Likewise.
33125 (fvar-tracking-uninit): Likewise.
33127 2017-01-07 Jakub Jelinek <jakub@redhat.com>
33129 PR translation/79018
33130 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
33133 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
33136 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
33137 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
33138 TARGET_LONG_JUMP_TABLE_OFFSETS.
33139 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
33140 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
33141 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
33142 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
33143 * config/m68k/m68k.md (tablejump expander): Likewise.
33144 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
33145 TARGET_LONG_JUMP_TABLE_OFFSETS.
33146 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
33147 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
33149 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
33150 David Holsgrove <david.holsgrove@xilinx.com>
33152 * common/config/microblaze/microblaze-common.c
33153 (TARGET_EXCEPT_UNWIND_INFO): Remove.
33154 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
33156 * config/microblaze/microblaze.c (microblaze_must_save_register)
33157 (microblaze_expand_epilogue, microblaze_return_addr): Handle
33159 (microblaze_eh_return): New function.
33160 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
33161 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
33162 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
33163 * config/microblaze/microblaze.md (eh_return): New pattern.
33165 2017-01-06 Jakub Jelinek <jakub@redhat.com>
33167 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
33168 GCC_DIAGNOSTIC_STRINGIFY): Define.
33170 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
33172 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33174 * config/arm/arm.md (<mcrr>): New.
33176 * config/arm/arm.c (arm_arch5te): New.
33177 (arm_option_override): Set arm_arch5te.
33178 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
33180 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
33181 (arm_mcrr_qualifiers): ... this. New.
33182 (MRRC_QUALIFIERS): Define to...
33183 (arm_mrrc_qualifiers): ... this. New.
33184 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
33186 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
33187 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
33188 (MRRCI, mrrc, MRRC): New.
33189 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
33190 VUNSPEC_MRRC2): New.
33192 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33194 * config/arm/arm.md (<mcr>): New.
33196 * config/arm/arm.c (arm_coproc_builtin_available): Add
33197 support for mcr, mrc, mcr2 and mrc2.
33198 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
33199 (arm_mcr_qualifiers): ... this. New.
33200 (MRC_QUALIFIERS): Define to ...
33201 (arm_mrc_qualifiers): ... this. New.
33202 (MCR_QUALIFIERS): Define to ...
33203 (arm_mcr_qualifiers): ... this. New.
33204 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
33206 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
33207 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
33208 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
33209 VUNSPEC_MRC2): New.
33211 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33213 * config/arm/arm.md (*ldc): New.
33217 * config/arm/arm.c (arm_coproc_builtin_available): Add
33218 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
33219 (arm_coproc_ldc_stc_legitimate_address): New.
33220 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
33221 'qualifier_const_pointer'.
33222 (LDC_QUALIFIERS): Define to...
33223 (arm_ldc_qualifiers): ... this. New.
33224 (STC_QUALIFIERS): Define to...
33225 (arm_stc_qualifiers): ... this. New.
33226 * config/arm/arm-protos.h
33227 (arm_coproc_ldc_stc_legitimate_address): New.
33228 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
33229 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
33230 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
33231 stc2, stcl, stc2l): New.
33232 * config/arm/constraints.md (Uz): New.
33233 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
33234 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
33235 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
33236 VUNSPEC_STC2L): New.
33238 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33240 * config/arm/arm.md (<cdp>): New.
33241 * config/arm/arm.c (neon_const_bounds): Rename this ...
33242 (arm_const_bounds): ... this.
33243 (arm_coproc_builtin_available): New.
33244 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
33245 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
33246 (CDP_QUALIFIERS): Define to...
33247 (arm_cdp_qualifiers): ... this. New.
33249 (arm_expand_builtin_args): Add case for 6 arguments.
33250 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
33251 (arm_const_bounds): ... this.
33252 (arm_coproc_builtin_available): New.
33253 * config/arm/arm_acle.h (__arm_cdp): New.
33255 * config/arm/arm_acle_builtins.def (cdp): New.
33257 * config/arm/iterators.md (CDPI,CDP,cdp): New.
33258 * config/arm/neon.md: Rename all 'neon_const_bounds' to
33259 'arm_const_bounds'.
33260 * config/arm/types.md (coproc): New.
33261 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
33262 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
33263 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
33264 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
33266 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33268 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
33269 (UBINOP_QUALIFIERS): New.
33271 (acle_builtin_data): New. Change comment.
33272 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
33273 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
33274 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
33275 arm_acle_builtins.def.
33276 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
33277 (arm_init_acle_builtins): New.
33278 (CRC32_BUILTIN): Remove.
33279 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
33280 crc32cb, crc32ch and crc32cw.
33281 (arm_init_crc32_builtins): Remove.
33282 (arm_init_builtins): Use arm_init_acle_builtins rather
33283 than arm_init_crc32_builtins.
33284 (arm_expand_acle_builtin): New.
33285 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
33286 * config/arm/arm_acle_builtins.def: New.
33288 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33290 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
33291 (arm_builtin_datum): ... this.
33292 (arm_init_neon_builtin): Rename to ...
33293 (arm_init_builtin): ... this. Add a new parameters PREFIX
33294 and USE_SIG_IN_NAME.
33295 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
33296 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
33297 'arm_builtin_datum'.
33298 (arm_init_vfp_builtins): Likewise.
33299 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
33300 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
33301 (arm_expand_neon_args): Rename to ...
33302 (arm_expand_builtin_args): ... this. Rename builtin_arg
33303 enum values and differentiate between ARG_BUILTIN_MEMORY
33304 and ARG_BUILTIN_NEON_MEMORY.
33305 (arm_expand_neon_builtin_1): Rename to ...
33306 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
33307 values, arm_expand_builtin_args and add bool parameter NEON.
33308 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
33309 (arm_expand_vfp_builtin): Likewise.
33310 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
33312 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33314 PR middle-end/77484
33315 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
33316 * predict.c (tree_estimate_probability_bb): Reverse direction of
33317 polymorphic call predictor.
33319 2017-01-06 David Malcolm <dmalcolm@redhat.com>
33321 * passes.c (execute_one_pass): Split out pass-skipping logic into...
33322 (determine_pass_name_match): ...this new function and...
33323 (should_skip_pass_p): ...this new function.
33325 2017-01-06 Nathan Sidwell <nathan@acm.org>
33327 * ipa-visibility.c (function_and_variable_visibility): Reformat
33328 comments and long lines. Remove extrneous if.
33329 * symtab.c (symtab_node::make_decl_local): Fix code format.
33330 (symtab_node::set_section_for_node): Fix comment typo.
33332 2017-01-06 Martin Liska <mliska@suse.cz>
33335 * lra-constraints.c: Rename invariant to lra_invariant.
33336 * predict.c (set_even_probabilities): Initialize e to NULL.
33338 2017-01-05 Martin Sebor <msebor@redhat.com>
33340 PR tree-optimization/78910
33341 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
33342 (format_integer): Correct off-by-one error in the handling
33343 of precision with negative numbers in signed conversions..
33345 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
33347 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
33349 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33351 PR tree-optimization/71016
33352 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
33353 factor_out_conditional_conversion. Formatting fix.
33354 (factor_out_conditional_conversion): Add cond_stmt argument.
33355 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
33356 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
33359 2017-01-05 David Malcolm <dmalcolm@redhat.com>
33361 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
33362 read-rtl-function.o, and selftest-rtl.o.
33363 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
33364 (selftest::aarch64_test_loading_full_dump): New function.
33365 (selftest::aarch64_run_selftests): New function.
33366 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
33367 selftest::aarch64_run_selftests.
33368 * config/i386/i386.c
33369 (selftest::ix86_test_loading_dump_fragment_1): New function.
33370 (selftest::ix86_test_loading_call_insn): New function.
33371 (selftest::ix86_test_loading_full_dump): New function.
33372 (selftest::ix86_test_loading_unspec): New function.
33373 (selftest::ix86_run_selftests): Call the new functions.
33374 * emit-rtl.c (maybe_set_max_label_num): New function.
33375 * emit-rtl.h (maybe_set_max_label_num): New decl.
33376 * function.c (instantiate_decls): Guard call to
33377 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
33378 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
33380 * gensupport.c (gen_reader::gen_reader): Pass "false"
33381 for new "compact" param of rtx_reader.
33382 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
33383 rather than an empty string for NULL strings.
33384 * read-md.c: Potentially include config.h rather than bconfig.h.
33385 Wrap include of errors.h with #ifdef GENERATOR_FILE.
33386 (have_error): New global, copied from errors.c.
33387 (md_reader::read_name): Rename to...
33388 (md_reader::read_name_1): ...this, adding "out_loc" param,
33389 and converting "missing name or number" to returning false, rather
33391 (md_reader::read_name): Reimplement in terms of read_name_1.
33392 (md_reader::read_name_or_nil): New function.
33393 (md_reader::read_string): Handle "(nil)" by returning NULL.
33394 (md_reader::md_reader): Add new param "compact".
33395 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
33396 (md_reader::read_file): New method.
33397 * read-md.h (md_reader::md_reader): Add new param "compact".
33398 (md_reader::read_file): New method.
33399 (md_reader::is_compact): New accessor.
33400 (md_reader::read_name): Convert return type from void to file_location.
33401 (md_reader::read_name_or_nil): New decl.
33402 (md_reader::read_name_1): New decl.
33403 (md_reader::m_compact): New field.
33404 (noop_reader::noop_reader): Pass "false" for new "compact" param
33406 (rtx_reader::rtx_reader): Add new "compact" param.
33407 (rtx_reader::read_rtx_operand): Make virtual and convert return
33408 type from void to rtx.
33409 (rtx_reader::read_until): New decl.
33410 (rtx_reader::handle_any_trailing_information): New virtual function.
33411 (rtx_reader::postprocess): New virtual function.
33412 (rtx_reader::finalize_string): New virtual function.
33413 (rtx_reader::m_in_call_function_usage): New field.
33414 (rtx_reader::m_reuse_rtx_by_id): New field.
33415 * read-rtl-function.c: New file.
33416 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
33417 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
33418 (selftest::verify_three_block_rtl_cfg): New decl.
33419 * read-rtl-function.h: New file.
33420 * read-rtl.c: Potentially include config.h rather than bconfig.h.
33421 For host, include function.h, memmodel.h, and emit-rtl.h.
33422 (one_time_initialization): New function.
33423 (struct compact_insn_name): New struct.
33424 (compact_insn_names): New array.
33425 (find_code): Handle insn codes in compact dumps.
33426 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
33427 (bind_subst_iter_and_attr): Likewise.
33428 (add_condition_to_string): Likewise.
33429 (add_condition_to_rtx): Likewise.
33430 (apply_attribute_uses): Likewise.
33431 (add_current_iterators): Likewise.
33432 (apply_iterators): Likewise.
33433 (initialize_iterators): Guard usage of apply_subst_iterator with
33434 #ifdef GENERATOR_FILE.
33435 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
33436 (md_reader::read_mapping): Likewise.
33437 (add_define_attr_for_define_subst): Likewise.
33438 (add_define_subst_attr): Likewise.
33439 (read_subst_mapping): Likewise.
33440 (check_code_iterator): Likewise.
33441 (rtx_reader::read_rtx): Likewise. Move one-time initialization
33443 (one_time_initialization): New function.
33444 (rtx_reader::read_until): New method.
33445 (read_flags): New function.
33446 (parse_reg_note_name): New function.
33447 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
33448 Handle reuse_rtx ids.
33449 Wrap iterator lookup within #ifdef GENERATOR_FILE.
33450 Add parsing support for RTL dumps, mirroring the special-cases in
33451 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
33452 values, and calling handle_any_trailing_information.
33453 (rtx_reader::read_rtx_operand): Convert return type from void
33454 to rtx, returning return_rtx. Handle case 'e'. Call
33455 finalize_string on XSTR and XTMPL fields.
33456 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
33457 "(nil)" values were omitted. Call the postprocess vfunc on the
33459 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
33460 class ctor. Initialize m_in_call_function_usage. Call
33461 one_time_initialization.
33462 * rtl-tests.c (selftest::test_uncond_jump): Call
33463 set_new_first_and_last_insn.
33464 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
33465 * selftest-rtl.c: New file.
33466 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
33467 (selftest::get_insn_by_uid): New decl.
33468 * selftest-run-tests.c (selftest::run_tests): Call
33469 read_rtl_function_c_tests.
33470 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
33471 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
33474 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
33476 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
33477 operands in a special way. Assert that pos+len <= mode precision.
33479 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33481 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
33482 3 argument Alias with unlimited for the negative form.
33483 (fno-vect-cost-model): Removed.
33485 2017-01-05 Martin Liska <mliska@suse.cz>
33487 * hsa-gen.c (gen_hsa_divmod): New function.
33488 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
33490 2017-01-05 Martin Liska <mliska@suse.cz>
33493 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
33496 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33498 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
33499 small constant length operands.
33501 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33503 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
33504 between loop iterations.
33506 2017-01-05 Martin Liska <mliska@suse.cz>
33509 * gimplify.c (gimplify_decl_expr): Compare to
33510 asan_poisoned_variables instread of checking flags.
33511 (gimplify_target_expr): Likewise.
33512 (gimplify_expr): Likewise.
33513 (gimplify_function_tree): Conditionally initialize
33514 asan_poisoned_variables.
33516 2017-01-04 Jeff Law <law@redhat.com>
33518 PR tree-optimizatin/78812
33519 * rtl.h (contains_mem_rtx_p): Prototype.
33520 * ifcvt.c (containts_mem_rtx_p): Move from here to...
33521 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
33522 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
33523 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
33524 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
33526 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33528 * input.c (assert_char_at_range): Default-initialize actual_range.
33530 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33532 * df-scan.c (df_ref_create_structure): Make regno unsigned,
33533 to match the caller.
33535 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33537 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
33538 insns after final jump in test to emit dummy move.
33540 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33542 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
33543 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
33545 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33547 * multiple_target.c (create_dispatcher_calls): Init e_next.
33548 * tree-ssa-loop-split.c (split_loop): Init border.
33549 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
33552 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
33557 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
33558 (altivec_register_operand): Do not return true if the operand
33559 contains a SUBREG mixing SImode and SFmode.
33560 (vsx_register_operand): Likewise.
33561 (vsx_reg_sfsubreg_ok): New predicate.
33562 (vfloat_operand): Do not return true if the operand contains a
33563 SUBREG mixing SImode and SFmode.
33564 (vint_operand): Likewise.
33565 (vlogical_operand): Likewise.
33566 (gpc_reg_operand): Likewise.
33567 (int_reg_operand): Likewise.
33568 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
33569 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
33570 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
33572 (rs6000_emit_move_si_sf_subreg): New helper function.
33573 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
33574 fixup SUBREGs involving SImode and SFmode.
33575 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
33576 numbers for the new peephole2 optimization.
33577 (peephole2 for SFmode unions): New peephole2 to optimize cases in
33578 the GLIBC math library that do AND/IOR/XOR operations on single
33579 precision floating point.
33580 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
33581 target macros to say whether we need to avoid SUBREGs mixing
33583 (TARGET_ALLOW_SF_SUBREG): Likewise.
33584 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
33585 (UNSPEC_SI_FROM_SF): Likewise.
33586 (iorxor): Change spacing.
33587 (and_ior_xor): New iterator for AND, IOR, and XOR.
33588 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
33589 (movdi_from_sf_zero_ext): Likewise.
33590 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
33591 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
33592 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
33593 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
33594 (fms<mode>4): Likewise.
33595 (fnma<mode>4): Likewise.
33596 (fnms<mode>4): Likewise.
33597 (nfma<mode>4): Likewise.
33598 (nfms<mode>4): Likewise.
33600 2017-01-04 Marek Polacek <polacek@redhat.com>
33603 * doc/invoke.texi: Document -Wpointer-compare.
33605 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33607 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
33610 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
33611 descriptions for -gdwarf-5 and emit them as uleb128 instead of
33614 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
33617 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
33618 documentation of the powerpc_popcntb_ok attribute.
33619 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
33620 code to issue warning messages if a requested CPU configuration is
33621 not supported by the binary (assembler and loader) toolchain.
33622 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
33623 made to define a built-in function that has been disabled.
33624 (paired_init_builtins): Add assertion to prevent ICE if attempt is
33625 made to define a built-in function that has been disabled.
33626 (altivec_init_builtins): Add comment explaining why definition
33627 of the DST built-in functions is not preceded by an assertion
33628 check. Add assertions to prevent ICE if attempts are made to
33629 define an altivec predicate or an abs* built-in function that has
33631 (htm_init_builtins): Add comment explaining why definition of the
33632 htm built-in functions is not preceded by an assertion check.
33634 2017-01-04 Jeff Law <law@redhat.com>
33636 PR tree-optimizatin/67955
33637 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
33638 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
33639 the points-to solution does not include pt_null. Use DECL_PT_UID
33642 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
33644 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
33645 Use gen_int_mode instead of gen_lopwart for const_int operands.
33647 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33649 PR tree-optimization/71563
33650 * match.pd: Simplify X << Y into X if Y is known to be 0 or
33651 out of range value - has low bits known to be zero.
33653 2017-01-04 Alan Modra <amodra@gmail.com>
33655 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
33656 * configure: Regenerate.
33657 * config.in: Regenerate.
33659 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33662 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
33663 a substring of the message, but strcmp with the whole message. Ifdef
33664 ENABLE_NLS, translate the message first using dgettext.
33666 2017-01-03 Jeff Law <law@redhat.com>
33668 PR tree-optimizatin/78856
33669 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
33670 (mark_threaded_blocks): Remove code to truncate thread paths that
33671 cross multiple loop headers. Instead invalidate the cached loop
33672 iteration information and handle case of a thread path walking
33673 into an irreducible region.
33675 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
33678 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
33679 assertions. Add support for doing the signbit if the IEEE 128-bit
33680 floating point value is in a GPR.
33681 * config/rs6000/rs6000.md (Fsignbit): Delete.
33682 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
33683 Update the length attribute if the value is in a GPR.
33684 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
33685 the sign or zero extension instruction, since the value is always 0/1.
33686 (signbit<mode>2_dm2): Delete using <Fsignbit>.
33689 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
33690 extracting SImode to a GPR register so that we can generate a
33691 store, limit the vector to be in a traditional Altivec register
33692 for the vextuwrx instruction.
33694 2017-01-03 Ian Lance Taylor <iant@google.com>
33696 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
33698 2017-01-03 Martin Sebor <msebor@redhat.com>
33700 PR tree-optimization/78696
33701 * gimple-ssa-sprintf.c (format_floating): Correct handling of
33702 precision. Use MPFR for %f for greater fidelity. Correct handling
33704 (pass_sprintf_length::compute_format_length): Set width and precision
33705 specified by asrerisk to void_node for vararg functions.
33706 (try_substitute_return_value): Adjust dump output.
33708 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
33710 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
33712 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
33714 * doc/invoke.texi (SPARC options): Document -mlra as the default.
33715 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
33716 -mlra/-mno-lra was passed to the compiler.
33718 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
33720 PR rtl-optimization/65618
33721 * emit-rtl.c (try_split): Move initialization of "before" and
33722 "after" to just before the call to emit_insn_after_setloc.
33724 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
33726 * doc/md.texi (Standard Names): Remove reference to Java frontend.
33728 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
33730 * dwarf2out.c (gen_enumeration_type_die): When
33731 -gno-strict-dwarf, add a DW_AT_encoding attribute.
33733 2017-01-03 Jakub Jelinek <jakub@redhat.com>
33735 PR tree-optimization/78965
33736 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
33737 Change first argument from const call_info & to call_info &. For %n
33738 set info.nowrite to false.
33740 PR middle-end/78901
33741 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
33742 possibly throwing calls.
33744 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
33745 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
33746 and fns handling, rather than in a separate case SSA_NAME.
33748 2017-01-02 Jeff Law <law@redhat.com>
33750 * config/darwin-driver.c (darwin_driver_init): Const-correctness
33751 fixes for first_period and second_period variables.
33753 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
33756 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
33757 (*insvqi_1): New insn pattern.
33758 (*insvqi_1_mem_rex64): Ditto.
33759 (*insvqi_2): Ditto.
33760 (*insvqi_3): Rename from *insvqi.
33762 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
33764 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
33766 * doc/cfg.texi (Edges): Remove reference to Java.
33767 (Maintaining the CFG): Ditto.
33769 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33771 PR middle-end/77674
33772 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
33773 transparent aliases.
33775 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33777 PR middle-end/77484
33778 * predict.def (PRED_CALL): Update hitrate.
33779 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
33780 * predict.c (tree_estimate_probability_bb): Split CALL predictor
33781 into direct/indirect/polymorphic variants.
33783 2017-01-01 Jakub Jelinek <jakub@redhat.com>
33785 Update copyright years.
33787 * gcc.c (process_command): Update copyright notice dates.
33788 * gcov-dump.c (print_version): Ditto.
33789 * gcov.c (print_version): Ditto.
33790 * gcov-tool.c (print_version): Ditto.
33791 * gengtype.c (create_file): Ditto.
33792 * doc/cpp.texi: Bump @copying's copyright year.
33793 * doc/cppinternals.texi: Ditto.
33794 * doc/gcc.texi: Ditto.
33795 * doc/gccint.texi: Ditto.
33796 * doc/gcov.texi: Ditto.
33797 * doc/install.texi: Ditto.
33798 * doc/invoke.texi: Ditto.
33800 Copyright (C) 2017 Free Software Foundation, Inc.
33802 Copying and distribution of this file, with or without modification,
33803 are permitted in any medium without royalty provided the copyright
33804 notice and this notice are preserved.